# GRAVITATIONAL PHYSIOLOGY, AGING AND MEDICINE

EDITED BY : Nandu Goswami, Olivier White, Jack J. W. A. van Loon, Andreas Roessler and Andrew Blaber PUBLISHED IN : Frontiers in Physiology

#### Frontiers eBook Copyright Statement

The copyright in the text of individual articles in this eBook is the property of their respective authors or their respective institutions or funders. The copyright in graphics and images within each article may be subject to copyright of other parties. In both cases this is subject to a license granted to Frontiers. The compilation of articles constituting this eBook is the property of Frontiers.

Each article within this eBook, and the eBook itself, are published under the most recent version of the Creative Commons CC-BY licence. The version current at the date of publication of this eBook is CC-BY 4.0. If the CC-BY licence is updated, the licence granted by Frontiers is automatically updated to the new version.

When exercising any right under the CC-BY licence, Frontiers must be attributed as the original publisher of the article or eBook, as applicable.

Authors have the responsibility of ensuring that any graphics or other materials which are the property of others may be included in the CC-BY licence, but this should be checked before relying on the CC-BY licence to reproduce those materials. Any copyright notices relating to those materials must be complied with.

Copyright and source acknowledgement notices may not be removed and must be displayed in any copy, derivative work or partial copy which includes the elements in question.

All copyright, and all rights therein, are protected by national and international copyright laws. The above represents a summary only. For further information please read Frontiers' Conditions for Website Use and Copyright Statement, and the applicable CC-BY licence.

ISSN 1664-8714 ISBN 978-2-88963-273-2 DOI 10.3389/978-2-88963-273-2

### About Frontiers

Frontiers is more than just an open-access publisher of scholarly articles: it is a pioneering approach to the world of academia, radically improving the way scholarly research is managed. The grand vision of Frontiers is a world where all people have an equal opportunity to seek, share and generate knowledge. Frontiers provides immediate and permanent online open access to all its publications, but this alone is not enough to realize our grand goals.

### Frontiers Journal Series

The Frontiers Journal Series is a multi-tier and interdisciplinary set of open-access, online journals, promising a paradigm shift from the current review, selection and dissemination processes in academic publishing. All Frontiers journals are driven by researchers for researchers; therefore, they constitute a service to the scholarly community. At the same time, the Frontiers Journal Series operates on a revolutionary invention, the tiered publishing system, initially addressing specific communities of scholars, and gradually climbing up to broader public understanding, thus serving the interests of the lay society, too.

### Dedication to Quality

Each Frontiers article is a landmark of the highest quality, thanks to genuinely collaborative interactions between authors and review editors, who include some of the world's best academicians. Research must be certified by peers before entering a stream of knowledge that may eventually reach the public - and shape society; therefore, Frontiers only applies the most rigorous and unbiased reviews.

Frontiers revolutionizes research publishing by freely delivering the most outstanding research, evaluated with no bias from both the academic and social point of view. By applying the most advanced information technologies, Frontiers is catapulting scholarly publishing into a new generation.

### What are Frontiers Research Topics?

Frontiers Research Topics are very popular trademarks of the Frontiers Journals Series: they are collections of at least ten articles, all centered on a particular subject. With their unique mix of varied contributions from Original Research to Review Articles, Frontiers Research Topics unify the most influential researchers, the latest key findings and historical advances in a hot research area! Find out more on how to host your own Frontiers Research Topic or contribute to one as an author by contacting the Frontiers Editorial Office: researchtopics@frontiersin.org

# GRAVITATIONAL PHYSIOLOGY, AGING AND MEDICINE

Topic Editors:

Nandu Goswami, Medical University of Graz, Austria Olivier White, INSERM U1093 Cognition, Action et Plasticité Sensomotrice, France Jack J. W. A. van Loon, Vrije Universiteit Amsterdam, Netherlands Andreas Roessler, Medical University of Graz, Austria Andrew Blaber, Simon Fraser University, Canada

Image Credit: Dominika Kalcher (http://www.kado.co.at/).

Citation: Goswami, N., White, O., van Loon, J. J. W. A., Roessler, A., Blaber, A., eds. (2020). Gravitational Physiology, Aging and Medicine. Lausanne: Frontiers Media SA. doi: 10.3389/978-2-88963-273-2

# Table of Contents


Marie Barbiero, Célia Rousseau, Charalambos Papaxanthis and Olivier White


Marc Kermorgant, Florian Leca, Nathalie Nasr, Marc-Antoine Custaud, Thomas Geeraerts, Marek Czosnyka, Dina N. Arvanitis, Jean-Michel Senard and Anne Pavy-Le Traon


Ajay K. Verma, Da Xu, Amanmeet Garg, Anita T. Cote, Nandu Goswami, Andrew P. Blaber and Kouhyar Tavakolian

*129 Human Locomotion in Hypogravity: From Basic Research to Clinical Applications*

Francesco Lacquaniti, Yury P. Ivanenko, Francesca Sylos-Labini, Valentina La Scaleia, Barbara La Scaleia, Patrick A. Willems and Myrka Zago


David A. Green and Jonathan P. R. Scott


Olivier White, Jean-Louis Thonnard, Philippe Lefèvre and Joachim Hermsdörfer

*224 Alterations in Leg Extensor Muscle-Tendon Unit Biomechanical Properties With Ageing and Mechanical Loading*

Christopher McCrum, Pamela Leow, Gaspar Epro, Matthias König, Kenneth Meijer and Kiros Karamanidis


Yunfang Gao, Yasir Arfat, Huiping Wang and Nandu Goswami

*255 Efficacy of Stochastic Vestibular Stimulation to Improve Locomotor Performance During Adaptation to Visuomotor and Somatosensory Distortion*

David R. Temple, Yiri E. De Dios, Charles S. Layne, Jacob J. Bloomberg and Ajitkumar P. Mulavara


Zeynep Masatli, Michael Nordine, Martina A. Maggioni, Stefan Mendt, Ben Hilmer, Katharina Brauns, Anika Werner, Anton Schwarz, Helmut Habazettl, Hanns-Christian Gunga and Oliver S. Opatz


Martina A. Maggioni, Paolo Castiglioni, Giampiero Merati, Katharina Brauns, Hanns-Christian Gunga, Stefan Mendt, Oliver S. Opatz, Lea C. Rundfeldt, Mathias Steinach, Anika Werner and Alexander C. Stahn


Andreas Roessler, Johannes Reichmuth, Joern Rittweger and Nandu Goswami

*364 Stress Related Shift Toward Inflammaging in Cosmonauts After Long-Duration Space Flight*

Judith-Irina Buchheim, Sandra Matzel, Marina Rykova, Galina Vassilieva, Sergey Ponomarev, Igor Nichiporuk, Marion Hörl, Dominique Moser, Katharina Biere, Matthias Feuerecker, Gustav Schelling, Detlef Thieme, Ines Kaufmann, Manfred Thiel and Alexander Choukèr


Elena Tomilovskaya, Tatiana Shigueva, Dimitry Sayenko, Ilya Rukavishnikov and Inessa Kozlovskaya

# Editorial: Gravitational Physiology, Aging and Medicine

Nandu Goswami 1,2 \*, Jack J. W. A. van Loon<sup>3</sup> , Andreas Roessler <sup>1</sup> , Andrew P. Blaber <sup>4</sup> and Olivier White<sup>5</sup>

<sup>1</sup> Division of Physiology, Otto Loewi Research Center for Vascular Biology, Immunology and Inflammation, Medical University of Graz, Graz, Austria, <sup>2</sup> Department of Health Sciences, Alma Mater Europaea Maribor, Maribor, Slovenia, <sup>3</sup> Department Oral and Maxillofacial Surgery/Pathology, Amsterdam UMC, Vrije Universiteit Medical Center, Academic Centre for Dentistry Amsterdam (ACTA), Amsterdam, Netherlands, <sup>4</sup> Department of Biomedical Physiology and Kinesiology, Simon Fraser University, Burnaby, BC, Canada, <sup>5</sup> INSERM UMR1093-CAPS, Université Bourgogne Franche-Comté, UFR des Sciences du Sport, Dijon, France

Keywords: spaceflight, aging, bedrest, falls, deconditioning, immersion, countermeasures

**Editorial on the Research Topic**

#### **Gravitational Physiology, Aging and Medicine**

#### Edited and reviewed by:

Geoffrey A. Head, Baker Heart and Diabetes Institute, Australia

\*Correspondence: Nandu Goswami nandu.goswami@medunigraz.at

#### Specialty section:

This article was submitted to Integrative Physiology, a section of the journal Frontiers in Physiology

Received: 26 September 2019 Accepted: 08 October 2019 Published: 25 October 2019

#### Citation:

Goswami N, van Loon JJWA, Roessler A, Blaber AP and White O (2019) Editorial: Gravitational Physiology, Aging and Medicine. Front. Physiol. 10:1338. doi: 10.3389/fphys.2019.01338 Physiological deconditioning changes induced by spaceflight are similar to those that occur in aging, thus leading to, for example, a greater incidence of syncope and falls and to a decrease in quality of life. The Research Topic "Gravitational Physiology, Aging and Medicine" examines effects of gravitational changes on human physiology, with applications to geriatrics and clinical medicine. The Research Topic also aimed at promoting national and international networks that include activities used to stimulate student-oriented learning.

The presence of gravity is known to influence human physiology and behavior. For instance, gravitational loads are required to maintain a healthy musculoskeletal system. Indeed, upon standing, ∼500 mL of blood move into the lower limbs within seconds. This can compromise venous return, and hence blood pressure, leading to falls. Gravity also influences the way we interact with the environment because it provides a strong reference to align multimodal information such as vision and proprioception, critical for balance. Fortunately, the body, through neural mechanisms, compensates rapidly to postural changes, acting to maintain mean arterial pressure and prevent orthostatic intolerance.

The microgravity environment of spaceflight causes cardiovascular, neurovestibular and musculoskeletal changes, including bone loss and muscle atrophy. The observed physiological changes lead to deconditioning and impaired responses to gravitational stress on return to Earth. These changes share important common features with the deconditioning and impaired functions due to the aging process. Such changes that are seen in astronauts as well as in older persons lead to greater incidence of orthostatic intolerance. In elderly persons, this can lead to falls which are associated with head injuries and/or bone fractures. Such injuries and fractures are often associated with long-term immobilization and, consequently, further deconditioning, a vicious cycle from which patients may not be able to recover. Several papers in this Research Topic specifically address the parallel mechanisms between spaceflight deconditioning and aging (Goswami; Goswami et al.; Strollo et al.; Siamwala et al.).

Furthermore, studies aimed at understanding spaceflightinduced deconditioning often use ground-based analogs such as bedrest confinement and wet and dry water immersion (Tomilovskaya et al.). Such studies provide unique insights into the role of short and long term bedrest as well as immersion on physiological responses (discussed in Tomilovskaya et al.; de Abreu et al.; Kermorgant et al.; Šarabon et al.; Stavrou et al.; Gennser et al.). The knowledge gained from those investigations is important for developing countermeasures against deconditioning induced by spaceflight and aging. Such countermeasures are discussed in Maggioni et al., Temple et al., White et al., and Marusic et al.

Bed confinement is a paramount problem in older persons. Hence, data from bedrest studies are useful for understanding the deconditioning effects of this kind of immobilization in this population, which is a rapidly growing segment of the overall population. Bedrest can lead to postural control deficiencies and orthostatic hypotension which are major contributors to falls in the elderly (see Goswami). Naturally, it can be expected that integrating information inherited from space environments and ground-based models of deconditioning will provide novel perspectives and innovative approaches for expanding knowledge in both space physiology and aging medicine. For example, scientific insights and methodologies developed in space science research of orthostatic intolerance can be exploited to study cardiovascular, cerebrovascular, and postural sensory motor control systems in males and females (reviewed in Evans et al.). This Research Topic is a perfect example of translational and multidisciplinary research.

# TOPICS COVERED

This Research Topic received 45 papers for peer-review. Following rigorous reviewing of abstracts and full papers, and after careful screening of revised manuscripts, 36 were selected for publication. The areas covered include aspects related to physiology of gravitational (un-)loading, effects of bedrest confinement, dry immersion, as well as aging-related physiological deconditioning. In addition, this Research Topic has examined the ways in which knowledge of life in space and the countermeasures developed in space to overcome spaceflight-induced effects—can help life on Earth. Like the possible application of large radius artificial gravity for research, possible treatment and training (van Loon et al., 2012). While most of the contributions were based on studies of responses in human participants, some used animal models to investigate more fundamental mechanisms (Gambara et al.; Giuliani et al.; Ling et al.). This Research Topic included both original research articles and reviews. The collection of articles, which also examine sex-based diiferences in physiological responses (Masatli et al.; White et al., 2019; Evans et al.), can be broadly classified as:

1. Overaching reviews and original articles related to **acute and chronic effects of spaceflight.** These papers included those that examined stress-related shift toward inflammaging in long-duration spaceflight (Buchheim et al.), effects of hypogravity on human locomotion (Lacquaniti et al.), including arm kinematics and postural strategy in whole-body reaching movements that occur in microgravity (Macaluso et al.), and venous function changes during long duration spaceflights (Fortrat et al.). Spinal health during loading and unloading that occurs in spaceflight (Green and Scott) as well as biomechanical and cardiopulmonary responses to partial gravity (Richter et al.) are also addressed.


# AUTHOR CONTRIBUTIONS

NG, JL, AR, AB, and OW were responsible for this Research Topic design, invitation to potential authors, reviews of abstracts, and submitted manuscripts and finally, in preparation of the editorial.

# ACKNOWLEDGMENTS

We thank the authors who contributed toward this Research Topic.

# REFERENCES


responses in men and women. Eur. J. Appl. Physiol. 119, 951–960. doi: 10.1007/s00421-019-04084-y

**Conflict of Interest:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Goswami, van Loon, Roessler, Blaber and White. This is an openaccess article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Orthostatic Challenge Shifts the Hemostatic System of Patients Recovered from Stroke toward Hypercoagulability

Gerhard Cvirn<sup>1</sup> , Markus Kneihsl <sup>2</sup> , Christine Rossmann<sup>1</sup> , Margret Paar <sup>1</sup> , Thomas Gattringer <sup>2</sup> , Axel Schlagenhauf <sup>3</sup> , Bettina Leschnik <sup>3</sup> , Martin Koestenberger <sup>3</sup> , Erwin Tafeit <sup>1</sup> , Gilbert Reibnegger <sup>1</sup> , Irhad Trozic<sup>4</sup> , Andreas Rössler <sup>4</sup> , Franz Fazekas <sup>2</sup> and Nandu Goswami <sup>4</sup> \*

1 Institute of Physiological Chemistry, Medical University of Graz, Graz, Austria, <sup>2</sup> Department of Neurology, Medical University of Graz, Graz, Austria, <sup>3</sup> Department of Pediatrics, Medical University of Graz, Graz, Austria, <sup>4</sup> Gravitational Physiology, Aging and Medicine Research Unit, Institute of Physiology, Medical University of Graz, Graz, Austria

#### Edited by:

Ovidiu Constantin Baltatu, Anhembi Morumbi University, Brazil

#### Reviewed by:

Yutang Wang, Federation University Australia, Australia Michele Salanova, Charité Universitätsmedizin Berlin, Germany

\*Correspondence:

Nandu Goswami nandu.goswami@medunigraz.at

#### Specialty section:

This article was submitted to Integrative Physiology, a section of the journal Frontiers in Physiology

Received: 12 September 2016 Accepted: 06 January 2017 Published: 07 February 2017

#### Citation:

Cvirn G, Kneihsl M, Rossmann C, Paar M, Gattringer T, Schlagenhauf A, Leschnik B, Koestenberger M, Tafeit E, Reibnegger G, Trozic I, Rössler A, Fazekas F and Goswami N (2017) Orthostatic Challenge Shifts the Hemostatic System of Patients Recovered from Stroke toward Hypercoagulability. Front. Physiol. 8:12. doi: 10.3389/fphys.2017.00012

Aims: The objective of our study was to assess the effects of orthostatic challenge on the coagulation system in patients with a history of thromboembolic events and to assess how they compared with age-matched healthy controls.

Methods: Twenty-two patients with histories of ischemic stroke and 22 healthy age-matched controls performed a sit-to-stand test. Blood was collected prior to- and at the end of- standing in the upright position for 6 min. Hemostatic profiling was performed by determining thrombelastometry and calibrated automated thrombogram values, indices of thrombin generation, standard coagulation times, markers of endothelial activation, plasma levels of coagulation factors and copeptin, and hematocrit.

Results: Orthostatic challenge caused a significant endothelial and coagulation activation in patients (Group 1) and healthy controls (Group 2): Plasma levels of prothrombin fragment F1+2 were increased by approximately 35% and thrombin/antithrombin-complex (TAT) increased 5-fold. Several coagulation variables were significantly altered in Group 1 but not in Group 2: Coagulation times (CTs) were significantly shortened and alpha angles, peak rate of thrombin generation (VELINDEX), tissue factor (TF) and copeptin plasma levels were significantly increased (comparison between standing and baseline). Moreover, the shortening of CTs and the rise of copeptin plasma levels were significantly higher in Group 1 vs. Group 2 (comparison between groups).

Conclusion: The coagulation system of patients with a history of ischemic stroke can be more easily shifted toward a hypercoagulable state than that of healthy controls. Attentive and long-term anticoagulant treatment is essential to keep patients from recurrence of vascular events.

Keywords: thrombin, thrombosis, tissue factor, stroke/prevention, standing, orthostasis, coagulation, thromboembolism

# INTRODUCTION

Several studies have reported that an orthostatic challenge activates the coagulation system. Both active and passive standing are associated with (i) pooling of blood in the legs, (ii) increased transmural pressure, (iii) increased shear stress, as well as (iv) endothelial activation in the lower extremities, associated with (v) release of procoagulant glycoproteins, e.g., Tissue Factor (TF) (Masoud et al., 2008; Cvirn et al., 2012; Haider et al., 2013). All these studies have been performed in healthy young controls. Despite coagulation activation, no cases of thrombosis have been reported in these persons. It, therefore, appears that an orthostatic challenge, induced by either active or passive standing, does not provoke thromboembolic events in healthy controls. However, orthostatic challenge-induced coagulation activation might constitute a problem for patients at an already elevated risk for vascular events, for example, ischemic stroke.

We, therefore, wanted to investigate the influence of orthostatic challenge on the coagulation system in patients with a history of ischemic stroke. In our study, 22 older patients, who had recovered from ischemic stroke, performed a sit-tostand test, a suitable model for orthostatic challenge. Twenty-two healthy age-matched controls served as controls.

Hemostatic profiling was carried out in both whole blood and platelet poor plasma samples. The clot formation process was monitored using TF-triggered thrombelastometry. Thrombin generation was assessed by means of calibrated automated thrombography (CAT), and by the determination of plasma levels of prothrombin fragment 1+2 (F1+2) and thrombinantithrombin complexes (TAT). For further assessment of orthostatic challenge-induced coagulation activation, we measured prothrombin times (PT), activated partial thromboplastin times (APTT), plasma levels of FII, FVII, FVIII, protein C, and protein S. In order to evaluate orthostatic challenge-induced endothelial activation, we determined plasma levels of TF and of tissue-plasminogen activator as well as of nitric oxide. Since nitric oxide has a very short half-life, it is very difficult to determine nitric oxide content in whole blood samples. We therefore measured nitrite + nitrate levels, the major stable metabolites of nitric oxide, which can be quantitatively determined (Romitelli et al., 2007).

It has been shown that standing is associated with hemoconcentration due to the transfer of intravascular fluid from the blood to the surrounding tissue (Hinghofer-Szalkay and Moser, 1986; Jacob et al., 2005). We therefore determined hematocrit values in pre- and post-orthostatic challenge blood samples to compare the hemoconcentration effect of orthostatic loading (standing) in both groups. Plasma volume changes were calculated from hematocrit, as described by Masoud et al. (2010).

Moreover, copeptin levels have been shown to be sensitive surrogate markers for the release of arginine-vasopressin, which is associated with orthostatic challenge (Goldsmith, 1989; Choi et al., 2015). We therefore determined levels of copeptin in preand post-orthostatic challenge plasma samples of the two groups in order to evaluate the orthostatic challenge-response on the hormonal levels.

# MATERIALS AND METHODS

# Subjects and Experimental Design

Twenty-two patients who had recovered from ischemic stroke and 22 healthy controls performed a sit-to-stand test. The patients had recovered from (their first) mild stroke syndromes (NIHSS: 1–3), which occurred not longer than 12 months before the measurements that were carried out in this study. The patients and healthy controls characteristics are shown in **Table 1**.

After 5 min of sitting still, a blood sample (baseline) was collected from the antecubital vein. Subsequently, test subjects were assisted into the upright positon for 6 min; another blood sample was then withdrawn from the vein (standing sample). While standing, test subjects kept their eyes open and did not alter foot placement.

Experiments were carried out in a room with minimal ambient noise. The room temperature was maintained between 23 and 25◦C and the experiments were carried out between 7 and 11 a.m. All control subjects underwent basic neurological assessment prior to our investigations.

The Ethics Committee of the Medical University of Graz, Austria approved this study (EK-Nr. 25-551 ex 12/13). Both patients and healthy controls provided informed written consent before taking part in the study.

# Blood Sampling

# Baseline Samples

Twelve milliliters of venous blood were collected in pre-citrated Vacuette <sup>R</sup> tubes, which contained 3.8% sodium citrate (Greiner Bio-one GmbH, Kremsmünster, Austria). Additionally, 4 mL of venous blood was collected into EDTA- and aprotinin-treated tubes to determine copeptin concentrations. Haematocrit and thrombelastometry measurements were performed in citrated whole blood samples. Subsequently, the remaining whole blood underwent centrifugation at 2000 g for 20 min in order to prepare platelet poor plasma samples. The remaining measurements were performed in platelet poor plasma samples.

### Standing Samples

After the subjects spent 6 min in an upright position, venous blood samples were withdrawn in the same wayand measurements performed- as described for the baseline samples.

# TF-Triggered Thrombelastometry Assay

The clot formation process was monitored using the thrombelastometry coagulation analyser (ROTEM <sup>R</sup> 05, Matel Medizintechnik, Austria). The following were measured: CT (Coagulation time): which is time from adding trigger to formation of initial fibrin formation; Clot formation time (CFT):

**Abbreviations:** APTT, Activated partial thromboplastin time; ETP, Endogenous thrombin potential; CAT, calibrated automated thrombography; CFT, clot formation time; CT, coagulation time; F 1+2, prothrombin fragment 1+2; INR, international ratio; MCF, maximum clot firmness; PT, prothrombin time; SD, standard deviation; TAT, thrombin/antithrombin-complex; TF, tissue factor; VELINDEX, peak rate of thrombin generation.

#### TABLE 1 | Participants' characteristics.


A total of 44 individuals performed a sit-to-stand test. Group 1 comprised of 22 older patients who had recovered from ischemic stroke, 22 healthy age-matched controls served as controls (Group 2). P-values were calculated by means of the Fischer's exact test. ASA, acetylsalicylic acid; MCI, myocardial infarction; NIHSS, National Institutes of Health Stroke Scale.

which is the time taken until the amplitude reaches 20 mm; Maximum clot firmness (MCF): which reflects clot stability; Alpha angle: which indicates the velocity of fibrin built-up and cross-linking. For this procedure, the volume of blood sample required was 340 µL and clot formation was initiated by adding 40 µL of "trigger solution" (containing 0.35 pmol/L TF and 3 mmol/L CaCl2, final concentration) to 300 µL of citrated whole blood (Sørensen et al., 2003).

# Thrombin Generation Assessments via the Automated Fluorogenic Measurements

CAT obtained from Thrombinoscope BV (Maastricht, the Netherlands) was used to monitor Thrombin generation curves (Hemker et al., 2003). Following laboratory values were determined: Lag Time: which is the time preceding the thrombin burst; Endogenous thrombin potential (ETP) and peak height (Peak): Either of these reflect the amount of thrombin being generated; ttPeak: time to peak; VELINDEX [peak thrombin/(peak time − lag time)]: the peak rate of thrombin formation; StartTail: the time at which no free thrombin is measurable. Low amounts of trigger (5 pmol/L of TF) were used for sensitive detection of thrombin formation.

# Standard Laboratory Tests

BM/Hitachi 917 (Roche, Austria) was used to determine the following: PT: expressed as International Normalized Ratio, INR; APTT; plasma levels of FII; FVII; FVIII; protein C; and protein S. ELISA kits (Behring Diagnostics GmbH, Marburg, Germany) were used to measure F 1+2 levels in the plasma and TAT. TF plasma levels were determined by applying ACTICHROME Tissue Factor ELISA and tissue-plasminogen activator concentrations were determined using the IMUBIND tissue-plasminogen activator ELISA kit (American Diagnostica, Pfungstadt, Germany).

# Haematocrit and Blood Cell Counts

Sysmex KX-21 N Automated Haematology Analyzer from Sysmex (Illinois, USA) was used to determine hematocrit and blood cell counts.

# Nitrite+Nitrate Plasma Levels

Two hundred microliters of plasma (baseline and standing samples) were diluted with double-distilled water (1:2, vol/vol) and loaded on pre-conditioned anion exchange columns (Chromabond SB, Macherey-Nagel, Düren, Germany). After a washing step with double-distilled water nitrite and nitrate were eluted with 1 mL 0.5 mol/L sodium chloride. In the eluates, nitrite and nitrate were determined simultaneously by means of HPLC analysis according to a previously published method (Romitelli et al., 2007) but with some modifications. Briefly, the HPLC consisted of a L-2200 autosampler, two L-2130 HTA pumps, and a L-2450 diode array detector (all: VWR Hitachi, VWR, Vienna, Austria). Separation was performed on a Hypersil ODS column (5 µm; 250 × 4 mm I.D.) with 10.0 min isocratic elution (buffer A: 0.1 mol/L NaH2PO4, pH = 5.5, containing 5.9 mmol/L tetrabutylammonium hydrogensulphate) followed by a linear gradient to 20% buffer B (buffer B: 0.1 mol/L NaH2PO4, pH = 5.5, containing 5.9 mmol/L tetrabutylammonium hydrogensulphate/acetonitrile, 3:1, vol/vol) within another 10 min. The injection volume of standard and sample solutions was 40 µL. The absorbance at 205 nm was recorded. Data

acquisition and subsequent analysis was done with the EZchrom Elite (VWR) program. Retention time was ∼7.80 min for nitrite and ∼14.5 min for nitrate.

# Concentration of Copeptin

Levels of copeptin were determined in heparinized plasma samples by means of a sandwich enzyme immunoassay from Wuhan USCN Business Co., Ldt (Houston, USA).

# Statistics

IBM SPSS Statistics 23 (IBM, New York, USA) program was used for statistical analyses. Differences in **Table 1** between group 1 (patients recovered from stroke) and group 2 (healthy, agematched controls) were determined by means of the Fischer's exact test. The variables that were normally distributed were tested by the Kolmogorov-Smirnov test and the Shapiro-Wilk test. Values of p < 0.05 were considered as significant deviations from the normal distribution. The median and the interquartile range were calculated and presented in **Tables 2**–**4** since most variables were not normally distributed. Furthermore, to investigate the significance of differences before and after orthostatic challenge, the Wilcoxon test was used for not normally distributed variables. If variables were normally distributed, a T-test for paired samples was applied. Finally,

TABLE 2 | Effects of orthostatic challenge on thrombelastometry and thrombin generation values.


Effects of orthostatic challenge in 22 older patients who had recovered from ischemic stroke (Group 1) compared with 22 healthy age-matched controls (Group 2). Data are expressed as median [interquartile range = Q1–Q3)]. P-values were calculated by means of the t-test for paired samples (a) in case of normally distributed variables; otherwise the Wilcoxon test (b) was applied.

to compare the changes between the patients and age-matched healthy controls (**Figures 1**, **2**), the difference between the standing and the baseline value was calculated for each variable (for example: CT-difference = CT-baseline − CT-standing). To investigate the significance of these differences between the patients and the control group, the T-test for independent samples was applied in case of normal distribution. Otherwise, if the new variables were not normally distributed, the Mann-Whitney U-test (Tafeit et al., 2003) was used.

# RESULTS

Detailed anthropometric measurements of all test subjects are provided in **Table 1**. Although, a significantly higher number of patients who had recovered from ischemic stroke were treated with antiplatelet (acetylsalicylic acid

TABLE 3 | Effects of orthostatic challenge on selected coagulation times and on indicators of endothelial activation.


Effects of orthostatic challenge in 22 older patients who had recovered from ischemic stroke (Group 1) compared with 22 healthy age-matched controls (Group 2). Data are expressed as median [interquartile range = (Q1–Q3)]. P-values were calculated by means of the t-test for paired samples (a) in case of normally distributed variables; otherwise the Wilcoxon test (b) was applied. APTT, activated partial thromboplastin time; TF, tissue factor; tPA, tissue-plasminogen activator.

#### TABLE 4 | Effects of orthostatic challenge on haematocrit and on copeptin plasma levels.


Effects of orthostatic challenge in 22 older patients who had recovered from ischemic stroke (Group 1) compared with 22 healthy age-matched controls (Group 2). Haematocrit data are presented as mean ± SD, while copeptin data are expressed as median [interquartile range = Q1–Q3)]. P-values were calculated by means of the t-test for paired samples (a) in case of normally distributed variables; otherwise the Wilcoxon test (b) was applied.

and/or clopidogrel) and hypotensive drugs, the basal levels of coagulation parameters were not significantly different between the two groups with the exception of tissue-plasminogen activator. The basal tissue-plasminogen activator levels were significantly higher in the recovered stroke-patients group (10.1 vs. 7.3 ng/mL, medians, p = 0.023). Unsurprisingly, the number of previous vascular events and of vascular risk factors was greater in patients when compared with controls.

FIGURE 1 | Orthostatic challenge-induced shortening of thrombelastometry-derived CTs in patients recovered from stroke (Group 1) vs. healthy, age-matched controls (Group 2). A boxplot of CT-differences for the two groups is shown. P-values were calculated by means of the Mann-Whitney U-test. "\*" is an outlier.

FIGURE 2 | Orthostatic challenge-induced rise of copeptin plasma levels in patients recovered from stroke (Group 1) vs. healthy, age-matched controls (Group 2). A bar diagram of copeptin-differences for the two groups presenting means ± SDs is shown. P-values were calculated by means of the t-test for independent samples.

# Effects of Orthostatic Challenge on Thrombelastometry Values

Two thrombelastometry values indicated an orthostatic challenge induced activation of the coagulation parameters in patients but not in healthy controls: CTs were shortened significantly and alpha angles were raised in the post standing whole blood samples (**Table 2**). The shortening of CTs (difference between standing and baseline) was significantly higher in the stroke patients (Group 1) compared with healthy controls (Group 2), shown in **Figure 1**. CFT and MCF values were not altered by the orthostatic challenge in either group.

# Effects of Orthostatic Challenge on CAT Values

CAT measurements indicated that orthostatic challenge, in the patients but not in healthy controls, enhances the capability of plasma to generate thrombin. VELINDEX values were significantly increased in the post standing samples in patients- indicating an increased velocity of thrombin formationcompared to baseline samples (**Table 2**). No significant influence of orthostatic challenge on all other CAT values (Lag time, ETP, Peak, ttPeak, and StartTail) was observed in both groups. Similarly, no significant differences were observed by comparing the coagulation changes between the two groups.

# Effect of Orthostatic Challenge on Levels of F1+2 and TAT

F1+2 and TAT plasma levels indicate that orthostatic challenge is associated with significant thrombin formation in both groups. F1+2 as well as TAT values were markedly increased in the standing samples compared to baseline in patients, and approximately to the same extent in controls (**Table 2**).

# Effects of Orthostatic Challenge on Standard Coagulation Times

PT values indicated that orthostatic challenge is associated with coagulation activation: PTs (expressed as INR) were significantly shorter in the post challenge samples compared to baseline in both groups. **Figure 3** depicts a paired graph showing the PT changes in the stroke recoverd patients and healthy controls (**Figures 3**, **4**; Kimura et al., 2011). APTTs were significantly shortened by orthostatic challenge in healthy controls but not in patients (**Table 3**). No significant differences were observed by comparing the changes between the two groups.

# Effects of Orthostatic Challenge on Coagulation Factor Levels

Orthostatic challenge did not affect FII, FVII, FVIII, protein C and protein S in both groups (data not shown).

# Effects of Orthostatic Challenge on Endothelial Activation

Orthostatic challenge appears to cause activation of the endothelium in both patients and healthy controls. Tissueplasminogen activator plasma values were raised in the standing samples compared to the baseline levels in healthy controls

but not in patients with previous ischemic stroke, whereas TF concentrations were increased in the post standing sample in patients but not in the healthy controls (**Table 3**). Nitrite + nitrate plasma levels were not altered by orthostatic challenge in either group (data not shown). As stated above, the basal levels of tissue-plasminogen activator were significantly higher in the recovered stroke patients group. No significant differences of tissue-plasminogen activator or TF were observed by comparing changes between the two groups.

# Effects of Orthostatic Challenge on Haematocrit/Plasma Volume

Orthostatic challenge was associated with hemoconcentration in healthy controls but not in patients. The hematocrit was increased in the post standing samples in healthy controls (**Table 4**). Correspondingly, the percent changes in plasma volume (calculated from hematocrit, according to Masoud et al., 2010) were greater in controls vs. patients (5.34 vs. 2.50%, p = 0.002). No significant differences of hematocrit were observed by comparing the changes between the two groups.

# Effects of Orthostatic Challenge on Copeptin Plasma Levels

The response to orthostatic challenge on the hormonal level was significantly higher in patients compared with healthy controls. The copeptin levels were greater in the post standing samples in patients but not in healthy controls (**Table 4**). The rise of copeptin levels (difference between standing and baseline) was significantly higher in the recovered stroke patients (Group 1) compared with healthy controls (Group 2), shown in **Figure 2**.

# DISCUSSION

In the present study we quantified coagulation changes in response to orthostatic challenge (a sit-to-stand test) in patients who had recovered from ischemic stroke as compared with healthy, age-matched controls. Although, significantly more stroke-patients were treated with antiplatelet drugs (acetylsalicylic acid and/or clopidogrel), approximately same basal levels of thrombelastometry-derived coagulation times in both groups were found in our study. While acetylsalicylic acid has been shown to possess only a limited capability to affect thrombelastometry values (Trentalange and Walts, 1991; Luddington, 2005), clopidogrel can effectively prolong platelet-fibrin clot formation (Gurbel et al., 2007). Marked effects were found in patients loaded with 300 to 600 mg. We assume that this effect could not be observed in our study since our post-stroke patients were receiving only 75 mg of clopidogrel.

Our data show that orthostatic challenge is associated with activation of the coagulation system. In both groups of test subjects, prothrombin fragment 1+2 as well as thrombin/ antithrombin-complex were significantly increased and prothrombin times were significantly shortened in the post standing samples compared to baseline to approximately the same extent. Activated partial thromboplastin times (median) were also shortened due to the orthostatic challenge, reaching significance only in the healthy control group. These findings are in agreement with two other studies, which investigated the effects of still standing on the coagulation cascade (Masoud et al., 2008, 2010). While the results from those studies were based on stationary standing periods that varied from 15 to 60 min, we observed that a short 6 min standing period also leads to a significant activation of the coagulation system. Furthermore, we observed that this activation of the coagulation is significantly more pronounced in patients than in healthy controls. Clotting times, determined by means of thrombelastometry, were significantly shorter in the standing samples compared with baseline in patients recovered from stroke but not in healthy controls. This indicates that orthostatic challenge is associated with a shift toward hypercoagulability in patients who had recovered from ischemic stroke, but not in healthy controls. Accordingly, alpha angles were significantly increased (compared to baseline) by orthostatic challenge in patients but not in healthy controls, indicating that orthostatic challenge apparently leads to enhancement of fibrin built-up and cross-linking in post-stroke patients but not in healthy controls. The changes were approximately the same by comparing both groups.

Moreover, peak rate of thrombin generation values, determined by means of calibrated automated thrombography, were greater in the standing samples in patients but not in healthy controls. This also indicates a shift toward hypercoagulability in patients recovered from stroke: orthostatic challenge seems to be associated with accelerated thrombin formation. Again, the changes were approximately the same when both groups were compared.

Furthermore, we found that orthostatic challenge significantly elevates plasma levels of Tissue Factor (a marker of endothelial activation, Tilley and Mackman, 2006) in patients compared to baseline. This also indicates an orthostatic challengeinduced shift toward hypercoagulability in the stroke-patients group, as Tissue Factor is the most important trigger of the coagulation cascade (Manly et al., 2011). Again, the changes were approximately the same across the two groups.

Plasma levels of tissue-plasminogen activator (another marker of endothelial activation) in healthy controls increased due to orthostatic challenge compared to baseline, indicating increased fibrinolytic activity in the standing samples in this group. However, we found no increase of tissue-plasminogen activator in the patients group compared to baseline, indicating that orthostatic challenge fails to activate the fibrinolytic system in this group (Mazzolai et al., 2002). Again, the changes were approximately the same across both groups.

Levels of another marker of endothelial activation, nitric oxide (determined as nitrite + nitrate plasma levels), were apparently not affected by orthostatic challenge. An explanation might be that the elevated basal levels of nitrite + nitrate in plasma (approximately 60 µM) (Rashid et al., 2003) mask the relatively low amounts of nitric oxide (most likely in the nanomolar range) produced in response to shear stress (Tian et al., 2010).

Interestingly, orthostatic challenge was associated with significant hemoconcentration in healthy controls (∼5%) but not in the patient group. Apparently, the microvascular permeability in the stroke-patients group is significantly lower than in the control group. An explanation might be that 73% of the patients, who had recovered from ischemic stroke, but only 32% of the healthy controls, were treated with hypotensive drugs (primarily angiotensin-converting enzyme inhibitors). Angiotensin-converting enzyme inhibitors could have led to the decreases in the microvascular permeability via suppression of angiotensin II formation (Pupilli et al., 1999; Newton et al., 2005). Hematocrit changes were approximately the same when comparing both groups.

Our results also indicate that recovered stroke-patients show more hormonal responses to orthostatic challenge than healthy controls: Plasma levels of copeptin were greater in the standing samples (in comparison to baseline) in patients but not in controls. Copeptin is a valid surrogate marker for argininevasopressin release. Due to the arginine-vasopressin being a synergistic component of the hypothalamo-pituitary-adrenal axis, levels of arginine-vasopressin also reflect the participants orthostatic stress response (Katan et al., 2009; Masoud et al., 2010). Copeptin changes were significantly higher in Group 1 compared with Group 2 of test subjects.

We have recently observed that there is some degree of autonomic dysfunction–as reflected in the heart rate variability measurements–during orthostatic challenge even after 1 year of recovery from stroke (Rodriguez et al., in press). As autonomic dysfunction changes have been associated with orthostatic intolerance and clot formation at the site of the vascular injury (Lopez-Vilchez et al., 2009), it is not surprising that a greater tendency toward clot formation was seen in our stroke patients. Our results are also in agreement with the work of Davidoff and colleagues (Davydov et al., 2015).

In conclusion, our data indicate that orthostatic challenge, i.e., a simple sit-to-stand test, leads to a shift toward hypercoagulability in healthy controls, and, to a significantly higher extent, in patients who had recovered from ischemic stroke. In addition to other studies showing coagulation activation after prolonged quiet standing (15 up to 60 min), we show herein that significant coagulation activation occurs even after 6 min of still standing. In our study, protein C plasma levels remained unchanged by orthostatic challenge; this is in agreement with the findings of Masoud et al. (2008, 2010). Thus, it appears that the anticoagulant protein C pathway fails to neutralize the shift toward hypercoagulability (Masoud et al., 2008, 2010).

On the assumption that this simple sit-to-stand test is a valid method that exposes subjects to a procoagulant challenge, our data suggest that the coagulation system of patients with a history of ischemic stroke can be more easily shifted toward a hypercoagulable state than that of healthy controls. Thus, our findings may explain the frequent recurrence of thrombotic events in recovered stroke-patients (Becattini et al., 2012). Our data emphasize the importance of accurate, regular and long-term anticoagulant treatment of patients with a history of thrombosis. It has to be stated that, by absolute numbers, the shift toward a hypercoagulable state in the strokepatients group is, although statistically significant, relatively small. All coagulation values remained within their reference values, as reported in a previous study (Lamprecht et al., 2013).

Notably, we show herein that even a short-term orthostatic challenge leads to significantly shortened standard coagulation times, particularly Prothrombin Times. With specific regard to the coagulation system, we could not find any literature that deals with the amount of time a patient should remain seated before providing blood coagulation assessments. From previous studies it is known that orthostatic challenge induced fluid shifts need about 30 min to return to baseline supine values (Hagan et al., 1978; Hinghofer-Szalkay et al., 2008; Goswami et al., 2012). Furthermore, using head-up-tilt and lower-body-negative pressure (Cvirn et al., 2012), we observed that endothelial activation (as assessed by tissue factor levels) returns to baseline levels around 20 min after orthostatic challenge. As patients are often standing/walking before they reach the consultation room, we recommend that when carrying out assessment of coagulation related parameters, patients should be allowed to rest in a seated or supine position for at least 20–30 min. This will help eliminate the influence of changes in posture on such tests as Prothrombin time.

# WHAT IS KNOWN ON THIS TOPIC


# WHAT THIS PAPER ADDS

• Orthostatic challenge causes a significant coagulation activation in patients who have had ischemic stroke more than 1 year prior.


# AUTHOR CONTRIBUTIONS

All authors have made significant contributions in data collection, analysis and interpretation of the findings. All

# REFERENCES


co-authors have been involved in writing the final version of the manuscript.

# ACKNOWLEDGMENTS

We thank Martina Mairold, Gerd Kager, Gerhard Ledinski, Floranda Lugaliu, Andreas Jantscher, and Rebecca Ruedl for technical assistance. This study was supported by a grant from the "Franz Lanyar-Stiftung P#393."


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2017 Cvirn, Kneihsl, Rossmann, Paar, Gattringer, Schlagenhauf, Leschnik, Koestenberger, Tafeit, Reibnegger, Trozic, Rössler, Fazekas and Goswami. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Microgravity-Induced Transcriptome Adaptation in Mouse Paraspinal longissimus dorsi Muscle Highlights Insulin Resistance-Linked Genes

Guido Gambara1, 2, Michele Salanova1, 2, Stefano Ciciliot 3, 4, Sandra Furlan<sup>5</sup> , Martina Gutsmann1, 2, Gudrun Schiffl1, 2, Ute Ungethuem<sup>6</sup> , Pompeo Volpe<sup>7</sup> , Hanns-Christian Gunga<sup>8</sup> and Dieter Blottner 1, 2 \*

<sup>1</sup> Center of Space Medicine Berlin, Charité Universitätsmedizin Berlin, Berlin, Germany, <sup>2</sup> Institute of Anatomy, Charité Universitätsmedizin Berlin, Berlin, Germany, <sup>3</sup> Venetian Institute of Molecular Medicine, University of Padova, Padova, Italy, <sup>4</sup> Department of Medicine, University of Padova, Padova, Italy, <sup>5</sup> Institute of Neuroscience Consiglio Nazionale Delle Ricerche, Padova, Italy, <sup>6</sup> Laboratory of Functional Genomics, Charité Universitätsmedizin Berlin, Berlin, Germany, <sup>7</sup> Dipartimento di Scienze Biomediche, University of Padova, Padova, Italy, <sup>8</sup> Department for Physiology and Centre for Space Medicine, Charité Universitätsmedizin Berlin, Berlin, Germany

#### Edited by:

Nandu Goswami, Medical University of Graz, Austria

#### Reviewed by:

Yuji Ogura, St. Marianna University School of Medicine, Japan John Joseph McCarthy, University of Kentucky, USA

> \*Correspondence: Dieter Blottner dieter.blottner@charite.de

#### Specialty section:

This article was submitted to Integrative Physiology, a section of the journal Frontiers in Physiology

Received: 14 February 2017 Accepted: 18 April 2017 Published: 05 May 2017

#### Citation:

Gambara G, Salanova M, Ciciliot S, Furlan S, Gutsmann M, Schiffl G, Ungethuem U, Volpe P, Gunga H-C and Blottner D (2017) Microgravity-Induced Transcriptome Adaptation in Mouse Paraspinal longissimus dorsi Muscle Highlights Insulin Resistance-Linked Genes. Front. Physiol. 8:279. doi: 10.3389/fphys.2017.00279 Microgravity as well as chronic muscle disuse are two causes of low back pain originated at least in part from paraspinal muscle deconditioning. At present no study investigated the complexity of the molecular changes in human or mouse paraspinal muscles exposed to microgravity. The aim of this study was to evaluate longissimus dorsi adaptation to microgravity at both morphological and global gene expression level. C57BL/N6 male mice were flown aboard the BION-M1 biosatellite for 30 days (BF) or housed in a replicate flight habitat on ground (BG). Myofiber cross sectional area and myosin heavy chain subtype patterns were respectively not or slightly altered in longissimus dorsi of BF mice. Global gene expression analysis identified 89 transcripts differentially regulated in longissimus dorsi of BF vs. BG mice. Microgravity-induced gene expression changes of lipocalin 2 (Lcn2), sestrin 1(Sesn1), phosphatidylinositol 3-kinase, regulatory subunit polypeptide 1 (p85 alpha) (Pik3r1), v-maf musculoaponeurotic fibrosarcoma oncogene family protein B (Mafb), protein kinase C delta (Prkcd), Muscle Atrophy F-box (MAFbx/Atrogin-1/Fbxo32), and Muscle RING Finger 1 (MuRF-1) were further validated by real time qPCR analysis. In conclusion, our study highlighted the regulation of transcripts mainly linked to insulin sensitivity and metabolism in longissimus dorsi following 30 days of microgravity exposure. The apparent absence of robust signs of back muscle atrophy in space-flown mice, despite the overexpression of Atrogin-1 and MuRF-1, opens new questions on the possible role of microgravity-sensitive genes in the regulation of peripheral insulin resistance following unloading and its consequences on paraspinal skeletal muscle physiology.

Keywords: skeletal muscle, gene expression, microgravity, BION-M1, microarray, spaceflight, disuse, insulin resistance

**19**

# INTRODUCTION

Low back pain is a common concern for crew members in short or long term duration space flight. A complete retrospective study by Kerstman and co-workers showed that among 772 astronauts of the U.S space program, 382 were positive to low back pain (LBP), also called space adaptation back pain (SABP) (Kerstman et al., 2012). Back pain onset in the major part of the analyzed cases was reported during the first days of space flight (day 1–12; Wing et al., 1991). The pathophysiology of microgravity induced back pain has been previously investigated and it is likely to be discogenic and somatic. Microgravity abolishes the physiological loads of the spine with a consequent increase of body-length determined essentially by an intervertebral disk (IVD) swelling and an adaptation of the thoracic and lumbar spine curvature. Consequently the disk expansion can stimulate type IV mechanoreceptors contributing to lumbar back pain onset (Sayson and Hargens, 2008; Sayson et al., 2013).

Spinal stability in response to the gravity load is maintained through a stabilizing system assembled by a passive component (vertebrae, disks, and ligaments), an active component (muscle and tendons), and the nervous system. In this view, paraspinal muscles and tendons generate force required for the stability of the spine. The impairment of any component of this stabilizing system can induce spine instability, consequentially generating back pain (Panjabi, 1992a,b). The lack of gravity forces in the microgravity environment unavoidably impairs the muscle involved in posture and stability both at structural and functional level, potentially contributing directly, or indirectly to the onset of back pain. The intensity of SABP is usually mild or moderated, but it could also impact the activity of astronauts in space missions, increasing the risk of musculoskeletal injury, or other traumas (Scheuring et al., 2009). Moreover, it has also been shown that the risk of IVD herniation is increased in astronauts, particularly within the first year after landing, indicating the potentially risky effect of reloading after microgravity exposure and suggesting the application of specific behavioral/training protocols for re-adaptation to gravity in crew members in spaceflight (Johnston et al., 2010; Belavy et al., 2016). Therefore, further efforts are needed to better understand the effect of microgravity on spine and paraspinal muscles and to design effective new countermeasures able to prevent paraspinal myofascial and neuromuscular deconditioning.

Only few investigations were performed so far on the effect of microgravity on paraspinal skeletal muscles. LeBlanc et al. (1995, 2000) investigated the effect of spaceflight on different muscles of shuttle/Mir crew members, concluding that both short term (8 and 17-days) and medium term (16–28 weeks) flights significantly reduced the intrinsic back muscle volume. Recently, the functional cross sectional area of lumbar paraspinal muscle has been found significantly reduced in one astronaut following 6-months exposure to microgravity (Hides et al., 2016). Moreover, there is even less evidence about the effect of microgravity on the gene or protein expression in back muscle. For example, 14-days spaceflight reduced the number of type I muscle fibers, while the expression of 70 kDa heat shock protein, t complex polypeptide 1 and other mitochondrial proteins was upregulated in paraspinal rat muscle. On the other hand, myocyte-specific enhancing factor 2C and aldolase resulted downregulated following spaceflight (Yamakuchi et al., 2000). Recently, the E3-ligase MuRF-1 was found to be increased in longissimus dorsi of mice exposed to microgravity for 30 days (Mirzoev et al., 2014).

In 2013 male mice were flown onboard of the Russian BION-M1 biosatellite to evaluate the effect of 30 days microgravity exposure on different bio-parameters in vivo and in vitro. Ogneva I.V. and co-worker investigated changes in the cortical cytoskeleton structure in skeletal muscle of the BION-M1 flown mice, showing significant alteration in the content of alphaactinin-1 and beta-actin respectively in soleus and tibialis anterior muscles (Ogneva et al., 2014). Among the BION-M1-based studies, it has been also observed a fiber type shift from slow to fast and a decrement of titin and nebulin proteins in gastrocnemius and tibialis anterior of flown mice (Ulanova et al., 2015).

The present study provides for the first time the global gene expression profile of longissimus dorsi in mice exposed to microgravity, which is central in further understanding of the structural, molecular and metabolic mechanism regulating vertebrate paraspinal muscle adaptation to spaceflight.

# MATERIALS AND METHODS

# Ethical Approval

All procedures including animal were performed in agreement with the European Convention for the Protection of Vertebrate Animals used for Experimental and Other Scientific Purposes (Strasbourg, 18.03.1986). Institutional Animal Care and Use Committee (IACUC) of MSU Institute of Mitoengineering (Protocol 35, 1 November, 2012) and Biomedical Ethics Commission of Institute for Biomedical Problems (IBMP), Moscow (protocol 319, 4 April, 2013) approved the study protocol related to the BION-M1 mission.

# Animals

C57BL/N6 male mice (22–25 g) were obtained from the Animal Breeding Facility Branch of Shemyakin & Ovchinnikov Institute of Bioorganic Chemistry, Russia. Mice were transported to the animal facility of Moscow State University (Institute of Mitoengineering) for training and selection and 19–20 weeks old mice were used in all the experiments. In detail, mice were randomly divided in 3 groups: BION Flown (BF), mice to be flown aboard the BION-M1 biosatellite exposed for 30 days to microgravity, BION Ground (BG) mice housed for 30 days in the same biosatellite habitat on ground, and Flight Control (FC) mice housed in the animal facility during the duration of the spaceflight.

# Sample Preparation and Transportation

Mouse skeletal muscle tissue sampling of all the experimental groups (longissimus dorsi and tongue) and freezing was performed by our Russian partners on site (IMBP, Moscow, Russia, contract # Bion-M1/2013 between RF SRC-IMBP and the Charité Berlin, Germany). In particular post-flight tissue sampling was done within 12–14 h at the IMBP after landing in Kazakhstan. All frozen samples were delivered deeply frozen in dry ice and further processed in our laboratory.

# Immunohistochemistry and Morphological Analysis

Cryosections (8 µm thickness) of mouse longissimus dorsi (n = 2) were cut with a Leica cryostat (CM 1850, Leica Microsystems, Bensheim, Germany), mounted on charged slides, stored frozen at −80◦C. Immunofluorescence with anti-MyHC isoform monoclonal antibodies were performed as previously described (Gambara et al., 2017): BA-D5 that recognizes type 1 MyHC isoform; SC-71 for type 2A MyHC isoform; BF-F3, for type 2B MyHC isoform (DSHB, Developmental Studies Hybridoma Bank, University of Iowa, Iowa City, IA). The sections were stained also using an anti-dystrophin antibody (Santa Cruz Biotechnology, Santa Cruz, CA) to visualize myofiber membranes. As secondary antibody, goat anti rabbit Alexa-635 and goat anti-mouse Alexa-555, goat antimouse Alexa-488 and goat anti-mouse Alexa 405-conjugated secondary antibody diluted to a final concentration of 1 µg/ml were used. For mouse-derived monoclonal primary and secondary antibodies, the MOM Ig blocking reagent (Vector Laboratories, Burlingame, CA, USA) was used to block mouse IgG background. Immunofluorescence were analyzed with an epifluorescence microscope (Axioplan; Zeiss, Oberkochen, Germany) equipped with a Cool Snap digital camera (Visitron Systems GmbH, Puchheim, Germany) and with 4-channel laser confocal microscope (TCS SP-8 with supersensitive HyD SP GaAsP detectors, Leica Microsystems, Germany). Digitized images were acquired with MetaVue software (Meta Series 7.5.6; system ID: 33693; Molecular Devices, Sunnyvale, CA, USA) and LAS X SP8 control software (Leica). Cross-sectional area (CSA) of the different myofiber types was semi-automatically measured by means of ImageJ 1.45 g (NIH, freeware imaging software).

# RNA Extraction and Sample Target Preparation

Total RNA was isolated from mouse longissimus dorsi (n = 5 for each group: BF, BG, and FC) and tongue (n = 3 for BF and BG; n = 2 for FC) muscles of each experimental group (BF, BG, and FC) by means of RNeasy micro Kit (Qiagen, Hilden, Germany). Frozen tissue samples were pulverized in liquid nitrogen and lysates were prepared in lysis buffer. Tissue lysate was centrifuged and the supernatant was used for RNA phenol/chloroform extraction. Single-step method of RNA isolation by acid guanidinium thiocyanate-phenolchloroform extraction was performed. The aqueous layer was mixed with an equal volume of 70% ethanol and total RNA was extracted using RNeasy spin columns according to the manufacturer's protocol. RNA integrity was checked by 2100 Bioanalyzer (Agilent technologies, PA, USA). The amplification and labeling of the RNA samples were carried out according to the manufacturer's instructions (Affymetrix, Santa Clara, CA). Briefly, total RNA was quantified by and checked by analysis on a LabChip (BioAnalyzer, AGILENT Technologies, Santa Clara, CA). The GeneChip <sup>R</sup> 3 ′ IVT Express Protocol is based on the Eberwine or reverse transcription method (in vitro transcription, IVT). Starting from 100 nanogram total RNA, first strand DNA was synthesized, containing a T7 promotor sequence and then converted into a double-stranded DNA. The double strand DNA serves as template in the subsequent in vitro transcription (IVT) reaction. This amplification step generates Biotin labeled complementary RNA (cRNA). After cleanup the biotin-modified RNA was fragmented by alkaline treatment. Fifteen microgram of each cRNA sample was hybridized for 16 h at 45◦C to an Affymetrix GeneChip Mouse 430A 2.0 Array. Arrays were washed and stained with streptavidin-phycoerythrin solutions using a fluidics station according to the protocols recommended by the manufacturer. Finally, probe arrays were scanned at 1.56-µm resolution using the Affymetrix GeneChip System confocal scanner 3000. Affymetrix Mouse Genome 430A 2.0 Array includes 22,600 probes sets to evaluate the expression level of more than 14,000 well-characterized mouse genes.

# Microarray Data Analysis and Pathway Analysis

Partek <sup>R</sup> Genomics Suite <sup>R</sup> 6.6 software was used for data analysis, applying Robust Multichip Average algorithm (RMA) for data normalization. To find differentially expressed genes, Analysis of Variance (ANOVA) was applied (p < 0.05). In the analysis of differentially regulated genes, a step up false discovery rate (FDR) of 5% was applied. Lists of differentially regulated genes included transcripts with fold changes major than 2 or minor than −2. Mouse Genome 430A 2.0 Array Probe Set Annotations was applied. Microarray data were uploaded in Gene Expression Omnibus (GEO) repository, accession number: GSE94381.

Gene ontology and pathway analysis were performed using the Functional Annotation Clustering module of DAVID v6.7 (The Database for Annotation, Visualization and Integrated Discovery). Gene enrichment were considered significant with an EASE score <0.05 (modified Fisher's exact test).

# Quantitative PCR Validation

Quantitative PCR was performed by SYBR Green method from total RNA isolated from longissimus dorsi (n = 5) and tongue muscle (n = 3) of each experimental group as described above. Briefly, 400 ng of RNA were converted to cDNA by using random hexamers and SuperScript <sup>R</sup> VILOTM (Invitrogen) following the manufacturer's instructions. Specific primers for qPCR were already published (Sandona et al., 2012) or designed using Primer3 software (http://frodo.wi.mit.edu/, Whitehead Institute for Biomedical Research). Their thermodynamic specificity was determined using BLAST sequence alignment (NCBI) and vector NTI <sup>R</sup> software (Invitrogen). Oligonucleotide primers used are listed in **Table S1**. The reaction mix consists of 10 µl of 2x iQ SYBR Green Supermix (Bio-Rad), 0.3 pmol/µl primers, 8 ng of cDNA, and DNase/RNase-free water up to 20 µl. The PCR parameters were initial denaturation at 95◦C for 30 s followed by 40 cycles of 10 s at 95◦C and 30 s at the corresponding annealing temperature (55–59◦C) for acquisition of fluorescence signal. A melting curve was generated by the iQ5 software (Biorad) following the end of the final cycle for each sample, by continuous monitoring the SYBR Green fluorescence throughout the temperature ramp from 65◦ to 99◦C in 0.5 s increments. All samples were run in triplicate, in parallel for each individual muscle sample and simultaneously with RNA-negative controls. Cyclophilin A (Ppia), glyceraldehyde 3 phosphate dehydrogenase (Gapdh), pyruvate carboxylase (Pcx), and Beta-actin (Actb) were tested as candidate reference genes being the latter the most stable to normalize Ct-values by 1Ct method. Same data were obtained if Ppia, Gapdh, or Pcx were used as housekeeping controls (data not shown).

# Statistics

Data were analyzed by GraphPad software and expressed as means ± SE. Statistical differences between groups were determined by unpaired t-test. Differences were considered statistically significant at the p < 0.05 level of confidence.

# RESULTS

# Morphological Analysis Showed Moderate Signs of Muscle Atrophy in longissimus dorsi of Mice Exposed to 30 Days of Microgravity

To assess whether microgravity exposure induced atrophy in longissimus dorsi of spaceflown mice (BF) compared to ground controls (BG and FC), haematoxylin-eosin (H.E.), and immunofluorescence analysis were performed. H.E. staining confirmed the absence of histopathological alterations, such as central nuclei, immune cell infiltration, or myofiber degeneration, in all experimental groups (BF, BG, and FC; data not shown).

To evaluate microgravity-induced myofiber cross sectional area (CSA) changes and phenotype shift (slow-to-fast) in skeletal muscle fibers, we performed multiple immunofluorescence staining using antibodies recognizing specific myosin heavy chain isoforms (MyHC I, 2A, 2B) and the sub sarcolemma marker dystrophin. As shown in **Figure 1**, no changes in CSA were observed in longissimus dorsi of spaceflown mice compared to ground controls. The comparison of the frequency distribution of CSAs in longissimus dorsi revealed no differences between the experimental groups (**Figure S1A**). Similar results were obtained calculating myofiber minimum Feret diameter (**Figure S1B**). In regard to myofiber type composition, we observed an increase in the percentage (1%) of fast 2B fibers in only one of the spaceflown mice. Given the low number of animals analyzed (n = 2) for immunohistochemistry, these results suggest that 30 days of microgravity exposure may not be sufficient to induce robust CSA changes in mouse longissimus dorsi, though the onset of the phenotypic myofiber shift in the paraspinal muscle is already detectable.

# Microgravity-Induced Gene Expression Adaptation in longissimus dorsi

To analyse the adaptation to microgravity of the paraspinal muscle transcriptome, we performed an expression profile analysis of longissimus dorsi from space flown (BF) vs. ground control mice (BG: mice housed in the same biosatellite habitat on ground, and FC: mice housed in the animal facility). A total of 15 mouse longissimus dorsi (BF n = 5, BG n = 5, and FC n = 5) were analyzed. To identify genes differentially regulated in muscle of space flown mice, we compared the muscle transcriptome of BF vs. BG, FC vs. BG and BF vs. FC. The BF vs. BG and BF vs. FC comparisons (flown mice vs. ground controls) reflected the gene expression adaptation in skeletal muscle exposed to microgravity, while the comparison of the two ground controls (FC vs. BG) was needed to rule out possible gene expression changes originating from different housing conditions used in the present study protocol.

Affimetrix data were analyzed applying a False Discovery Rate (FDR) cut-off of 5% and a Fold Change (FC) cut-off < −2 and > 2. Venn diagram in **Figure 2** shows the number of genes significantly differentially regulated in longissimus dorsi (BF vs. BG, FC vs. BG, and BF vs. FC). As expected, the higher number of differentially regulated genes were found comparing BF vs. BG (89 transcripts) and BF vs. FC (68 transcripts), indicating that microgravity exposure induced an adaptation of gene expression in longissimus dorsi of spaceflown mice (BF) compared to ground controls (BG and FC). On the other hand, the lower number of genes (33 genes) differentially regulated comparing the two ground controls (FC vs. BG), only 2 of which overlapping with transcripts found in BF vs. BG, indicated that the housing condition in the BION-M1 biosatellite on its own did not affect gene expression changes observed in muscle of flown mice compared with ground control (BG).

As shown in **Figure 3**, the two-way hierarchical clustering analysis centered on genes significantly differentially regulated in BF vs. BG revealed some similarity in the gene expression of the two ground controls (BG and FC) compared to longissimus dorsi muscles of microgravity exposed mice (BF) (**Figure 3**). Thus, BG and FC samples resulted in the same gene cluster arrangement.

One major aim of this study was to identify molecular players involved in longissimus dorsi adaptation process induced by exposure to microgravity environment. Thus, we focused the present analysis only on BF vs. BG comparison. Within the complete list of 89 genes differentially regulated in BF vs. BG (**Table 1**), the expression of 19 genes was downregulated, while 70 genes were upregulated. Next we performed gene ontology (GO) analysis to identify genes specifically linked to biological functions or molecular pathways. Database for Annotation, Visualization and Integrated Discovery (DAVID) analysis identified genes linked to four main functional clusters (**Table 2**): transcription regulation, transporter activity, ErbB signaling pathway, RNA recognition motif and other genes of interest (classified as other/unknown in **Table 2**). All the genes included in these functional clustering were significantly and differentially regulated in BF vs. BG, while only few of such genes were differentially regulated significantly compared to the two ground controls (BG vs. FC) or with a fold change < −2 or > 2, thus confirming that the observed changes in gene expression were specifically induced by microgravity exposure (**Figure 4**).

# Real Time PCR Validation of Microgravity-Induced Gene Expression Changes in longissimus dorsi

Quantitative real time PCR (qPCR) was performed to validate gene expression changes observed in longissimus dorsi following 30 days spaceflight (n = 5 for each group: BF, BG, and FC). The choice of genes to be validated was essentially centered on their p-values, fold changes (> 2.0 or < −2.0), and the potential association with skeletal muscle pathophysiology. Due to the low amount of tissue used for RNA extraction, we evaluated only the expression of 10 genes by qPCR in flown mice (BF) compared to ground controls (BG and FC): lipocalin 2 (Lcn2), sestrin 1(Sesn1), kelch-like ECH-associated protein 1(Keap1), glycogen synthase kinase 3 beta (Gsk3b), phosphatidylinositol 3 kinase, regulatory subunit polypeptide 1 (p85 alpha) (Pik3r1), vmaf musculoaponeurotic fibrosarcoma oncogene family protein

B (Mafb), synaptojanin 2 (Synj2), protein kinase C delta (Prkcd), Muscle Atrophy F-box (MAFbx/Atrogin-1/Fbxo32), and Muscle RING Finger 1 (MuRF-1). **Table S1** includes the sequence of the primer used in qPCR analysis. Four housekeeping genes were selected as reference to calculate the delta Ct of the selected genes: Actb, beta actin; Ppia, cyclophilin A; Gapdh, glyceraldehyde-3-phosphate dehydrogenase and Pcx, pyruvate carboxylase. PCR data normalized using Actb confirmed that Lcn2, Sesn1, Pik3r1, Mafb, Prkcd, Atrogin-1, and MuRF-1 were differentially regulated significantly in longissimus dorsi of BF compared to BG mice (Lcn2 p = 0.0062, Sesn1 p = 0.0090, Pik3r1 p = 0.036, Mafb p = 0.0094, Prkcd p = 0.0030, Atrogin-1 p = 0.0003, and MuRF-1 p = 0.0029), confirming the reliability of the Affymetrix data analysis (**Figure 5**). Identical results were obtained using Gapdh, Ppia, and Pcx as housekeeping genes (Data not shown).

# Tongue Muscle as in-Flight Negative Control of Microgravity Transcriptome Adaptation

To investigate whether microgravity exposure affected gene expression also in muscles that are constantly "activity loaded" also in spaceflight, we analyzed the transcriptome of mouse tongue from spaceflown mice (BF) compared to ground control (BG). Affymetrix analysis of BF vs. BG (applying a FDR <5% and fold change < −2 and > 2 cut-off) showed that only 27 genes were differentially regulated in the tongue muscle of spaceflown mice. **Table S2** shows the complete list of differentially regulated genes in tongue. The two way hierarchical clustering analysis centered on genes differentially regulated in BF vs. BG showed similarity in the gene expression of the BF and FC groups (**Figure 6**), suggesting that in this particular non-appendicular visceral skeletal muscle gene expression regulation has a

relatively low sensitivity to microgravity exposure compared to appendicular or somatic skeletal muscle in murine vertebrates.

# DISCUSSION

Our study investigated for the first time paraspinal muscle adaptation in 30 days space-flown mice both at morphological and gene expression level. At present, only few studies analyzed the effect of microgravity exposure on paraspinal skeletal muscle, and to our knowledge no study actually quantified the changes

#### TABLE 1 | Longissimus dorsi transcriptome adaptation to microgravity exposure.


(Continued)

### TABLE 1 | Continued


Complete list of genes found differentially regulated in longissimus dorsi of BF vs. BG mice, meeting FDR < 0.05 and fold change < −2 and > +2 criteria.

in either myofiber cross sectional area or fiber type composition neither in spaceflight crew member or space-flown rodent paraspinal back muscles. So far human studies on lumbar back muscle are mostly based on clinical imaging (magnetic resonance imaging) and physiological measurements (e.g., strength test). On the other hand, the effects on human paraspinal muscle deconditioned by the spaceflight analog bed rest have been already investigated to test efficacy of countermeasures able to prevent or reduce the impact of extended disuse or unloading of the spine. In a 60-days 6◦ head-down tilt (HDT) bed rest study, resistive exercise with or without superimposed whole body vibration reduced the whole muscle cross sectional area (CSA) loss observed in different paraspinal functional muscle groups such as lumbar multifidus (medial group), lumbar erector spinae (lateral group) and the anterior and medial hip group with the psoas (Belavy et al., 2010). Moreover, it has been recently reported


#### TABLE 2 | Genes differentially regulated in longissimus dorsi of spaceflown mice linked to main functional clusters identified by DAVID.

List of genes linked to five different gene clusters (transcriptional regulation, transported regulation, ErbB signaling pathway, RNA recognition motif, and other/unknown) differentially regulated in longissimus dorsi of BF vs. BG mice. Gene transcripts meeting FDR < 0.05 and fold change < −2 and > +2 criteria are included.

that lower body negative pressure (LBNP) treadmill and flywheel significantly reduced total lumbar paraspinal muscle loss (76% represented by erector spinae) induced by 60 days of 6◦ HDT bed rest in women (Holt et al., 2016). Unfortunately, paraspinal back muscle biopsies, which would allow a more accurate view on disuse-induced structural and molecular tissue changes, may not be feasible from bed rest participants or even crew in Space because of two major constraints: firstly, the paraspinal back muscle shows some anatomical and functional complexity, for example subdivision into several muscular subgroups (e.g., medial deeper and lateral superficial columns), and secondly, there are obvious ethical constraints and risks of invasive biopsy sampling procedure on the back muscles of the participant's trunk staying supine for longer periods in bed rest.

However, the results reported in this study are based on the quantification of myofibers CSA and type composition of longissimus dorsi tissue samples from mice exposed to microgravity for as long as 30 days, showing that spaceflight only moderately affected the myofiber CSA and only slightly increased the amount of type 2B myofibers in longissimus dorsi of spaceflown mice compared to ground controls. These results suggest that, despite elevated levels of Atrogin-1 and MuRF-1 found in muscle of space-flown mice, robust morphological changes distinctive of paraspinal muscle deconditioning are likely just to be set on in mice exposed for 30 days to microgravity. Our present data on mice partially disagree with those of Yamakuchi and coworkers, who described a "qualitative" decrease in the number and volume of type 1 MyHC positive myofibers in paraspinal rat muscle exposed to microgravity for 14 days (Yamakuchi et al., 2000). On the other hand, we expected to find some lower effects of microgravity adaptation in mouse longissimus dorsi compared to the robust changes previously observed in crewmembers after spaceflight or long term bed rest participants, considering the obvious differences in paraspinal muscle loading between bipedal humans and four-legged vertebrates on ground, and in particular during body movement adaptations of both species observed in the microgravity environment. Moreover, another possible source of discrepancy between our results and human studies might originate by the anatomical complexity of back muscles and the type of analysis performed. The major part of the human studies analyzing paraspinal muscle deconditioning focuses on imaging analysis (MRI) of functional groups of muscle in the lumbar region, such as the erector spinae (including components of different muscles such as multifidus, longissimus dorsi, iliocostalis, and spinalis dorsi (Belavy et al., 2010; Holt et al., 2016).

Though, the apparent lack of an atrophic structural phenotype in mouse longissimus dorsi (i.e., almost constant myofiber CSA with little slow-to-fast myofiber shift) was also confirmed by the global gene expression and gene ontology (GO) analysis, by which no genes linked to the sarcomere structure were found differentially regulated in space-flown mice. Surprisingly, transcripts coding for two ubiquitin ligases, Atrogin-1 and MuRF-1, involved in skeletal muscle atrophy (Bodine et al., 2001) were upregulated in space-flown mice compared with ground controls. Since Atrogin-1 and MuRF-1 expression levels have never been previously investigated in longissimus dorsi

FIGURE 5 | Real time qPCR analysis of selected genes differentially regulated in longissimus dorsi of space-flown mice compared to ground controls. Expression levels of lipocalin 2 (Lcn2), sestrin 1(Sesn1), kelch-like ECH-associated protein 1(Keap1), glycogen synthase kinase 3 beta (Gsk3b), phosphatidylinositol 3-kinase, regulatory subunit polypeptide 1 (p85 alpha) (Pik3r1), v-maf musculoaponeurotic fibrosarcoma oncogene family protein B (Mafb), synaptojanin 2 (Synj2), protein kinase C delta (Prkcd), Muscle Atrophy F-box (Atrogin-1), and Muscle RING Finger 1 (MuRF-1) were evaluated by real-time quantitative PCR in longissimus dorsi of space-flown mice (BF, n = 5) and ground controls (BG and FC, each n = 5). Graph shows ∆C<sup>t</sup> ± SEM; \*\*\*p < 0.0009, \*\*p < 0.0095, and \*p < 0.05.

following unloading, different mechanisms could explain the overexpression of these two ubiquitin ligases without robust signs of muscle atrophy. For example, it is known that in rodent skeletal muscles the over-expression of transcripts coding for Atrogin-1 and MuRF-1 occurs relatively early following unloading and decreases back to baseline level within 15 days (Bodine et al., 2001; Bodine and Baehr, 2014). Thus, these data suggest that the expression of these two genes, usually considered an early event in the onset of the atrophic phenotype, is instead a relatively late event (30 days) in longissimus dorsi of space flown mice. Another possible explanation could involve a muscle specific mechanism regulating the translation of both transcripts. For example, it has been previously shown that miR-23 a suppresses the translation of both MAFbx/atrogin-1 and MuRF-1 in a 3\_-UTR-dependent manner (Wada et al., 2011). Unfortunately, in the absence of time course experiments needed to evaluate Atrogin-1 and MuRF-1 expression in longissimus dorsi of space flown mice and the onset of back muscle atrophy, we can only speculate about the gap between the expression of these two genes and the translation of the corresponding transcripts. Further investigations are obviously needed to understand the role of Atrogin-1 and MuRF-1 in the regulation of the phenotype in mouse paraspinal muscle.

We recently investigated the morphological changes induced by 30 days microgravity in mouse soleus and extensor digitorum longus (EDL) (Gambara et al., 2017): reduced CSAs were observed in type I, IIa, IIb, and IIx myofibers in soleus of mice flown aboard of the biosatellite BION-M1 whereas no changes were observed in EDL, thus largely confirming the notion that soleus, the postural slow type muscle of lower murine limb, was highly responsive to microgravity unloading as compared to EDL, the fast type murine muscle. Moreover, the analysis of myofiber phenotype showed a myofiber type shift from slow to fast mainly in soleus muscle of flown mice (Gambara et al., 2017). Since the myofiber type composition of mouse longissimus dorsi is prevalently fast, as in the case of EDL, minor atrophying effect of microgravity on this specific back muscle was expected.

Among the genes linked to the transcriptional regulation that we found differentially regulated after microgravity exposure, Mafb (v-maf musculoaponeurotic fibrosarcoma oncogene family protein B) was found to be highly downregulated. Mafb (named also as Kreisler or Krml) is known to be essential for the hindbrain development and defects in the organization of facial motor neurons were found in Mafb mutant mice (McKay et al., 1997). More recently it has been demonstrated that Mafb has a pivotal role in the development of Duane syndrome, a congenital disorder generated by an aberrant cranial innervation (Park et al., 2016). To our knowledge, a role for Mafb in paraspinal muscle physiology has never been described yet, thus further investigation is needed to elucidate the function of Mafb in this particular context.

Focusing mainly on genes that were validated by real time qPCR, almost all these genes are directly or indirectly linked to insulin signaling and sensitivity in skeletal muscle, highlighting the impact of spaceflight on glucose metabolism in the longissimus dorsi back muscle. It is well-known that both spaceflight and bed rest induce a set of subclinical diabetogenic effects in humans, such as insulin resistance, decreased glucose tolerance, increased glucose level in plasma and reduced insulin (Tobin et al., 2002). Recently Rudrappa and co-workers reviewed the potential reciprocal influence of insulin resistance and muscle disuse atrophy (Rudrappa et al., 2016), concluding that further studies are needed to better understand the causative role of insulin resistance in determining the atrophic phenotype in skeletal muscle.

Among the genes differentially regulated in longissimus dorsi of spaceflown mice, Lipocalin-2 transcripts (Lcn2) were the most upregulated (Fold change = 11.06, BF vs. BG comparison) found in our study. This gene encodes for a small secreted protein that it is known to be involved in the onset of aging and obesity-induced systemic insulin resistance (Law et al., 2010). However, it has been shown that the phosphorylation of insulin receptor and Akt, and insulin-stimulated glucose uptake were not significantly altered in skeletal muscle of Lcn2 knockout mice compared to wild-type mice. These results may be in part explained by the low basal level of Lcn2 expression in murine skeletal muscle (Yan et al., 2007). On the other hand, Guo and co-workers showed opposite results, that is Lcn2 deficiency potentiated diet-induced insulin resistance in mice (Guo et al., 2010). Interestingly, Lcn2 protein level was found to be increased in sera of individuals following bed rest (Rucci et al., 2015), therefore based on our results we may speculate that Lcn2 protein could be at least in part secreted in a soluble form in blood by skeletal muscle tissue similar to the known myokines.

In our study, we also identified another well-known regulator of cell metabolism, Sestrin 1 (Sesn1), that we found to be upregulated in longissimus dorsi of spaceflown mice compared to ground controls. Sestrins are a family of stress activated proteins involved in the metabolism of reactive oxygen species (ROS), in the regulation of autophagy, and in insulin signaling. More in detail, Sestrins increase AMPK activation, consequentially inhibiting mTORC1 activity, known to induce insulin resistance through the inhibition insulin receptor substrates and the consequent reduction of phosphoinositide-3-kinase (PI3K)/Akt signaling (Lee et al., 2013). Moreover, it has been shown that only Sestrin 3 (Sesn 3) was upregulated in vastus lateralis of type 2 diabetes patients compared to healthy individuals but its role is more likely to be linked to skeletal muscle differentiation than regulating glucose and lipid metabolism. On the other hand, hydrogen peroxide increased only mRNA level of Sesn 1 and 2 in human myotubes, suggesting that these two isoforms are specifically involved in the response to oxidative stress in skeletal muscle (Nascimento et al., 2013). With regard to p85 alpha PI3K regulatory subunit (Pik3r1), we found upregulation of this transcript in longissimus dorsi of the spaceflown mice. This result agrees with the upregulation of Pik3r1 found in gastrocnemius of mice exposed to microgravity for 11 days (Allen et al., 2009). PI3K is known to be crucial for insulininduced glucose transport (Hara et al., 1994) and it has been shown that in bed rest, for example, the glucose transporter GLUT4, Akt signaling, and Glycogen synthase activity are equally reduced (Bienso et al., 2012). Interestingly, Terauchi and coworkers showed an increased insulin sensitivity and increase in glucose transport in skeletal muscle of Pik3r1−/<sup>−</sup> mice due to the compensatory upregulation of the other two isoforms p55 and p50 alpha (Terauchi et al., 1999). To better understand the role of Pik3r1 in the onset or prevention of insulin resistance in disused skeletal muscle, the expression of all Pik3r1 isoforms should be further investigated in microgravity or other experimental models of unloading.

In the current study, we also found that protein kinase C delta (Prkcd) transcript were significantly upregulated in longissimus dorsi of spaceflown mice. This result supports the crucial role of Prkcd in skeletal muscle insulin sensitivity. In fact, an increase in the Prkcd levels related to the aging has been previously described, and the muscle-specific Prkcd knockout improved aging-related decline of glucose tolerance and insulin resistance (Li et al., 2015).

Interestingly, two of the validated genes differentially regulated in longissimus dorsi of space-flown mice and linked to insulin resistance, Lcn2 and Sesn1, and another gene coding for Mafb were previously identified as overexpressed genes in both soleus and EDL of mice flown for 30 days onboard the BION-M1 biosatellite (Gambara et al., 2017), suggesting that the regulation of only few genes following microgravity exposure may not be strictly muscle specific. On the other hand, microarray analysis of space-flown murine soleus and EDL compared to ground controls showed that the larger number of differentially regulated genes was highly muscle specific: only 24 differentially regulated genes were in common between the two functionally different muscles (Gambara et al., 2017). In the present study the tongue muscle was used as reference internal control of microgravity exposure as the tongue represents a visceral skeletal muscle with obviously constant activity load. A typical example of tongue loading is performed during daily nutritional activities in microgravity (i.e., water and food uptake) as recently proposed for mice masticatory muscle (masseter, MA). MA insensitivity to microgravity was explained by the protective effect against mass loss due to continuing chewing-induced loading following 13 days spaceflight onboard the STS-135 Space shuttle (Philippou et al., 2015). Further studies are however needed to show if tongue muscle also shares similar atrophy mechanisms than somatic/appendicular skeletal muscle following long-duration spaceflight.

In summary, a number of genes linked to insulin sensitivity and metabolism of skeletal muscle were found significantly dysregulated in the mouse longissimus dorsi back muscle following 30 days of microgravity exposure compared to ground control animals. The apparent absence of robust signs of muscle atrophy in paraspinal muscles following 30 days of microgravity opens new questions on the potential roles of these genes in the onset of peripheral insulin resistance following unloading and its consequences on vertebrate skeletal muscle structure and function. Finally, in the present study microgravity-induced transcriptional adaptation of mouse paraspinal muscles has been investigated by means of global gene expression profiling in space-flown mice, providing the basis for a deeper understanding of the complex mechanisms of molecular adaptation to microgravity in vertebrate back muscles as potential cause of lower back pain and spine destabilization processes following whole body unloading conditions in different clinical settings, bed rest, and spaceflight.

# AUTHOR CONTRIBUTIONS

Conceptualization: DB and MS; Data curation: GG, MS, and UU; Formal analysis: GG, UU, and SC; Funding acquisition: DB; Investigation:MS,DB,andGG;Methodology:GS,MG,UU,andSF; Project administration: DB and MS; Resources: DB; Supervision: HG; Validation: UU and SF; Writing—original draft: GG and DB; Writing—review and editing: GG and MS, DB, and PV.

# FUNDING

This work was supported from grants of the Department of Economics and Technology of the German Government (BMWi) through the German AeroSpace Board, Deutsches Zentrum für Luft- und Raumfahrt (DLR), e.V. Bonn, Germany (grant # 50 WB821, 1121, and 1421 to DB). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

# ACKNOWLEDGMENTS

The BION-M1 mission was organized in 2013 by the Russian Space Agency (Roskosmos) in cooperation with the Institute of Biomedial Problems (IMBP); Moscow, Russia. We herewith acknowledge the operational support of the IMBP, in particular Professor B. Shenkman and his group, Moscow, Russia, for expert tissue dissection, sample freezing and tissue sharing between the IMBP and the Charité Berlin.

# REFERENCES


# SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fphys. 2017.00279/full#supplementary-material

Figure S1 | CSA heterogeneity and Myofiber minimum Feret diameter. (A) comparison of the frequency distribution of CSAs in longissimus dorsi of the three experimental groups (Figure S1). (B) Quantification of the minimum Feret diameter in longissimus dorsi of mice from the three experimental groups (BF, BG, and FC, each n = 2) shown as scatter plots.

Table S1 | Quantitative PCR primers and conditions. HK, Reference genes; bp, expected product size; T◦ , annealing temperature. <sup>∗</sup> Sandona et al. (2012) Adaptation of Mouse Skeletal Muscle to Long-Term Microgravity in the MDS Mission. PLoS ONE 7(3):e33232.

Table S2 | Microgravity-induced gene expression changes in the tongue muscle of space -flown vs. ground control mice. List of genes differentially regulated in the tongue muscle (internal reference control) of BF vs. BG mice, meeting FDR < 0.05 and fold change < −2 and > +2 criteria. The comparisons between all the three experimental groups are shown.


causes duane syndrome, aberrant extraocular muscle innervation, and innerear defects. Am. J. Hum. Genet. 98, 1220–1227. doi: 10.1016/j.ajhg.2016. 03.023


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2017 Gambara, Salanova, Ciciliot, Furlan, Gutsmann, Schiffl, Ungethuem, Volpe, Gunga and Blottner. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Coherent Multimodal Sensory Information Allows Switching between Gravitoinertial Contexts

Marie Barbiero1, 2, Célia Rousseau1, 2, Charalambos Papaxanthis 1, 2 and Olivier White1, 2 \*

<sup>1</sup> Université de Bourgogne Franche-Comté, Cognition Action et Plasticité Sensorimotrice UMR1093, Dijon, France, <sup>2</sup> Institut National de Santé et de Recherche Médicale, Cognition Action et Plasticité Sensorimotrice UMR1093, Dijon, France

Whether the central nervous system is capable to switch between contexts critically depends on experimental details. Motor control studies regularly adopt robotic devices to perturb the dynamics of a certain task. Other approaches investigate motor control by altering the gravitoinertial context itself as in parabolic flights and human centrifuges. In contrast to conventional robotic experiments, where only the hand is perturbed, these gravitoinertial or immersive settings coherently plunge participants into new environments. However, radically different they are, perfect adaptation of motor responses are commonly reported. In object manipulation tasks, this translates into a good matching of the grasping force or grip force to the destabilizing load force. One possible bias in these protocols is the predictability of the forthcoming dynamics. Here we test whether the successful switching and adaptation processes observed in immersive environments are a consequence of the fact that participants can predict the perturbation schedule. We used a short arm human centrifuge to decouple the effects of space and time on the dynamics of an object manipulation task by adding an unnatural explicit position-dependent force. We created different dynamical contexts by asking 20 participants to move the object at three different paces. These contextual sessions were interleaved such that we could simulate concurrent learning. We assessed adaptation by measuring how grip force was adjusted to this unnatural load force. We found that the motor system can switch between new unusual dynamical contexts, as reported by surprisingly well-adjusted grip forces, and that this capacity is not a mere consequence of the ability to predict the time course of the upcoming dynamics. We posit that a coherent flow of multimodal sensory information born in a homogeneous milieu allows switching between dynamical contexts.

Keywords: multisensory information, feedback, switching, grip force, human centrifuge, gravity, internal model

# INTRODUCTION

Consider a worker whose job is to sort Christmas packages of varying size and weight into bins, bags, or slots. Each of these packages will have different inertial properties and will impose different loads on the arm. The physical properties of these objects are not fixed but vary according to a given statistical distribution that depends both on object properties and on the sequence of planned movements. Despite the fact that variability occurs on a

#### Edited by:

Ovidiu Constantin Baltatu, Anhembi Morumbi University, Brazil

#### Reviewed by:

Joyce McClendon Evans, University of Kentucky, USA Luzia Grabherr, Centre Hospitalier Universitaire Vaudois (CHUV), Switzerland

\*Correspondence: Olivier White olivier.white@u-bourgogne.fr

#### Specialty section:

This article was submitted to Integrative Physiology, a section of the journal Frontiers in Physiology

Received: 20 January 2017 Accepted: 21 April 2017 Published: 11 May 2017

#### Citation:

Barbiero M, Rousseau C, Papaxanthis C and White O (2017) Coherent Multimodal Sensory Information Allows Switching between Gravitoinertial Contexts. Front. Physiol. 8:290. doi: 10.3389/fphys.2017.00290 movement basis, this context is predictable in the sense that the worker can estimate the upcoming mechanical properties based on visual cues. If the worker carries out this task for a prolonged time, s/he will adjust her/his motor plan according to the object and action. In other words, motor adaptation and context switching will occur (Kawato, 1999).

Studies regularly use robotic devices to perturb the dynamics of motor tasks. This allows testing of how specific parameters such as stiffness (Descoins et al., 2006), viscosity (Shadmehr and Mussa-Ivaldi, 1994) and inertia (Wang and Sainburg, 2004) are taken into account by the central nervous system to plan efficient actions. Robot-based investigations highlighted limitations of the brain to concurrently learn different task dynamics (Gandolfo et al., 1996; Conditt et al., 1997; Karniel and Mussa-Ivaldi, 2002), even when the expected dynamics are made fully predictable through the use of explicit cues, such as the association of a color to a direction of a forthcoming perturbation (Krakauer et al., 1999; Osu et al., 2004). In other contexts, however, the motor system is quite capable of learning different dynamics. If one moves the arm alone or the arm linked to an unfamiliar object, two parallel predictive strategies are formed by the brain (Kluzik et al., 2008). The same observation has been reported with different objects and one or two hands (Ahmed et al., 2008; White and Diedrichsen, 2008). Furthermore, this efficient concurrent learning is also possible if control policies or predictive strategies—are associated to different contexts, such as a leftward or rightward perturbing force field (White and Diedrichsen, 2013). Whether participants can or cannot switch between contexts critically depends on experimental details.

Motor adaptation has also been probed using other approaches. For instance, parabolic flights and human centrifuges provide unique means to alter the whole gravitational or gravitoinertial environment. In the former, the participant is immersed into a repeated gravitational profile (e.g., 1, 1.8, 0, 1.8 g and back to 1 g, where 1 g is Earth gravity). Human centrifuges allow programming an arbitrary gravitoinertial environment (e.g., staircase function from 1 to 3 g). In contrast to conventional robotic experiments, where only the hand is perturbed, parabolic flight and rotating-room environments plunge the subject into a radically new setting. Nearly perfect adaptation of motor responses in those challenging environments were observed in dexterous manipulation (Augurelle et al., 2003; White et al., 2005; Göbel et al., 2006; Mierau et al., 2008; Crevecoeur et al., 2009b), arm movements (Papaxanthis et al., 1998; White et al., 2008) and more realistic tasks (Steinberg et al., 2015).

In previous investigations involving movements in altered gravitoinertial environment, the dynamic consequences of actions only depended on time. In other words, external constraints were constant in the Euclidian space, but could vary according to a predefined experimental schedule and/or selfgenerated movements. A question arises as to whether adaptation observed in the above studies is a mere consequence of the fact gravitoinertial profiles vary over time and can be predicted? A structural decoupling between underlying variables—space and time—may highlight different time scales of adaptation. Here, we test the ability of participants to adapt and switch between very unusual dynamical contexts generated by rotation of a short-arm human centrifuge. Space and time are decoupled because the gravitoinertial vector can vary significantly along a short movement amplitude. In other words, this also means that local gravity will be different according to where the object is in space, independently of time (see Methods). Twenty participants cyclically moved an object along the head-to-foot body axis, aligned with the gravitoinertial vector, induced by rotation of the centrifuge. We measured adaptation through the robust paradigm of grip force adjustments to load force (Westling and Johansson, 1984; Jaric et al., 2005). When moving an object with a precision grip configuration (thumb opposing the index finger), the brain must estimate the dynamical consequences of the movement on the tangential destabilizing force (load force). This is necessary to estimate the required grasping force (grip force) and avoid accidental slips. Many studies have shown very good adjustments of grip force to a variety of physical object parameters (mass, texture, shape, friction) or environments (force fields, gravitational fields). We created different dynamics by instructing participants to perform the movements at three different paces. These contextual sessions were interleaved such that we could simulate concurrent learning at two time scales (within a session and between sessions). We speculate that failure to adapt grip force like in other similar experiments would underline the fact that time alone is not sufficient to predict the forthcoming dynamics. In contrast, successful grip force adaptation would demonstrate the flexibility of the nervous system to efficiently switch between these truly novel dynamics.

# MATERIALS AND METHODS

# Participants and Ethical Considerations

Twenty healthy, un-trained, non-obese, non-smoking men (n = 10, 29.5 ± 5.3 years old, 178.9 ± 4.6 cm, BMI 25.1 ± 2.0 kg/m<sup>2</sup> ) and women (n = 10, 27.6 ± 4.6 years old, 165.1 ± 4.8 cm, BMI 21.9 ± 1.9 kg/m<sup>2</sup> ) without histories of vasovagal syncope or cardiovascular problems took part in this protocol. Each participant received a comprehensive medical examination by a medical doctor from MEDES (French Institute for Space Medicine and Physiology) prior to participation. Inclusion criterial were: age between 20 and 40 years old, BMI < 30 kg/m<sup>2</sup> , normal clinical examination, normal electrocardiogram and arterial pressure, signed consent and enrolled in the French social security system. Those already participating in another biomedical test, who did not comply with any of the above inclusion criteria or under medication for 8 days before the experiment were not retained for this experiment. The experiment could be interrupted at any time upon participants' request or his/her health status under constant monitoring by a medical supervisor. We had to interrupt the experiment during the last session (see Experimental procedures) for five female participants who showed signs of motion sickness. This did not impact our results since we designed the experiment in such a way to accumulate a large data set in a short amount of time. Consequently, we had only slightly less data (30% drop out) for these five participants.

The experiment took place at MEDES, Toulouse (France). None had previously experienced hypergravity in a short arm human centrifuge (SAHC) and the preparatory visit did not include a familiarization session in order to keep them naïve with respect to this new environment. The study was conducted in accordance with the ethical practices stipulated in the Declaration of Helsinki (1964). Ethics approval was obtained by MEDES (2014-A00212-45). All volunteers signed the informed consent form, which is stored at MEDES.

# Experimental Procedures

The participant laid on a horizontal bed and was monitored with heart rate and arterial pressure systems using non-invasive photoplethysmography (Portapres: FMS, the Netherlands). The Portapres finger cuff was placed on the resting hand during the task. Her/his head rested on a thin pillow and her/his feet contacted a rigid metallic platform. The participant was then equipped with headphones in order to maintain contact with the operator in the control room. Visual feedback of the environment was prevented by placing an opaque ventilated box above the head.

Participants underwent three centrifugation sessions, each lasting 5 min (**Figure 1**). These sessions were separated by 10-min breaks during which participants rested supine and quietly while the centrifuge was idle. During centrifugation and following a signal from the operator, participants performed rhythmic upper arm movements in the sagittal plane with an instrumented object held in precision grip. The device recorded the 3-d forces and torques (mini40 force-torque sensor, 0.04 kg, ATI Industrial Automation, NC, USA). A 3 d accelerometer was also embedded in the object (TSD109C Tri-Axia, BIOPAC, ±5 g, 0.017 kg, CA, USA). All signals were continuously sampled at 200 Hz through a DAQ board (NI USB 6211, National Instruments,

Austin, TX) and stored on a computer laptop strapped on the centrifuge. Participants were trained to produce trajectories parallel to the long (head-to-foot) body axis. Movement pace was provided by a metronome that emitted 2 auditory signals per cycle, one at the top and one at the bottom of the trajectory. The rhythm was controlled by the operator and routed via headphones to the subject's ears. Three paces (Slow = 0.7 Hz, Medium = 1 Hz and Fast = 1.3 Hz) were presented twice each for 30 s (3 paces × 2 repetitions × 30 s = 3 min). Pauses of about 20 s separated movement conditions in order to prevent fatigue and also to ensure that a good contact was maintained with the participant. Pace order was randomized and counterbalanced across participants. At the end of each session, the centrifuge went back to idle position and the subject was debriefed.

# Short Arm Human Centrifuge Configuration

Previous experiments extensively tested grip force adaptation to load force (LF) when either mass (m), acceleration (a) or gravity (g) were altered, separately or in combination (White et al., 2005; White, 2015), LF (t) = mg + ma(t). In this experiment, we set out to investigate how grip force is adjusted to load force when the gravitoinertial resultant also explicitly depended on position, LF (x,) = mg(x) + ma(t).

A short arm human centrifuge offered a unique opportunity to separate out the effects of time and space on the adaptation process of grip force to load force. Indeed, in contrast to a long radius human centrifuge, the resultant between centripetal acceleration induced by the rotation of the centrifuge and veridical gravity varies more for a given amplitude of movement close to the rotation axis than far from the rotation axis. In other words, gravitoinertial gradients are larger when approaching the center of rotation. Consider a point mass m situated at a horizontal distance R from the axis of rotation (**Figure 1**). This object is rotated at a constant angular velocity ω = 2π T , with T being the period of rotation of the centrifuge, and is moving at a constant velocity v = 2πR T , tangent to the circular trajectory. This mass is subjected to both a constant gravitational acceleration directed downward (**Figure 1**, g) and to the centripetal acceleration (**Figure 1**, a<sup>c</sup> = v 2 R ). Therefore, when m is translated by a distance x along the radius, the magnitude of the gravitoinertial vector (**Figure 1**, Gz(x), in units of g) is given Gz (x) = 1 g q 16π<sup>4</sup> (R+x) 2 <sup>T</sup><sup>4</sup> <sup>+</sup> <sup>g</sup> 2 . We identified the centrifuge and geometrical parameters that maximized the gravitoinertial gradient. In other words, we adjusted T, movement space [R, R + x] and bed inclination angle such that <sup>∂</sup>Gz(x) ∂z was maximal. We also had to take both ethical and technical constraints into account as some values of these parameters either could not be handled by the centrifuge or would have generated strong motion sickness. Details of this mathematical optimization process are presented in the Appendix of Supplementary Material. The centrifuge completed one revolution in 2.09s, the bed was tilted 24◦ downward and the elbow was positioned at 1.39 m from the axis of rotation. This configuration allowed us to induce a 0.4 g-gradient between both extremes of the hand trajectory (**Figure 1**, P<sup>T</sup> and PL)

TABLE 1 | Resultant dynamics at five points along the head-to-foot body axis (tilted 24◦ downward) placed in the centrifuge (one revolution in 2.09 s).


The elbow was positioned at 1.39 m from the axis of rotation. The first column denotes positions as illustrated in Figure 1 (PH, head; P<sup>T</sup> , top of trajectory; P<sup>E</sup> , elbow; PL, lower part of trajectory; and P<sup>F</sup> , feet). The next columns report, for each point: horizontal distance from the axis of rotation (X-distance), magnitude of centripetal acceleration (|Centripetal Acc|, horizontal vector in Figure 1), magnitude (|Gz|) and direction (dir(Gz) of the gravitoinertial resultant (oblique vector in Figure 1).

which is a very strong perturbation and unnatural. The five positions (PH, PT, PE, PL, P<sup>F</sup> in **Figure 1**) were subjected to different gravitoinertial vectors. **Table 1** reports for each point, its horizontal distance from the axis of rotation, the magnitude of the centripetal acceleration and the magnitude and direction of the gravitoinertial vector.

# Model of the Task

In this section, we develop a simple model of the task that allows us to identify differences between acceleration signals when we take into account the effects of the centrifuge or not. Portions of cycles for which these differences are the largest are of particular interest. Indeed, we expect grip forces to be proportional to the real inertial variations.

Participants moved a small object (mass = 0.057 kg) in a noninertial reference frame along a straight tilted trajectory in the sagittal plane. The accelerometer embedded in the instrument recorded the resultant vector of three accelerations: (1) the Earth constant gravitational attraction, (2) a centripetal acceleration due to the rotation of the centrifuge and (3) the acceleration induced by the movement of the object by the participant. Therefore, the load force that had to be counteracted is given by:

$$\overrightarrow{LF} = m\left(\vec{\mathcal{g}} + \vec{G}\_z(\varkappa) + \vec{a}\_m\right) \tag{1}$$

The first term is constant both in direction and magnitude. The second term varies in amplitude in function of the radial distance x from the axis of rotation. In this section, we quantify how the third term interacts with the two others and we model how pace affects the time course of the acceleration signal within a cycle, and for the three experimental paces.

Let us define a Cartesian reference frame centered on PE, with the x-axis and y-axis pointing rightward and upward, respectively. Rhythmic movements were performed on a straight line between P<sup>L</sup> and PT, starting at the neutral position, i.e., between P<sup>L</sup> and PT. The vectors (x, y), (x˙, y˙) and (x¨, y¨) denote position, velocity and acceleration, respectively. These trajectories are well described with sine waves, both for the x and y components:

$$\begin{cases} \mathbf{x}\left(t\right) = \mathbf{x}\_i + \frac{1}{2} \left(\mathbf{x}\_f - \mathbf{x}\_i\right) \left(\sin 2\pi ft + \frac{1}{2}\right) \\ \mathbf{y}\left(t\right) = \mathbf{y}\_i + \frac{1}{2} \left(\mathbf{y}\_f - \mathbf{y}\_i\right) \left(\sin 2\pi ft + \frac{1}{2}\right) \end{cases} \tag{2}$$

The parameters x<sup>i</sup> , xf , y<sup>i</sup> and y<sup>f</sup> are the initial (subscript i) and final (subscript f) positions in x and y and f is the frequency of movement. Velocity and acceleration are obtained by successive derivations of Equation (2):

$$\begin{cases} \dot{x}\left(t\right) = \pi f \left(\chi\_f - \chi\_i\right) \cos 2\pi ft\\ \dot{y}\left(t\right) = \pi f \left(\chi\_f - \chi\_i\right) \cos 2\pi ft \end{cases} \tag{3}$$

and

$$\begin{cases} \ddot{\mathbf{x}}\ (t) = 2\pi^2 f^2 \left(\chi\_i - \chi\_f\right) \sin 2\pi ft\\ \ddot{\mathbf{y}}\ (t) = 2\pi^2 f^2 \left(\chi\_i - \chi\_f\right) \sin 2\pi ft \end{cases} \tag{4}$$

One can now easily calculate the respective acceleration vectors involved in Eqaution (1):

$$\begin{array}{l} \vec{\text{g}} = (0, -\text{g})\\ \vec{G}\_z(\mathbf{x}) = \begin{pmatrix} \frac{4\pi^2(\mathbb{R} + \mathbf{x})}{T^2}, \mathbf{0} \end{pmatrix} \\ \vec{a}\_m = \begin{pmatrix} \vec{a}\_{m\mathbf{x}}, \vec{a}\_{m\mathbf{y}} \end{pmatrix} \end{array} \tag{5}$$

The centripetal acceleration <sup>G</sup><sup>E</sup> <sup>z</sup> and the acceleration generated by the participant depend on object position. **Figure 2** (left column) depicts, for each pace (three rows) the resultant acceleration with (kE<sup>g</sup> <sup>+</sup> <sup>G</sup><sup>E</sup> <sup>z</sup>(x) + Eamk, red dotted trace) and without (kE<sup>g</sup> + Eamk, blue trace) taking into account the effects of the rotation of the centrifuge. It shows that there are differences between pace conditions but also within the time course of a single cycle. The largest differences, in proportion to the total amplitude of acceleration, are 52% for the fast pace, 28.8% for the medium pace and 5.1% for the slow pace. Interestingly, the contribution of <sup>G</sup><sup>E</sup> <sup>z</sup> strongly depends on the phase of the cycle, especially for the two fastest paces. The subtraction between the traces depicted in **Figure 2** (right column) magnifies how the rotation of the centrifuge contributes to total acceleration and, hence, load force. The largest differences occur at 76.6, 75.2, and 75.5% from cycle onset for fast, medium and slow paces, respectively (vertical cursors). These instants correspond to the lowest part of the trajectory.

# Data Processing and Analysis

Force and acceleration signals were smoothed with a zero phaselag autoregressive filter (cutoff 10 Hz). A trial was defined as a series of cyclic movements. On average, per trial, participants performed 19.5 cycles for 0.7 Hz (SD = 6.9), 20.9 cycles for 1 Hz (SD = 2.2) and 26.3 cycles for 1.3 Hz (SD = 2.5). The largest number of cycles common to all conditions was 17. We analyzed trials and cycles separately. Furthermore, since load force varied differently within a cycle whether we take into account the effects of the rotation or not, we also analyzed four phases of the cycle.

Quantile-quantile plots were used to assess normality of the data. Repeated measures ANOVA was conducted on cycle frequency, grip forces and on the regression coefficients between grip force and load force. When relevant, we assessed the

effects of Session (1, 2, or 3), Frequency (0.7, 1, or 1.3 Hz), Repetition (1 or 2), Cycle (1 to n) and Phase (1, 2, 3, or 4) on the above variables. Post-hoc comparisons were made using Fischer least significant differences (LSD). Paired ttests of individual subject means were used to investigate differences between conditions on the above variables. Data processing and statistical analyses were done using Matlab (The Mathworks, Chicago, IL). We report partial eta-squared values for significant results (p < 0.05) to provide indication on effect sizes.

# RESULTS

Participants cyclically moved an instrumented object along the long body axis aligned with the gravitoinertial direction during rotation in a human centrifuge. Here, we challenged the limits of the adaptation capacity of the motor system by assessing how participants controlled grip force when load force comprised a gravitoinertial component that varied explicitly with local vertical position. The generation of such dynamics can only be tested in a short arm human centrifuge.

We verified that participants adopted a pace that matched the instructions. We used a Fast Fourier Transform to extract the main frequency component of the acceleration profile for each trial. A 2-way ANOVA confirmed a main effect of Frequency [F(2,139) = 143.5, p < 0.001, η 2 <sup>p</sup> <sup>=</sup> 0.66] and Session [F(2,139) <sup>=</sup> 3.4, p = 0.038, η 2 <sup>p</sup> <sup>=</sup> 0.02] on real movement frequency. Paired t-tests revealed no difference between actual and theoretical rhythms for 1 and 1.3 Hz [both t(18) = 1.8, p = 0.084], but faster paces for the slowest condition [0.79 vs. 0.7 Hz, t(18) = 3.1, p = 0.006, η 2 <sup>p</sup> <sup>=</sup> 0.35].

Frequency, acceleration and load forces are linked through Equations (1) and (4). A 3-way ANOVA (factors: Frequency, Session, and Cycles) revealed higher peaks of acceleration in high frequency conditions [F(2, 2,276) = 144.1, p < 0.001, η 2 <sup>p</sup> <sup>=</sup> 0.11 <sup>η</sup> 2 <sup>p</sup> <sup>=</sup> 0.11], which also induced larger peak load forces [F(2, 2,276) = 144.1, p < 0.001, η 2 <sup>p</sup> <sup>=</sup> 0.11<sup>η</sup> 2 <sup>p</sup> <sup>=</sup> 0.11]. As reported previously (Flanagan and Wing, 1995, 1997), participants adopted grip forces proportional to peak load forces, as revealed by proportional peak grip forces [F(2, 2,276) = 15, p < 0.001, η 2 <sup>p</sup> <sup>=</sup> 0.01<sup>η</sup> 2 <sup>p</sup> <sup>=</sup> 0.01].

A first question arises as to how the tight link between grip and load forces was affected by Frequency and whether it was influenced by time. To quantify this relationship, we calculated, for each cycle of movement, the best linear fit between these two time series (Flanagan and Wing, 1995; Hejdukova et al., 2002; Zatsiorsky et al., 2005). Participants accomplished the task for three sessions (Session), each frequency was repeated twice per session (Repetition) and each repetition involved at least 17 cycles (Cycles). We could therefore analyze adaptation at three different time scales. The 4-way ANOVA revealed significant increases of gains [F(2, 3,606) = 31.1, p < 0.001, η 2 <sup>p</sup> <sup>=</sup> 0.02] and offsets [F(2, 3,606) = 7.9, p < 0.001, η 2 <sup>p</sup> < 0.01] with Frequency. Furthermore, gains significantly increased across Session [F(2, 3,606) = 4.6, p = 0.01, η 2 <sup>p</sup> < 0.01] and Repetition [F(1, 3,606) = 10.6, p = 0.001, η 2 <sup>p</sup> < 0.01]. In contrast, offset significantly decreased across Session [F(2, 3,606) = 22.4, p < 0.001, η 2 <sup>p</sup> <sup>=</sup> 0.01] and Repetition [F(1, 3,606) <sup>=</sup> 66, <sup>p</sup> <sup>&</sup>lt; 0.001,

η 2 <sup>p</sup> <sup>=</sup> 0.02]. However, we did not observe any effect of Cycle on these two parameters [gains: F(16, 3,606) = 0.8, p = 0.714; offset: F(16, 3,606) = 0.7, p = 0.761]. To sum up, while Frequency induced larger slopes and safety margins, participants tended to optimize the task by simultaneously increasing the gain and lowering grip force. This adaptation occurred within a trial but not between trials.

Parameters of a linear regression do not provide indications on goodness of fit. Therefore, we pushed our analyses one step further by considering the cross-correlation between grip and load forces within each cycle. This procedure provided an estimate of the overall synergy between the two forces. Correlations quantified how well grip and load force profiles matched, which indicated the accuracy of scaling of grip force. Time-shifts provided a measure of the asynchrony between the two forces. A positive time-shift signaled an anticipatory grip force. A 4-way ANOVA reported significant effects of Frequency [F(2, 2,759) = 42.4, p < 0.001, η 2 <sup>p</sup> <sup>=</sup> 0.03] and Session [F(2, 2,759) <sup>=</sup> 12.7, p < 0.001, η 2 <sup>p</sup> <sup>=</sup> 0.01] on this best correlation coefficient (**Figure 3A**). A post hoc t-test revealed that fast pace induced better correlations than slow [t(17) = 4.3, p = 0.001 η 2 <sup>p</sup> <sup>=</sup> 0.52] and medium paces [t(17) = 5.6, p < 0.001, η 2 <sup>p</sup> <sup>=</sup> 0.65] and that the two slower paces were not different [t(18) = 1.4, p = 0.189]. Furthermore, the time-shift (**Figure 3B**) increased across Session [F(2, 2,759) = 17.2, p < 0.001, η 2 <sup>p</sup> < 0.01] and from Repetition 1 to Repetition 2 [F(1, 2,759) = 11.6, p < 0.001, η 2 <sup>p</sup> <sup>&</sup>lt; 0.01] but not across Frequency [F(2, 2,759) <sup>=</sup> 0.2, <sup>p</sup> <sup>=</sup> 0.807]. Therefore, the synergy between grip force and load force improved across Session, participants adopting a more predictive behavior underlined by increasing time-shifts.

Centrifugation added a position-dependent acceleration component that contributed to the total inertial force, resulting in an unusual perturbation. **Figure 4A** depicts simulated load force over normalized time when the model takes into account the three sources of accelerations (i.e., constant gravity, cyclic movement and centripetal accelerations). It shows that the amplitude of the signal was proportional to frequency. **Figure 4B** also shows simulated data but without taking into account the effects of the rotation. The model predicts very different patterns of acceleration and, hence, load force, if we include or not the effects of the centrifugation. Actual load force traces (**Figure 4C**, averaged normalized cycles across all conditions) clearly resemble the model that includes all acceleration terms (**Figure 4A**). In particular, the three amplitudes were significantly different between Frequency [F(2, 139) = 15.8, p < 0.001, η 2 <sup>p</sup> <sup>=</sup> 0.18] while the average load forces were similar [F(2, 139) = 0.1, p = 0.872]. Furthermore, modeled load force traces intersected at 25 and 75% from cycle onset which is very close to what we observed in real data (28.6 and 74.5%).

Participants should have anticipated the actual load force profile by adjusting grip force. Data show that participants exerted grip forces that paralleled the actual load forces and not the one they might have predicted without taking into account the effects of the centripetal acceleration (**Figure 4D**). A natural question arises as to whether the behavior observed in **Figure 4D** was reached immediately upon exposure to the environment or needed time to settle. To quantify this, we formed five blocks of continuous cycles and plotted averaged force traces across blocks. **Figure 5** depicts these five averaged traces for load force (**Figure 5A**) and grip force (**Figure 5B**). It shows first that load force traces overlap well (**Figure 5A**). In contrast, grip forces exhibit a continuous progression between early (**Figure 5B**, dark lines) and late grip force traces (**Figure 5B**, light lines). While amplitudes gradually decreased across Blocks [F(4, 684) = 15.8, p < 0.001, η 2 <sup>p</sup> <sup>=</sup> 0.04], the occurrences of minimal grip forces shifted sooner in the cycle.

To deepen these analyses, we focused on the normalized time at which the minimal forces were reached. We found that minimal load forces occurred on average 48.2% after cycle onset (**Figure 5C**) and did not vary between Session [F(2, 263) = 2.4, p = 0.09], Repetition [F(1, 263) = 2.7, p = 0.103] or Blocks [F(4, 263) = 2.2, p = 0.119]. In contrast, the same analysis conducted on grip forces reported an initial skewness of 57% in grip force profiles (**Figure 5D**, Session 1, Blocks 1-2) that gradually decreased with Session. We confirmed this observation statistically. The 3-way ANOVA reported a main effect of Session [F(2, 263) = 8.6, p < 0.001, η 2 <sup>p</sup> <sup>=</sup> 0.04], Repetition [F(1, 263) <sup>=</sup> 29.4, <sup>p</sup> <sup>&</sup>lt; 0.001 , <sup>η</sup> 2 <sup>p</sup> <sup>=</sup> 0.07] and Blocks [F(1, 263) = 4.4, p < 0.005, η 2 <sup>p</sup> <sup>=</sup> 0.03] on this minimum grip force. Altogether, this demonstrates that a subtle modification of grip force occurs over time to match the actual and novel perturbation.

# DISCUSSION

The purpose of this study was to test whether the successful adaptation usually reported in altered gravitoinertial environments is a consequence of the ability to predict the time course of the perturbation or results from a more complex process. Put differently, we tested participants' ability to switch

and adapt to a gravitoinertial field induced by a short-arm centrifuge that explicitly varied with position. Apart from following the prescribed rhythmic tone, there were no further accuracy requirements. We addressed these questions by using the well-established grip force/load force coupling paradigm.

Motor adaptation to different dynamical contexts has been widely documented (Wolpert et al., 2011; Wolpert and Flanagan, 2016). To probe motor learning, scientists use robot-based paradigms to perturb a task with fixed and repeatable structures. For example, in a seminal study, Shadmehr and Mussa-Ivaldi used a robot to apply mechanical forces to the hand which revealed powerful error-based learning in the motor system (Shadmehr and Mussa-Ivaldi, 1994). In all investigations, the dynamics produced by the robot had a clear dependency on some movement parameters, such as the speed of the subject's hand. Furthermore, in most experiments, only the end effector or the upper limb is perturbed by the robot. Importantly, in these cases, the sensory system remains unaffected. When a motor error occurs, it is most likely attributed to the effector that sensed the perturbation by an unusual or uncontrollable phenomenon (White and Diedrichsen, 2010).

Parabolic flights, rotating-rooms and underwater settings allow circumventing these limitations as they coherently immerse participants in a new dynamical context. Adaptation of motor responses has been reported following changes in gravity during parabolic flights (Hermsdörfer et al., 1999; Augurelle et al., 2003; Mierau et al., 2008; Crevecoeur et al., 2009a), in gravitoinertial environments (Dizio and Lackner, 1995; Nowak

et al., 2004; Göbel et al., 2006) and underwater (Macaluso et al., 2016). However, exposures were either constant or occurred in a reproducible manner and could eventually be predicted.

In the present experiment, we report a successful motor adaptation of grip force with load force in yet another context. A change of frequency induced larger accelerations and hence load forces. Participants followed the instruction generally well and could move the object at the correct frequency. Previous reports demonstrated the versatility of the motor system to match load forces even when movement pace is higher than 1 Hz (Flanagan et al., 1993; Zatsiorsky et al., 2005) or when load force frequency is multiplied by a factor 2 in weightlessness (Nowak et al., 2000; Augurelle et al., 2003). They were, however, slightly faster for the slowest pace during early exposure as shown previously (Augurelle et al., 2003; White et al., 2008). Further, the nature of the linear regression between load force and grip force changed with frequency as revealed by larger gains and offsets and better correlation coefficients. Offsets reflect the net grip force predicted by the linear model when load force is zero and can therefore be interpreted as a safety margin (Johansson and Westling, 1984, 1988; Cole and Johansson, 1993). Consistent with our results, previous work reported that gains decreased and offsets increased with movement frequency (Zatsiorsky et al., 2005; White, 2015). We found values of correlation coefficients compatible with other experiments (Flanagan et al., 1993). Finally, time-shifts that quantify feed forward processes were not affected by frequency.

Our paradigm allowed breaking down the experiment into different time scales. Our study used three sessions separated by 10-min pauses. Each pace was presented randomly twice per session and each trial was composed of a series of 10–20 cycles of movements. Increased gains, decreased offsets, improved correlation and more positive time-shifts between load and grip forces, all revealed that learning occurred over sessions and repetitions but not over contiguous cycles of movement. Despite the very stressful environment—5 participants (20%) became motion sick and could not complete the experiment—, grip to load force coordination improved over time. Noteworthy, grip forces were unnecessarily large (10–11 N) considering the light object mass. The presence of disease (Hermsdorfer et al., 2003), high complexity (Krishnan and Jaric, 2010), variability (Hadjiosif and Smith, 2015), or fatigue (Emge et al., 2013) usually translate in a deterioration of the above parameters.

Sessions were separated by idle time and repetitions were randomly interleaved. Participants performed context switches between conditions. Blocks, instead, were a succession of cycles within the same dynamical context. Interestingly, we did not observe forgetting between switches, which indicates participant's abilities to adjust their control early in the trial. In contrast, the capacity of the central nervous system to learn different task dynamics in different contexts has been proved to be limited (Gandolfo et al., 1996; Karniel and Mussa-Ivaldi, 2002) even when the change of direction of a perturbation is made fully predictable through the use of an alternating sequence (Conditt et al., 1997) or a predictive visual cue (Osu et al., 2004). Our data show that the central nervous system is capable of switching between different dynamics even when they contain highly unfamiliar components, such as a positiondependent gravitoinertial term. This adds to the list of previously observed experimental contexts in which switching is made possible (Cothros et al., 2006; Nozaki et al., 2006; White and Diedrichsen, 2013). One fundamental difference between our experimental context and those using robotic approaches and rotating chairs is the fact that the participants are completely immersed into a new environment. Indeed, healthy participants tested in robotic studies are endowed with somatosensory signals from the reaching arm while the rest of the body is not affected by the new dynamics. In contrast, some centrifuge investigations placed the subject's head aligned with the vertical axis of rotation, therefore preventing information from the vestibular system to contribute to motor adjustments (DiZio and Lackner, 2001; Nowak et al., 2004). It was indeed shown that deviations of the hand remain uncorrected when the patient's head is fixed in space during trunk rotations. However, adaptation occurred when the head moved with the trunk (Guillaud et al., 2011). It was proposed that vestibular signals may influence all stages of the sensorimotor pathway from a desired movement goal down to specific motor-unit innervation (Bockisch and Haslwanter, 2007). Neuroimaging protocols using small amplitudes of movements (Rousseau et al., 2016a), visual gravitational cues (Indovina et al., 2005) or resting states analyses in astronauts (Demertzi et al., 2016) reported the critical role of a vestibular network that may process gravity-relevant information in action planning and execution. However, different this novel dynamic is, we posit the switching is also made possible because low level multisensory signals are coherently affected which allows adaptation. We speculate the same phenomenon occurs during parabolic flights, when participants are exposed to a series of gravitational environments or underwater, when neutral buoyancy is exerted on body segments as opposed to body center of mass (Macaluso et al., 2016).

While learning a new task in different gravitational fields is surprisingly fast, sometime is necessary for the motor system to adjust subtle parameters underlying the action. One such parameter is the bias induced by gravitational and visual verticality. In reaching hand movements, the arm spends proportionately less of the total time to accelerate upward compared with downward and horizontal movements (Papaxanthis et al., 1998; Gaveau et al., 2016). It is now accepted that in order to save muscular effort, the brain integrates the assistive role of gravity to slow an upward movement and to accelerate a downward movement (Papaxanthis et al., 2003; Rousseau et al., 2016b). This translates into directional kinematic asymmetries. The same bias is responsible for the persistence of larger grip forces when moving an object upward compared to downward in weightlessness (White et al., 2012).

Here, the switching we observed was not incomplete. Whereas participants could produce stereotyped trajectories from the outset (**Figure 5A**), one subtle feature in the grip force profile needed time to settle (**Figure 5B**). Indeed, grip force cycles were asymmetric, exhibiting a minimum later in the movement cycle. In other words, participants produced a movement that was only efficiently mastered at the end of the experiment. This time parameter gradually adjusted across sessions and repetitions, with a forgetting only observable between the last block of Session 1 and the first block of Session 2. This behavior contrasts with the fact that people can learn to predict the consequences of their actions before they can learn to control them (Flanagan et al., 2003). We speculate that it is not the case here because the state of the sensorimotor system itself is altered by the environment. Although coherent, flows of sensory information are new and more time is necessary to accomplish fine adjustments.

To sum up, we have shown that the motor system can switch between different dynamical contexts never experienced before and that this is not a mere consequence of the ability to predict the time course of this new dynamics. Our results further confirm that the brain integrates the effects of the gravitoinertial environment to perform optimal actions and does not consider these effects as disturbances. Furthermore, our findings show that consistent sensory information born in a homogeneous context and from all sensory organs convey signals that can be efficiently processed by the brain to define a control policy and execute an action. We speculate that learning of new challenging motor tasks could be sped up by providing coherent multimodal sensory feedback, which has consequences when designing efficient rehabilitation protocols. Indeed, one recommendation for neurorehabilitation would be to provide to the patient multiple sensory inflows in parallel (e.g., vision, touch and audition) and not only one at a time. A straightforward prediction is that providing irrelevant multimodal sensory information should, instead, negatively impact learning and rehabilitation.

# AUTHOR CONTRIBUTIONS

OW: designed the experiment; MB, CR, and OW: recorded the data using the human centrifuge; MB: analyzed the data; MB and OW, wrote the manuscript; CP, provided feedback on the manuscript.

# ACKNOWLEDGMENTS

We wish to thank the MEDES team for their excellent efforts in helping us with the passing of the ethical agreement, recruitment of participants and for carrying out the experiment on the centrifuge. We are also grateful to the participants who took part in these demanding experiments. The authors declare no conflict of interest. This research was supported by the "Centre National d'Etudes Spatiales" grant 4800000665 (CNES), the "Institut National de la Santé et de la Recherche Médicale" (INSERM), the "Conseil Général de Bourgogne" (France) and the "Fonds Européen de Développement Régional" (FEDER).

# SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fphys. 2017.00290/full#supplementary-material

# REFERENCES


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2017 Barbiero, Rousseau, Papaxanthis and White. This is an openaccess article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Human Biomechanical and Cardiopulmonary Responses to Partial Gravity – A Systematic Review

Charlotte Richter 1, 2, Bjoern Braunstein2, 3, 4, Andrew Winnard<sup>5</sup> , Mona Nasser <sup>6</sup> and Tobias Weber 1, 7 \*

*<sup>1</sup> Space Medicine Office (HRE-AM), European Astronaut Centre Department (HRE-A), Cologne, Germany, <sup>2</sup> Institute of Biomechanics und Orthopaedics, German Sport University, Cologne, Germany, <sup>3</sup> Centre for Health and Integrative Physiology in Space, Cologne, Germany, <sup>4</sup> German Research Centre for Elite Sport, Cologne, Germany, <sup>5</sup> Faculty of Health and Life Sciences, Northumbria University, Newcastle upon Tyne, United Kingdom, <sup>6</sup> Peninsula Dental School, Plymouth University, Plymouth, United Kingdom, <sup>7</sup> KBRwyle, Wyle Laboratories GmbH, Science, Technology and Engineering Group, Cologne, Germany*

#### Edited by:

*Nandu Goswami, Medical University of Graz, Austria*

#### Reviewed by:

*Marcel Egli, Lucerne University of Applied Sciences and Arts, Switzerland Marco Aurelio Vaz, Federal University of Rio Grande do Sul (UFRGS), Brazil Joyce McClendon Evans, University of Kentucky, United States Davide Susta, Dublin City University, Ireland*

> \*Correspondence: *Tobias Weber tobias.weber@wylelabs.de*

#### Specialty section:

*This article was submitted to Integrative Physiology, a section of the journal Frontiers in Physiology*

Received: *24 April 2017* Accepted: *28 July 2017* Published: *15 August 2017*

#### Citation:

*Richter C, Braunstein B, Winnard A, Nasser M and Weber T (2017) Human Biomechanical and Cardiopulmonary Responses to Partial Gravity – A Systematic Review. Front. Physiol. 8:583. doi: 10.3389/fphys.2017.00583* The European Space Agency has recently announced to progress from low Earth orbit missions on the International Space Station to other mission scenarios such as exploration of the Moon or Mars. Therefore, the Moon is considered to be the next likely target for European human space explorations. Compared to microgravity (µg), only very little is known about the physiological effects of exposure to partial gravity (µg < partial gravity <1 g). However, previous research studies and experiences made during the Apollo missions comprise a valuable source of information that should be taken into account when planning human space explorations to reduced gravity environments. This systematic review summarizes the different effects of partial gravity (0.1–0.4 g) on the human musculoskeletal, cardiovascular and respiratory systems using data collected during the Apollo missions as well as outcomes from terrestrial models of reduced gravity with either 1 g or microgravity as a control. The evidence-based findings seek to facilitate decision making concerning the best medical and exercise support to maintain astronauts' health during future missions in partial gravity. The initial search generated 1,323 publication hits. Out of these 1,323 publications, 43 studies were included into the present analysis and relevant data were extracted. None of the 43 included studies investigated long-term effects. Studies investigating the immediate effects of partial gravity exposure reveal that cardiopulmonary parameters such as heart rate, oxygen consumption, metabolic rate, and cost of transport are reduced compared to 1 g, whereas stroke volume seems to increase with decreasing gravity levels. Biomechanical studies reveal that ground reaction forces, mechanical work, stance phase duration, stride frequency, duty factor and preferred walk-to-run transition speed are reduced compared to 1 g. Partial gravity exposure below 0.4 g seems to be insufficient to maintain musculoskeletal and cardiopulmonary properties in the long-term. To compensate for the anticipated lack of mechanical and metabolic stimuli some form of exercise countermeasure appears to be necessary in order to maintain reasonable astronauts' health, and thus ensure both sufficient work performance and mission safety.

Keywords: partial gravity, lunar gravity, martian gravity, biomechanics, energetics, exercise countermeasures

# INTRODUCTION

It is almost 50 years since July 1969, when Apollo 11 Astronauts Neil Armstrong and "Buzz" Aldrin were the first human beings to set foot on the Moon. The Apollo missions can still be regarded as one of the most exceptional endeavors in human history, not only from an engineering and technology perspective but also from a medical and physiological point of view. It was shown that the human body can adapt to extreme environments outside of Earth's protecting atmosphere and its gravitational field with an acceleration of 9.8 ms−<sup>2</sup> (also referred to as 1 g). The Apollo Astronauts were able to live and work in micro- and partial gravity without experiencing any significant medical problems, neither during their (relatively short) missions nor upon their return to Earth (Berry, 1974).

In 2016 the Director General of the European Space Agency (ESA) introduced the agency's plans for the era after the planned decommissioning of the International Space Station (ISS) in 2024. The plans included going back to the Moon to set up a permanent habitat on its surface and/or a Cis-Lunar space station orbiting the Moon (Foing, 2016). It is thought that a progressively staggered approach using the proposed Lunar base will allow safer development and testing of hardware and procedures, toward the ultimate goal of a human space mission to Mars (Horneck et al., 2003; Goswami et al., 2012).

Astronauts exposed to microgravity (µg) experience physiological deconditioning (referred to as "space deconditioning"), in particular with regards to the physiological systems sensitive to mechanical loading such as the cardiovascular, pulmonary, neurovestibular, and musculoskeletal systems (Baker et al., 2008). In order to attenuate these effects, current ISS Crew members exercise every day for 2.5 h including preparation time. Current exercise devices used on the ISS are a cycle ergometer and a treadmill for cardiovascular exercise (∼1 h) as well as an advanced resistive exercise device (ARED) for strength training (∼1.5 h) (Loehr et al., 2015; Petersen et al., 2016). Despite the extensive use of exercise countermeasures, astronauts still return from 6 months ISS missions showing space deconditioning effects. Examples of these effects include decreased calf muscle volume and power, loss of bone mineral density and reduction of peak oxygen uptake (Trappe et al., 2009; Moore et al., 2014; Sibonga et al., 2015).

It is understandable that in the past, medical divisions of space agencies have mainly set their foci of interest on the physiological effects of µg, to optimize operational procedures, to better understand the effects of µg on the human body and to mitigate undesirable and harmful effects. Consequently, compared to the bulk of literature and knowledge generated on the physiological effects of µg, the consequences of immediate and chronic partial gravity exposure (µg < partial gravity <1 g) as present on the Moon (0.16 g) or Mars (0.38 g), are somewhat understudied (Horneck et al., 2003; Goswami et al., 2012; Widjaja et al., 2015).

Nonetheless, despite the fact that the knowledge gained through real partial gravity exposure during the Apollo missions and through partial gravity analogs is sparse, a first step to direct future research and to help to better understand physiological effects of partial gravity should be to gather and synthesize all available information of experiences made in the past. Logically, valuable sources of information are the medical data, records and publications of the Apollo missions conducted in the 1960s and 1970s with up to 75 h of continuous partial gravity exposure (Johnston and Hull, 1975; Kopanev and Yuganov, 1975) as well as various terrestrial partial gravity simulations (**Figure 1**; Shavelson, 1968; Davis and Cavanagh, 1993; Sylos-Labini et al., 2014; Salisbury et al., 2015).

The aim of this work was therefore to review all available information in order to quantify cardiopulmonary and biomechanical changes expected to occur in partial gravity environments (0.1–0.4 g). The objectives of the study were to:


Using the highest standard available to perform systematic reviews (www.cochrane.org) the synthesized information presented here shall help to identify knowledge gaps and develop a better understanding of medical issues that future astronauts will face when returning to the Moon and eventually advancing to Mars. Moreover, this systematic review seeks to provide a working reference for experts designing evidence-based exercise countermeasures for a Lunar habitat and future long-duration exploration missions beyond the Moon.

# MATERIALS AND METHODS

The present systematic review was conducted following the guidelines of the Cochrane Collaboration (Higgins and Green, 2011).

Additionally, the PRISMA (preferred reporting items for systematic reviews and meta-analyses) checklist was used to ensure transparent and complete reporting (Liberati et al., 2009).

# Search Strategy

A range of keywords, grouped by main search terms, was used in various combinations to search the following databases for English language articles: Pubmed, Web of Science, Cochrane Collaboration Library, Institute of Electrical and Electronics Engineers database as well as ESA's "Erasmus Experiment Archive," the National Aeronautics and Space Administration's

**Abbreviations:** µg, Microgravity; ARED, Advanced Resistive Exercise Device; bpm, Beats per Minute; BW, Body Weight; BWS, Body Weight Support (System); RCT, Randomized Controlled Trial; CoM, Center of Mass; DLR, German Aerospace Centre; EMG, Electromyography; ESA, European Space Agency; EVA, Extravehicular Activity; g, Gravitational Acceleration; GRF, Ground Reaction Force; <sup>h</sup> , Hopping; HUT, Head-Up Tilt; ISS, International Space Station; LBPP, Lower Body Positive Pressure; MeSH, Medical Subject Headings; m/s, Meter per Second; NASA, National Aeronautics and Space Administration; pg, Partial Gravity; PICOS, Eligibility Criteria (Population, Intervention, Control, Outcome, Study Type; PRISMA, Preferred Reporting Items for Systematic Reviews and Meta-Analyses; PTS, Preferred Walk-to-Run Transition Speed; <sup>r</sup> , Running; <sup>s</sup> , Skipping; <sup>w</sup>, Walking.

(according to ESA 1st Joint European Partial Gravity Parabolic Flight Campaign, 2011).

(NASA) "Life Science Data Archive" and "Technical Reports Server" and the German Aerospace Centre's (DLR) database "elib".

The literature search was performed in March and April 2016 according to the search strategy shown in **Table 1**. No restrictions to publication dates were applied. For ESA's, NASA's, and DLR's internal data archives, the search strategy was altered and specifically tailored due the inability to use "Boolean logic" in these databases. For the latter archives, only keywords of the search term "partial gravity" and/or one of the other synonyms (as listed in **Table 1**, search number 1) were used and all relevant records concerning biomechanics and/or the cardiopulmonary system were downloaded.

# Criteria for Considering Studies for this Systematic Review

The following eligibility criteria, which specify the types of included populations, interventions, control conditions, outcomes and study designs (PICOS) were applied.

# Population

The main target group for the present systematic review were astronauts. However, since most of the included studies were simulation studies, healthy terrestrial people with no gender restrictions were included as well.

### Interventions

Apollo missions 11–17 with Lunar surface time and various terrestrial partial gravity simulation models (**Figure 1**) were included (see list below). Variations in terms used for the different methods were at this point disregarded.


TABLE 1 | Search strategy.


*Keywords were combined using the Boolean operators and grouped by main search terms. Medical Subject Headings (MeSH) as a comprehensive controlled vocabulary for the purpose of indexing journal articles and books in the life sciences were included in the search strategy. In the Pubmed advanced search builder either 'Title/Abstract' or 'All Fields' was used. The combined search allows to screen databases for various combinations of main search terms and their keywords.*

Only gravity levels from 0.1 up to 0.4 g were reviewed. Due to the varying gravity levels investigated in the reviewed studies, out of this range three different "gravity-groups" were determined with gravity conditions expressed either as the physical gravitational constant "g" or as percent of body weight (BW), applied body weight support (BWS) or degree of head-up tilt angles (HUT):


# Control Conditions

Terrestrial gravity (1 g) and microgravity (µg) were used as control conditions.

### Outcomes

To be included, studies had to contain outcomes linked to energetics and/ or biomechanics. A full list of outcome parameters is presented in **Table 2**.

### Study Designs

All types of experimental studies were included.

# Data Collection and Analysis Study Selection

Studies were screened by the lead author and one other independent reviewer using the Rayyan web application (https:// rayyan.qcri.org/) (Elmagarmid et al., 2014). The initial screening was performed using titles and abstracts. Considering the main research question of the present study (which human biomechanical and cardiopulmonary changes occur due to partial gravity exposure?) relevant articles were included. Articles were excluded if titles and/or abstracts were considered as clearly irrelevant. This was the case if titles and abstract did not reveal a direct link to the previously defined PICOS. Any uncertainties of study inclusion or exclusion were discussed consulting a third expert reviewer. Full-text articles were obtained in case the initial screening was unclear and were downloaded for all other included studies. After screening the full-text resources a further round of exclusion took place. The complete systematic literature screening and exclusion process is illustrated in **Figure 2**.

### Data Extraction and Management

Data extraction from each study was performed using an adapted version of the Cochrane Collaboration's 'Data collection form for intervention reviews: RCTs and non-RCTs', version 3, April 2014 (RCT: randomized controlled trial).

#### TABLE 2 | Outcome parameters for studies to be included.


*Outcomes are divided into the two main groups "energetics" and "biomechanics" and further into subgroups with more specific outcome measurements.*

## Assessment of Risk of Bias in Included Studies

The Cochrane Collaboration's risk of bias analysis tool was used to assess the quality of included studies. Uncertainties were discussed with a third reviewer. As the study types of included studies were mainly case series without a separated control group [for study classification see also: "2009 Updated Method Guidelines for Systematic Reviews in the Cochrane Back Review Group" by Furlan et al. (2009)] and with small sample sizes, often not much information to allow for objective judgment was provided. A "+" stands for low risk, "−" for high risk and "?" for unclear risk. For the studies that are case series, no risks of random sequence generation and allocation concealment can be assessed (NA–not applicable, see **Table 3**).

## Quality Appraisal of Technical Principles to Simulate Partial Gravity

There is limited high quality research on changes in energetics and biomechanics in humans due to exposure to partial gravity. Main problems are logistical limitations, limited numbers of participants and a diversity of simulation models. Therefore, no tools for assessing partial gravity methodological quality are available except for the approach of Chappell and Klaus (2013) who characterized models allowing locomotion of being good or poor in reproducing factors associated with partial gravity (Chappell and Klaus, 2013). Since there was a lack of completeness in Chappell and Klaus (2013) it was decided for this review to develop a new rating scale of included technical principles with partial gravity parabolic flights set as a gold standard (see **Table 4**). The underlying assumption of this tool is how well the simulation study reflects the reality. This can provide an indicative rating how well the simulation study results are transferable to real human partial gravity missions. This tool is piloted in the present review to highlight which studies may have a greater rigor in simulating partial gravity but it is important to consider that no further empirical studies on its validity and reliability were performed.

# Data Analysis

Main changes across all outcome measures are presented in six different tables (**Supplementary Tables**). There are three tables for cardiopulmonary changes and three tables for biomechanical changes presenting outcomes of the three defined gravity ranges (Lunar, in between and Martian -gravity). Changes from either terrestrial gravity and/or microgravity as control conditions are presented with arrows. An up (↑) or down (↓) arrow was set as soon as minimal changes of the mean were presented either as values or as figures (visual observation) or if the authors stated that values were in- or decreasing (even if not statistically significant or if no statistics were performed), meaning that there is an upward or downward trend. Arrows marked with an asterisk (∗) indicate that there were statistically significant differences from control with P < 0.05. Arrows marked with a hash tag (#) indicate that only partly statistical significant differences were found, for example if in general values increased but only for men significantly. A horizontal arrow (→) was used if no visual differences were detected, values were the same or if the authors stated that there were no changes. A swung dash (∼) was used in case of inconsistent results for example if two participants showed results in opposite direction.

Effect sizes (for data available either presented in included articles or obtained from authors after requested) were calculated between partial gravity conditions and 1 g. The effect sizes were then bias corrected using weighted (accounting for n = sample size) pooled standard deviations as per Hedge's g method (Ellis, 2010). Effect sizes in **Figures 5**–**9** are presented as Hedge's g:

$$\text{Hodge's g} = \frac{\text{sample mean 2 -- sample mean 1}}{\text{pooled standard deviation of sample 1 and 2(weight)}}$$

In the absence of previously reported and validated minimal clinically meaningful changes on which to base conclusions, standardized mean changes between comparisons groups were defined. As there are currently no direct empirical studies for astronauts to demonstrate the thresholds suggested by Hopkins et al. (2009) were used. Thresholds of 0.1, 0.3, 0.5, 0.7, and 0.9 were defined as small, moderate, large, very large and extremely large effects between two comparison groups (Hopkins et al., 2009). This enabled conclusions to be based upon the estimated size of the effect between g-levels. The level for the confidence interval for the effect size comparisons was set to 95%. The most meaningful effect sizes are presented in plots to highlight the areas where medical operations will need to focus attention ahead of the missions taking place.

# RESULTS

# Description of Studies

The study selection process and reasons for exclusion are summarized in **Figure 2**. The initial search identified 1,323 citations of which 244 were confirmed to be duplicates. Therefore, 1,079 titles and abstracts were screened and further 969 studies excluded which did not meet the eligibility criteria. After reading the remaining 110 full-text articles, further 54 studies were excluded for various reasons (see description on the right side of the flow chart in **Figure 2**). Initially, 56 studies met the inclusion criteria but 13 of them were excluded after being defined as not suitable considering the protocol, methodology, control condition, or time points of data acquisition.

The final 43 included studies were mainly case series studies except for the case report of Waligora and Horrigan (1975) and the study of Baranov et al. (2016) who conducted a randomized controlled trial. Apart from the two latter publications, all other included studies investigated different levels of partial gravity without a separated control group. Depending on the technical principles used to simulate partial gravity and a different terminology, the authors expressed partial gravity either as percent of body weight, percent of body weight support, degree of head-up tilt or as a specific gravity level (g). For a uniform designation, **Figure 3** helps to translate different units into the gravitational acceleration "g", as it will be the standard unit used within this review.

**Figure 4** summarizes the applied gravity levels within the in the PICOS defined gravity range (0.1–0.4 g) as well as the simulation model used of each included study. As shown in **Figure 4**, the majority of studies (n = 29) were conducted in the range of Lunar gravity (0.1– 0.2 g). Out of these 29 studies, 18 applied actual Lunar gravity of 0.16 g. Seventeen studies were conducted in the range of Martian gravity and nine applied the actual value for Martian gravity. In the range between Lunar and Martian gravity 10 studies applied 0.25 g. Nine studies used µg as a comparison whereas 41 studies compared their outcomes to 1 g


*Authors' judgement about each methodological quality item of each included study.*

+*: Low risk;* −*: High risk; ?: Unclear risk; NA: Not applicable.*


*results in the number of total points per method and therefore defines the overall ranking. Partial gravity parabolic flight was set as a gold standard.*

*all x's* 

 *were* 

 *as per* 

 *accurate* 

 *x:* 

 *not accurate; xx:* 

 *accurate* 

 *xxx: very accurate.* 

 *sum* 

(under consideration that studies could use µg and/or 1 g as control conditions).

Nineteen studies used vertical body weight support systems ( ), indicating that this was the most often used simulation model to generate partial gravity conditions. Additionally, in descending order, nine studies used head-up tilt ( ◦), six studies tilted body weight support systems ( •), five studies partial gravity parabolic flights ( ), four studies lower body positive pressure ( ) and two studies centrifugation ( ♦). Physiological data of Lunar surface explorations during Apollo missions ( ) were presented in five studies.

The age of participants across the studies ranged from 18–63 years. Studies recruited predominantly men. Taken together 197 men and 88 women participated in total. For 19 adults gender was not indicated. The highest number of participants within one study was 21, the lowest number was two.

Some of the investigated partial gravity simulation models did not allow movements. When movements were not possible (e.g., head-up tilt) or required, posture within the experimental protocols was semi-supine (a body position where the participant lies on his/her back but is not completely horizontally) sitting or standing. Included locomotion types were walking ( <sup>w</sup>), running ( r ), skipping ( s ), and hopping ( h ) at different velocities or the preferred walk-to-run transition speed (PTS).

# Methodological Quality of Included Studies

The overall risk of bias (see **Table 3**) was very low. Most studies were case series and did not have control groups, therefore, some aspects of the Cochrane risk of bias tool (which was designed for controlled clinical trials) were not relevant. This includes randomization and allocation concealment. In addition to this, many studies failed to give sufficient detail to assess their potential risk of bias including blinding of participants, personnel and outcome assessment. Therefore, only conclusions about incomplete outcome data and selective reporting could be drawn.

The majority of included studies were case series. Hence, the evidence/taxonomy of study designs of included studies is very low (IV, where V is the lowest level of evidence) using "Oxford Centre for Evidence-based Medicine" (March, 2009) guidelines (Phillips et al., 1998).

The number of participants in the included studies was comparably low and therefore often no adequate statistical analysis in consistency with the research question was performed. This reduces the quality of most of the included studies with respect to the authors' research question.

# Results of Changes in Outcome Parameters

All characteristics of included studies and changes from 1 g and/or µ g of relevant outcome parameters due to exposure to partial gravity are presented in **Supplementary Tables 1** – **6** .

In the following the clinical relevance of available data is presented. Main effects of outcome parameters and their bias corrected effect sizes (Hedge's g) are depicted in **Figures 5** – **9**. Please note that since the scale we used to define effect sizes (as per Hopkins et al. (2009): small, moderate, large, very large and extremely large) we mainly found extremely large effect sizes, referring to effect sizes larger than 0.9. This means that differences of effect sizes within the category "extremely large" can be really great (as presented in **Figures 5** – **9**).

TABLE

4


appraisal

of

included

technical

principles

to

simulate

the

effects

of

partial

gravity

on

the

various

physiological

and

biomechanical

outcome

measures.

are shown in gray. The exact values for Lunar and Martian gravity and each unit are depicted through solid diamonds and circles. BW, Body weight (in %); BWS, Body

# Main Effects and Effect Sizes of Cardiopulmonary Changes in Partial Gravity

weight support (in %); HUT, Head-up tilt (in degree).

In the following, if effects were similar in direction and magnitude, then these effects were generalized and body postures and simulation models were not further considered.

Heart rate, stroke volume, cost of transport, efficiency (except of the hopping condition in Pavei and Minetti, 2015) as measured in Lunar and Martian gravity conditions revealed the most pronounced changes compared to 1 g (**Figures 5**, **6**).

For cardiac output as measured using the lower body positive pressure model, Kostas et al. (2014) presented large effects in Lunar and Martian gravity. The effects for cardiac output as measured during head-up tilt revealed moderate changes in Lunar and small changes in Martian gravity compared to 1 g.

For blood pressure parameters, inconsistent results between the different included studies were found. Widjaja et al. (2015) presented extremely large effects for diastolic and systolic blood pressure values measured during a Lunar and Martian gravity parabolic flight. The study of Kostas et al. (2014) presented mostly small effects for different blood pressure parameters using two different simulation models. Using the lower body positive pressure model, effects for total peripheral resistance were reported to be moderate (Lunar vs. 1 g) and large (Martian vs. 1 g). For thoracic impedance, data published by Kostas et al. (2014) revealed moderate changes in Lunar and Martian gravities compared to 1 g.

For 0.25 g only effect sizes of net metabolic rates are presented (**Figure 8**). Teunissen et al. (2007) found extremely large effects during running while data from Grabowski et al. (2005)reported a small effect during walking in 0.25 g compared to 1 g.

# Main Effects and Effect Sizes of Biomechanical Changes in Partial Gravity

In all of the three defined gravity ranges (**Figures 7**–**9**) vertical and forward work as well as total internal, external and mechanical work are the most reduced parameters compared to 1 g indicating extremely large effects (>0.9).

For the biomechanical parameter recovery (ability of the human body to safe energy by behaving like a pendulum-like system), especially in Martian gravity different effects and direction of changes were found ranging from small to extremely large changes depending on locomotion modes and velocities (**Figure 9**).

For joint kinematics only effect sizes for Lunar gravity compared to 1 g are presented and indicate reductions with extremely large effects for hip and knee range of motion using the tilted and vertical body weight support systems. For ankle range of motion, effect sizes cover the whole range from small to extremely large (**Figure 7**).

Most of the spatio temporal parameters showed extremely large effects in all defined gravity ranges (**Figures 7**–**9**). Examples of these effects include increased swing phase and cycle duration in Lunar gravity, increased Froude number and decreased preferred walk-torun transition speeds in Lunar and Martian gravity. One exception for the overall extremely large reduced stride frequency was found in the study of Pavei et al. (2015) for walking in Lunar and Martian gravity (small effect) using the vertical body weight support system. In partial gravity, stride length is mostly reduced during walking (Donelan and Kram, 1997) and increased during running (Donelan and Kram, 2000; Cutuk et al., 2006) with extremely large effect sizes. Stride length data from Ivanenko et al. (2002) indicate moderate to extremely large effects in 0.25 g depending on the walking speed

(**Figure 8**). Duty Factor is reduced in partial gravity compared to 1 g but beside extremely large effects, also moderate effects are presented by Donelan and Kram (1997) for walking at fixed Froude numbers in 0.25 g.

Ground reaction forces (GRF) (except relative values) and impulses are reduced in partial gravity compared to 1 g and involve extremely large effects. Contrary, the time to impact peak force is increased in 0.25 g and Martian gravity but again with an extremely large effect.

# DISCUSSION

# Summary of Main Results

The main findings of this study were the heterogeneity of results across studies, the extremely large effect sizes within a wide range of effect sizes, the low quality of applied methodologies as well as the discovery of a significant lack of knowledge concerning long-term adaptations in partial gravity. The longest continuous exposure to partial gravity reported in one of the included studies was a period of 2 weeks, with 9.6◦ head-up tilt during daytime and 0◦ supine position during the nights (Baranov et al., 2016). The reasons for the heterogeneous findings across studies can be explained as follows: (1) The included studies reported a wide range of ages; (2) Studies were performed with both male and female participants. For example Evans et al. (2013) found different significant results in diastolic blood pressure for males and females; (3) The presentation of data varies from study to study. While some authors reported absolute values, some others reported relative values using different normalization reference values; (4) Gravity levels were inconsistent between studies because not all studies used the exact gravity levels of 0.16 g for the Moon and 0.38 for Mars; (5) Durations of partial gravity exposure varied depending on the used simulation model [e.g., 25–30 s of partial gravity exposure during parabolic flights (Aerts et al., 2012) vs. 6 h head-up tilt (Lathers et al., 1990, 1993; Lathers and Charles, 1994)]; (6) Different velocities and locomotion types or postures (e.g., walking vs. running or standing vs. sitting) were used in the different protocols. Donelan and Kram (1997, 2000) found significant different results for relative stride length at same speed depending on walking or running

protocols; (7) Varying experimental conditions were reported between studies. While most of the included studies tested their participants in sport clothes, others performed their experiments with space suits, very likely leading to restrictions/alterations of movements (Spady and Krasnow, 1966; Robertson and Wortz, 1968; Spady and Harris, 1968); (8) Diverse experimental set ups were used (e.g., rolling vs. a fixed trolley for the vertical body weight support system); (9) Various partial gravity simulation models were included but all of them have certain limitations. For instance not all simulation models are suitable to expose the whole body to a partial gravity environment or to simulate realistic hemodynamic changes (e.g., vertical body weight support systems) and therefore the impact on the cardiovascular system may vary from model to model (see Section Validity of included partial gravity simulators).

In all included studies the cardiopulmonary parameters heart rate, oxygen consumption, respiratory rate, expired minute volume, (net) metabolic rate, locomotion efficiency, cost of transport and bioelectrical thoracic impedance revealed either decreasing trends or significant reductions with decreasing gravity levels. On the other hand, stroke volume seems to increase with decreasing gravity levels. For blood pressure parameters, no consistent results were found. Some studies reported increasing values (Cutuk et al., 2006; Evans et al., 2013; Kostas et al., 2014) and some other studies reported decreasing values (Cardus, 1996; Chang et al., 1996; Aerts et al., 2012; Kostas et al., 2014; Widjaja et al., 2015) or unchanged values (Schlabs et al., 2013) in response to changing gravity levels from 1 g. However, effect sizes for most of the cardiac as well as metabolic outcomes and for two exceptions concerning blood pressure parameters were extremely large.

Data obtained during the Apollo missions 11–17 reveal that during actual Lunar surface explorations mean heart rates were 90–100 beats per minute (bpm) with maximum values of 160 bpm (Kopanev and Yuganov, 1975). Metabolic rates had a total mean of 234 kilocalories per hour within a total time of 159 h of Lunar Extravehicular Activities (EVA; Waligora and Horrigan, 1975). Importantly, data from the Apollo missions have to be interpreted with caution as Apollo astronauts were restricted in their movements through their space suits, making it impossible to compare Apollo data to most of the data obtained in lab conditions.

The biomechanical data of the included studies duty factor, vertical impact loading rate, active force peaks, peak vertical and horizontal impulses, horizontal and vertical work as well as the resultant total external, internal and mechanical work per unit distance decreased significantly with decreasing gravity levels. The preferred walk-torun transition speed revealed a decreasing trend while recovery of mechanical energy during walking, range of motion for hip and knee angles, stance phase duration, ground contact time, stride frequency and (net/normalized) vertical peak GRF mostly showed decreasing trends and partly significant reductions with decreasing gravity levels. All included studies presented increasing trends for the Froude number, vertical spring stiffness and with one exception (during walking; Sylos Labini et al., 2011) also for swing phase duration and stride length. Further, a significant increase for time to impact force peak (vertical GRF) with decreasing gravity levels

was shown. For Electromyography (EMG) amplitude and activation patterns inconsistent results were reported, with studies reporting changes in all directions depending on locomotion velocity (Ivanenko et al., 2002) or no (abrupt) changes at all (Sylos Labini et al., 2011).

The largest effect sizes were associated with parameters influencing the center of mass oscillation such as internal, external and mechanical work and for GRF and impulses where extremely large effects were presented. These outcomes together with cardiac and metabolic parameters are therefore the main areas that operational guidelines and decision making need to consider. Future research should attempt to address these same issues ahead of upcoming exploration missions to minimize risks to the astronauts.

# Quality of Evidence

### Validity of Included Partial Gravity Simulators

Terrestrial partial gravity simulation models seek to simulate reduced gravity and its impact on human physiology as close as possible to the actual Lunar or Martian environment. A main problem is the lack of "real" partial gravity data and therefore, it is almost impossible to validate current partial gravity simulation models. Nevertheless, a quality appraisal of included technical principles to create partial gravity conditions was performed by the authors (see **Table 4**). Partial gravity as created through parabolic flights was set as a gold standard with the highest possible rating. The reason for this is that parabolic flights create partial gravity that affects all physiological systems similar to "real" partial gravity on the surface of the Moon or Mars. Obviously, considering the very short exposure times during parabolic flights, the model validity must only refer to immediate physiological adaptations. Slow reacting systems cannot be studied using parabolic flights and require different models. Thus, considering the aim of the present quality appraisal all methods were rated as per how accurate they can mimic the effects of partial gravity for relevant physiological and biomechanical categories. The ratings are based on the advantages and limitations of the included simulation models and were performed in agreement with physiological and biomechanical experts from ESA's Space Medicine Office and from the German Sport University.

As shown in **Table 4**, all models are suitable to manipulate GRF very accurately whereas cardiovascular responses are dependent on the posture of the body (e.g., degree of body tilt) or systems that promote fluid shifts (e.g., lower body positive pressure). Kinematics were only rated as quite accurate because movements are influenced by the set-up of the included simulation systems which may limit natural friction-free movements (e.g., using rubber cords or exoskeletons). As GRF can be mimicked very precisely

and kinematics are quite accurate, biomechanics can be investigated in all included simulation models quite accurately. This includes muscle activation patterns with the one exception that compared to the suspension systems, lower body positive pressure treadmills are probably closer to "real" partial gravity. This is because of the free moving limbs when walking/running on a lower body positive pressure treadmill and thus both the stance and the swinging legs are exposed to partial gravity at all times. Respiratory and metabolic properties were rated as quite accurate but not perfect due to the movement constraining nature of all partial gravity simulators.

Different simulation models affect different physiological systems in different ways. This may explain, why for cardiopulmonary outcomes within this review, mainly partial gravity parabolic flights, head-up tilt or lower body positive pressure models were used. In agreement with the rating performed by Chappell and Klaus (2013) the vertical body weight support system was only used to investigate metabolic or respiratory changes but not for cardiovascular properties.

Biomechanical outcomes within this review were mainly investigated using a vertical body weight support system as it is a valid method to reduce GRF while almost preserving natural movements. Results from different models may therefore vary and comparisons between models should be made with caution.

### Quality of Statistical Analyses of Included Studies

Statistical analyses of included studies were in many cases deficient or not performed at all. The reason for this is probably the often very limited number of participants without normal distributed data. Therefore, the sample sizes probably failed to provide adequate power to draw conclusions about all outcome parameters using traditional significance testing. Furthermore, almost all studies had no separate control group and several studies did not involve both genders equally. Unfortunately, in most of the included studies means and standard deviations for the experimental as well as for control conditions were not presented and had to be requested. Hence, the authors of this article were limited by the data available and in some cases only visual inspection of figures was possible. Additionally, statistics sometimes failed to address the research questions of this study and therefore

some of the presented p-values could not be used. In other cases no post-hoc tests were performed indicating the direction of changes or the significance level alpha was set to a rather liberal alpha = 0.1 (Lathers et al., 1990, 1993; Lathers and Charles, 1994).

# Overall Completeness and Applicability of Evidence

Not all of the outcomes defined in the PICOS have been investigated in the included studies. Some parameters such as the arteriovenous oxygen difference are missing and diverse respiratory parameters (except of oxygen consumption) are very sparse being only investigated in one study (Robertson and Wortz, 1968). The same can be said for venous hemodynamics. Morphological parameters such as fiber type composition, muscle fiber length, physiological and anatomical cross sectional areas, muscle pennation angles, tendon function and material properties as well as bone mineral density are completely missing but are important indicators for physiological deconditioning and very relevant for space flight operations. Obviously, changes of these parameters can only be investigated during long-term exposure to partial gravity, and as already mentioned there is a lack of long-term partial gravity studies. Furthermore, muscle force, angular velocities and joint torques have not been investigated but are important measures for the mechanical strain in the musculoskeletal system. Of all included technical principles to simulate partial gravity only supine suspension systems are missing. Despite a lack of important outcome parameters, the 43 included studies were overall sufficient to address the objectives of this review, even if in some cases it was necessary to "read between the lines" and to filter relevant results for this systematic review.

# Potential Bias in the Review Process

The strict methodology of this review with clearly defined inclusion criteria as well as a comprehensive search strategy minimized the potential for bias.

The literature research was hindered by the design of some databases. Databases such as the Erasmus Experiment Archive of ESA do not offer "advanced search methods" and had therefore to be searched manually. Obviously, this increases the risk of failing to include relevant studies. Furthermore, misleading or wrong terminology may have led to the undesired exclusion of relevant studies. For example "reduced gravity" was often used as a synonym for µg and not for partial gravity.

In some cases the authors had to obtain data from figures instead of numeric tables (e.g., from conference presentation slides of Cowley et al., 2015). Possibly, this could have introduced a potential bias, as the measurements on figures and detection of (visual) changes are not 100% accurate and could vary from person to person. Therefore, smallest differences from control were defined as changes, even if statistically no significant results were reported. In the result **Supplementary Tables 1**–**6**, arrows without an asterisk indicate this fact and should be interpreted with caution.

# Agreements and Disagreements with Other Studies

Due to the comprehensive literature research there are almost no experimental studies left to which the present results can be compared. Therefore, also computational models/simulations were included for this comparison to see if experimental data show similar changes in magnitude and direction as modeled data.

## Comparison with Modeled Data

The kinematic model of Raichlen (2008) which predicts the effects of gravity on human locomotion matches the data as presented in this review. The author calculated, that relative stride lengths at 2 meters per second (m/s) as well as the Froude number at walk-to-run transition speed increases with decreasing gravity (Raichlen, 2008). For Froude number, the same increasing tendency was estimated from audio transcripts and video clips of Lunar EVA's as well as by the astronauts and space suit characteristics by Carr and Mcgee (2009). The latter study also found out, that wearing a spacesuit appears to lower the Froude number and the walk-to-run transition will occur at lower velocities. Therefore, they suggest the introduction of an "Apollo number" (Froude number divided by mass) to capture the effects of spacesuit self-support (Carr and Mcgee, 2009). Ackermann and van Den Bogert (2012) predicted values for different locomotion types in partial gravity through a computational simulation using a realistic musculoskeletal model. They calculated reduced vertical GRF for each gait type compared to terrestrial gravity which is in agreement with the present findings. They also predicted skipping as the preferred gait mode in Lunar gravity because their results suggest that skipping is more efficient and less fatiguing compared to walking or running (Ackermann and van Den Bogert, 2012).

Keller and Strauss (1992) predicted bone mineral density changes in partial gravity using modeled data. They predicted a weekly loss of 0.39% for bone mineral density in a Lunar- and a loss of 0.22% in a Martian gravity environment. Unfortunately, no included study of the present review investigated changes in bone mineral density. Nevertheless, it should be pointed out that this model predicted that bone mineral density loss will not be prevented in a partial gravity environment. Provided the mathematical modeling is accurate, it appears that planetary stay times could be extended from ∼100 weeks on Moon to 3 years on Mars (based on the assumption that until then a reduction of 66% in strength is attained) before a weakened skeleton could create serious hazards during the stresses of re-entry and returning to terrestrial gravity (Keller and Strauss, 1992). Obviously, the latter assumption represents a pure scientific point of view and is certainly not in line with medical operations guidelines.

## Comparison with Other Review Articles

There are only very few reviews about partial gravity research but none of them used such a systematic and comprehensive search strategy as this study. Davis and Cavanagh (1993) focused on biomedical issues related to human locomotion in partial gravity. They summarized (using data from up to 3 included experiments and calculations from a ballistic walking model) that swing phase as well as stance phase duration is increasing with decreasing gravity whereas cadence and walking velocity is decreasing. The latter also affected peak vertical GRF which were decreased in reduced gravity (Davis and Cavanagh, 1993). This is in agreement with the findings of the present review with the exception of stance phase duration which was reported to decrease in all of the included studies (Ivanenko et al., 2002, 2011; Sylos Labini et al., 2011; Cowley et al., 2015).

Shavelson (1968) summarized findings of different studies on metabolic rate and concluded that in four out of five studies metabolic rate is decreasing with decreasing gravity unless high mechanical work and external forces are required (Shavelson, 1968). Their findings are in agreement with the present results.

The review article of Sylos-Labini et al. (2014) included mainly studies which have been investigated in this systematic review but without a comparable systematic approach and not fully in agreement with the present outcome parameters as defined in the PICOS. The present outcomes cover a wider spectrum of parameters which are considered as operationally relevant by ESA's Space Medicine Office, in particular for future planetary explorations. Finally, the review article of Sylos-Labini et al. (2014) did not cover the whole range of available literature about biomechanics in partial gravity.

### Studies That "Slipped through"

Despite the fact that we have applied a comprehensive strategy, there were a few relevant studies that escaped our search. This could have been for the following reasons: Studies were published after the period of this literature search or studies did not cover included keywords or were not listed in included databases. These studies were either found through random online searches or were cross-referenced in one of the included studies. Data of these studies were not extracted for the present review however the findings of these studies and the findings of this systematic review are compared in the following. The study of Ruckstuhl et al. (2009) that compared gait parameters and heart rate as measured using lower body positive pressure or vertical body weight support (33% BW) during different walking speeds (0.5–1.2 m/s) was not found during the present research process. Nevertheless, it fits with the defined PICOS and results are in agreement with findings of this systematic review. In their study, heart rate, stride frequency and duty factor decreased significantly with decreasing gravity levels. For Martian gravity, normalized stride length was not found in the present results but Ruckstuhl et al. (2009) presented a significant reduction in their results. For leg angle at touch down they showed a significant increase compared to terrestrial gravity. Further, Ruckstuhl et al. (2009) compared lower body positive pressure and vertical body weight support systems and found no significant differences for the gait parameters but did for heart rate (Ruckstuhl et al., 2009). This is in agreement with the conclusions of the present quality appraisal of included technical principles to simulate partial gravity (see **Table 4**).

One of the most recent studies about musculoskeletal changes due to partial gravity exposure is the study of Ritzmann et al. (2016) which is not included in this systematic review because it was not published during the time of the present literature search. The authors measured biomechanical parameters of a bouncing movement (often referred to as skipping) during a partial gravity parabolic flight (Mars and Moon parabolas). Their results show a reduction of peak vertical GRF, rate of force development and vertical impulse with decreasing gravity (Ritzmann et al., 2016). This is in agreement with the results presented in this systematic review whereas joint angles and EMG can hardly be compared to the present results because bouncing movements and normal walking or running are quite different. The main conclusion of the study by Ritzmann et al. (2016) was that subjects are able to keep their motor control patterns. They suggest that muscle activity in changed gravity environments can be anticipated (shown in a decline in activation amplitudes before touchdown) and resulting muscle forces can be properly adjusted.

# Relationship and Interplay between Biomechanical and Cardiopulmonary Outcome Parameters

Exposure to partial gravity reduces body weight and therefore external forces acting on the human body (**Figure 10**). This can be seen in the reduced vertical GRF with a reduced first impact and second active force peak as well as a reduced rate of force development. As the area under the force-time curve becomes smaller, also impulses are reduced. Additionally, the time of exposure to impact forces becomes less as stance phase duration, ground contact times and duty factor decrease. As a consequence, it is likely that the reduced mechanical stimuli (supported by extremely large effects) associated with walking and running in Moon and Mars gravity conditions will not be sufficient to fully maintain terrestrially optimal bone mineral density and muscle mass in the long-term. Further, due to partial gravity-induced mechanical unloading, the mechanical work that is necessary to move the body becomes less. In the present data this becomes apparent in the reduction of horizontal and vertical work per distance, resulting in a reduced total external work. Together with a reduced total internal work necessary to rotate and accelerate limbs, the total mechanical work is decreased in partial gravity environments as can be seen in the extremely large effects. Most likely this explains the reduced load on the cardiopulmonary system in reduced gravity. For instance, heart rate and oxygen consumption correlate with work performance and therefore it does not surprise that these parameters are decreased in partial gravity. Rates of oxygen consumption and carbon dioxide production are measured to estimate (net) metabolic rates and as oxygen consumption decreases it seems to be logical that metabolic rate also decreases with decreasing gravity levels. If the mass specific metabolic rate is divided by speed, the net cost of transport can be calculated. The relative metabolic cost of transport at similar velocities is therefore also reduced in partial gravity environments. This means that less physical effort is necessary to move the body. As locomotion efficiency (defined as the total mechanical work divided by cost of transport) is reduced as well as both total mechanical work and cost of transport are reduced in partial gravity, total mechanical work must be reduced by a greater extent than cost of transport. Under consideration that partial gravity leads to a thoracic fluid shift, as indicated by the reduced thoracic impedance and the increased venous emptying volume, a higher blood volume in the region of the heart is very likely to lead to an increase in stroke volume. If stroke volume increases more than heart rate decreases, cardiac output must be increased (as found in the results) and might compensate for the reduction in heart rate.

# Relevance for Future Human Space Explorations and Countermeasure Developments

### Anticipated Consequences of Reduced Mechanical and Metabolic Stimuli in Partial Gravity

As described above, reduced impact forces due to partial unloading may lead to reductions of the work necessary to move the human body. This in turn may have detrimental long-term effects on the cardiopulmonary system, likely resulting in a loss of work performance capacity. Due to a reduction of important mechanical and metabolic stimuli the body is set into a "fake" resting state, affecting physiological systems and in the worst case resulting in physiological degeneration beyond (long-term-) mission threatening levels. It is very important that EVA's of the astronauts are completed without exhaustion and that their physical well-being is maintained for reasons of health, safety and mission success.

From an operational perspective it would be highly desirable to know minimum thresholds and exposure times to certain gravity levels that are needed to maintain relevant physiological systems (Horneck et al., 2003; Goswami et al., 2012). These systems will presumably react differently to similar gravity levels and therefore it is very unlikely that one minimum gravity threshold is sufficient to maintain all physiological systems equally. It can be anticipated from linear regression analyses that for some systems the lack of sufficient mechanical physiological stimuli becomes less severe as gravity increases. Some studies showed that there is a strong correlation between heart rate (Schlabs et al., 2013), oxygen consumption (Schlabs et al., 2013), (net) metabolic rate (Farley and McMahon, 1992; Teunissen et al., 2007), peak vertical ground reaction force (Ivanenko et al., 2002; Schlabs et al., 2013) and the simulated gravity levels in the range between 1 g and µg (with R <sup>2</sup> > 0.88 for all tested correlations). Therefore, exposure to Moon and Mars gravities might be less severe compared to physiological deconditioning as experienced in µg.

### Requirements for Exercise Countermeasure Concepts in Partial Gravity

To compensate for the anticipated loss in performance capacity some form of supplementary exercise will most likely be required. The slogan "use it or lose it" describes the adaptation process in a very simple way (Corcoran, 1991), and may also be applied to partial gravity environments.

As pointed out, reduced external forces acting on the body seem to be a main problem because a reduction of mechanical stimuli could also account for a reduction in metabolic stimuli. Therefore, exercise countermeasures should provide an individual, comprehensive training and especially focus on applying Earthlike GRF.

GRF can be modified/increased through increased locomotion velocity (Davis and Cavanagh, 1993), external applied horizontal forces (Chang et al., 2001) or reactive jumps (Kramer et al., 2010). Davis and Cavanagh (1993) provide the example that running at 4 m/s at 80% body weight creates the same magnitude of vertical GRF as running at 2.9 m/s at 100% body weight. Moreover, Chang et al. (2001) found out, that running at 0.38 g with 20% of additional applied horizontal forces increases impact force peaks to magnitudes equal or greater than those observed during running at Earth gravity. Furthermore, jumps induce high impact forces and internal muscle forces that are necessary for the deformation of bone and thus provide an osteo- and muscleprotective stimulus (Rittweger et al., 2000; Ebben et al., 2010). At the same time, plyometric exercise can be very exhaustive and therefore high workload- or high intensity interval protocols could induce cardiovascular responses, preventing cardiovascular deconditioning (Arazi et al., 2012, 2014). The workgroup of Kramer (University of Konstanz, Germany) invented a new sledge jump system which allows after some practice almost natural reactive jumps in reduced gravity (Kramer et al., 2010, 2012). This seems to be a promising countermeasure as it provides myogenic as well as osteogenic stimuli while only short exercise durations are necessary.

Unfortunately, there is no "one-fits-all"- strain level to maintain bone mineral density because it is affected by skeletal location and other systemic factors such as age, gender, and genetic background (Ruff et al., 2006). This can also be said for muscle mass, as the active tension required to induce hypertrophy or prevent atrophy is very likely to vary as it is subordinate to the complex "response matrix" of the respective subject (Toigo and Boutellier, 2006). Nevertheless, Frost (2004) refers to a bone's genetically determined modeling threshold strain range (1,000–1,500 microstrain; ∼2 kg/mm<sup>2</sup> ), within and above which formation of new bone exceeds resorption of bone mineral (Frost, 2004). Therefore, exercise countermeasures should aim at exposing the bones to up to 1,000 microstrain to at least maintain its strength. The study results of Peterman et al. (2001) reveal that bone strain magnitudes in the distal tibia are linearly related to GRF (R <sup>2</sup> > 0.7) (Peterman et al., 2001) which supports the authors suggestion that exercise countermeasures should focus on applying Earth like GRF as experienced during running and jumping.

# CONCLUSION AND OUTLOOK

This systematic review provides insights into the current state of research about human biomechanical and cardiopulmonary responses to partial gravity exposure. The synthesized results presented here suggest a lack of sufficient metabolic and mechanical stimuli when humans are exposed to partial gravity as can be seen in the extremely large effects of most of the presented outcomes. It can be anticipated that partial gravity environments as present on the Moon or on Mars are not sufficient to preserve all physiological systems to a 1 g standard if not addressed through adequate countermeasures. Therefore, to maintain astronaut's health, safety and performance capacity smart and evidence-based exercise countermeasure systems are needed. The main goal of these systems should be to re-create Earth-like GRF. Considering the smaller habitat/vehicle size to be used in future exploration missions, countermeasure devices should be as compact as possible but still target the musculoskeletal and cardiopulmonary systems equally. Bulky exercise machines as currently used on the ISS (e.g., ARED, cycle ergometer or treadmills) will not be an option for these missions.

The methodological quality of the vast majority of the available/included studies is too low to generate a compeling evidence. Future research is needed and should address physiological long-term effects of partial gravity exposure. Moreover, future studies should help defining minimal gravity thresholds and exposure times needed to maintain relevant physiological systems.

# AUTHOR CONTRIBUTIONS

All authors listed have made a substantial, direct and intellectual contribution to the work, and approved it for publication.

# ACKNOWLEDGMENTS

This systematic literature review is part of the "moon gym-project" that has the aim to better understand the physiological consequences of partial gravity exposure and to develop evidence-based concepts for new exercise countermeasures for future exploration missions. The "moon gym-project" is part of ESA's program "Spaceship EAC" which is an initiative investigating innovative technological and operational concepts in support of ESA's exploration strategy. This work was further supported by the Space Medicine Office of the European Space Agency (ESA). Patrick Jäkel, André Rosenberger, Nora Petersen, Dr. David Green, Dr. Aidan Cowley, Frits de Jong, Dr. Guillaume Weerts, Henny Vollmann as well as Dr. Jonathan Scott shall receive special thanks for their support throughout the project.

# SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fphys. 2017.00583/full#supplementary-material

Supplementary Table 1 | Cardiopulmonary changes in Lunar gravity.


Supplementary Table 4 | Biomechanical changes in Lunar gravity.

Supplementary Table 5 | Biomechanical changes in 0.25 g.

Supplementary Table 6 | Biomechanical changes in Martian gravity.

Supplementary Tables 1–6 | In the first row, information about authors, their used simulation model (HDT: head-down tilt), posture or locomotion (<sup>w</sup> walking; r running; <sup>s</sup> skipping; <sup>h</sup> hopping; PTS moving at preferred walk-to-run transition speed) during the intervention as well as number of participants and control conditions are presented. Terrestrial gravity as a control condition indicates that changes from 1 g to partial gravity have been considered, whereas µg as a control indicates that changes from µg to partial gravity have been considered. In the left column the different outcome parameters are listed and assigned to categories. Arrows indicate an increasing (↑), decreasing (↓), or stable (→) trend. Asterisks (∗) indicate significant differences from control conditions (*P* < 0.05) and hashtags (#) refer to partly significant differences.

# REFERENCES


Berry, C. A. (1974). Medical legacy of Apollo. Aerosp. Med. 45, 1046–1057.


gravity. J. Appl. Physiol. 110, 781–788. doi: 10.1152/japplphysiol.007 99.2010


Trappe, S., Costill, D., Gallagher, P., Creer, A., Peters, J. R., Evans, H., et al. (2009). Exercise in space: human skeletal muscle after 6 months aboard the International Space Station. J. Appl. Physiol. 106, 1159–1168. doi: 10.1152/japplphysiol.91578.2008

Waligora, J. M., and Horrigan, D. J. (1975). "Metabolism and heat dissipation during Apollo EVA periods," in Biomedical Results of Apollo, eds R. S. Johnston, L. F. Dietlein and C. A. Berry (Washington, DC: National Aeronautics and Space Administration), 115–128.

Widjaja, D., Vandeput, S., Van Huffel, S., and Aubert, A. E. (2015). Cardiovascular autonomic adaptation in lunar and martian gravity during parabolic flight. Eur. J. Appl. Physiol. 115, 1205–1218. doi: 10.1007/s00421-015-3118-8

**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2017 Richter, Braunstein, Winnard, Nasser and Weber. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Altered Venous Function during Long-Duration Spaceflights

Jacques-Olivier Fortrat <sup>1</sup> \*, Ana de Holanda<sup>1</sup> , Kathryn Zuj <sup>2</sup> , Guillemette Gauquelin-Koch<sup>3</sup> and Claude Gharib<sup>4</sup>

<sup>1</sup> UMR Centre National de la Recherche Scientifique, Faculté de Médecine d'Angers, 6214 Institut National de la Santé et de la Recherche Médicale, 1083 (Biologie Neurovasculaire et Mitochondriale Intégrée), Angers, France, <sup>2</sup> Faculty of Applied Health Sciences, University of Waterloo, Waterloo, ON, Canada, <sup>3</sup> Centre Nationale d'Etudes Spatiales, Paris, France, <sup>4</sup> Faculté de Médecine Lyon Est, Université Claude Bernard Lyon 1, Lyon, France

Aims: Venous adaptation to microgravity, associated with cardiovascular deconditioning, may contribute to orthostatic intolerance following spaceflight. The aim of this study was to analyze the main parameters of venous hemodynamics with long-duration spaceflight.

Edited by:

Olivier White, Université de Bourgogne Franche Comté, France

#### Reviewed by:

Eun Bo Shim, Kangwon National University, South Korea Capelli Carlo, Norwegian School of Sport Sciences, Norway J. Thomas Cunningham, Univerity of North Texas Health Science Center, United States

> \*Correspondence: Jacques-Olivier Fortrat jofortrat@chu-angers.fr

#### Specialty section:

This article was submitted to Integrative Physiology, a section of the journal Frontiers in Physiology

Received: 21 April 2017 Accepted: 29 August 2017 Published: 12 September 2017

#### Citation:

Fortrat J-O, de Holanda A, Zuj K, Gauquelin-Koch G and Gharib C (2017) Altered Venous Function during Long-Duration Spaceflights. Front. Physiol. 8:694. doi: 10.3389/fphys.2017.00694 Methods: Venous plethysmography was performed on 24 cosmonauts before, during, and after spaceflights aboard the International Space Station. Venous plethysmography assessed venous filling and emptying functions as well as microvascular filtration, in response to different levels of venous occlusion pressure. Calf volume was assessed using calf circumference measurements.

Results: Calf volume decreased during spaceflight from 2.3 ± 0.3 to 1.7 ± 0.2 L (p < 0.001), and recovered after it (2.3 ± 0.3 L). Venous compliance, determined as the relationship between occlusion pressure and the change in venous volume, increased during spaceflight from 0.090 ± 0.005 to 0.120 ± 0.007 (p < 0.01) and recovered 8 days after landing (0.071 ± 0.005, arbitrary units). The index of venous emptying rate decreased during spaceflight from −0.004 ± 0.022 to −0.212 ± 0.033 (p < 0.001, arbitrary units). The index of vascular microfiltration increased during spaceflight from 6.1 ± 1.8 to 10.6 ± 7.9 (p < 0.05, arbitrary units).

Conclusion: This study demonstrated that overall venous function is changed during spaceflight. In future, venous function should be considered when developing countermeasures to prevent cardiovascular deconditioning and orthostatic intolerance with long-duration spaceflight.

Keywords: blood volume, cardiovascular deconditioning, microgravity, venous plethysmography

# INTRODUCTION

In a standing posture, gravity induces peripheral venous pooling that, when excessive, can lead to orthostatic intolerance and fainting (Rowell, 1993; Fedorowski et al., 2012; Raj, 2014). For cosmonauts returning to Earth, the risk of orthostatic intolerance and fainting is greater in part due to hypovolemia resulting from adaptation to the microgravity environment (Gharib et al., 1988; Blomqvist et al., 1994; Fortney et al., 1996; Coupé et al., 2011). However, this hypovolemia is moderate and cannot fully explain the observed orthostatic intolerance in cosmonauts following spaceflight (Blomqvist and Stone, 1983; Fortney et al., 1996).

**67**

Other adaptations to microgravity have been identified that could contribute to orthostatic intolerance including changes in cardiac and baroreflex autonomic control (Shen et al., 1988; Eckberg and Fritsch, 1992; Hughson et al., 1994). While these adaptations are thought to primarily involve the arterial system, it should be noted that veins also possess autonomic adrenergic receptors that modulate functions including pooling capacity (Gelman, 2008). Previous work, using ultrasound imaging, has shown that venous morphology is altered during spaceflight (Arbeille et al., 2001, 2015). Although commonly used in daily medical practice, ultrasound imaging does not provide a global assessment venous functions (Donnelly et al., 2000). Therefore, it is still unknown whether spaceflight results in alterations to venous autonomic control or general venous function that could contribute to orthostatic intolerance.

Venous functions are complex and include factors such as filling and emptying properties, efficiency of the muscular venous pump, as well as microvascular filtration (Stewart, 2003; Krishnan et al., 2009). Venous occlusion plethysmography is one method that has been proposed to assess venous function (Skoog et al., 2015). Studies using this method have shown that the filling function of veins is altered during simulated and real shortterm spaceflight (Bungo, 1989; Louisy et al., 2001; Besnard et al., 2002). Therefore, the purpose of the current study was to use this method to assess venous function with long-duration spaceflight. It was hypothesized that long duration spaceflight would result in alterations in venous function that may lead to greater pooling in an upright posture and reduced orthostatic tolerance.

# MATERIALS AND METHODS

# Subjects

Twenty-four male cosmonauts from the Russian space program were studied between 2009 and 2015 during spaceflights (124– 192 days) aboard the International Space Station. Their mean ± SD anthropometric characteristics were as follows: age 44.3 ± 6.1 years, weight 82.6 ± 6.7 kg, height 1.77 ± 0.05 m, body mass index 26.4 ± 2.3 kg/m<sup>2</sup> . Their physical activity was not controlled before flight but each performed 2 h of exercise daily aboard the International Space Station as detailed elsewhere (Petersen et al., 2016). Data collection was performed by the medical team of the Institute for Bio-Medical Problems (IMBP, Moscow, Russia, see Section Acknowledgments) as part of the cosmonauts' regular medical supervision. This study was carried out in accordance with the recommendations of the Institutional Review Board of IMBP and all subjects gave written informed consent in accordance with the Declaration of Helsinki.

# Protocol

Data were collected on six testing days with two sessions occurring before spaceflight, two during the flight, and two after returning to Earth. The initial pre-flight session occurred more than 2 months before spaceflight (B > 2) with the second session occurring <2 months before the flight (B < 2). An early spaceflight session was conducted during the first 3 months of flight (F < 3) and a late flight session was conducted after the cosmonaut had spent 3 months in space (F > 3). Post-flight sessions occurred within 15 h of landing and 8 days after landing (L0 and L8, respectively). Each cosmonaut took selfmeasurements during the flight after having undergone training for the procedure before the flight. On Earth, all measurements were conducted with the cosmonaut in a supine position with the measurement leg supported at heart level according to the standard procedures (Stewart, 2003; Krishnan et al., 2009). During spaceflight, cosmonauts were in a "free floating" position with the knee slightly bent and the thigh in weak abduction.

# Calf Volume

Calf volume (CV) was determined by means of calf circumference measurements (measuring tape) using the method described by Thornton et al. (1992). Briefly, nine calf circumference measurements, distributed along the leg in predetermined positions were performed. The calf section between two adjacent circumference measurements was considered to be a truncated cone in order to convert the measured distances into a volume. The calf volume was the sum of the volume of the 8 truncated cones. A calf volume measurement was repeated before every plethysmography session.

# Air Plethysmography

Venous function was determined using an Air Plethysmograph APG <sup>R</sup> 1000 (ACI Corporation, San Marcos, CA, USA) that was modified for use in a microgravity environment. The device consists of a long tubular air cuff, positioned around the lower leg, that was inflated to a pressure of about 6 mmHg by an air pump. Throughout testing, the pressure in the cuff was constantly measured reflecting variations in leg volume.

Venous hemodynamics were assessed according to the procedure previously described (Stewart, 2003; Krishnan et al., 2009) with the determination of standard venous plethysmography parameters (Boccalon et al., 1987; Neglén and Raju, 1995; Louisy et al., 1997, 2001; Krishnan et al., 2009; Lattimer et al., 2014; Shiraishi, 2014). Venous occlusion was performed using a thigh cuff with changes in leg volume determined at five levels of venous occlusion; 20, 30, 40, 50, and 60 mmHg (**Figure 1**). For each level of venous occlusion, an "n" shaped curve was obtained with an increase in calf volume followed by a plateau with occlusion and a decrease in calf volume to pre-occlusion values with thigh cuff deflation (**Figure 1**). Venous occlusion was applied long enough to reach the plateau of the "n" shape curve as visually estimated by the operator. Venous occlusion lasted 3–5 min. Four points where marked by visual inspection by a trained operator (JOF): (a) the beginning of the volume increase, (b) the relative stabilization of the volume after the increase, (c) the beginning of the deflation, and (d) the relative stabilization of the volume after the deflation (point noted a, b, c, and d, respectively, **Figure 1**).

The initial fast increase (first 20 s), linked to arterial inflow and an arterial filling velocity, was assessed as the change in calf volume over the first 20 s of venous occlusion (aV in ml/min; Louisy et al., 1997; Shiraishi, 2014). The absolute volume increase at the plateau was used as a determination of venous filling function (1Vmax-a in ml, Louisy et al., 1997, 2001).

FIGURE 1 | Example of the venous plethysmography curve. (A) A whole session that includes several venous occlusion steps at increasing pressure (20, 30, 40, 50, and 60 mmHg, from left to right). (B) Example of an occlusion step showing the points to determined plethysmography variables: a, b: start and end points for calf vein filling measurements; c, d: start and end points for calf vein emptying measurements.

Venous filling was also determined with respect to resting calf volume [1Vmax-r, in percentage, that is (1Vmax-a/CV) <sup>∗</sup> 100] for the determination of venous capacitance (Louisy et al., 1997, 2001). This value was plotted against venous occlusion pressure (**Figure 2**) with the slope of the relationship providing an indication of venous compliance (Neglén and Raju, 1995; Krishnan et al., 2009,). Finally, venous distensibility was assessed as the Venous Filling Index (VFI, the mean filling velocity of 90% of the 1Vmax-a, in ml/min; Louisy et al., 2001; Shiraishi, 2014).

The drift at the end of the plateau was quantified as the slope of the line passing through the second and the third marked points (b and c on **Figure 1**, arbitrary units). With the venous occlusion cuff inflated, this drift is due to microvascular filtration increasing calf volume (Stewart, 2003; Krishnan et al., 2009; **Figure 3**).

Venous emptying is characterized by an initial fast emptying followed by a slower emptying. The initial part, dependent on venous elasticity and resistance to venous outflow (Boccalon et al., 1987; Louisy et al., 1997), was assessed as the emptying rate of 50% of pooled venous volume (VER50%, in ml/s). The slower emptying, mainly dependent on resistance to venous outflow, was quantified as the emptying rate of 90% of pooled venous volume (VER90%, in ml/s, Lattimer et al., 2014).

## Statistics

Data are presented as means ± SD. Each plethysmography session was analyzed as previously described (Stewart, 2003; Krishnan et al., 2009). Briefly, values of each of the

FIGURE 2 | Venous compliance. Venous compliance is assessed through the pressure/volume relationship. The diagram is drawn using the relative filling volume (1Vmax-r, in percentage) and the venous occlusion pressure during two whole plethysmography sessions on the same cosmonaut. The first session occurred more than 2 months before space flight (B > 2) and the second one during the flight but before its third month (F < 3). A whole plethysmography session included five levels of venous occlusion (x-axis). Equations of the linear regressions are mentioned on the graph.

plethysmography parameters were plotted against venous occlusion pressures to determine the slopes of the regression lines (see **Figure 2**). Slopes of regression lines were then compared using a between-period analysis of variance (ANOVA) after a Barlett's test for equality of variances. When appropriate, a post-hoc t-test for paired data with Bonferroni correction

volume. The slope of the dashed line provides a quantification of this drift and

is used to assess microvascular filtration.

was applied (Prism 5.01, GraphPad Software, San Diego, CA, USA). All cosmonauts did not perform all plethysmography sessions due to operational limitations, therefore, a repeated measurement ANOVA was not performed. Statistical significance was set at p < 0.05.

# RESULTS

One hundred and three plethysmography sessions were performed during this study (**Table 1**). Post-flight data were only collected on a small number of cosmonauts due to the late introduction of these measures into the study and operational limitations. Calf volume measures and venous plethysmography results for each testing day are presented in **Table 1**.

Calf volume decreased during space flight and remained unchanged throughout the flight. Recovery to pre-flight volume began shortly after landing and was completed 8 days later. Absolute filling volume was not significantly altered during spaceflight or recovery from flight. However, relative to calf volume, venous filling significantly increased early during spaceflight and tended to remain elevated later in flight. Upon return to Earth, relative venous filling remained elevated on L0, but had recovered by L8. Venous distensibility, assessed through VFI, initially increased during the flight but recovered to pre-flight values later in flight. Following flight, VFI showed an exaggerated recovery on L0 and recovered to pre-flight values on L8.

Initial venous emptying (VER50%) was not significantly changed during or after spaceflight while the late venous emptying (VER90%) was decreased on the first flight session with a further decrease later in flight. Additionally, this parameter did not return to preflight levels on either L0 or L8. A slight increase in microvascular filtration was seen early in spaceflight but was not changed on any other testing day.

All the venous filling parameters had the same pattern of changes during space flight that was different from the one of calf volume and venous emptying parameters. This pattern showed a large change during the initial part of the space flight and a trend toward recovery of pre-flight values during the second part of the space flight (**Table 1**). Calf volume showed a large change that was maintained during the whole space-flight while the venous emptying parameters showed continuously increasing changes (**Table 1**). The pattern of change was similar between venous filling parameters and microvascular filtration (**Table 1**).

# DISCUSSION

The purpose of this study was to investigate venous function before and during long-duration spaceflight. Consistent with the hypothesis, results indicated alterations in venous functions with adaptation to microgravity. Changes were seen with both venous filling and emptying but different patterns in responses were noted that did not completely parallel changes in calf volume.

Reduced calf volume leading to "bird legs" is a well-known result of spaceflight (Blomqvist et al., 1994) and was noted in the current study. It is generally believed that this reduction in calf volume is primarily due to muscular atrophy (Blomqvist et al., 1994). Vein function is strongly linked to muscle mass due to the actions of the muscle pump and the influences of muscle on venous transmural pressure (Atkov and Bednenko, 1992). However, upon return to Earth, venous function tended to recover after 8 days whereas muscle mass recovery requires additional time (6–8 weeks, Atkov and Bednenko, 1992). Moreover, evidence of lower limb muscle atrophy during spaceflight has mainly been obtained from animal studies during which animals were completely inactive and food intake was uncontrolled (Atkov and Bednenko, 1992). Today, cosmonauts perform daily exercise countermeasures and close attention is paid to food intake (Petersen et al., 2016). Recent work has alternatively focused on spinal muscle adaptation (Hides et al., 2016) as leg muscle atrophy is not readily apparent with the current countermeasures used. Therefore, it is unlikely that changes in leg muscle mass contributed to the changes in calf volume and venous function observed in this study.

Calf volume changes were likely the result of fluid shifts during spaceflight as volume was seen to rapidly recover upon return to Earth. However, venous blood shift alone cannot fully explain the large calf volume changes suggesting the involvement of tissues and interstitial volumes. In general, these fluid shifts undoubtedly influenced venous functions. However, venous function showed an adaptation to these shifts since venous function tended to recover toward pre-flights value after 3 months of spaceflight. In 1998, White and Blomqvist proposed a model to explain the initial cardiovascular adaptations to spaceflight which included a substantial redistribution of fluid and pressure throughout the body that differed from that seen in Earth based spaceflight simulations. Results from the current study and recent longduration spaceflight investigations also support the idea of fluid redistribution throughout the body not only throughout the cardiovascular system but also within tissues and interstitial spaces (Baisch, 1994; Verheyden et al., 2010; Norsk et al., 2015).

Venous plethysmography demonstrated a decrease in VER90% that indicated a decrease in venous resistance. The decrease in venous resistance is also consistent with the overall vasorelaxation observed during space flight (Norsk et al., 2015). Alteration in autonomic nervous control of venous functions with spaceflight could explain the decrease in venous resistance. Recent studies have, however, challenged the notion of reduced sympathetic activity with spaceflight suggesting that adaptations likely reflect sympathoexcitation (Verheyden et al., 2010; Mandsager et al., 2015; Norsk et al., 2015). Norsk et al. (2015) observed that the increase in cardiac output during long duration spaceflights is more than previously observed during short duration spaceflights. Similarly, we demonstrated a decrease in venous resistance during spaceflight with a further decrease later in flight (VER90%, **Table 1**). Decreased venous resistance promotes venous return and might explain the increased cardiac output observed by Norsk et al. (2015). Alteration of venous resistance and cardiac output are likely to be the result of the body fluid redistribution suggested by White and Blomqvist (1998).

The small but significant effect of spaceflight on microvascular filtration contrasts with the strong effects on venous filling and


TABLE 1 | Plethysmography data during long-term spaceflight.

Measurements were conducted more than 2 months (B > 2) and <2 months before flight (B < 2), during the first 3 months of spaceflight (F<3), after 3 months of flight (F > 3), on landing day (L0), and 8 days after landing (L8). Table shows the values (mean ± SD) for the number of cosmonauts tested (n), calf volume (CV), arterial filling speed (aV), maximal filling volume (∆Vmax-a), maximal filling volume as a percentage of calf volume (∆Vmax-r), Venous Filling Index (VFI), emptying rate of 50% of pooled venous volume (VER50%), emptying rate of 90% of pooled venous volume (VER90%), and microvascular filtration (µ filtration). Calf volumes are reported in liters while all other values are the slopes of the regression lines between measurements and venous occlusion pressure and are reported in arbitrary units. Values that are statistically different from B > 2, B < 2, F < 3, and F > 3 are denoted by \*, +, \$, and § respectively. Single, double, and triple symbols represent statistical significance at p < 0.05, p < 0.01, and p < 0.001, respectively.

emptying functions. The pattern of change is, however, similar between venous filling function and microvascular filtration and the lack of change could be due to the large standard deviations of the filtrations assessment. Stewart (2003) showed that microvascular filtration function is changed in patients with a postural orthostatic tachycardia syndrome (POTS) suggesting that increased microvascular filtration could be related to the development of orthostatic intolerance. However, not all cosmonauts experience orthostatic intolerance after spaceflight and few of those with orthostatic intolerance exhibit symptoms of POTS (Coupé et al., 2011). Further studies are needed to determine whether microvascular filtration can be used as a measure for identifying cosmonauts who will experience orthostatic intolerance and POTS after long-duration spaceflight.

The current study utilized venous plethysmography for the assessment of venous function. However, complete assessments of vein function requires measurements that are difficult to conduct on Earth and even more difficult in a microgravity environment. In addition to venous plethysmography assessments, measures of venous blood volume and central, and peripheral venous pressure are also required to fully explore venous function (Gelman, 2008). While the current study demonstrated changes in venous properties with long-duration spaceflight, questions remain regarding mechanism involved and potential functional consequences of these changes.

In conclusion, the current study demonstrated that both venous filling and emptying functions are altered during longduration spaceflight. While partially associated with changes in calf volume, the changes in venous function may indicate a redistribution of fluid unique to microgravity adaptations. Understanding changes in venous function with microgravity exposure may help in the development if future countermeasures to protect against cardiovascular deconditioning and the development of orthostatic intolerance with long-duration spaceflight.

# AUTHOR CONTRIBUTIONS

JF: Analysis; JF, Ad, and KZ: Drafting of the work. JF, Ad, KZ, GG, and CG: Data interpretation, revising the work critically for important intellectual content, and final approval of the version to be published.

# ACKNOWLEDGMENTS

We would like to thank the cosmonauts for taking part in this study. We thank A. Kotovskaya, G. Fomina, and A. Salnikov from the Institute for Bio-Medical Problems of Moscow for sharing the data, for the in-depth scientific discussions, and for critical review of the manuscript. We thank D. Chaput, J. C. Lloret, and L. Nguyen from Centre National d'Études Spatiales (Toulouse, France) who set-up the Cardiomed device. We would also like to thank the Centre Hospitalier Universitaire d'Angers (Angers University Hospital) and the Centre de Recherche Clinique of the Centre Hospitalier-Universitaire d'Angers (Angers University Hospital Clinical Research Centre) for their help with the administrative procedures for biomedical research projects. Finally, we extend our thanks to M. Kantamirova who overcame the language barrier. JF benefits from the support of the Centre National d'Études Spatiales (National Centre for Spatial Studies) (CNES, grant # 2014/4800000763).

# REFERENCES

Arbeille, P., Fomina, G., Roumy, J., Alferova, I., Tobal, N., and Herault, S. (2001). Adaptation of the left heart, cerebral and femoral arteries, and jugular and femoral veins during short- and long-term head-down tilt and spaceflights. Eur. J. Appl. Physiol. 86, 157–168. doi: 10.1007/s004210100473

Arbeille, P., Provost, R., Zuj, K., and Vincent, N. (2015). Measurements of jugular, portal, femoral, and calf vein cross-sectional area for the assessment of venous blood redistribution with long duration spaceflight (Vessel Imaging Experiment). Eur. J. Appl. Physiol. 115, 2099–2106. doi: 10.1007/s00421-015-3189-6


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2017 Fortrat, de Holanda, Zuj, Gauquelin-Koch and Gharib. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Falls and Fall-Prevention in Older Persons: Geriatrics Meets Spaceflight!

#### Nandu Goswami 1, 2 \*

*<sup>1</sup> Gravitational Physiology, Aging and Medicine Research Unit, Institute of Physiology, Medical University of Graz, Graz, Austria, <sup>2</sup> Department of Health Sciences, Alma Mater Europea University, Maribor, Slovenia*

This paper provides a general overview of key physiological consequences of microgravity experienced during spaceflight and of important parallels and connections to the physiology of aging. Microgravity during spaceflight influences cardiovascular function, cerebral autoregulation, musculoskeletal, and sensorimotor system performance. A great deal of research has been carried out to understand these influences and to provide countermeasures to reduce the observed negative consequences of microgravity on physiological function. Such research can inform and be informed by research related to physiological changes and the deterioration of physiological function due to aging. For example, head-down bedrest is used as a model to study effects of spaceflight deconditioning due to reduced gravity. As hospitalized older persons spend up to 80% of their time in bed, the deconditioning effects of bedrest confinement on physiological functions and parallels with spaceflight deconditioning can be exploited to understand and combat both variations of deconditioning. Deconditioning due to bed confinement in older persons can contribute to a downward spiral of increasing frailty, orthostatic intolerance, falls, and fall-related injury. As astronauts in space spend substantial amounts of time carrying out exercise training to counteract the microgravity-induced deconditioning and to counteract orthostatic intolerance on return to Earth, it is logical to suggest some of these interventions for bed-confined older persons. Synthesizing knowledge regarding deconditioning due to reduced gravitational stress in space and deconditioning during bed confinement allows for a more comprehensive approach that can incorporate aspects such as (mal-) nutrition, muscle strength and function, cardiovascular (de-) conditioning, and cardio-postural interactions. The impact of such integration can provide new insights and lead to methods of value for both space medicine and geriatrics (Geriatrics meets spaceflight!). In particular, such integration can lead to procedures that address the morbidity and the mortality associated with bedrest immobilization and in the rising health care costs associated with an aging population demographic.

# Edited by:

*Brian James Morris, University of Sydney, Australia*

#### Reviewed by:

*Slade T. Matthews, University of Sydney, Australia Angela J. Grippo, Northern Illinois University, United States*

\*Correspondence: *Nandu Goswami nandu.goswami@medunigraz.at*

#### Specialty section:

*This article was submitted to Integrative Physiology, a section of the journal Frontiers in Physiology*

Received: *07 July 2017* Accepted: *04 August 2017* Published: *11 October 2017*

#### Citation:

*Goswami N (2017) Falls and Fall-Prevention in Older Persons: Geriatrics Meets Spaceflight! Front. Physiol. 8:603. doi: 10.3389/fphys.2017.00603*

Keywords: immobilization, orthostatic intolerance, countermeasures, spaceflight, aging, bedrest, fraility, falls

# INTRODUCTION

This paper outlines a general overview and comparison of spaceflight medicine, the experience of human aging, as well as important connections and parallels between spaceflight physiology and the aging process. Consideration of these connections and parallels can be exploited to improve human health for both astronauts living in microgravity and for older persons on earth. Included here are the following aspects: (1) Gravity and its effects on physiological systems; (2) Spaceflight induced physiological effects; (3) Aging induced physiological effects; (4) Earth-based simulation of spaceflight deconditioning compared with the experience of bedrest immobilization in older persons; (5) Maintenance of astronaut health: current countermeasures against the negative effects of microgravity in space; and (6) Benefits of life in space and how it can help life on Earth: Geriatrics meets spaceflight! These aspects are now outlined in the sections below.

# GRAVITY AND ITS EFFECTS ON PHYSIOLOGICAL SYSTEMS

During the course of natural evolution, as species moved from water to land, the effects of gravity needed to be countered. Evolving land animals, especially those that changed posture from a horizontal to vertical orientation, including humans, needed to develop new systems to regulate fluid distribution and blood flow, to ensure structural support as well as to maintain postural stability and facilitate locomotion (see Morey-Holton, 2003).

The extent to which gravity influences physiology is often underestimated. Gravity affects many systems and structures such as, size and location of internal organs (e.g., heart). Lillywhite (1988) investigated these factors in detail. He reported that snakes that live in trees, on land or in the sea have different locations of the heart, which in turn, affects their gravity tolerance (as assessed by loss of consciousness with increasing gravitation loading as provided by centrifugation). For instance, during centrifugation, the greatest gravity tolerance was seen in tree snakes (in whom the heart is closest to the brain) while the least tolerance was seen in water snakes (in whom the heart is located in the central part of the body and far away from the brain; Lillywhite et al., 1997).

In humans, physiological systems have evolved in such a way as to compensate for the effects of gravity. For example, even though the heart is located below the brain during standing, the force of contraction of the heart and heart rate are adequate to maintain blood flow to the brain. Furthermore, during standing, the pooling of blood in the legs due to the gravitational force directed earthward is countered by the muscle pump in the lower limbs, by one-way venous valves as well as via the "respiratory pump." Physiological systems in healthy humans are well adapted to gravitational changes induced by standing (orthostasis) and thus changes in posture do not normally lead to any significant problem. This ability to stand without physiological problems is termed orthostatic tolerance.

# SPACEFLIGHT INDUCED PHYSIOLOGICAL EFFECTS

The powerful influences of gravity (or the lack of it) on physiological systems can be seen when gravity is reduced (e.g., the microgravity environment of spaceflight). Spaceflight produces observable changes in the physiology of humans, many of which are almost impossible to control during missions, and all of them lead to time-dependent adaptation processes (Blaber et al., 2013; Hargens et al., 2013). These adaptive changes show inter-individual differences (Goswami et al., 2013). Entering microgravity leads to the alteration of physiological processes, that depend significantly on gravity, and introduces physiological functional changes such as, adaptations in cardiovascular control functions, cerebral perfusion (Blaber et al., 2013; Goswami et al., 2013), and changes in the musculoskeletal system and a reduction in sensorimotor system performance.

Antonutto and di Prampero (2003) reported that cardiovascular deconditioning in spaceflight is a significant problem, which is related to the time spent in microgravity. The homeostatic adjustments in the cardiovascular system in spaceflight include alterations in autonomic regulation, decrease in blood pressure and decreases in blood, plasma, and interstitial fluid volumes (Gazenko et al., 1986). The shift of fluid toward the head is followed by a reflex-induced decrease in total blood volume (Levy and Talbot, 1983). Over a few hours, a new equilibrium is established for blood returning to the heart with decreased stroke volume and diastolic blood pressure (Antonutto and di Prampero, 2003).

Returning to normal gravity on Earth has many important effects, including increased heart rate, decreased vasoconstriction (increased venous pooling), a reduction in exercise capacity (as quantified by heart rate responses and oxygen consumption during specified work loads, Hamilton, 2008), and decreased orthostatic tolerance (Buckey et al., 1996). Orthostatic intolerance involves exaggerated increase of heart rate, fatigue, light-headedness, dizziness, impairment of performing physical and mental tasks, and often leads to syncope during upright posture. In particular, such orthostatic intolerance is a potential threat to crew safety when re-entering a gravitational field in general. The impact of decreased orthostatic tolerance is not trivial, in that ∼30% of astronauts experience orthostatic intolerance following even a short duration spaceflight (Buckey et al., 1996). This number grows to 80% after long duration flights (Meck et al., 2004). The reduction of orthostatic tolerance is an extremely important aspect of cardiovascular deconditioning following actual space flight (Blaber et al., 2011), and can also be observed in several important clinical problems in Earth based medicine. Orthostatic intolerance seems to be common in both American (Bungo and Johnson, 1983) and Russian crew members (Gazenko et al., 1981) returning from space. Reduced tolerance to lower body negative pressure (LBNP) has also been observed (Blomqvist, 1983).

Spaceflight leads to reductions in heart function and changes in heart rate responses, lowering of the total blood volume, and even changes in venous compliance in different vascular beds due to blood pooling. All these factors, either individually or in combination, may be responsible for the decreased venous return that occurs on standing (Watenpaugh and Hargens, 1996) thus leading to post-spaceflight orthostatic intolerance (Blaber et al., 2011). Furthermore, it has been suggested that autonomic neural control alterations, including those of the baroreflex may induce most of the responses seen in the cardiovascular system post-spaceflight (Mano, 2005).

# AGING INDUCED PHYSIOLOGICAL EFFECTS

Important parallels can be found between deconditioning due to spaceflight and deconditioning and other physiological changes due to aging. For instance, in humans living on Earth and especially in older persons, hypovolemia and reduced cerebral blood flow, cerebral or peripheral vascular disease, metabolic or endocrine disorders, autonomic neuropathy, or cardiac arrhythmias may result in syncope (dizziness and loss of consciousness) when standing up. For older persons such reduced tolerance to upright posture (orthostatic intolerance) is a common condition that can lead to falling down and injury.

Frail older adults often have decreases in their functional and physiological reserves, as well as malnutrition, which particularly makes them vulnerable to diseases and falls (detailed in Martínez-Velilla et al., 2015). Furthermore, the negative consequences of aging such as, illness or injury due to falls often require admission to hospital. However, the immobilization that occurs during hospitalization is itself a major factor in physiological deconditioning and functional decline and can contribute to a downhill spiral of increasing frailty, orthostatic intolerance and increased risk and incidence of falls (Mühlberg and Sieber, 2004). A systematic review by Heinrich et al. (2010) indicates that fully 0.85–1.5% of all health care costs are dedicated only to falls and their consequences. Furthermore, given that older patients spent up to 83% of hospital admission lying in bed (Lazarus et al., 1991; Brown et al., 2009; Pedersen et al., 2013), confinement in bed during hospitalization—and its effects, for example, orthostatic intolerance —represents a central challenge in the care of the vulnerable older population in any acute care hospital across the whole world.

# EARTH-BASED SIMULATION OF SPACEFLIGHT DECONDITIONING COMPARED WITH BEDREST CONFINEMENT IN OLDER PERSONS

Bedrest has been used routinely as an analog for studying the consequences of weightlessness on body systems as seen during space flight (Pavy-Le Traon et al., 2007; Jost, 2008; Goswami et al., 2015a). Sandler and Vernikos reviewed the first 25 years of bedrest studies in 1986 and proposed inactivity as a tool to study space flight effects (Sandler and Vernikos, 1986).

The bedrest study protocol, in which subjects lie in supine position over variable time periods, is a highly controllable experimental set-up which provides excellent possibilities to investigate physiological function changes during lowered gravitational stress (Trappe et al., 2007; Arzeno et al., 2013; Cvirn et al., 2015; Florian et al., 2015; O'Shea et al., 2015). Bedrest is also an important platform that allows for the testing of (potential) countermeasures to overcome the physiological deconditioning effects of lowered gravitational loading due to microgravity as well as the effects of bedrest deconditioning during hospitalization (Schneider et al., 2009; Waha et al., 2015). Insights and countermeasures to combat deconditioning obtained from these experiments would also be of great value in addressing the problem of increasing frailty in older persons which could arise due to bedrest confinement.

# MAINTENANCE OF ASTRONAUT HEALTH: CURRENT COUNTERMEASURES AGAINST THE NEGATIVE EFFECTS OF MICROGRAVITY

For maintenance of their health in space—and to counter difficulties of re-entry into Earth's atmosphere as well as postflight orthostatic intolerance—astronauts carry out exercise regimes comprised of resistive exercise and physical training of up to 2.5 h per day for 6–7 days per week (Hackney et al., 2015; Petersen et al., 2016). Petersen et al. (2016) recently reported that since the advanced resistive exercise device (ARED)—which provides an opportunity to increase the prescription of resistive training—was introduced at the international space station (ISS) 8 years ago, the resistance exercise component of total in-flight exercise has increased but cycle ergometry and treadmill running has decreased. The usage of ARED is particularly important since cycle ergometry performed in space does not seem to increase workload despite the need to achieve this in microgravity. This could be due to technical and biomechanical factors associated with exercising in space, which provides additional physiological and technical constraints. Similarly, when carrying out treadmill running exercise in space, crewmembers exercise with static loads of only between 70 and 80% of their bodyweight (detailed in Petersen et al., 2016). Overall, the current devices for exercise available on the ISS and the physical activity training regime appear to better address some of the issues related to astronaut health.

Just like physical activity, nutrition is important for maintenance of human health on Earth and in space. Aspects such as inadequate nutrition intake, vitamin D status, and oxidative damage need to be considered in human spaceflight (Smith et al., 2005). However, only up to 70% of energy intake requirement is typically met in space crew members (Smith et al., 2009, 2013). Unsurprisingly, inadequate food/energy intake in space leads to body weight loss and could affect, muscular, cardiovascular, and endocrine systems (Smith et al., 2009). For instance, caloric restriction in obese humans following weight reduction therapy, in fasted pilots and from animal experiments have shown that even caloric restriction in moderate amounts can affect fluid homeostasis, as well as lead to decreases in blood volume, heart rate, blood pressure, and norepinephrine and cardiovascular function (Florian et al., 2015). In addition, caloric/fat restriction has been shown to lead to alterations in baroreflex function, altered vascular smooth muscle tone, and consequently, decreased orthostatic tolerance (Florian et al., 2015).

Recent missions, however, have shown that crew energy intake can be maintained at > 90% of the required amount (Smith et al., 2012). Nutrition alone, however, has not been shown to be effective in preventing lower limb muscle volume or strength loss. Adequate energy intake + resistance exercise training during long-term spaceflight has been shown to improve lean muscle tissue, reduce muscle volume loss and muscle strength loss (Trappe et al., 2007) and to reduce fat (Smith et al., 2012). Furthermore, nutrition and exercise in space have been shown to help in counteracting bone loss (Smith et al., 2012): Bone resorption in space was accompanied by bone formation.

Potential nutritional countermeasures that may help in bone protection during spaceflight include (Smith et al., 2013):


As we embark on long-term exploration of deep space, nutrient intake must be optimized. Therefore, research aimed at optimizing/supplementing the effects of physical exercises via additional nutritional supplementations and/or pharmacological interventions is continuing (detailed in Smith et al., 2012; Hackney et al., 2015). Additionally based on data from ground based (bedrest) studies, dietary countermeasures such as decreasing sodium intake (currently >5 g Sodium/ day is ingested, Smith et al., 2012) and how protein intake affects bone health are also currently being investigated in space.

# LIFE IN SPACE AND HOW IT CAN HELP LIFE ON EARTH: GERIATRICS MEETS SPACEFLIGHT!

From the above discussion, it can be seen that many key problems of aging and age-related deconditioning have parallels with those problems confronted during spaceflight and thus synergies can be drawn between aging and spaceflight. For instance, older persons spend up to 80% of their time in hospital bed-confined, which, in turn, effects their physiological functions in a manner similar to spaceflight deconditioning. The strategies used in space for maintenance of astronaut health could also be used for the benefit of life on Earth, especially in older persons (Geriatrics meets spaceflight!). For instance, just as in space, caloric restriction on Earth has been shown to slow the aging process, increase stress resistance and delay the onset of age-related diseases, and even provide cardioprotection (Han and Ren, 2010).

In the following section, aspects related to bedrest confinement induced physiological deconditioning as well as the need to provide rapid interventions (countermeasures) to alleviate the deconditioning are discussed, as well as how current countermeasures used in space could be potentially used in geriatrics.

# Aging, Immobilization, and Early Remobilization

Ambulation is a prerequisite for mobility, and mobility is a primary requirement for autonomy and quality of life. However, toward the end of an individual's lifespan, reduced muscle function due to, for example, sarcopenia and dynapenia impair ambulatory function. The addition of injuries or illnesses renders people often unable to ambulate, and, therefore, muscles become deprived of the habitual stimulating signals. Combination of sarcopenia and disuse atrophy causes profound muscle wasting (Kortebein et al., 2008; Suetta et al., 2009), and the deconditioning process is a medical risk in itself that causes numerous adverse events (Brown et al., 2004). These consequences, related to reduced ambulation and increased bed confinement, begin immediately after admission to hospital, and deficits in the Activities of Daily Living (ADL) can be seen even on the second day of bedrest confinement in older people (Hirsch et al., 1990). If the deconditioning passes below the threshold of independent ambulation, then the deficits in muscular strength and power become perpetuated. Such muscle deficits are difficult to recover from in older patients, as most of them have generalized inflammation and metabolic disorders.

To effectively address this challenge, it is necessary to identify and intervene in pathways that link age-associated frailty, undernutrition and other factors to the effects of immobilization as for example in the cycle depicted in **Figure 1**.

During bedrest confinement, immediate action is therefore required to break the vicious circle of immobilization and muscle wasting: effective remobilization has to start as early as possible during and after hospital admission. Intervening in this cycle requires a holistic multifactorial approach that takes into

account key factors such as, (mal-) nutrition, (de-) conditioning, muscle loss, cardiovascular, and vestibular effects; all of these could contribute toward immobilization induced orthostatic intolerance. It is important to note that since astronauts returning to Earth from the microgravity environment of spaceflight show similar physiological deconditioning as bed-confined persons, effective management of these effects is important for both astronauts and older persons.

# Addressing Both Ambulation and Falls in Older Persons

Falls in older people are particularly frequent during hospital stays or in the weeks and months thereafter (Mahoney, 1998). While compromised muscle function undoubtedly contributes to falls in old age, there are many other reasons for the increasing number of falls in older people (Blain et al., 2016; Bousquet et al., 2017). One of the key factors responsible for falls in older persons is postural hypotension (Weiss et al., 2004). Furthermore, it is well known that bed rest confinement induces hypovolemia (Convertino, 2007) and orthostatic intolerance (Dittmer and Teasell, 1993), the latter being perceived as dizziness upon standing up. Of note, 40% of falls in nursing homes are related to posture changes from supine to standing (Rapp et al., 2012). Also, anti-hypertensive and diuretic medicine treatment is a known cause of falls in older people (Gangavati et al., 2011). It is very likely, therefore, that the excessive risk of falls after hospital discharge is attributable to compromised cardiopostural control and impaired cerebral perfusion. Therefore, there is a need to study both falls in ambulatory persons and falls following bedrest confinement, and assess the interrelationship between cardio-postural control and the risk of falls.

For maintenance of standing balance, the ability of the body to detect postural disturbances and respond accordingly is required. These abilities deteriorate with increasing age, thus predisposing older people to imbalance and greater risk of falls. Aging is also associated with worsening of the somatosensory and motor systems functions, which in turn, can lead to poor static standing balance. A research model to measure and assess the cardiovascular-postural system as an integrated and interacting system, which can be used to determine the effectiveness of both the postural and cardiovascular systems (both of which are severely impacted by immobilization, particularly in older persons) has been developed (Blaber et al., 2009; Goswami et al., 2012). Current knowledge proposes the measurement and tracking of changes in the two systems and their interaction in order to detect and assess changes due to immobilization that can impact cardiovascular, postural, or interactive processes in older persons. This assessment strategy, carried out via a sit to stand test followed by gait monitoring, will allow for the delivery of specific therapies to improve ambulatory recovery and also provide feedback on the effectiveness of physical interventions during immobilization.

As most astronauts also experience orthostatic intolerance post-spaceflight, there is a need to develop countermeasures that can alleviate orthostatic intolerance in the deconditioned returning astronaut. Astronauts replace spaceflight-induced plasma loss via salt ingestion shortly before landing to prevent post-spaceflight orthostatic intolerance (Campbell and Charles, 2015). Similarly, replenishment of plasma volume loss using salt tablets and fluid loading following 12 days of 6-degree head down bed rest has been shown to prevent post-bed-rest orthostatic hypotension (Waters et al., 2005). However, the impact of plasma volume loss due to bedrest confinement during hospitalization is currently not well-documented and needs to be further explored. It should be proposed that bed-confined older persons should also undergo regular assessment of the plasma electrolytes and depending on the duration of bedrest confinement (Stuempfle and Drury, 2007)—be provided replacement of the plama volume losses that occur.

# Countering the Impact of Immobilization during Hospitalization in Older Persons

When older patients are bed confined, a rapid decline in muscle mass, bone mass, and functionality sets in (Singh et al., 2008; Martínez-Velilla et al., 2015). If remobilization starts early, then the declines incurred are small enough to allow recovery (Singh et al., 2008). If intervention is started later, then recovery is usually incomplete and patients can be left with chronic reduced function (Singh et al., 2008; Martínez-Velilla et al., 2015). In many cases unfortunately, remobilization starts too late and patients permanently lose their independence and autonomy; this is associated with higher morbidity and mortality (Singh et al., 2008).

The interaction of pre-existing factors such as, malnutrition and/or sarcopenia with the effects of immobilization poses complex challenges to managing health in older persons. Furthermore, innovative methods and care procedures implemented during immobilization need to be extended to the period after hospitalization and return to the community to maximize recovery and reduce frailty, to restore mobility and physical activity, and to avoid falls and new hospitalizations. This important dimension has not been investigated yet but knowledge in this regard is urgently needed. An optimal sequence of intervention steps is described in the box below (**Figure 2**).

Remobilization of older people in the acute care setting is delayed in many hospitals. Physical exercise interventions are often started too late and based upon care provider experience and non-standard decision making, and often started when there is already a substantial muscle mass and function loss.

FIGURE 2 | A schema showing the need for patient screening, in-hospital interventions during immobilization and extension of interventions during follow-up in the community.

Many older patients pass beyond the "point" of no return for sustainable physical functioning, thereby entering the vicious circle of de-conditioning and hospital re-admissions and further dependency care. Therefore, there is a need to start the interventions as soon as possible (Singh et al., 2008; Martínez-Velilla et al., 2015). Intervention combinations could incorporate physical exercise, nutrition and cognitive-behavioral training. Based on above observations regarding orthostatic intolerance and deconditioning in astronauts, it is reasonable to expect insights from space medicine in regards to these issues.

# Physical Activity and Exercise Training in Older Persons

In older persons physical activity appears to be beneficial for both physical and mental well being (Olanrewaju et al., 2016). However, there is almost no literature related to the effects of physical activity during bedrest confinement in older persons. While manual physiotherapy is routinely used in bedrest-confined patients, and is highly beneficial in preventing immobilzation-induced contractures and possible thrombotic events, the extent to which this is beneficial in terms of muscle strength and muscle function has not been studied in detail.

Older ambulatory persons often have access to exercise machines, which are often difficult to use in bed-confined patients. Such devices as powerplates and/or vibration plates are routinely used in ambulatory geriatric care. However, currently such devices and interventions are not used in bed-confined patients. Proposing resistive vibration exercises for bed-confined older persons, which have been shown to maintain muscle strength and function in experimental bedrest studies, of varying durations, in young bedrested persons (Schneider et al., 2009) provides an important example of how evidence from ground based analogs of spaceflight could be used in geriatrics. Schneider et al. (2009) reported that carrying out alternating aerobic and resistive exercise during bedrest confinement in young women leads to restoration of the aerobic capacity. This effect, however, is not seen with interventions that only involve nutritional supplementation.

# Role of Nutrition: Current Evidence Related to (mal-) Nutrition in Older Persons

There is growing evidence that there is a strong association between risk of frailty and inadequate intake of food in elderly people (Martone et al., 2013). In a large multinational study with more than 4,000 older participants (mean age of 82.3 years: 75.2% female), malnutrition prevalence was 22.8% (Kaiser et al., 2010). Malnutrition prevalence varied in different settings: prevalence in rehabilitation > in hospital > in nursing home > in community. When examining the combined database, the "at risk" group prevalence was up to 46% (Kaiser et al., 2010). In a German malnutrition study prevalence at hospital admission was even higher: between 35 percent (60–70 years) and 60 percent (>70 years) were already malnourished or at risk for malnutrition (Pirlich et al., 2006).

Malnutrition is characterized by unintended weight loss, loss of appetite, loss of muscle mass and bone mass. It results in functional decline, increased morbidity, preterm dependency, more frequent readmission after hospital discharge, early dependency and institutionalization, and finally increased mortality. Together, these factors result in increased costs for healthcare systems. Low dietary intake is one of the most important factors of malnutrition (Saunders et al., 2015). The vast majority of literature shows that sufficient amount of energy and key nutrients such as, protein, vitamin D, and calcium may help to improve muscle function and to maintain muscle and bone mass (European Commission, 2012; Saunders et al., 2015).

# Role of Nutritional Therapy in Older Persons

The evidence regarding the beneficial role of nutrition alone is limited (Schneider et al., 2009). A recent Cochrane review reported that nutritional therapy can reduce healthcare costs but overall the evidence from the studies is too heterogeneous and of limited quality for concluding whether malnutrition or its treatment helps in reducing re-admissions (Muscaritoli et al., 2016). Strandberg et al. (1985) reported that nutritional therapy, along with resistance training, improves muscle mass in older persons. A further study showed that high protein diet, provided by lean red meat, at 1.3 g/ kg/day increases lean tissue mass and muscle strength when it is complemented with resistive training in older women (Daly et al., 2014).

# Role of Cognitive Training in Preventing FCD in Sedentary Older Persons

Cognitive interventions have been used as countermeasures to prevent functional decline and even to promote functional outcomes, especially mobility-related improvements in sedentary seniors (Verghese et al., 2010). Indeed, usage of a computerized cognitive training (CCT) protocol, was shown in a recent study to prevent possible bed rest-associated decline in physiological function as well as cognitive changes (Goswami et al., 2015b; Marusic et al., 2016). The total bedrest confinement in the Goswami et al. (2015b) and Marusic et al. (2016) study—lasting up to 2 weeks—was carried out in healthy 55–65 years olds. CCT intervention was effective in improving cognitive function at the end of bedrest as well as improved dual-task walking condition and reduced gait variability (Marusic et al., 2016). The latter functional changes could potentially lead to a reduction in number of falls following prolonged bedrest. Moreover, CCT during bedrest confinement in older persons also prevented decreases in vascular function changes (Goswami et al., 2015b). Since vasculature function is an important contributor to bedrest-induced orthostatic intolerance, and/or cardiovascular diseases, Goswami et al. (2015b) propose that doing CCT in the immobilization state in older persons could lead to a reduction in the bedrest-induced effects on the vasculature.

# PERSPECTIVES

Bed rest studies, in addition to being directly relevant for spaceflight and often employed in spaceflight related studies, can be designed and used for investigation of a number of physiological conditions that arise due to aging or are related to a diverse set of situations involving surgery, injury, or chronic debilitating diseases. This example illustrates how a powerful synergy of information can be developed by connecting the physiological responses to microgravity and the consequences of bed-confinement, reflecting the parallels that can be drawn between aging, microgravity, and immobilization (Vernikos and Schneider, 2010).

There is also a need to better understand the pathways that link healthy status to frailty and functional physical and cognitive decline (FCD). This is important as previously independent senior persons can suddenly be hospitalized and become dependent. Thus, there is a need to reduce progression to frailty among the pre-frail and intervene in persons at risk to avoid frailty [European Innovative Partnership Active

# REFERENCES


Healthy Aging (EIP-AHA) report, 2013]. The interventions related to physical activity and life style modifications discussed in this review could be used for developing guidelines and strategies for preventing the negative consequences of bedrestconfinement and for reducing frailty and falls in older persons. Resistive vibration exercises—which have been shown to maintain muscle strength and function in ground based analogs of spaceflight—could be proposed for bed-confined older persons (Schneider et al., 2009). This is an important example of how evidence from ground based analogs of spaceflight could be used in geriatrics. The integration of knowledge of human physiology under conditions of microgravity and changes in physiology due to the aging process allows for each of these physiological conditions to provide new insights into the other and allows for development of new intervention strategies.

# AUTHOR CONTRIBUTIONS

The author confirms being the sole contributor of this work and approved it for publication.

# ACKNOWLEDGMENTS

I wish to thank Dr. Jerry Joseph Batzel of the Medical University of Graz for carefully going through the manuscript and for his useful suggestions.


**Conflict of Interest Statement:** The author declares that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

The reviewer SM and handling Editor declared their shared affiliation, and the handling Editor states that the process nevertheless met the standards of a fair and objective review.

Copyright © 2017 Goswami. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Impacts of Simulated Weightlessness by Dry Immersion on Optic Nerve Sheath Diameter and Cerebral Autoregulation

Marc Kermorgant <sup>1</sup> , Florian Leca<sup>2</sup> , Nathalie Nasr 1, 3, Marc-Antoine Custaud<sup>4</sup> , Thomas Geeraerts 2, 5, Marek Czosnyka6, 7, Dina N. Arvanitis <sup>1</sup> , Jean-Michel Senard1, 3 and Anne Pavy-Le Traon1, 3 \*

<sup>1</sup> UMR Institut National de la Santé et de la Recherche Médicale 1048, Institut des Maladies Métaboliques et Cardiovasculaires, Toulouse, France, <sup>2</sup> Department of Anesthesiology and Intensive Care, University Hospital of Toulouse, Toulouse, France, <sup>3</sup> Department of Neurology and Institute for Neurosciences, University Hospital of Toulouse, Toulouse, France, <sup>4</sup> BNMI, UMR Institut National de la Santé et de la Recherche Médicale 1083, UMR Centre National de la Recherche Scientifique 6214, Centre de Recherche Clinique, University Hospital of Angers, Angers, France, <sup>5</sup> Toulouse NeuroImaging Center, UMR 1214, Institut National de la Santé et de la Recherche Médicale, Université Toulouse III-Paul Sabatier, Toulouse, France, <sup>6</sup> Brain Physics Laboratory, Division of Neurosurgery, Department of Clinical Neurosciences, Cambridge University Hospital, Cambridge, United Kingdom, <sup>7</sup> Institute of Electronic Systems, Warsaw University of Technology, Warsaw, Poland

Edited by: Nandu Goswami, Medical University of Graz, Austria

### Reviewed by:

Jie Liu, Tangdu Hospital, Fourth Military Medical University, China Antonio Longo, Università degli Studi di Catania, Italy

> \*Correspondence: Anne Pavy-Le Traon pavy-letraon.a@chu-toulouse.fr

#### Specialty section:

This article was submitted to Integrative Physiology, a section of the journal Frontiers in Physiology

Received: 12 July 2017 Accepted: 25 September 2017 Published: 12 October 2017

#### Citation:

Kermorgant M, Leca F, Nasr N, Custaud M-A, Geeraerts T, Czosnyka M, Arvanitis DN, Senard J-M and Pavy-Le Traon A (2017) Impacts of Simulated Weightlessness by Dry Immersion on Optic Nerve Sheath Diameter and Cerebral Autoregulation. Front. Physiol. 8:780. doi: 10.3389/fphys.2017.00780 Dry immersion (DI) is used to simulate weightlessness. We investigated in healthy volunteers if DI induces changes in ONSD, as a surrogate marker of intracranial pressure (ICP) and how these changes could affect cerebral autoregulation (CA). Changes in ICP were indirectly measured by changes in optic nerve sheath diameter (ONSD). 12 healthy male volunteers underwent 3 days of DI. ONSD was indirectly assessed by ocular ultrasonography. Cerebral blood flow velocity (CBFV) of the middle cerebral artery was gauged using transcranial Doppler ultrasonography. CA was evaluated by two methods: (1) transfer function analysis was calculated to determine the relationship between mean CBFV and mean arterial blood pressure (ABP) and (2) correlation index Mxa between mean CBFV and mean ABP.ONSD increased significantly during the first day, the third day and the first day of recovery of DI (P < 0.001).DI induced a reduction in Mxa index (P < 0.001) and an elevation in phase shift in low frequency bandwidth (P < 0.05). After DI, Mxa and coherence were strongly correlated with ONSD (P < 0.05) but not before DI. These results indicate that 3 days of DI induces significant changes in ONSD most likely reflecting an increase in ICP. CA was improved but also negatively correlated with ONSD suggesting that a persistent elevation ICP favors poor CA recovery after simulated microgravity.

Keywords: transcranial Doppler, cerebral autoregulation, optic nerve sheath diameter, intracranial pressure, dry immersion

# INTRODUCTION

Due to the difficulties to perform in-flight experiments in addition to restricted opportunities of spaceflight, models are used on Earth to simulate the effects of microgravity. Dry immersion (DI) is a model of ground-based simulated microgravity (Pavy-Le Traon et al., 2007) where the subject is immersed and separated from the water with waterproof fabric (Navasiolava et al., 2011).This

**82**

Kermorgant et al. Intracranial Pressure after Dry Immersion

method allows for a rapid fluid migration toward the upper body. DI also removes afferent signals from the support zones and triggers a drop in the activity of the postural muscle system (Gevlich et al., 1983; Navasiolava et al., 2011; Watenpaugh, 2016). Adaptation of the cardiovascular system to microgravity is complex and its underlying mechanisms are not thoroughly understood. Exposure to real or simulated microgravity induces a redistribution of body fluids toward the upper part of the body (Charles and Lathers, 1991). This cranial redistribution observed in astronauts after exposure to microgravity is likely responsible for the elevated intracranial pressure (ICP) reported after longduration flights (Nelson et al., 2014). The determination of the optic nerve sheath diameter (ONSD) can be used as an indirect marker of increased ICP (Geeraerts and Dubost, 2009; Dubost et al., 2011) and changes in ONSD have been shown to be a reliable surrogate of changes in ICP. Visual impairment has been reported in some astronauts exposed to long duration spaceflights. Indeed, some changes such an increase in ONSD, posterior globe flattening and optic nerve protrusion suggest a potential intracranial hypertension (Kramer et al., 2012), resulting in the Vision Impairment and Intracranial Pressure syndrome (Nelson et al., 2014).Therefore, studies are currently performed to examine and understand the effect of exposure to microgravity on ICP. Exposure to microgravity may also affect cerebral autoregulation (CA). Impaired CA has been proposed as a contributing factor to orthostatic intolerance reported after real or simulated microgravity. However, in-flight data on CA are controversial. An impaired CA could induce a reduction in the phase frequency between cerebral blood flow (CBF) and blood pressure and may lead to presyncope (Blaber et al., 2011). It has been shown that astronauts which spent long-duration flights on the International Space Station had an impaired CA with a decrease in cerebrovascular CO<sup>2</sup> reactivity (Zuj et al., 2012). In contrast, a study performed on astronauts in a 1 and 2-week spaceflight showed an improved CA with a decreased low frequency (LF) gain (Iwasaki et al., 2007).

The aim of the study was to assess the changes in ONSD after exposure to simulated microgravity by DI and to determine whether these changes may impact CA.

# MATERIALS AND METHODS

# Subjects

Twelve healthy male volunteers participated in this study. None of the participants were smokers or took any medical treatment or drugs. All participants were informed about clinical assessment and gave their written consent. Participants selected for the experiments did not exhibit acute or chronic pathologies, which could affect the physiological data. The volunteers had normal clinical and paramedical examination and laboratory tests (hematology and blood chemistry). This Clinical Trial was conducted in accordance with the principles laid down by the 18th World Medical Assembly (Helsinki, 1964) and approved by the Ethics Committee (CPP Sud-Ouest Outre-Mer I) and the French Health Authorities. The study was conducted by the Institute for Space Medicine and Physiology (MEDES-IMPS) in Toulouse, France.

# General Protocol

The volunteers ensued an ambulatory control period (BDC), 3 days of DI (DI 1–DI 3) and 2 days of recovery (R 0– R+1). Except in DI periods, all subjects remained in ambulatory conditions. During DI, the subject was immersed slowly into the water in supine position (**Figure 1**). The water temperature was automatically set to 32–34.5◦C (thermoneutral) and adjusted if needed for subject's comfort. The study was carried out in a quiet room and the air temperature was approximately 24◦C. All subjects remained continually under medical observation. The folds of fabric could be moved easily for the different recordings without affecting the experimental conditions. Once a day and during 15 min, the subject was allowed to get out from the bath for hygiene procedures. The volunteers woke up at 6:30 a.m. and the light was switched off at 11:00 p.m. Each subject had a daily medical follow-up including ABP and heart rate (HR) measurements by permanent staff of MEDES. The flow chart is represented **Table 1**.

# Plasma Volume Measurements

The plasma volume was estimated by CO-rebreathing method (SpiCO <sup>R</sup> ; Blood tec, Bayreuth, Germany) in the morning (before breakfast) in resting supine position just before the first day of DI (DI 1) and immediately after the end of DI period, just before standing (R 0).

# Ocular Ultrasonography

Ocular examination was realized at rest in supine position during BDC, DI 1, DI 3, and R+1 by investigators trained for ocular ultrasonography. A thick layer of gel was applied over the closed upper eyelid. The probe was placed on the gel in the temporal area of the eyelid and adjusted to obtain an appropriate

FIGURE 1 | Dry immersion experiment.

TABLE 1 | Study flow chart.


BDC, baseline data collection; DI, dry immersion; R, recovery.

display of the optic nerve into the globe. The prospection was realized in a two-dimensional mode and ONSD was measured 3 mm behind the ocular globe. Right and left optic nerves were assessed and two measures were realized for each eye: a first measure in the transverse plane (horizontal probe) and a second measure in the sagittal plane (vertical probe). The final measure corresponds to the average of the four measures: horizontal right eye, vertical right eye, horizontal left eye and vertical left eye. All the measures of ONSD were validated by an expert (Thomas Geeraerts) blinded from the subject condition. The subjects were separated into 2 groups: "good recovery" and "poor recovery" groups. To define these groups, an arbitrary cut-off value ONSD was established when subjects recovered ONSD values below 20% between BDC and R+1.

# Transcranial Doppler Ultrasonography and Blood Pressure Measurements

The assessment of CA was realized in supine position and performed during the morning at BDC and R+1. A 2-MHz Doppler probe (EZ-DOP, DWL, Germany) maintained by a headset was placed close to the temporal window in order to obtain signal from the right middle cerebral artery (MCA) and thus to assess relative CBF changes. The MCA was insonated unilaterally at a depth of 50–55 mm. The continuous ABP was non-invasively monitored by a photoplethysmographic monitor (Nexfin–B Meye, the Netherlands).The signals, cerebral blood flow velocity (CBFV) and arterial blood pressure (ABP) were synchronized, acquired with Biopac MP 150 and visualized on the screen of a PC.

# Analysis of Cerebral Autoregulation by Transfer Function Analysis

Beat-by-beat mean ABP and CBFV were linearly interpolated and resampled at 4 Hz for spectral analysis. Using fast Fourier transform with 50% superposition of segments (Welch algorithm), the mean CBFV and mean ABP time series, beforehand preprocessed, were transformed from the time domain to the frequency domain. A length of 100 s was chosen for data segments and these segments were passed through a Hanning window. The transfer function analysis is a mathematic model of the relationship between changes in input and output signals, respectively beat-to-beat mean ABP and mean CBFV. To predict to what extent ABP has an influence on CBFV, a cross-spectral analysis method was applied. This analysis enabled to obtain three parameters: coherence, gain and phase. The coherence function measured the fraction of output power that was explained by the input power at a given frequency. The coherence was between values 0 and 1. A value close to 1 indicated, a strong linear relationship between the two signals with high signal-to-noise ratio, whereas the coherence approximating with values near zero may suggest a nonlinear relationship, a low signal-to-noise ratio or other variables influencing variables. To validate gain and phase values, it is necessary to obtain coherence value over 0.5 to ensure measures robustness (Claassen et al., 2016).Over a specified frequency range, the gain reflects the relative amplitude between the changes in the two ABP and CBFV signals and the phase is considered as temporal relation between these signals (Zhang et al., 1998). The recordings with phase shift wrap-around were corrected by adding 2 π. The mean values of the transfer function (coherence, phase and gain) were calculated in very low frequency (VLF: 0.02–0.07 Hz), low frequency (LF: 0.07–0.20 Hz) and high frequency (HF: 0.20–0.35 Hz) ranges as previously defined (Zhang et al., 1998, 2009). The mechanisms of CA can be considered to reflect a high-pass filter that dampens slow fluctuations of blood pressure but allows for passing through of rapid oscillations, such as the pulsatile signals of the blood pressure waves. To assess dynamic CA, the previously calculated mean values of the transfer function were assessed in the VLF and LF bands as a measure of the transmission of blood pressure fluctuations on CBFV (Zhang et al., 1998).These signals were processed with the PDL software (Notocord Systems, France).

# Analysis of Cerebral Autoregulation by the Autoregulatory Index Mxa

CA was also evaluated using the correlation coefficient Mxa, established from the spontaneous variations of mean ABP and mean CBFV. The acquisition of these parameters was described previously. The Mxa index was assessed as follows: firstly, mean ABP and mean CBFV were calculated after specific signal filtering to reduce or remove the influence of noise or artifacts, according to the recent recommendations (Claassen et al., 2016); secondly, 60 consecutive 10-s periods were established to calculate the Pearson's correlation coefficients between mean ABP and mean CBFV and; thirdly, the resulting 60 Mxa correlation coefficient were finally averaged in order to obtain the autoregulatory index Mxa. A Mxa value close to 1 indicated that ABP variations affected changes in CBFV, thus determining a disturbed CA. A Mxa value nearby 0 stated that ABP variations did not impact CBFV variations, suggesting a normal CA (Czosnyka et al., 1996). The threshold of Mxa commonly used to characterize a CA impairment is >0.45 (Brady et al., 2010).

# Statistics

General hemodynamics parameters, transfer function analysis results, ONSD and Mxa data were expressed as mean ± SD. Paired-t test was used for comparisons of data between BDC and R+1. One-way repeated measures analysis of variance was used to compare ONSD values and general hemodynamics parameters and Dunnett method was used for post-hoc analysis. Unpaired-t test was performed to compare Mx values in "good recovery" and "poor recovery" groups at BDC-3 and R+1. Correlation Mxa-ONSD and coherence-ONSD were tested by Spearman rank correlation coefficient. All statistical analyses were performed with GraphPrism 7.00. Differences were considered as statistically significant when P < 0.05.

# RESULTS

The main characteristics of the 12 healthy male volunteers participating in this study are the following: mean ± SD at BDC, 32 ± 5 years; 177 ± 6 cm; 74 ± 8 kg.

The changes in mean systolic blood pressure (SBP), diastolic blood pressure (DBP), heart rate (HR) were measured at rest in supine position between BDC and R+1. The plasma volume measurements were realized between DI 1 and R 0. We noted no significant changes either in SBP or DBP whereas HR increased after DI (P < 0.001). DI induces a significant decrease (17%) in plasma volume (P < 0.001) in all subjects (**Table 2**).

At DI 3, measurements were only performed on 10 subjects (2 subjects were excluded for technical problems). ONSD value at BDC was the following: 4.64 ± 0.40 mm. In all subjects ONSD increased significantly by 28% during DI 1 (5.94 ± 0.67 mm; P < 0.001), by 30% during DI 3 (6.01 ± 0.49 mm; P < 0.001) and by 22% at R+1 (5.66 ± 0.69 mm; P = 0.002) (**Figure 2A**). As previously described, the 12 subjects were then allocated in "good recovery" group and in "poor recovery" group. In "good recovery" group, ONSD increased significantly by 27% during DI 1 (6.15 ± 0.59 mm; P = 0.007), by 21% during DI 3 (5.90 ± 0.41 mm; P < 0.001) and by 6% after R+1 (5.17 ± 0.47 mm; P = 0.025) vs. BDC (4.86 ± 0.30 mm). During R+1, ONSD was reduced compared with DI 1 (P = 0.016) and DI 3 (P = 0.002) (**Figure 2B**). In "poor recovery" group, ONSD increased significantly by 30% during DI 1 (5.73 ± 0.72 mm; P = 0.010), by 40% during DI 3 (6.18 ± 0.62; P = 0.007) and remained elevated by 40% during R+1 (6.16 ± 0.50 mm; P < 0.001) vs. BDC (4.41 ± 0.36 mm). There was no significant difference between DI 1 and R+1 and DI 3 and R+1 (respectively P = 0.254 and P = 0.410) (**Figure 2C**).

In VLF bandwidth, the gain and phase could not be taken into account because the coherence value was below 0.5. Phase shift was significantly elevated in LF bandwidth at R+1 compared with BDC (P = 0.014). None of the others transfer function analysis parameters were significantly different in VLF, LF, and HF components between BDC and R+1 (**Table 3**). There was no significant difference between "good recovery" and "poor recovery" groups (data not shown).

The Mxa index was significantly reduced at R+1 compared to BDC-3 (P = 0.009). After DI, Mxa was significantly decreased in "good recovery" group (P = 0.019) whereas no significant difference was noted in "poor recovery" group (P = 0.263) (**Table 4**). Mx values did not significantly different between "good recovery" and "poor recovery" groups both at BDC-3 (P = 0.088) and R+1 (P = 0.083) (data not shown).


DBP, diastolic blood pressure; DI, dry immersion; HR, heart rate; SBP, systolic blood pressure.

FIGURE 2 | Changes in optic nerve sheath diameter after dry immersion. Changes in optic nerve sheath diameter (ONSD) before (BDC), the first day (DI 1), the third day (DI 3) and after (R+1) dry immersion in all subjects (A), "good recovery" group (B) and "poor recovery" group (C). Individuals points with the changing curves are represented. P < 0.05 vs. BDC; P < 0.01 vs. BDC; P < 0.001 vs. BDC.


TABLE 3 | Transfer function analysis of cerebral autoregulation.

Data are expressed as mean ± standard deviation. BDC, baseline data collection; HF, high frequency; LF, low frequency; R+1, 1 day after dry immersion; VLF, very low frequency; \*P < 0.05.


Data are expressed as mean ± standard deviation. BDC, baseline data collection; R+1, 1 day after dry immersion.

Three subjects have been excluded due to the presence of several artifacts.

At R+1, Mxa was strongly correlated with ONSD (n = 11; P = 0.018) but not at BDC (n = 11; P = 0.193). One subject has been excluded due to the presence of several artifacts (**Figures 3A,B**). There was no relationship between transfer function coherence in LF band and ONSD at BDC (n = 9; P = 0.613); whereas, at R+1, a high positive correlation was observed (n = 9; P = 0.028). One subject has been excluded due to the presence of several artifacts and two others subjects because the coherence value was below 0.5 (**Figures 3C,D**).

# DISCUSSION

Our results indicate that DI: (1) could induce an ONSD enlargement similar to those observed in intracranial hypertension, (2) DI could improve CA, and (3) a poorer recovery of ONSD would be related to a less improvement in CA in supine position. To our knowledge, this is the first study assessing the relation between changes in ICP and CA after DI.

The reduction of the plasma volume is the result of a rapid fluid shift toward the upper body. These results are usually described in simulated microgravity (Vernikos et al., 1993). This redistribution may have other consequences and especially on ICP. Previous studies showed that the assessment of ONSD may be a good non-invasive method for detecting an elevated ICP (Geeraerts et al., 2007; Geeraerts and Dubost, 2009; Dubost et al., 2011). Our experiments reveal that DI induces an increase in ONSD with a fluctuating return to basic values. Our values are consistent to those found in other studies where 27 astronauts were exposed to microgravity (Kramer et al., 2012). A previous study demonstrated that ONSD values above 5.82 mm could reflect intracranial hypertension with a 90% probability (Geeraerts et al., 2008). Moreover, our ONSD values would be equivalent to an elevation around of 20 mmHg in ICP while the normal range is between 7 and 15 mmHg. Similar findings were found on intracranial hypertension (Geeraerts et al., 2008; Soldatos et al., 2008). One of the main factors of these changes induced by real and simulated microgravity exposure could be mainly driven by the thoraco-cephalic fluid shift. Consistently, it has been demonstrated in spaceflight that cephalad shift fluids could induce an elevated ICP (Leach et al., 1991; Heer and Paloski, 2006). However, the kinetics and the exact mechanisms involved remain unknown. This fluid shift would be the main causal factor of the ONSD enlargement and one of the underlying mechanisms would originate from direct transmission of elevated subarachnoid pressure from the intracranial to intraocular compartment through the perioptic subarachnoid space (Mader et al., 2011). Nevertheless, a recent study performed in astronauts showed that ICP measured directly, did not rise during parabolic flights (20 s microgravity period) (Lawley et al., 2017); but in contrast to our study, the ICP measurements were obtained in 90 degrees seated upright posture and obtained at the level of external acoustic meatus, which could have modified the baseline ICP values.

Most of the results of the literature are contradictory since some studies showed an unmodified, impaired or enhanced CA after actual or simulated microgravity exposure. However, the assessment of CA has never been studied during DI. In our study, we used two reference tests to assess CA the transfer function analysis and the autoregulatory index; these two methods provide information both in frequency (transfer function analysis) and time (autoregulatory index) domains. The concordant results with these two techniques strengthen our findings. The decrease in Mxa would mean that 3 days of DI could improve CA. Indeed, an increase in Mxa values reflects an alteration in CA and the threshold value of Mxa characterizing this impairment is >0.45 (Brady et al., 2010). The phase relationship between mean CBFV and mean ABP can be assessed to determine the state of CA. Nevertheless, subject's position may impact on transfer function analysis especially gain and phase functions. Indeed, significant differences could be encountered if experiments are performed in supine or seated position (Saul et al., 1989). It was previously shown that in healthy subjects, a positive phase shift was observed; whereas in patients with autoregulatory disorders, a negative phase was found (Diehl et al., 1995). In our findings, phase shift increases after DI which would mean a preserved or improved CA. These results confirm our findings for Mxa

values. Moreover, a previous study reported that during shortduration spaceflight, CA could be improved in astronauts. In the LF component, gain was significantly reduced after few weeks in space compared with preflight values (Iwasaki et al., 2007). The conclusions made in our study should be done carefully. Each technique used to assess CA is different and expressed different components of the cerebral pressure-flow relationship (Tzeng et al., 2012).

One assumption is that during the simulated microgravity, relatively applicable to spaceflight, could raise the responsiveness of cerebral vascular smooth muscle to changes of transmural pressure (Iwasaki et al., 2007). It has also been shown in 14 men that the dynamic CA was improved by a reduced transfer function gain in LF range during a 2-week spaceflight (Ogawa et al., 2009). A study showed that cerebrovascular resistance slope did not significantly differ before, during and after a 7-day head-down bed rest, depicting no significant alteration of CA in 8 healthy women (Pavy-Le Traon et al., 2002). Another study demonstrated that an acute head-down tilt test in 10 healthy subjects did not modify cross-spectral analysis parameters (Cooke et al., 2003). One explanation for differences found in CA would come from, in part, time duration experience. Short term duration could enhance CA while the long term experiences may alter CA. Indeed, a previous study showed that long duration spaceflight impaired dynamic cerebrovascular autoregulation accompanied by a reduction in cerebrovascular CO<sup>2</sup> reactivity (Zuj et al., 2012).

Coherence could be used to assess dynamic CA as suggested in a previous study. The intrinsic characteristic of cerebrovascular resistance implies that coherence values should be high in impaired CA situations and alternatively low in normal conditions (Panerai, 1998). Usually, CA is evaluated in VLF (0.02–0.07 Hz) and LF bandwidths (0.07–0.20 Hz) as previously described (Claassen et al., 2016). However, in our study, the VLF coherence value was below 0.5 so the gain and phase values were not sufficiently robust results. In our results, DI does not affect coherence but a positive correlation between coherence-ONSD is found, in addition to a correlation between Mx-ONSD. An altered CA is associated with head-injury with the existence of an elevated ICP (Czosnyka et al., 2001). Indeed, cerebral perfusion pressure (CPP) depends on two factors, ABP and ICP and their relationship can be established as follows: CPP = ABP – ICP. So, an elevation in ICP would imply a reduction in CPP which may provoke a vasodilation of cerebral vessels and probably a reduction in CBF (Rangel-Castillo et al., 2008). In these healthy volunteers, the enlargement of ONSD

was negatively correlated with CA improvement. Little is known about the involvement of an elevated ICP on cerebrovascular remodeling and the regulation of CBF. However, a study realized in astronauts showed that cerebrovascular autoregulation could be impaired and leading to a potential syncope after their return to Earth (Blaber et al., 2011; Zuj et al., 2012).

## Study Limitations

The study was performed with small numbers healthy subjects (n = 12) which could dampen the statistical significance of our results.

The use of transcranial Doppler necessitates the diameter of the MCA for an appropriate assessment of CBFV. This potential limitation is inherent to all CA studies using transcranial Doppler (Nasr et al., 2014).

The transfer function analysis of CA could induce bias in the findings. Phase and gain are well known to be affected by some variables (CO<sup>2</sup> changes, respiration rate. . . ). The choice of supine posture may also affect transfer function analysis; in particular gain and phase shift in the LF band compared with upright position. An increase in the sympathetic tone was described in the upright posture (Saul et al., 1989).

Despite the strong correlation existing between our dynamic CA metrics, established conclusions should be done carefully. Most dynamic CA metrics are not specifically related to each other. Each technique used would reflect different components of the cerebral pressure-flow relationship (Tzeng et al., 2012).

In conclusion, our study demonstrates for the first time that 3-day DI leads to an increase in ONSD, and enhanced the

# REFERENCES


CA, with the CA improvement reversely related to the increase of ONSD. However, the kinetics underlying this process is unknown.

# ETHICS STATEMENT

This study was carried out in accordance with the recommendations of "Comité de Protection des Personnes/CPP Sud-Ouest Outre-Mer I" with written informed consent from all subjects. All subjects gave written informed consent in accordance with the Declaration of Helsinki. The protocol was approved by the "Agence Française de Sécurité Sanitaire des Produits de Santé."

# AUTHOR CONTRIBUTIONS

FL, TG, MAC, and AP carried out the experiments. MK, FL, NN, MAC, TG, MC, DA, JS, and AP evaluated results and wrote the paper.

# FUNDING

This experiment was proposed and sponsored by Centre National d'Etudes Spatiales (CNES).

# ACKNOWLEDGMENTS

The authors thank the volunteers who participated in this research.

epidural blood patch: a preliminary report. Br. J. Anaesth. 107, 627–630. doi: 10.1093/bja/aer186


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2017 Kermorgant, Leca, Nasr, Custaud, Geeraerts, Czosnyka, Arvanitis, Senard and Pavy-Le Traon. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Multi-System Deconditioning in 3-Day Dry Immersion without Daily Raise

Steven De Abreu1†, Liubov Amirova1, 2†, Ronan Murphy <sup>3</sup> , Robert Wallace<sup>3</sup> , Laura Twomey <sup>3</sup> , Guillemette Gauquelin-Koch<sup>4</sup> , Veronique Raverot <sup>5</sup> , Françoise Larcher <sup>6</sup> , Marc-Antoine Custaud1, 7 \* and Nastassia Navasiolava<sup>7</sup>

<sup>1</sup> Mitovasc, UMR Institut National de la Santé et de la Recherche Médicale 1083, Centre National de la Recherche Scientifique 6015, Université d'Angers, Angers, France, <sup>2</sup> Russian Federation State Research Center, Institute of Biomedical Problems, Russian Academy of Sciences, Moscow, Russia, <sup>3</sup> Center for Preventive Medicine, School of Health and Human Performance, Dublin City University, Dublin, Ireland, <sup>4</sup> Centre National d'Etudes Spatiales, Paris, France, <sup>5</sup> Hospices Civils de Lyon, Lyon, France, <sup>6</sup> Laboratoire de Biochimie, Centre Hospitalier Universitaire d'Angers, Angers, France, <sup>7</sup> Centre de Recherche Clinique, Centre Hospitalier Universitaire d'Angers, Angers, France

#### Edited by:

Olivier White, INSERM U1093, Université de Bourgogne Franche Comté, France

#### Reviewed by:

Matteo Maria Pecchiari, Università degli Studi di Milano, Italy Alessandro Tonacci, Istituto di Fisiologia Clinica (CNR), Italy

> \*Correspondence: Marc-Antoine Custaud macustaud@chu-angers.fr

† These authors have contributed equally to this work.

#### Specialty section:

This article was submitted to Integrative Physiology, a section of the journal Frontiers in Physiology

Received: 17 July 2017 Accepted: 28 September 2017 Published: 13 October 2017

#### Citation:

De Abreu S, Amirova L, Murphy R, Wallace R, Twomey L, Gauquelin-Koch G, Raverot V, Larcher F, Custaud M-A and Navasiolava N (2017) Multi-System Deconditioning in 3-Day Dry Immersion without Daily Raise. Front. Physiol. 8:799. doi: 10.3389/fphys.2017.00799 Dry immersion (DI) is a Russian-developed, ground-based model to study the physiological effects of microgravity. It accurately reproduces environmental conditions of weightlessness, such as enhanced physical inactivity, suppression of hydrostatic pressure and supportlessness. We aimed to study the integrative physiological responses to a 3-day strict DI protocol in 12 healthy men, and to assess the extent of multi-system deconditioning. We recorded general clinical data, biological data and evaluated body fluid changes. Cardiovascular deconditioning was evaluated using orthostatic tolerance tests (Lower Body Negative Pressure + tilt and progressive tilt). Metabolic state was tested with oral glucose tolerance test. Muscular deconditioning was assessed via muscle tone measurement.

Results: Orthostatic tolerance time dropped from 27 ± 1 to 9 ± 2 min after DI. Significant impairment in glucose tolerance was observed. Net insulin response increased by 72 ± 23% on the third day of DI compared to baseline. Global leg muscle tone was approximately 10% reduced under immersion. Day-night changes in temperature, heart rate and blood pressure were preserved on the third day of DI. Day-night variations of urinary K<sup>+</sup> diminished, beginning at the second day of immersion, while 24-h K<sup>+</sup> excretion remained stable throughout. Urinary cortisol and melatonin metabolite increased with DI, although within normal limits. A positive correlation was observed between lumbar pain intensity, estimated on the second day of DI, and mean 24-h urinary cortisol under DI. In conclusion, DI represents an accurate and rapid model of gravitational deconditioning. The extent of glucose tolerance impairment may be linked to constant enhanced muscle inactivity. Muscle tone reduction may reflect the reaction of postural muscles to withdrawal of support. Relatively modest increases in cortisol suggest that DI induces a moderate stress effect. In prospect, this advanced ground-based model is extremely suited to test countermeasures for microgravity-induced deconditioning and physical inactivity-related pathologies.

Keywords: modeled weightlessness, physical inactivity, supportlessness, cardiovascular deconditioning, glucose intolerance, muscle tone, day-night variations, kaliuresis

Spaceflight induces physiological multi-system deconditioning which may impact astronauts efficiency and create difficulties upon their return to normal gravity (Nicogossian et al., 1993). Understanding the underlying mechanisms of this process and enhancement of countermeasures remains a challenge and major priority for manned space programs. Moreover, resultant data and experience may be used to resolve common earth-based chronic healthcare problems related to increased physical inactivity, for example poststroke patients, bedridden, paralyzed or immobilized subjects, sedentary people, aging etc. Experimental opportunities during actual spaceflight being limited, the appeal of ground-based simulations is obvious and paramount (Pavy-Le Traon et al., 2007). Dry immersion (DI) is one such prolonged microgravity model. It accurately reproduces most physiological effects of microgravity, including centralization of body fluids and hypokinesia (Kozlovskaia, 2008; Navasiolava et al., 2011a; Watenpaugh, 2016). The benefit of DI, compared to more widely-known and traditional head-down bed rest (HDBR) technique, is support unloading ("supportlessness"), a state akin to weightlessness, with water hydrostatic pressure equally distributed over the body surface, providing conditions similar to complete lack of structural support (Grigor'ev et al., 2004; Navasiolava et al., 2011a). DI promotes rapid gravitational deconditioning, exceeding for some systems (i.e. for neuromuscular system) the deconditioning induced by spaceflight itself (Navasiolava et al., 2011a). However, this DI method, developed and widely used in Russia, is not yet routine elsewhere. Moreover, all DI experiments performed until now included a short daily raise for personal hygiene procedures and weighing (Navasiolava et al., 2011a). Importantly, literature shows that this type of short daily orthostatic stimulation could act as a countermeasure (Greenleaf, 1984; Vernikos et al., 1996). Therefore, in order to eliminate this aberration, our novel DI protocol did not permit subjects to rise at all for 3 days, and a −6 ◦ head down position was maintained when the subjects were out of water, as is observed in strict bedrest protocols. Hence, this study is the first DI protocol specially conceived to assess integrative aspects of "strict" DI impact.

We aimed to study integrative response to 3-day strict DI, and to assess the extent of multi-system deconditioning with regard to cardiovascular, metabolic, muscular system, daynight changes in renal excretion, and adaptive capacities/ stress effect.

# MATERIALS AND METHODS

A total of twelve healthy non-athletic men aged 26 to 39 year. o. (age 32 ± 1.4 yr, weight 75 ± 2 kg, height 178 ± 2 cm, BMI 23.6 ± 0.4 kg/m<sup>2</sup> , maximal oxygen uptake VO˙ <sup>2</sup>max 39 ± 1.1 mL/min/kg) were included in the study. Subjects had no history of cardiovascular or other chronic diseases, and were not taking medication prior to the experiment. All subjects were informed about the experimental procedures and gave their written consent. The experimental protocol conformed to the standards set by the Declaration of Helsinki and was approved by the local Ethic Committee (CPP Sud-Ouest Outre-Mer I, France) and French Health Authorities (n◦ ID RCB: 2014-A 00904-43).

# General Protocol

The study was conducted at the MEDES space clinic, Toulouse, France. The experimental protocol lasted for 8 days: 3 days of ambulatory baseline measurement before immersion (B-3, B-2, B-1), 3 days (72 h) of dry immersion (DI1, DI2, DI3) and 2 days of ambulatory recovery (R0, R+1). Two subjects in two separate baths underwent DI simultaneously. Thermoneutral water temperature (32.5–33.5◦C) was continuously maintained. Light-off period was set at 23:00–07:00. General discomfort and lumbar pain under immersion were evaluated using a visual analog scale. Daily hygiene, weighing and some specific measurements required extraction from the bath. During these short out-of-bath periods, subjects maintained the −6 ◦ headdown position. Total out-of-bath supine time for the 72 h of immersion was 4.7 ± 0.16 h. Otherwise, during DI, subjects remained immersed in a supine position for all activities and were continuously observed by video monitoring. Blood pressure, heart rate (HR) and tympanic body temperature were measured twice daily at 07:00 and 19:00. Body weight was measured daily at 07:00. Onset and end of immersion both occurred at 09:00, therefore morning measurements on DI1 were performed before immersion, and on day R0-still under immersion. Before, during and after DI, water intake was ad libitum (measured), and diet was the same for all participants and standardized to body weight in energy and nutrients. Daily caloric intake was approximately 2,820 kcal for baseline and recovery and 2,270 kcal for the immersion period. Daily intake for sodium and potassium was approximately 3–4 g. Daily nutrition is detailed in supplementary data (Table S1). The timeline schematic of the global protocol is outlined in **Figure 1**.

This 3-day DI allowed for several protocols on different domains performed by 8 research groups. Some studies have already been published to-date (Treffel et al., 2016a,b; Arbeille et al., 2017; Demangel et al., 2017).

# Lower Body Negative Pressure (LBNP)-Tilt Test

Tilt testing with combined lower body negative pressure (LBNP) was chosen as the accepted standard for measuring orthostatic tolerance (Protheroe et al., 2013). The measurement was conducted in the morning in a temperature-controlled room (22 ± 0.6◦C) at baseline on B-2 and immediately following DI on R0

**Abbreviations:** ALT, Alanine Aminotransferase; aMT6s, 6-sulphatoxymelatonin; ANOVA, analysis of variance; AST, Aspartate Aminotransferase; AUC, area under curve; B, baseline; BMI, body mass index; BNP, blood natriuretic peptide; BR, bed rest; DI, dry immersion; DPV (%), plasma volume percent change; GGT, Gamma-Glutamyl Transferase; Hb, hemoglobin; Hct, hematocrit; HDBR, head down bed rest; HOMA-IR, homeostasis model assessment-insulin resistance index; HR, heart rate; hs-CRP, high-sensitivity C-reactive protein; LBNP, lower body negative pressure; MCHC, mean corpuscular hemoglobin concentration; MCV, mean corpuscular volume; R, recovery; RBC, red blood cells; SV, stroke volume; VO˙ <sup>2</sup>max, maximal oxygen uptake.

(first rising after DI). The subject remained supine for 20 min, after which supine data were recorded for 5 min. The tilt-table was then rotated to 80◦ for 15 min. After that, LBNP was applied with steps of −10 mmHg every 3 min. The test was stopped at LBNP −60 mmHg or earlier upon appearance of pre-syncopal signs, request to stop, systolic blood pressure ≤ 80 mmHg, HR < 50 bpm or > 170 bpm.

During the LBNP-tilt test, finger blood pressure (Nexfin, BMeye, USA) and standard ECG (Biopac, ECG 100C, USA) were recorded continuously. Orthostatic tolerance time, heart rate (HR), blood pressure (systolic, diastolic), stroke volume (SV), total peripheral resistance were estimated. Stroke volume and total peripheral resistance were evaluated from the blood pressure wave using the modelflow <sup>R</sup> method (Beatscope <sup>R</sup> software, TNO, the Netherlands). The state of autonomic nervous system was estimated via power spectrum analysis of heart rate variability. Sympathetic index, an indicator of sympatho-vagal balance, was calculated as the ratio of low-to-high frequency spectral power. Spontaneous baroreflex sensitivity was estimated using online software (televasc.fr).

# Progressive Tilt Test

Progressive tilt was used to progressively stimulate the cardiovascular system (volume receptors for low angles and baroreceptors for higher angles) and vestibular system. A head-up tilt test with progressive angles (0, 20, 45, and 80◦ ) was performed at baseline on B-3 and on the second recovery day (R+1, 24 h following the completion of the DI protocol). After 10 min of supine rest recording, the subject was tilted at 20◦ (5 min), 45◦ (5 min) and 80◦ (5 min), and then returned to the horizontal position.

During the progressive tilt test, HR, systolic and diastolic blood pressure, total peripheral resistance, stroke volume, sympathetic index, spontaneous baroreflex sensitivity were calculated and recorded.

# Blood Studies

Antecubital venous blood samples were collected before (B-3, B-2, B-1), during (DI1-evening, DI2, DI3, R0) and after DI (R+1, R+2). Blood sampling was performed in the morning before breakfast, except for DI1 when it was done in the evening (10h of DI).

Plasma and serum samples were analyzed for electrolytes (Na+, K+, Cl−), glucose, proteins, albumin, urea, and creatinine concentrations, high-sensitivity CRP, insulin, leptin, triglycerides, total cholesterol, and HDL-cholesterol. LDLcholesterol was calculated using the Friedewald formula. Homeostasis model assessment-insulin resistance index (HOMA-IR) was calculated as fasting insulin concentration (µU/mL) × fasting glucose concentration (mmol/L)/22.5.

Additionally, on B-3 and R+1 blood count, GGT, ALT, AST, alkaline phosphatase, total bilirubin and prothrombin time were assessed. Hb, Hct, renin, aldosterone and blood natriuretic peptide (BNP) were assessed on B-3, R0 and R+1.

# 75-g Oral Glucose Tolerance Test

Glucose tolerance tests were performed in the morning on B-1 and DI3 (48 h of immersion). Blood glucose and insulin were measured after an overnight fast before and 30, 60, 90, and 120 min after a 75-g glucose intake (consumption of glucose solution drink within 5 min).

The total area under the blood glucose curve (AUC-Glu) and insulin curve (AUC-Ins) during the 75-g oral glucose tolerance test were calculated using the trapezoidal rule.

# Urine Sampling

Urine pools were collected over 12 h (07:00–19:00 pool for "day" and 19:00–07:00 pool for "night") throughout the protocol. Light on/off periods were not taken into account for urine collection, therefore "night" pools included 4 h of light-on (07:00–23:00) and 8 h of light-off (23:00–07:00). Urine volume was measured, and aliquots stored at −80◦C. Partial water balance, defined as the difference between consumed water and urine volume, was calculated. Urine samples were analyzed for electrolytes (Na+, K <sup>+</sup>), cortisol and 6-sulphatoxymelatonin (aMT6s, a hepatic metabolite of melatonin usually used as a proxy measure of melatonin level).

# Biochemical Analyses

Active renin analysis was performed using a chemiluminescence immunoassay on the Liaison analyzer (DiaSorin). Plasma aldosterone was determined by a competition radioimmunoassay using a commercially available RIA kit (Immunotech, Beckman Coulter). Urinary cortisol and aMT6s assays were carried out by radioimmunoassay. All other variables from blood and urine samples were evaluated using the Architect c16,000 automated clinical chemistry analyzer (Abbott). Minimal detectable levels for aldosterone and BNP were 10 ng/L, for hs-CRP–0.1 mg/L. Results that were less than the minimal detectable level, were taken for half minimal value.

# Blood and Plasma Volume Measurement

Blood volume and plasma volume were estimated using the optimized CO-rebreathing method (Schmidt and Prommer, 2005) in the morning before breakfast on DI1 just before the onset of immersion and on R0 immediately at the end of immersion, in supine position. Additionally, percent change in plasma volume on R0 and R+1 vs. B-3 was calculated using Hb and Hct count (Dill and Costill formula): DPV (%) = 100 × [HbB (1−0.01Hcti)]/ [Hbi (1−0.01HctB)] − 100, where HbB and HctB are baseline Hb and Hct levels, and Hbi and Hcti are Hb and Hct on days R0 and R+1, respectively.

# Muscle Tone

Mechanical characteristics of muscles were determined using a hand-held myotonometer (MyotonPRO; Myoton Ltd, Estonia) before (B-1), 6 h following onset (DI1) and on day 3 of immersion (DI3), and in recovery period on R0 (6 h following the end of DI) and on R+1.

During the tests, the subject was out-of-water in a relaxed supine position (dorsal and ventral decubitus). The testing end of the myotonometer was kept perpendicular to the muscle's projection in the middle of the muscle belly, at the same muscle point throughout the study. The device was used in multiscan mode, where one measurement corresponded to the mean of 5 mechanical taps. Leg measurements were taken on the left m. rectus femoris, m. tibialis anterior, m. gastrocnemius lateralis and m. soleus. Trunk measurements were taken on the right and left m. splenius, m. trapezius, m. longissimus cervicis, m. longissimus thoracis and lumbar portion of m. multifidus. Since trunk data showed no difference between the right and left sides (all p > 0.05), results from right and left sides were averaged. For technical reasons, the data of 2 subjects (subjects C and D) have not been analyzed.

# Statistical Analysis

The values are presented as mean ± SEM. Statistical analysis was performed with Prism6 GraphPad. Data were compared by a one-way and two-way repeated measures ANOVA followed by post hoc Bonferroni test. Relationships between data were examined using the Pearson correlation coefficient (r). Detailed correlations are given in supplemental data (Table S2).

# RESULTS

# General Data

HR, blood pressure and body temperature remained within normal limits throughout the protocol. Baseline values of blood pressure and temperature were slightly higher in the evening. On the third day of immersion HR, blood pressure and body temperature did not differ significantly from B-1 baseline level (**Figure 2**). Body weight had decreased by 1–2 kg on the third day of immersion (**Table 1**).

# Body Fluids

## Diuresis, Water Intake and Partial Water Balance

Data are shown in **Table 1**. Water intake set ad libitum was decreased as anticipated on DI1, due to the known reset of water balance to lower level (Navasiolava et al., 2011c), but also on DI2 and DI3. This was likely due to discomfort during subject urination in the water tank, causing a voluntary reduction of water intake. Diuresis remained unchanged on DI1 despite a 30% reduction in water intake, therefore partial water balance decreased up to 700–800 mL and became negative on the first day of DI. On DI2 and DI3, water intake remained approximately 30% reduced, together with decrease in diuresis. At the recovery stage, water intake returned to baseline values, while diuresis remained diminished, with a compensatory increase in water balance.

# Blood and Plasma Volume

Blood volume decreased by 11% on R0 (from 6.45 ± 0.20 L to 5.74 ± 0.17 L, p < 0.001), while total Hb mass diminished by 3% (from 887 ± 33 g to 861 ± 33 g, p = 0.009). In relation to plasma volume, the CO technique showed a 16 ± 2% decrease at DI3 compared to DI1, and the Dill and Costill estimation showed a 14 ± 2% decrease vs. B-3. Hence, there was no significant difference between these two types of measurements. There were no changes in plasma volume at R+1 compared to baseline. Preimmersion plasma volume expressed in ml/kg correlated with initial VO˙ <sup>2</sup>max (Pearson r = 0.67; p = 0.017), and percent reduction in plasma volume also correlated with initial VO˙ <sup>2</sup>max (Pearson r = −0.76; p = 0.006), i.e., fitter subjects had greater relative plasma volume at baseline and greater hypovolemia under DI.

# Volume-Regulating Hormones

### (Renin-Angiotensin-Aldosterone System and BNP)

A 25% increase in renin was observed prior to the end of DI (morning of R0). Twenty-four hours after DI3, renin was expectedly increased by two-fold, and aldosterone by 3-fold. Brain Natriuretic Peptide (BNP) was below the minimum detectable level in all subjects on DI3 and R+1 compared to 8.3 ± 1.6 ng/L on B-3 before DI (**Table 2**).

# Cardiovascular Deconditioning Orthostatic Tolerance

Presyncopal LBNP-tilt test revealed a pronounced decrease in orthostatic tolerance with orthostatic tolerance time drop from 27 ± 1 min–on baseline to 9 ± 2 min–on R0. Before DI all subjects tolerated the first stage of test which

TABLE 1 | Body weight, water intake, diuresis, and partial water balance.

mean ± SEM. No significant difference on DI3 vs. B-1. #p ≤ 0.05 vs. Morning.


Values are mean ± SEM; \*p ≤ 0.05 vs. B-1.

consisted of an 80◦ tilt for 15 min. However, immediately after DI, 9 subjects out of 12 were unable to accomplish this tilt. Therefore, orthostatic tolerance time for non-finishers averaged at 4.95 ± 0.7 min. No significant correlation between percent decrease in plasma volume and post-immersion orthostatic tolerance time was found (Pearson r = −0.36; p = 0.28).

Tolerance for different LBNP steps is shown in **Figure 3**. Before DI, all subjects finished 10 mmHg LBNP step, a half tolerated 40 mmHg step, and one (subject E) accomplished the last step of 60 mmHg. After DI, only 2 of the 12 subjects finished the first LBNP step, and only 1 subject (subject E) finished the step of 20 mmHg. On R+1, all subjects finished the progressive tilt test.

## Hemodynamic and Autonomic Responses to LBNP-Tilt and Progressive Tilt

Hemodynamic and autonomic responses to LBNP-tilt and Progressive tilt tests are presented in **Figures 4**, **5**. Pre-immersion supine baseline for diastolic blood pressure, HR, total peripheral resistance and sympathetic index was greater for LBNP-tilt compared to progressive tilt, most likely due to stress caused by using the LBNP equipment. Before DI, upright position provoked expected changes in central hemodynamics and


Values are mean ± SEM; \*p ≤ 0.05 vs. baseline.

cardiac autonomic neural control (increased blood pressure, HR, total peripheral resistance and sympathetic index; decreased stroke volume and baroreflex sensitivity). These changes were progressive with progressive tilt.

Post-immersion supine measurements on R0 showed significant increases in diastolic blood pressure, HR and total peripheral resistance, a 2-fold increase in sympathetic index, and a decrease in stroke volume and baroreflex sensitivity. On R+1, supine measurements did not differ significantly from pre-immersion levels.

Upright measurements showed a decrease in systolic blood pressure, stroke volume and baroreflex sensitivity, accompanied by pronounced tachycardia on R0. Total peripheral resistance and sympathetic index, which were already increased in supine, failed to further increase with orthostasis. On R+1, upright position was still accompanied by greater tachycardia. Interestingly, while tilt-induced sympathetic activation before DI increased progressively with verticalisation [reaching maximum at maximal angle of tilt (80◦ )], on R+1 the maximal sympathetic index was observed at 45◦ (difference between angles 45 and 80◦ before DI, p = 0.07; after DI, p = 0.8).

# Metabolism

### Blood Variables Relevant to Metabolism

All fasting blood variables relevant to metabolism remained within physiological limits (**Table 2**). DI increased insulin levels, while insulin resistance (HOMA-IR) increased by 43 ± 11% on R0 vs. baseline. Total cholesterol and LDL cholesterol fraction were moderately increased under immersion. Triglycerides and HDL remained nearly unchanged.

## Oral Glucose Tolerance Test

The response to oral glucose tolerance test is shown in **Figure 6**. Fasting glucose was beneath the pre-diabetic threshold (6.1 mmol/L) in all volunteers, and was not altered by DI. Glucose tolerance was compromised by DI in 11 subjects out of 12, and in 3 subjects this impaired glucose tolerance reached pre-diabetic level (more than 7.8 mmol/L 2 h after having a glucose drink). Net insulin response (AUC-Ins) increased by 72 ± 23% on DI3 compared to baseline. There was also a 14 ± 5% increase in the net glucose response (AUC-Glu), while incremental AUC for glucose response increased twofold.

# Muscle Tone

Muscle tone variations under DI are shown in **Figure 7**. Global leg muscle tone, proxy measured by "frequency" parameter, was decreased by approximately 10% under immersion. This decrease was immediate (seen 6 h following the onset of DI) and especially pronounced for m. rectus femoris (results for our subjects are also outlined in Demangel et al., 2017). The tone of m. gastrocnemius lateralis significantly decreased on DI3. The tone of superficial muscles of the neck and upper trunk (m. trapezius and m. splenius) was not significantly modified under immersion. Behavior of deep back muscle tone had cervicolombar gradient. Tone immediately dropped in the upper part (12% for m. longissimus cervicis), shifting to slight decrease on D3 in the middle (m. longissimus thoracis) and was not modified in the lower part (lumbar portion of m. multifidus). Six hours following the end of DI, muscle tone was completely restored.

# Day-Night Variations in Urinary Excretion Urinary Sodium

At baseline, daytime excretion exceeded night-time excretion by approximately 20%. On DI3, day-night variations in Na<sup>+</sup> excretion were preserved (**Figure 8A**).

# Urinary Potassium

At baseline, day-night differences in K<sup>+</sup> excretion were clearly marked and stable, with 2.5 fold greater excretion in daytime. There was no effect on day-night difference in K<sup>+</sup> excretion after DI1. At the start of DI2, night excretion tended to increase while day excretion tended to decrease, showing an important drop in day-night difference reaching significance on R0 (**Figure 8B**).

Interestingly, 24 h excretion was stable for K<sup>+</sup> throughout the protocol, whereas 24 h Na<sup>+</sup> excretion varied significantly within immersion, peaking at the onset of about 40% over baseline and drop at the end of DI of about 60% beneath baseline.

# Urinary Cortisol

Urinary cortisol had pronounced day-night oscillations. Cortisol levels tended to decrease in the first 10 h of DI, then to increase on the night of DI1, however, these changes were within physiological limits. Day-night differences in cortisol excretion were smoothed at the onset of immersion, but restored on DI2. Cortisol level recovered immediately after DI (**Figure 9A**).

### Urinary aMT6s

Urinary aMT6s had pronounced day-night oscillations, which tended to accentuate during immersion mainly due to the increased night release. On R0, day-night differences in aMT6s

secretion were diminished, suggesting readaptation to normal conditions (**Figure 9B**).

# General Discomfort and Pain during DI

In general, the majority of subjects described the discomfort level experienced under DI as 30–45 out of 100. Some subjects reported pronounced discomfort at nighttime. We observed important inter-subject variance in auto-reported discomfort level, varying from 0 to 88. There was no significant difference in discomfort level between the different days of DI. Eleven out of the 12 subjects reported moderate back pain under DI, which was predominantly localized in the lumbar area for 10 subjects. Lumbar pain intensity as estimated on DI2 was 3.8 ± 0.7 on the scale of 1–10. There was a positive correlation between lumbar pain intensity and mean 24-h urinary cortisol during DI (199 ± 24 mmol/24 h) (Pearson r = 0.64; p = 0.02). As well, we observed a positive correlation between mean discomfort level and mean 24-h urinary melatonin metabolite under DI (40.5 ± 5.4 µg/24 h) (Pearson r = 0.67; p = 0.018).

# Blood Electrolytes, Blood Count, Liver-Related Biochemistry, hs-CRP

Data are detailed in **Table 2**. Plasma electrolytes showed modest changes without clinical relevance. Blood count remained unchanged except for a modest but highly significant decrease in RBC volume, accompanied by slight increase in MCHC on R+1 vs. B-3. Blood GGT, ALAT, ASAT, alkaline phosphatase, total bilirubin, prothrombin time and serum albumin were not modified on R+1 vs. B-3. CRP levels were unaltered by DI.

# DISCUSSION

# Main Findings

Orthostatic tolerance time dropped from 27 ± 1 min to 9 ± 2 min after DI. Impairment in glucose tolerance was significantly pronounced. Net insulin response increased by 72 ± 23% on DI3 compared to baseline. Global leg muscle tone was reduced by approximately 10% with the DI protocol. Day-night changes in temperature, heart rate and blood pressure were preserved on the third day of DI. Day-night levels of urinary K<sup>+</sup> were reduced, beginning on the second day of immersion, while 24-h K <sup>+</sup> excretion remained stable. Urinary cortisol and melatonin metabolite levels increased with DI, although within normal limits. A positive correlation between lumbar pain intensity estimated on DI2 and mean 24-h urinary cortisol under DI was observed.

# Need for Strict DI without Daily Rise

In the present study, we observed pronounced cardiovascular deconditioning after 3 days of strict DI. This deconditioning was characterized by a predisposition to orthostatic intolerance and tachycardia.

# Orthostatic Intolerance

Seventy-five percent of our subjects were intolerant to orthostasis on R0 after 3 days of strict DI. In comparison with literature data using chi-square test, orthostatic intolerance rate after strict bed rest without countermeasures does not differ significantly from our findings: 5/11 (45%)-after 4 days (p = 0.15), 4/6 (67%)-after 14 days (p = 0.7), 5/9 (56%)-after 28, or 30 days (p = 0.35), and 4/7 (57%)-after 42 days of HDBR (p = 0.4) (Pavy-Le Traon et al., 1999).

However, the reported rate of orthostatic intolerance following non-strict DI seems much less-approximately 15 to 40% of subjects for 1 to 7 days of DI. Indeed, after 24 h of DI, 2 subjects out of 10 were reported to be intolerant to the orthostatic test on R0 (p = 0.01), after 1.5 days of DI–1 out of 6 (p = 0.02) (Navasiolava et al., 2011a), after 3 days of DI–1 out of 6 (p = 0.02) (Iwase et al., 2000) and 0 out of 4 (p = 0.009) (Miwa et al., 1997), after 5-day DI–4 out of 7 (p = 0.4) (one was already intolerant before DI) (Coupé et al., 2013), and after 7 days of DI–2 out of 6 (p = 0.09) (Iarullin et al., 1987) and 1 out of 8 (p = 0.006) (Navasiolava et al., 2011b). The main differentiator of our DI was the maintenance of supine position throughout. Our data suggest that daily short periods out of the water tank in sitting or standing positions represent a powerful countermeasure against orthostatic intolerance induced by

DI. Periodic short gravitational stimuli appear to be effective countermeasures, maintaining gravitational tolerance. Vernikos et al. (1996) demonstrated a very efficient preventive effect of a short period in a standing position (for 2 h daily) against orthostatic intolerance after 4 days of HDBR. The beneficial effect of LBNP for 20 min/day during both spaceflight and bed rest was discussed in a review by Clement and Pavy-Le Traon (2004). They cite the evidence that daily centrifugation for 30– 45 min reduces most of the physiological markers associated with orthostatic intolerance. Thus, a daily short period of orthostatic stress is sufficient to reduce the risk of orthostatic intolerance after simulated weightlessness. Daily short term gravitational load might also promote the preservation of baroreflex sensitivity and autonomic balance.

### Tachycardia

Resting tachycardia measurements observed immediately after immersion at R0 suggest that even the supine position out of water represents an additional workload for subjects exposed to immersion. Strict DI induces more pronounced cardiovascular deconditioning than non-strict DI, even if daily rise is very short. Similar to our finding, an increase in both asleep and awake HR was documented following 5-to-10-day space mission (Fritsch-Yelle et al., 1996). The majority of previous DI studies did not report significant changes in resting supine HR following DI (Navasiolava et al., 2011a). Tachycardia, when measured in an upright position, demonstrated a 50% increase at the first rise following DI (125 ± 5 bpm vs. 83 ± 4 bpm). Previous studies reported a 25% increase in upright HR after 3-day DI (Iwase et al., 2000), 30%- after 5-day DI (Coupé et al., 2013), 40% and 48% after 7-day DI (Iarullin et al., 1987; Navasiolava et al., 2011a), and 50%- after 10-day DI (Panferova, 1976). Taken together, literature suggests that resting and upright tachycardia readings are emphasized and enhanced following strict DI compared to non-strict DI protocols.

# Sympathetic Regulation of Cardiovascular Functions

Sympathetic neural control is extremely important in maintaining blood pressure homeostasis against gravity (Mano, 2005). In our study, strict DI clearly affected cardiac sympathetic neural control and baroreflex sensitivity. Supine sympathetic index, which presumably reflects resting cardiac sympathetic activity, was more than 2-fold increased immediately after DI, suggesting an activation of sympathetic nervous system. Supine sympathetic index failed to further increase in response to tilt, unlike previous DI studies, such as that by Miwa et al. (1997), who found an increase in upright sympathetic index following 3-day DI. Observed alteration in the sympathetic index response to progressive tilt may be related to dysregulation of high–low pressure baroreceptors. Therefore, a decrease in sympathetic index may represent a factor in orthostatic intolerance. We found that upright sympathetic index on R0 tended to correlate with orthostatic tolerance time (Pearson r = 0.54; p = 0.068).

# Vasoconstriction to Orthostasis

Total peripheral resistance is an integrative characteristic of overall resistance of peripheral vasculature in the systemic circulation. Insufficient vasoconstriction (increase in total peripheral resistance) to orthostatic stimuli is an acknowledged major factor for microgravity-induced orthostatic hypotension (Zhang, 2001). Astronauts intolerant to orthostatism fail to adequately increase TPR when upright (Buckey et al., 1996). Increased resting total peripheral resistance observed in our study suggests greater basal vasoconstriction and therefore a decrease in the reserve of vasoconstriction, with limitation of the vasoconstrictive response. According to Convertino hypothesis (Convertino, 1999), diminished vasoconstrictive reserve may be the main mechanism of vasoconstrictor insufficiency in cases of orthostatic intolerance. The maximal capacity of vasoconstriction is not altered under microgravity (Convertino, 1999), but hypovolemia might induce an increase in baseline vasoconstriction and thus decrease the vasoconstrictive reserve. However we did not observe direct correlation between upright total peripheral resistance at R0 and orthostatic tolerance time (Pearson r = 0.47; p = 0.12), suggesting involvement of additional factors into orthostatic intolerance.

# Water-Sodium Balance and Plasma Volume

Resetting of water-sodium balance under DI, accompanied by acute BNP increase and renin-aldosterone suppression, usually occurs over the first to second day, with a subsequent return in hormonal concentrations to their usual level (Leach Huntoon et al., 1998; Navasiolava et al., 2011a,c; Coupé et al., 2013). We did not assess the initial hormonal responses. The observed moderate rise in renin at the end of DI may be related to the slight decrease in dietary sodium under DI. However we did not find direct correlation between the decrease in dietary sodium and the increase in renin at the end of DI (Pearson r = −0.36; p = 0.26). Similarly, Shulzhenko et al. (1980) reported a minor increase in plasmatic renin on DI7 and in aldosterone on DI5 and DI6. The observed decrease in plasma volume of 16% is in accordance with the data described in other DI studies, with 15–20% hypovolemia (Leach Huntoon et al., 1998; Navasiolava et al., 2011a; Coupé et al., 2013). The similarity in hypovolemia numbers obtained by the CO technique and the Dill and Costill estimation suggests the accuracy of the indirect method of plasma volume estimation by Hb and Hct count under DI. The observed greater reduction in plasma volume in fitter subjects was in line with Convertino (1996a).

## What May Be Responsible for This Decrease in Orthostatic Tolerance?

Hypovolemia appears to be the major contributor for the observed rapid cardiovascular impairment in DI. However, the degree of hypovolemia in our study is similar to that reported for non-strict DI (15%–Leach Huntoon et al., 1998; 17-18%–Coupé et al., 2013; 15%–Navasiolava et al., 2011c; 12-16%–Gogolev et al., 1980). Moreover, the percentage decrease in plasma volume did not directly correlate with orthostatic tolerance time (Pearson r = −0.36; p = 0.28). The degree of hypovolemia develops within the first 24 h and is not greatly dependant on the duration of actual or simulated microgravity, whereas the degree of orthostatic intolerance tends to increase with extension of microgravity. Aside from a reduction in plasma volume, other mechanisms including increased lower limb venous capacitance (Convertino, 1996b), leg muscle tone diminution, compromised sympathetic regulation (Convertino, 2002; Mano, 2005), myocardial function and baroreflex sensitivity (Engelke et al., 1996; Convertino, 2002) may also contribute to cardiovascular deconditioning following DI. Vascular impairment at macro- (Zhang, 2001) and microvascular (Coupé et al., 2009) level promoted by physical inactivity and reduced shear stress (de Groot et al., 2006), in addition to metabolic, hormonal and vestibular changes may also play a role.

# Rapid and Profound Metabolic Changes Induced by Strict DI

The negative metabolic effects of DI are mainly related to increased inactivity. DI rapidly impaired glucose metabolism and lipid profile, inducing a decrease in insulin sensitivity and dyslipidemia. The same changes were observed in bed rest experiments (Blanc et al., 2000; Hamburg et al., 2007; Bergouignan et al., 2010). Even short-time physical inactivity appears sufficient to impair metabolism (Hamburg et al., 2007; Coupé et al., 2013; this study).

Some studies suggest that fasting glucose is not modified by DI, and fasting insulin is somewhat increased following 7-day DI (Navasiolava et al., 2011a). However, the effect of DI on glucose tolerance has not previously been explicitly investigated. This study is believed to be the first to examine the effect of DI on oral glucose tolerance test in depth. Data in the published literature indicate that bedrest (BR) increases insulin response to glucose loading without deterioration in glucose tolerance (Kiilerich et al., 2011–non-strict 7-day BR in men with allowed sitting 5h/day, Blanc et al., 2000–7-day strict HDBR in women, Dirks et al., 2016–7-day strict BR in men, mixed-meal tolerance test) or with slight increase in the net glucose response of about 6% (Hamburg et al., 2007–5-day BR in men and women, Heer et al., 2014–21-day BR in men). Globally, in BR, an increased insulin response is able to prevent glucose increase, or at a minimum, this glucose increase remains relatively moderate. In our DI experiment, a shorter exposure (2 days) seems to induce a greater impairment in glucose tolerance (14%-increase in the net glucose response). However, caution should be employed when interpreting these results as our subjects showed pre-immersion response to oral glucose intake with a faster decline in glucose concentration compared to baseline of bedrest studies reported by Heer et al. (2014) and by Hamburg et al. (2007), whereas postimmersion response was much closer to that observed after those bedrest protocols. This difference in baseline measurements may be related to individual sensitivity of subjects and limits the possibility of direct comparison.

Increased blood glucose and dyslipidemia may compromise endothelial integrity and microvascular functions (Hamburg et al., 2007; Yuan et al., 2015), thus indirectly contributing to orthostatic intolerance. However, we did not observe any direct correlation between induced glycemia on DI3 and orthostatic tolerance time following DI (Pearson r = −0.32; p = 0.31).

# Day-Night Variations

Day-night variability in HR, blood pressure and body temperature on DI3 did not differ from baseline, suggesting a good adaptation to DI conditions. Circadian rhythm of K<sup>+</sup> excretion is one of the most consistent and stable physiological circadian fluctuations (Gumz et al., 2015). It is independent of cyclic potassium intake, adrenal hormones, changes in plasma potassium, renal nerves and sodium cyclic excretion. Physiologic circadian rhythm in potassium excretion has high amplitude and is driven by a brain oscillator. As suggested by Gumz and Rabinowitz (2013), fluctuations in potassium excretion are regulated by circadian fluctuations in K<sup>+</sup> transporter gene expression. "Predictive" signals from central nervous system (CNS) cyclically change kidney sensitivity for the excretory commands, thus preparing increase in renal excretion for the period of expected potassium intake with meal (Gumz et al., 2015).

In our study, circadian rhythm in potassium excretion was preserved, however we observed a delayed diminution in its amplitude. A possible rationale is that DI might induce a forward shift of day-night cycle (under DI the subjects have sleeping difficulties and might fall asleep later). Consequently, the 19:00– 07:00 urine pool would contain more "daily" urine than before, so night kaliuresis would appear to be elevated. However, night melatonin was not decreased as expected in case of increased "contamination" of night urine with daily portion. This suggests another potential explanation—a certain disturbance in circadian regulation itself. The observed delayed changes in renal K<sup>+</sup> excretion variance (and delayed restoration) might reflect a change in its "predictive"/feedforward regulation by homeostatic system of CNS (Gumz et al., 2015). Other studies have also shown that circadian rhythms could be modified by spaceflight (Fuller et al., 1994; Guo et al., 2014) and bed rest (Millet et al., 2001; Pavy-Le Traon et al., 2007; Liang et al., 2012). Bed rest did not change the excretion of K+, but altered the rhythmicity and circadian amplitude of Na<sup>+</sup> excretion (Millet et al., 2001; Liang et al., 2012).

# Stress Is Moderate in DI and Might Be Related to Back Pain

A relatively modest increase in cortisol suggests quite a moderate physiological stress effect of DI. During immersion, the increase in urinary cortisol level was more evident at night. It may be related to the greater perception of the stress factors inherent to DI at night time. Interestingly, in the first hours of immersion, a drop in cortisol levels was observed. In part, this is likely due to the fact that short time immersion is quite comfortable and relaxing, which is why the method is employed in a spa setting (Navasiolava et al., 2011a). The positive correlation between the change in urinary cortisol and self-reported lumbar pain level confirms the role of cortisol in response to stress under DI.

This lumbar pain was not accompanied by muscle spasm. Indeed, an increase in lower back muscle tone might be expected, as lumbar pain should induce a reflex contraction of the lumbar paravertebral muscles. However, direct support withdrawal immediately switches off the tone of postural muscles (Navasiolava et al., 2011a). We observed this immediate response for antigravity leg muscles and upper deep back, but not for lower deep back. Muscle tone stability in lower deep back could represent an averaged effect of immersion (which decreases the tone) and back pain (which increases tone). The deep muscles of the lower back, which play an important role in stabilizing the joints within the spine, may preserve their tone to counteract lumbar deformation associated with back pain.

# REFERENCES


Night increase in melatonin might aim to counteract the effect of unusual environment and promote sleeping. Positive correlation between urinary melatonin and global discomfort level under DI suggests the "efforts" of melatonin to counteract discomfort.

# CONCLUSION

Dry immersion (DI) induces an accelerated model of cardiovascular deconditioning in response to microgravity. The "support unloading" induced by DI provides rapid and profound cardiovascular deconditioning. This is exemplified by increased plasma volume loss, orthostatic intolerance, pronounced autonomic changes, pronounced metabolic impairment, rapid and profound decreases in muscle tone, and influence on circadian rhythms. DI is tolerated well enough despite backache, and shows rather moderate stress effect. Such rapid, profound and quickly reversible gravitational deconditioning renders the strict DI model extremely significant to test countermeasures for microgravity-induced deconditioning and physical inactivity-related pathologies.

# AUTHOR CONTRIBUTIONS

Conception and design of the study: MC, NN, GG, and RM. Data acquisition and sample analysis: SD, LA, RM, VR, FL, LT, and RW. Analysis and interpretation of results, drafting and revising the article: All authors.

# FUNDING

This dry immersion was supported by CNES (CNES grant number 2014 n◦ 4800000748).

# ACKNOWLEDGMENTS

We thank the volunteers and the stuff of MEDES for participation in this protocol at the MEDES space clinic in 2015.

# SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fphys. 2017.00799/full#supplementary-material

Blanc, S., Normand, S., Pachiaudi, C., Fortrat, J. O., Laville, M., and Gharib, C. (2000). Fuel homeostasis during physical inactivity induced by bed rest. J. Clin. Endocrinol. Metab. 85, 2223–2233. doi: 10.1210/jc.85.6.2223


water [in Russian, English summary]. Kosm. Biol. Aviakosm. Med. 21, 45–50.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2017 De Abreu, Amirova, Murphy, Wallace, Twomey, Gauquelin-Koch, Raverot, Larcher, Custaud and Navasiolava. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Sensorimotor Reorganizations of Arm Kinematics and Postural Strategy for Functional Whole-Body Reaching Movements in Microgravity

Thomas Macaluso<sup>1</sup> , Christophe Bourdin<sup>1</sup> , Frank Buloup<sup>1</sup> , Marie-Laure Mille1, 2, 3 , Patrick Sainton<sup>1</sup> , Fabrice R. Sarlegna<sup>1</sup> , Jean-Louis Vercher <sup>1</sup> and Lionel Bringoux <sup>1</sup> \*

<sup>1</sup> Aix Marseille Univ, CNRS, ISM, Marseille, France, <sup>2</sup> UFR STAPS, Université de Toulon, La Garde, France, <sup>3</sup> Department of Physical Therapy and Human Movement Sciences, Feinberg School of Medicine, Northwestern University, Chicago, IL, United States

#### Edited by:

Olivier White, INSERM U1093, Université de Bourgogne Franche Comté, France

#### Reviewed by:

Francesco Lacquaniti, Università degli Studi di Roma Tor Vergata, Italy Vaughan G. Macefield, Mohammed Bin Rashid University of Medicine and Health Sciences, United Arab Emirates

> \*Correspondence: Lionel Bringoux lionel.bringoux@univ-amu.fr

#### Specialty section:

This article was submitted to Integrative Physiology, a section of the journal Frontiers in Physiology

Received: 27 June 2017 Accepted: 05 October 2017 Published: 20 October 2017

#### Citation:

Macaluso T, Bourdin C, Buloup F, Mille M-L, Sainton P, Sarlegna FR, Vercher J-L and Bringoux L (2017) Sensorimotor Reorganizations of Arm Kinematics and Postural Strategy for Functional Whole-Body Reaching Movements in Microgravity. Front. Physiol. 8:821. doi: 10.3389/fphys.2017.00821 Understanding the impact of weightlessness on human behavior during the forthcoming long-term space missions is of critical importance, especially when considering the efficiency of goal-directed movements in these unusual environments. Several studies provided a large set of evidence that gravity is taken into account during the planning stage of arm reaching movements to optimally anticipate its consequence upon the moving limbs. However, less is known about sensorimotor changes required to face weightless environments when individuals have to perform fast and accurate goal-directed actions with whole-body displacement. We thus aimed at characterizing kinematic features of whole-body reaching movements in microgravity, involving high spatiotemporal constraints of execution, to question whether and how humans are able to maintain the performance of a functional behavior in the standards of normogravity execution. Seven participants were asked to reach as fast and as accurately as possible visual targets while standing during microgravity episodes in parabolic flight. Small and large targets were presented either close or far from the participants (requiring, in the latter case, additional whole-body displacement). Results reported that participants successfully performed the reaching task with general temporal features of movement (e.g., movement speed) close to land observations. However, our analyses also demonstrated substantial kinematic changes related to the temporal structure of focal movement and the postural strategy to successfully perform -constrained- whole-body reaching movements in microgravity. These immediate reorganizations are likely achieved by rapidly taking into account the absence of gravity in motor preparation and execution (presumably from cues about body limbs unweighting). Specifically, when compared to normogravity, the arm deceleration phase substantially increased. Furthermore, greater whole-body forward displacements due to smaller trunk flexions occurred when reaching far targets in microgravity. Remarkably, these changes of focal kinematics and postural strategy appear close to those previously reported when participants performed the same task underwater with neutral buoyancy applied to body limbs. Overall, these novel findings reveal that humans are able to maintain the performance of functional goal-directed whole-body actions in weightlessness by successfully managing spatiotemporal constraints of execution in this unusual environment.

Keywords: whole-body reaching, arm kinematics, postural strategy, sensorimotor adaptation, microgravity, parabolic flight, weightlessness

# INTRODUCTION

On Earth, humans' motor behavior takes place within the ubiquitous gravitational force field. Several previous work already reported that the gravity direction and intensity are taken into account for motor execution, both on focal and postural components. For instance, regarding vertical arm movements, kinematic differences have been revealed between upward and downward movements (i.e., executed against or toward the direction of gravity). Particularly for upward arm movements, the relative deceleration duration was shown to be longer than the relative acceleration duration, while the opposite was observed for downward arm movements (Papaxanthis et al., 1998, 2003). Such asymmetric bell-shaped velocity profiles would allow humans to take advantage of mechanical effects of gravity torque on the limb by passively decelerating/accelerating upward/downward movements (Gaveau et al., 2014). This assumption is supported by the analysis of muscle activation patterns during vertical arm movements (Papaxanthis et al., 2003) and the removal of this specific asymmetry for horizontal movements wherein the gravitational torques did not vary (Gentili et al., 2007; Le Seac'h and McIntyre, 2007). Furthermore, these direction-dependent kinematic asymmetries appeared early in movement execution suggesting that the gravity effects could be anticipated and integrated into motor planning (Gaveau and Papaxanthis, 2011). Noticeably, the focal part of the movement investigated by these previous work is executed within a postural context, which was also subject to the influence of gravity. On Earth, body posture has to deal with the gravitational force to avoid falling. Indeed, humans would try to actively maintain the vertical projection of the center of mass (CoM) inside the support surface (Massion, 1992; Vernazza et al., 1996; Massion et al., 2004). Thus, trunk bending or upper limb movements may act as internal sources of disturbance to equilibrium. To prevent both substantial CoM displacement and falling, compensatory displacements of hip and knee usually occur (Babinski, 1899; Crenna et al., 1987; Massion, 1992; Horak, 2006).

Overall, studies mentioned above clearly demonstrated that the gravitational force plays an important role into the motor planning and execution on Earth. More precisely, the velocity profiles of arm movements and the postural strategy seem to be relevant gravity-dependent kinematic markers of human motor behavior. What happens however when gravity is removed? Understanding the impact of weightlessness on human behavior is of critical importance for keeping efficient sensorimotor behavior during the forthcoming long-term space missions. Parabolic and space flights contexts are privileged by researchers to investigate the effects of microgravity exposure on motor control. Previous studies focusing on arm movements revealed that final accuracy decreased in microgravity as compared to normogravity observations (Bock et al., 1992; Fisk et al., 1993; Watt, 1997; Carriot et al., 2004; Bringoux et al., 2012) which is consistent with works on pointing movements into a new force field (Lackner and DiZio, 1994; Shadmehr and Mussa-Ivaldi, 1994; Goodbody and Wolpert, 1998; Bourdin et al., 2001, 2006; Lefumat et al., 2015). However, the way microgravity exposure impacts kinematic features remains unclear. Indeed, some authors observed a reduction of movement speed (Ross, 1991; Berger et al., 1997; Mechtcheriakov et al., 2002; Carriot et al., 2004; Crevecoeur et al., 2010) whereas others reported no significant changes as compared to normogravity (Papaxanthis et al., 2005; Bringoux et al., 2012; Gaveau et al., 2016). More interestingly, contrasting findings have been also reported concerning gravity-dependent kinematic markers based on the temporal organization of focal movement and postural behavior. Indeed, some studies of arm vertical movements performed during parabolic flights showed either a progressive disappearance of asymmetric velocity profiles (Papaxanthis et al., 2005; Gaveau et al., 2016) or conversely an increase of the relative deceleration duration (Bringoux et al., 2012) with respect to normogravity. Regarding postural control in microgravity, most previous work demonstrated the persistence of a terrestrial strategy by stabilizing the CoM displacements during internal disturbance, such as trunk bending or arm and leg raising (Massion et al., 1993, 1997; Mouchnino et al., 1996; Vernazza-Martin et al., 2000). However, during long-term exposure, Pedrocchi et al. (2002, 2005) reported significant shifts of CoM toward the moving leg on a same lateral lower limb raising task.

In these previous experiments, it should be noted that the focal and postural parts of movement were separately investigated, although both components are known to largely interact during functional motor behavior. Only few works have studied goaldirected whole-body reaching movements in microgravity and contradictory findings were reported. On the one hand, Patron et al. (2005) reported a decrease of the relative deceleration duration of arm movement associated to a stabilized CoM displacement in microgravity. On the other hand, Casellato et al. (2012) reported an invariance of the asymmetry of the hand velocity profile as compared to normogravity data, associated to a vertical CoM projection beyond the base of support. These discrepant findings may partly originate from inter-individual variability, as Casellato et al. (2016) recently observed different and highly variable behaviors regarding CoM stabilization on three astronauts onboard the ISS (long-term exposure). Most importantly, task-related concerns, especially target location, body limbs displacements and movement speed, could also explain these contradictory results. For instance, Patron et al. (2005) investigated postural influences on a reaching task toward targets close to the participant's feet with or without speed instructions, while Casellato et al. (2012) asked the participants to perform unconstrained forward hand movements toward targets located beyond arm's length. However, to the best of our knowledge, we are not aware of any study which has investigated goal-directed whole-body reaching movements requiring to be performed as fast and as accurate as possible in microgravity.

The present study thus aimed at characterizing kinematic features of goal-directed whole-body reaching movements in microgravity, involving high spatiotemporal constraints of execution, by comparing them to normogravity observations. The spatial requirements were defined in terms of target location and size, while the temporal requirements referred to the necessity of performing the movements as fast as possible within the accuracy constraints. To that aim, close versus far external visual targets were presented during microgravity episodes in parabolic flight. To reach far targets, additional whole-body displacement was required. For both targets, two different sizes of target area were presented. As indicated by studies mentioned above, task requirements must be accounted for when considering the impact of microgravity on motor behavior. Thus, the high spatiotemporal constraints of execution in the present study constitute a novel approach allowing us to investigate whole-body reaching movements through a more functional behavior in weightless environments, close to those performed by astronauts during their space missions. In other words, we question whether and how humans are able to maintain the performance of a functional behavior in the standards of normogravity execution. We predicted substantial changes of gravity-dependent kinematic markers reflecting the specific reorganizations of focal and postural components in microgravity as compared to normogravity.

# MATERIALS AND METHODS

# Participants

Seven right-handed (3 women and 4 men, mean age = 39 ± 6.9 years) participated in the experiment on a voluntary basis. They had no prior experience of microgravity exposure. As the present study is part of a scientific program studying human motor behavior in different force fields, participants were previously tested in normogravity and underwater for the same task as reported in Macaluso et al. (2016). None of the participants suffered from neuromuscular or sensory impairments, as confirmed by a medical examination prior to the experiment. Vision was normal or corrected by lenses. Before microgravity exposure, the participants received comfort medication (scopolamine) to avoid motion sickness. It has been demonstrated that its use for parabolic flights did not induce neuromuscular side-effects on sensorimotor control (Ritzmann et al., 2016). All the participants were naive as to the specific purpose of the experiment, which was authorized by the ANSM (French National Agency for Biomedical Security) and approved by the Committee for the Protection of Persons concerned (CPP). The participants gave their signed informed consent prior to the study in accordance with the Helsinki Convention.

# Experimental Setup

Circular targets were presented in front of participants standing upright and maintained to the ground structure by means of foot-straps (**Figure 1A**). They had to press their right index finger on the start push-button positioned alongside their body. The height of the start push-button was adjusted to each participant's height for initial posture standardization. Circular targets were oriented along the frontal plane and were positioned relative to participants' anthropometric features. Close targets were located at shoulder's height (i.e., the height of the target center corresponded to the horizontal projection of the height of the acromioclavicular joint in the sagittal plane) at a distance corresponding to arm length, allowing the participants to reach these targets without trunk displacement. Far targets were located 25 cm away and 20 cm below the close targets: in that case, participants had to perform additional trunk displacement to reach these targets (**Figure 1B**). For both target locations, the diameter was also manipulated through Light-Emitting Diodes (LEDs) equally distributed to define two target sizes: Small targets 4 cm or Large targets 10 cm (**Figure 1C**). Therefore, in this experiment, combining location and size corresponded to the presentation of four targets: CS (Close–Small), CL (Close–Large), FS (Close–Small), FL (Far–Large). Switching targets on and off were achieved by a homemade software (Docometre©) piloting a real-time acquisition/control system running at 10 kHz (ADwin-Gold©, Jäger, Lorsch, Germany).

Markers were positioned onto the participants' index, shoulder and hip. Markers position was recorded (i) in normogravity with a video motion capture system (LEDtype markers) composed of three cameras sampled at 60 Hz (resolution: 848 × 480 pixels); (ii) in microgravity by an optical motion capture system (infra-red active markers) at

200 Hz (Codamotion CXS and Active CodaHub, Charnwood Dynamics Ltd, Leicestershire, UK). Importantly, both acquisition systems yielded a similar accuracy in the definition of markers' position. Indeed, the data acquired in NormoG by the video motion capture system were processed using Direct Linear Transformation (Abdel-Aziz and Karara, 1971), to reach the accuracy level of the optical motion capture system used in MicroG (i.e., millimeter order). Moreover, according to the sampling theorem (Shannon, 1949), the sampling rates used in both environments are known to be sufficient to capture the whole range of velocities associated to biological motion, including fast reaching movements (Song and Godøy, 2016).

# Procedure

All participants were exposed to two environments: first in normogravity ("NormoG") before the parabolic flight campaign, then in microgravity ("MicroG"). The MicroG environment was achieved in the A-310 ZERO-G aircraft chartered by the French Centre National d'Etudes Spatiales (CNES) and Novespace for parabolic flight studies, during the campaign #125, including 3 days of flight. For each flight, the aircraft ran a sequence of 30 parabolas. Parabolic maneuver was composed of three distinct phases: 24 s of hypergravity (1.8 g, pull-up phase) followed by 22 s of microgravity (0 g) before a second period of 22 s of hypergravity (1.8 g, pull-out phase). Each parabola was separated by 1 min of normogravity (1 g, steady flight phase).

Positions of the start push-button and the targets were adjusted for each participant, then an initial calibration of targets was performed along the Z vertical axis (i.e., defining positions relative to arm movement elevation). Before each trial, participants had to stand upright, the arms outstretched along the body, and the right index pressing the start push-button. When one of the targets was illuminated, participants were asked to perform a reaching movement toward the target while keeping the arm outstretched. Reaching movements had to be performed as fast as possible while primarily respecting accuracy constraints related to the target area. Each trial was validated when the index fingertip reached the target. The final position had to be maintained until target extinction (3 s after movement onset) which prompted the participants to return to the starting position.

Participants performed 10 pointing movements toward each of the four targets for a total of 40 trials per experimental session in each environment. In the MicroG environment, these 40 trials were presented during 10 successive parabolas for each participant, thus including four trials per parabola. The targets were presented in a pseudorandom order, which was counterbalanced between the participants. Each session included three specific blocks of four trials in which the order of target presentation was the same. These blocks were presented in the initial, middle and final part of the session (corresponding to the 1, 5, and 10th parabola in MicroG) to assess the potential evolution of motor performance during each session, which lasted about 25 min.

# Data Processing

Data presented below describe behavioral features of reaching movements in the sagittal plane and some of them are detailed in Macaluso et al. (2016). First, we analyzed the fingertip trajectory, success rate (index fingertip within a given target area), index final deviation from target center, reaction time (RT), movement duration (MD), and mean tangential velocity (Vmeanendpoint). The index final deviation was measured as the mean absolute distance of the final position of the index fingertip relative to the target center along the Z vertical axis. For each trial, the time elapsed between target illumination and the release of the start push-button by the participants defined RT. Index position in the sagittal plane was filtered (digital second-order dual-pass Butterworth filter; cutoff frequency 6 Hz in NormoG and 10 Hz in MicroG) and differentiated to obtain the endpoint tangential velocity in m.s−<sup>1</sup> . Regarding the different sampling rates of acquisition systems used in both environments, we found that the cutoff frequencies mentioned above were the most suitable to reflect the raw data in normo and microgravity. The movement onset was defined as the time when the index tangential velocity reached 1.5% of its peak. Conversely, movement end was defined when the tangential velocity dropped below 1.5% of its peak.

The focal component of whole-body reaching movements was analyzed by considering the arm angular elevation over time (i.e., angle evolution of the extended arm around the shoulder with respect to its initial orientation). Arm angular elevation was computed from the index and shoulder XZ raw data, filtered (digital second-order dual-pass Butterworth filter; cutoff frequency 6 Hz in NormoG and 10 Hz in MicroG) and differentiated to obtain the arm angular velocity profile. From this velocity profile, the peak velocity (PVang in deg.s−<sup>1</sup> ) and the relative angular deceleration duration (rDDang, defined as the duration between PVang and movement end, expressed in % of movement duration to facilitate comparison between both environments) were extracted. Arm angular velocity profile was also differentiated to obtain arm angular acceleration profile, informing on early changes in motor execution which may give an insight upon the planning stage of focal movement. From this acceleration profile, peak acceleration (PAang in deg.s−<sup>2</sup> ) and time to peak acceleration (TPAang expressed in ms to precisely estimate the occurrence of motor changes) were extracted.

In parallel, the postural component involved in the wholebody reaching movements (particularly to reach the far target) was analyzed by considering trunk displacement. This latter was illustrated by the final angular position of trunk (hip-shoulder segment) relative to vertical (β<sup>f</sup> trunk: trunk flexion in deg) at arm movement end, and by the forward displacement of participants' shoulder and hip (translation along the horizontal plane in mm). Shoulder and hip movement onset/end in the sagittal plane were defined as the time when the tangential velocity respectively reached/dropped below 1.5% of its peak.

Statistical analyses were based on mean comparisons. Repeated-measures analyses of variance (ANOVAs) were performed to compare the means of kinematic parameters mentioned above after having ensured that the assumption of normality was not violated (Kolmogorov-Smirnov test). Newman-Keuls tests were used for post-hoc analyses and the significance threshold was set at.05 for all statistical tests.

# RESULTS

# Potential Learning Effects

We conducted prior analyses to investigate the potential learning effects during a single session (40 trials). Repeated-measures ANOVAs including 2 Environment (NormoG, MicroG) × 2 Target Location (Close, Far) × 2 Target Size (Small, Large) × 3 Block (Initial, Middle, Final) were initially performed on all the selected parameters of the study. The results did not show any significant main effect of Block or any interaction with the other factors (p > 0.05). To specifically exclude the presence of any adaptive processes in MicroG environment, we conducted complementary analyses comparing a specific set of trials occurring during the 1, 5, and 10th parabola (see Material and Methods). Repeated-measures ANOVAs including 3 Parabola (1, 5, and 10th parabola) × 4 Target Presentation (CS, CL, FS, FL) did not reveal any significant main effect of Parabola or any interaction with the other factor on all the selected parameters (p > 0.05). Thus, the reported values did not significantly change throughout the experiment.

# Upper-Limb Displacement

First of all, we investigated arm movement toward the targets in each environment. **Figure 2** illustrates mean endpoint trajectories (i.e., index fingertip) in the sagittal plane observed for a typical participant when reaching close and far targets. It shows that spatial characteristics of endpoint motion were impacted by the microgravity environment.

# Success Rate and Index Final Deviation

Overall, participants successfully performed the task. Indeed, success rate was 100% in NormoG and 95.42 ± 8.99% in MicroG. In this latter environment, only the Small targets were sometimes missed (CS and FS). The ANOVA performed on success rate revealed no significant main effect of the experimental conditions (Environment: p = 0.06; Target Location: p = 0.39; Target Size: p = 0.06) and no significant interaction between these factors (Environment × Target Location: p = 0.39; Environment × Target Size: p = 0.06; Target Location × Target Size: p = 0.39). Moreover, the ANOVA conducted on the index final deviation yielded no main effect of the experimental conditions (Environment: p = 0.10; Target Location: p = 0.97; Target Size: p = 0.06) but showed a significant interaction between Environment × Target Size [F(1, 6) = 8.49; p < 0.05]. While no significant difference appeared between both environments when reaching Small targets (p > 0.05), the mean distance between the final position of the index and the target center when reaching Large targets was significantly higher in MicroG as compared to NormoG (13.04 ± 6.07 mm vs. 7.41 ± 2.96 mm; p < 0.01). No significant interaction between the other factors was revealed (Environment × Target Location: p = 0.69; Target Location × Target Size: p = 0.32).

# Reaction Time (RT)

The ANOVA performed on RT (mean = 326 ± 70 ms) revealed no significant main effect of the experimental conditions (Environment: p = 0.48; Target Location: p = 0.23; Target Size: p = 0.43) and no significant interaction between these factors (Environment × Target Location: p = 0.19; Environment × Target Size: p = 0.23; Target Location × Target Size: p = 0.52).

# Movement duration (MD) and Mean Tangential Velocity (Vmeanendpoint)

The ANOVA conducted on MD only yielded a significant main effect of Target Location [F(1, 6) = 166.21; p < 0.001]; MD was longer when reaching Far targets (0.73 ± 0.17 s) as compared to Close targets (0.58 ± 0.16 s). No other significant main effect or interaction was found with regard to the other factors (Environment: p = 0.07; Target Size: p = 0.11; Environment × Target Location: p = 0.35; Environment × Target Size: p = 0.26; Target Location × Target Size: p = 0.59).

The ANOVA conducted on Vmeanendpoint revealed significant main effects of Target Location [F(1, 6) = 24.05; p < 0.01] and Target Size [F(1, 6) = 11.30; p < 0.05]. Vmeanendpoint was higher when reaching Close targets (1.94 ± 0.39 m.s−<sup>1</sup> vs. 1.66 ± 0.31 m.s−<sup>1</sup> , respectively for Close and Far targets). Vmeanendpoint was also higher when reaching Large targets (1.83 ± 0.39 m.s−<sup>1</sup> vs. 1.76 ± 0.37 m.s−<sup>1</sup> , respectively for Large and Small targets). No other significant main effect or interaction was found with regard to the other factors (Environment: p = 0.52; Environment × Target Location: p = 0.14; Environment × Target Size: p = 0.76; Target Location × Target Size: p = 0.91).

typical subject in the sagittal plane in MicroG (dotted line), and NormoG (solid line) for the Close and Far targets. Gray lines represent the positive and negative standard deviations of the mean index trajectories.

To sum up, microgravity did not significantly affect the performance of whole-body reaching movements without substantially disrupting the general temporal outputs of endpoint displacement and the success rate. Then, we investigated the temporal organization of the focal component illustrated by the arm angular elevation over time.

## Temporal Organization of Arm Angular Elevation

**Figure 3A** illustrates mean arm angular velocity profiles for a typical participant when reaching Close and Far targets in each environment. It shows that the MicroG environment impacts the temporal structure of the velocity profile [reflected by the analysis of rDDang , see below Velocity profile: peak angular velocity (PVang) and relative angular deceleration duration (rDDang)] without substantially changing its amplitude. As reported below, this modulation could derive from changes of the temporal structure and amplitude of the acceleration profile [as suggested by the analysis of TPAang and PAang , see below Acceleration profile: peak angular acceleration (PAang) and time to peak angular acceleration (TPAang)].

# **Velocity profile: peak angular velocity (PV**ang **) and relative angular deceleration duration (rDD**ang **)**

The ANOVA conducted on PVang only revealed a significant main effect of Target Location [F(1, 6) = 58.74; p < 0.001]. PVang was higher when reaching Close targets (397.98 ± 68.77 deg.s−<sup>1</sup> ) as compared to Far targets (327.55 ± 48.39 deg.s−<sup>1</sup> ). No other significant main effect or interaction was found with regard to the other factors (Environment: p = 0.29; Environment × Target Location: p = 0.08; Environment × Target Size: p = 0.73; Target Location × Target Size: p = 0.51).

The ANOVA conducted on rDDang revealed significant main effects of Environment [F(1, 6) = 48.54; p < 0.001], Target Location [F(1, 6) = 20.91; p < 0.01] and Target Size [F(1, 6) = 7.38; p < 0.05]. Importantly, rDDang was substantially higher in MicroG as compared to NormoG (**Figure 3B**). Overall, rDDang was higher when reaching Far targets (69.65 ± 7.69%MD vs. 60.80 ± 7.63%MD, respectively for Far and Close targets) and Small targets too (65.63 ± 9.05%MD vs. 64.83 ± 8.71%MD, respectively for Small and Large targets). No significant interaction was found between these factors (Environment × Target Location: p = 0.22; Environment × Target Size: p = 0.54; Target Location × Target Size: p = 0.44).

## **Acceleration profile: peak angular acceleration (PA**ang **) and time to peak angular acceleration (TPA**ang **)**

The ANOVA performed on PAang revealed significant main effects of Environment [F(1, 6) = 9.30; p < 0.05] and Target Location [F(1, 6) = 73.70; p < 0.001]. PAang was higher in MicroG than NormoG (**Figure 3C**) and also higher when reaching Close targets (3661.18 ± 1332.30 deg.s−<sup>2</sup> ) as compared to Far targets (3175.85 ± 1265.36 deg.s−<sup>2</sup> ). No other significant main effect or interaction was found with regard to the other factors (Target Size: p = 0.54; Environment × Target Location: p = 0.23; Environment × Target Size: p = 0.99; Target Location × Target Size: p = 0.98).

The ANOVA conducted on TPAang also yielded significant main effects of Environment [F(1, 6) = 7.43; p < 0.05] and Target Location [F(1, 6) = 8.92; p < 0.05). Importantly, TPAang was lower in MicroG than in NormoG (**Figure 3D**) and also lower when reaching Far targets (54 ± 18 ms) as compared to Close targets (62 ± 20 ms). No other significant main effect or interaction was found with regard to the other factors (Target Size: p = 0.06; Environment × Target Location: p = 0.92; Environment × Target Size: p = 0.42; Target Location × Target Size: p = 0.95).

In summary, microgravity exposure influenced the temporal structure of arm angular elevation by decreasing the time to peak acceleration, thus leading to an increase of the relative deceleration duration as compared to NormoG. These modifications did not affect the maximal velocity of arm elevation in MicroG as compared to NormoG, presumably because of a higher maximal acceleration reached earlier during movement execution. The next part will focus on the postural component involved in whole-body reaching movements, particularly when reaching Far targets.

# Trunk Displacement

#### Final Angular Position of Trunk Relative to Vertical (βf trunk)

The ANOVA performed on β<sup>f</sup> trunk revealed a main effect of Target Location [F(1, 6) = 264.09; p < 0.001] and a significant interaction between Environment × Target Location [F(1, 6) = 24.74; p < 0.01]. Interestingly, while no significant difference appeared between both environments when reaching Close targets (p > 0.05), mean β<sup>f</sup> trunk was significantly lower when reaching Far targets in MicroG as compared to NormoG (p < 0.001; **Figure 4**).

### Shoulder and Hip Forward Displacement

Unsurprisingly in both environments, no noticeable forward translation was detected for shoulder and hip when reaching Close targets (located at participants' arm length, see Material and Methods). Therefore, we subsequently led our analysis on the shoulder and hip forward displacement occurring when reaching Far targets.

The ANOVA conducted on shoulder displacement yielded a significant main effect of Environment [F(1, 6) = 183.78; p < 0.001]. Shoulder displacement in MicroG (448.28 ± 25.37 mm) was significantly higher than in NormoG (285.54 ± 36.18 mm). The ANOVA performed on hip displacement revealed significant main effects of Environment [F(1, 6) = 20.94; p < 0.01] with higher displacement in MicroG (185 ± 84.28 mm) as compared to NormoG (38.68 ± 39.49 mm). The ANOVA also revealed a main effect of Target Size [F(1, 6) = 9.09; p < 0.05] and a significant interaction between Environment × Target Size [F(1, 6) = 7.34; p < 0.05]. While no significant difference appeared between Small and Large targets in NormoG, mean hip displacement in MicroG was higher when reaching Large target (191.66 ± 86.15 mm) as compared to Small target (178.50 ± 88.70 mm).

Overall, these analyses highlight that the postural component varied during whole-body reaching movements mainly as a function of the Environment and Target Location. In MicroG,

reaching Far targets involved smaller trunk bending associated to larger forward displacements of the shoulder and hip, as compared to NormoG. In the next section, we will discuss the main focal and postural features reported above and will propose possible interpretations for these observations.

# DISCUSSION

The present study aimed at characterizing kinematic features of goal-directed whole-body reaching movements in microgravity involving high spatiotemporal constraints of execution, with respect to normogravity observations. Our original experimental design enabled us to investigate reaching movements performed as fast as possible toward targets of different sizes and locations in both environments. Our data revealed stabilized motor features throughout microgravity exposure. While some of them are associated to the preservation of general temporal outputs with respect to land observations (e.g., movement speed), we found substantial changes in gravity-dependent kinematic markers reflecting the reorganization of focal and postural components. These points will be developed in the following sections.

# Prompt Reorganization of Motor Behavior in Microgravity

Although the participants never experienced microgravity exposure before the present experiment, we did not find any significant evolution in the reported variables across the successive trials. Thus, we failed to show the presence of sensorimotor adaptation during the experiment which would indeed have led to more progressive changes across

the repetition of reaching movements (Lackner and DiZio, 1994; Shadmehr and Mussa-Ivaldi, 1994). Rather, we observed a prompt reorganization of some movement features (see section Microgravity Is Accounted for into the Planning of Focal Movement) which took place at the earliest onset of exposure. From previous work conducted in parabolic flights, the occurrence of adaptive processes on reaching movements is not clear. Indeed, some studies reported slow progressive changes of kinematics across parabolas (Papaxanthis et al., 2005; Gaveau et al., 2016) whereas others observed rapid behavioral stabilization or no significant change during exposure (Patron et al., 2005; Bringoux et al., 2012; Casellato et al., 2012). In our study, one hypothesis related to the parabolic flight context can be advanced to explain this immediate stabilization of motor behavior. Before the 30 parabolas achieved for experimental acquisition, the aircraft performed one parabola to allow participants discover the parabolic maneuver. Moreover, since 3 participants were tested during each flight (see Material and Methods), two of them had even more time to experience microgravity exposure. Although we ensured that no reaching movements were performed by the participants out of the experiment, this preliminary although short exposure before data acquisition would enable the participants to develop prior expectancies about how it feels to move in these novel environments. Moreover, microgravity episodes of parabolic flights induced a global modification of the force field applied to the whole-body before initiating each trial. Thus, in this context, the participants accessed the new dynamic properties of the environment prior to movement onset (Barbiero et al., 2017) which might be sufficient to rapidly update their internal model for sensorimotor planning and execution, hence leading to an immediate motor reorganization (Wolpert and Kawato, 1998; Wolpert and Ghahramani, 2000). It has been shown indeed that the initial state of the sensorimotor system is primarily used to adjust the internal representations necessary to perform upcoming movements (Starkes et al., 2002; Flanagan et al., 2006; White et al., 2012, Rousseau et al., 2016). Here, the limb proprioception could contribute to detect the gravity release at the level of muscles and joints, through muscle spindles and Golgi tendon organs identified as load receptors related to gravity force field (Dietz et al., 1992). The following sections aim at discussing the stabilized motor features observed in microgravity for whole-body reaching.

# Preservation of Functional Reaching Movements within Normogravity Standards

The high spatiotemporal constraints of execution in the present study enabled us to investigate functional whole-body reaching movements in microgravity. In this line, our data did not reveal any significant difference between MicroG and NormoG environments in terms of movement duration, mean and peak velocity during movement execution. The absence of effect of the environment on these variables may reflect a tendency to keep the average movement speed in the range of normogravity experience. To that aim, the participants may have reduced the safety margin related to the final reaching accuracy in microgravity. Indeed, while the preservation of movement speed was not detrimental to reaching performance (i.e., the high success rate observed in MicroG, > 95%, was not significantly different from NormoG), the distance between the endpoint final position and the target center was significantly higher in MicroG when reaching large targets. Thus, in this task, the participants tended to maintain speed over accuracy margin (Woodworth, 1899) for a still successful performance without gravity. Keeping the average speed and reaching performance within normogravity standards were though not at the expense of movement preparation duration, since the reaction time remained also unaffected by the environment. Hence, alleviating gravity before movement execution did not impact the time allocated for motor planning. Nevertheless, we will detail in the following parts some evidence for substantial qualitative reorganizations, notably in focal and postural components of the reaching movement, which helped maintain the functionality of motor behavior in microgravity.

# Microgravity Is Accounted for into the Planning of Focal Movement

On Earth, kinematics of arm movement elevation has been welldescribed in terms of asymmetric bell-shaped velocity profiles (Papaxanthis et al., 1998; Gentili et al., 2007). Classically, the relative deceleration duration appears longer than the relative acceleration duration, suggesting that gravity is accounted for during motor planning to act as an assistive force for decelerating upward movements (Papaxanthis et al., 2003; Gaveau and Papaxanthis, 2011). The way gravity is integrated into motor planning has been recently formalized by a Minimum Smooth-Effort model (Gaveau et al., 2016), in line with the optimal control theory minimizing absolute work and jerk (Berret et al., 2011; Gaveau et al., 2014).

In our experiment, the substantial increase of the relative deceleration duration in MicroG constitutes the most salient feature of motor reorganization concerning the focal part of the reaching movement. The asymmetry was thus notably amplified in microgravity without changing the amplitude of peak velocity. Such reorganization, consistent across subjects as shown in the Supplementary Figure 1A, appears as a direct consequence of an earlier peak acceleration observed during motor execution (∼47 ms). Recent work established that the shortest feedbackbased corrections of EMG (electromyographic) patterns during arm reaching occur at ∼60 ms when considering a limb disturbance with no changes of target location (Scott, 2016 for a review). We therefore hypothesize that the kinematic changes promptly observed in microgravity following arm movement onset are based on feedforward control mechanisms, directly expressed in the motor intention (Gaveau and Papaxanthis, 2011). In other words, we argue that the CNS could predict the effect of gravity release on the moving segments and could subsequently integrate the novel dynamics associated to a weightless environment into motor planning. As there is no external force to help braking upward movements in microgravity (neglecting the air friction forces), the participants had to actively counteract the inertial force of their moving limbs, presumably by increasing the antagonist muscle activations (Bonnard et al., 1997). In this context, a longer deceleration phase may reflect a greater use of feedback processes (Chua and Elliott, 1993; Sarlegna et al., 2003; Terrier et al., 2011). This greater retroactive control would enable the participants to better manage the speed reduction of their reaching movements, especially when approaching the target, to maintain final accuracy. Increased asymmetry between acceleration and deceleration phases was also reported when removing gravitational shoulder torque before arm movement onset (Rousseau et al., 2016) and with additional loads placed on the arm (Gaveau et al., 2011). Such reorganization in kinematics may thus illustrate a cautious strategy accounting for force/inertia uncertainties in unusual context. In this line, previous studies demonstrated that the lack of information prior to movement onset strongly affects motor planning (Bringoux et al., 2012; Rousseau et al., 2016), presumably to face unexpected or erroneous sensorimotor estimates during subsequent movement execution in unfamiliar environments (Brooks et al., 2015).

Contradictory with the present findings, some other studies reported a progressive disappearance of asymmetric velocity profiles in microgravity (Papaxanthis et al., 2005; Gaveau et al., 2016). Unlike our experiment, movement accuracy was not a primary constraint in these previous work where the braking phase of arm movements was not crucial during motor execution to correctly perform the task. Alternatively, when the participants had to perform reaching movements "as accurately as possible" in microgravity, Bringoux et al. (2012) also observed a longer deceleration phase as compared to normogravity exposure. Interestingly in our study, such reorganization of motor planning for arm reaching in MicroG was not detrimental to movement duration: the longer deceleration phase was compensated by higher peak acceleration in microgravity. This compensatory increase of peak acceleration may also represent a specific reorganization of the movement in a given environment as it likely exploits the absence of gravity torque at movement onset to efficiently trigger the initial impulse. Additionally, as discussed in the following section, the planning of focal movement was not the only component to be modified during whole-body reaching movements in microgravity.

# Efficient Postural Strategy for Reaching Without Gravity

Kinematics collected from the trunk clearly supports two different postural strategies as a function of the gravity environment while reaching far targets placed beyond arm's length. Under normogravity, our analyses revealed a significant forward trunk bending expressed by large shoulder displacement associated to very small hip displacement in space. This feature is typical of a "hip strategy" (Horak and Nashner, 1986), through which the postural component supporting the focal part of movement is also used to prevent falling (Massion, 1992). This posturokinetic organization would thus reduce the displacement of the CoM by using compensatory mechanisms (Massion, 1992; Vernazza et al., 1996) and would favor equilibrium maintenance at the expense of mechanical energy minimization and joint smoothness maximization (Hilt et al., 2016).

Alternatively, the second postural strategy specifically observed in microgravity was illustrated by very small trunk bending associated to larger shoulder and hip displacement from vertical. This organization would resemble the "ankle strategy" evoked by Horak and Nashner (1986), though with greater whole-body forward displacement. In MicroG environment, the participants were indeed not constrained by gravitational force, allowing for a vertical CoM projection outside the base of support. This observation is consistent with others reporting that postural control in weightlessness is predominantly managed at the ankle level (Clement et al., 1984; Clément and Lestienne, 1988). On Earth, this posturokinetic strategy decreases the equilibrium safety margin but the risk of falling is greatly minimized in microgravity. The participants might therefore adopt the strategy which would allow them to reduce the degrees of freedom (Bernstein, 1967), helping minimize the mechanical energy expenditure and maximize joint smoothness (Hilt et al., 2016). In line with the optimal control theory (Berret et al., 2011; Gaveau et al., 2014), the combination of these cost functions would enable the postural component to support more efficiently the focal part of the reaching movement in weightless environment. Despite methodological differences with our study, Casellato et al. (2012) also reported whole-body forward displacement when performing unconstrained bimanual reaching (i.e., natural pace and uncontrolled accuracy). In line with their observations, our data may also support the existence of an "oversimplification" of postural control to perform a functional behavior when facing high spatiotemporal constraints of execution in microgravity. Moreover, unlike previous observations of Casellato et al. (2016) in long-term weightlessness, the postural strategy observed in the present study was not subjected to large inter-individual variability. Overall, individual trends for both focal and postural observations are clear and systematic (as illustrated in the Supplementary Figure 1). Remarkably, the stabilized motor features observed in microgravity in the present study appear close to those previously reported when participants performed the same task underwater with neutral buoyancy applied to body limbs (Macaluso et al., 2016). The following section discusses the behavioral analogies observed in both environments.

# Behavioral Analogies between Neutral Buoyancy Underwater and Microgravity

As the present study is part of a scientific program studying human motor behavior in different force fields, the same participants were previously tested underwater in the same task for comparison purpose. Specifically, they were immersed in a prototypical submersible simulated space suit (AquaS environment; Macaluso et al., 2016), to apply neutral buoyancy at the level of body limbs. As in the present study conducted 2 years later, we did not find any significant evolution in the reported variables across the successive trials performed underwater. Rather, we observed immediate reorganizations at the earliest onset of exposure excluding the presence of adaptive processes during the experiment. As in MicroG, AquaS environment also implied initial exposure before data acquisition related to the installation of participants on the pointing structure. Thus, participants were submitted to global modifications of the force field applied to the whole-body before trial execution. This observation extends the hypothesis provided in section Prompt Reorganization of Motor Behavior in Microgravity. When participants accessed the new "unweighting" properties of a given environment before performing the first reaching movements, they could promptly reorganize their motor behavior. Most interestingly, the changes of focal and postural components of reaching movements in MicroG are close to those observed underwater in AquaS. Indeed, the increase of the relative arm deceleration duration and the decrease of trunk flexion when reaching far targets appear strikingly comparable (see the Supplementary Figure 2). In other words, the participants adopted analogous temporal structure of arm movements and almost similar postural strategy to perform whole-body reaching movements in these different environments. In so far as these two parameters are known to be gravity-dependent kinematic markers (see Introduction), and as AquaS and MicroG environments attempted to reproduce a weightless context, we hypothesize that these very close motor strategies would be mainly due to whole-body unweighting. It suggests that a fine control of neutral buoyancy underwater across the whole-body segments would tend to better simulate microgravity when considering the execution of sensorimotor tasks. Further studies are obviously required to challenge this hypothesis, especially to better investigate the effects of viscous force on motor control.

# CONCLUSION

The present study provides clear and original evidence that participants could successfully perform goal-directed wholebody reaching movements involving high spatiotemporal constraints in a novel environment, such as microgravity, by immediately reorganizing focal and postural control strategies compared to normogravity. Moreover, these substantial modifications occurred in motor planning at the very beginning of weightless exposure which strongly suggests that the effects of the absence of gravity were anticipated and integrated by CNS. Overall, our novel findings highlight that humans are able to maintain the performance of functional goaldirected whole-body actions in weightlessness in the standards of normogravity observations by successfully managing spatiotemporal constraints of execution in this unusual environment. Interestingly, our previous work reported similar kinematic features of whole-body reaching movements performed underwater when neutral buoyancy was rigorously applied at the level of each body limb (Macaluso et al., 2016). Therefore, this suggests that comparable initial state estimates and subsequent motor reorganizations could arise from unweighting the body at the level of body skin, muscles and joints, irrespective of the presence of gravity-related vestibular cues. Further experiments are of course mandatory to investigate this challenging hypothesis, which may be crucial for instance in astronauts training underwater, where gravitational field still acts at the level of the vestibular system.

# ETHICS STATEMENT

This study was carried out in accordance with the recommendations of the ANSM (French National Agency for Biomedical Security) with written informed consent from all subjects. All subjects gave written informed consent in accordance with the Declaration of Helsinki. The protocol was approved by the CPP.

# AUTHOR CONTRIBUTIONS

TM designed and performed experiments, analyzed data and wrote the paper; CB wrote the paper; FB designed experiments and analyzed data; MM wrote the paper; PS designed and performed experiments, analyzed data; FS wrote the paper; JV wrote the paper; LB designed and performed experiments, wrote the paper.

# FUNDING

This work (SIMEXPLOR) was supported by the Centre National d'Etudes Spatiales (CNES) through the APR Grants and the CNRS (Centre National de la Recherche Scientifique). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

# ACKNOWLEDGMENTS

The authors wish to thank Novespace and CADMOS (CNES) for technical support, Patrick Sandor for medical inclusion of

# REFERENCES


the participants, Jean-François Bramard and Alexis Rosenfeld for video assistance and the participants who took part in this study.

# SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fphys. 2017.00821/full#supplementary-material


control of goal-directed arm movements. Exp. Brain. Res. 151, 524–535. doi: 10.1007/s00221-003-1504-7


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2017 Macaluso, Bourdin, Buloup, Mille, Sainton, Sarlegna, Vercher and Bringoux. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Non-linear Heart Rate and Blood Pressure Interaction in Response to Lower-Body Negative Pressure

Ajay K. Verma<sup>1</sup> , Da Xu<sup>2</sup> , Amanmeet Garg<sup>3</sup> , Anita T. Cote<sup>4</sup> , Nandu Goswami <sup>5</sup> , Andrew P. Blaber 1, 2 and Kouhyar Tavakolian1, 2 \*

*<sup>1</sup> Department of Electrical Engineering, University of North Dakota, Grand Forks, ND, United States, <sup>2</sup> Department of Biomedical Physiology and Kinesiology, Simon Fraser University, Burnaby, BC, Canada, <sup>3</sup> Department of Engineering Science, Simon Fraser University, Burnaby, BC, Canada, <sup>4</sup> School of Human Kinetics, Trinity Western University, Langley, BC, Canada, <sup>5</sup> Institute of Physiology, Medical University of Graz, Graz, Austria*

Edited by:

*Brian James Morris, University of Sydney, Australia*

#### Reviewed by:

*Alessandro Silvani, University of Bologna, Italy Alexander V. Ovechkin, University of Louisville, United States*

> \*Correspondence: *Kouhyar Tavakolian kouhyar@und.edu*

#### Specialty section:

*This article was submitted to Integrative Physiology, a section of the journal Frontiers in Physiology*

Received: *02 June 2017* Accepted: *20 September 2017* Published: *24 October 2017*

### Citation:

*Verma AK, Xu D, Garg A, Cote AT, Goswami N, Blaber AP and Tavakolian K (2017) Non-linear Heart Rate and Blood Pressure Interaction in Response to Lower-Body Negative Pressure. Front. Physiol. 8:767. doi: 10.3389/fphys.2017.00767* Early detection of hemorrhage remains an open problem. In this regard, blood pressure has been an ineffective measure of blood loss due to numerous compensatory mechanisms sustaining arterial blood pressure homeostasis. Here, we investigate the feasibility of causality detection in the heart rate and blood pressure interaction, a closed-loop control system, for early detection of hemorrhage. The hemorrhage was simulated via graded lower-body negative pressure (LBNP) from 0 to −40 mmHg. The research hypothesis was that a significant elevation of causal control in the direction of blood pressure to heart rate (i.e., baroreflex response) is an early indicator of central hypovolemia. Five minutes of continuous blood pressure and electrocardiogram (ECG) signals were acquired simultaneously from young, healthy participants (27 ± 1 years, *N* = 27) during each LBNP stage, from which heart rate (represented by RR interval), systolic blood pressure (SBP), diastolic blood pressure (DBP), and mean arterial pressure (MAP) were derived. The heart rate and blood pressure causal interaction (RR↔SBP and RR↔MAP) was studied during the last 3 min of each LBNP stage. At supine rest, the non-baroreflex arm (RR→SBP and RR→MAP) showed a significantly (*p* < 0.001) higher causal drive toward blood pressure regulation compared to the baroreflex arm (SBP→RR and MAP→RR). In response to moderate category hemorrhage (−30 mmHg LBNP), no change was observed in the traditional marker of blood loss i.e., pulse pressure (*p* = 0.10) along with the RR→SBP (*p* = 0.76), RR→MAP (*p* = 0.60), and SBP→RR (*p* = 0.07) causality compared to the resting stage. Contrarily, a significant elevation in the MAP→RR (*p* = 0.004) causality was observed. In accordance with our hypothesis, the outcomes of the research underscored the potential of compensatory baroreflex arm (MAP→RR) of the heart rate and blood pressure interaction toward differentiating a simulated moderate category hemorrhage from the resting stage. Therefore, monitoring baroreflex causality can have a clinical utility in making triage decisions to impede hemorrhage progression.

Keywords: hemorrhage, causality, baroreflex, heart rate, arterial blood pressure, central hypovolemia, blood loss

# INTRODUCTION

Hemorrhage, a physiological condition resulting in central hypovolemia; owing to loss of blood from the circulation which leads to inadequate tissue perfusion (Schiller et al., 2017), is a major cause of death in soldiers and civilians following a trauma (Søreide et al., 2007; Evans et al., 2010; Eastridge et al., 2012; Rhee et al., 2014). Additionally, postpartum hemorrhage is a recognized cause of maternal mortality (Hancock et al., 2015), especially in low-income countries (Say et al., 2014). The capability to track hemorrhage progression early, accordingly, can be indispensable toward impeding hemorrhage advancement via appropriate intervention (Beekley et al., 2008; Gerhardt et al., 2011). Reliance on arterial blood pressure as an early indicator of central blood loss has been ineffectual, due to various compensatory mechanisms maintaining arterial blood pressure homeostasis until the point of autonomic decompensation (Soller et al., 2008; Convertino et al., 2015). Blood pressure falls abruptly upon autonomic decompensation, followed by systemic hypotension, which, if not immediately compensated for, could result in tissue hypoperfusion (hemorrhagic shock; Victorino et al., 2003; Parks et al., 2006; Schiller et al., 2017).

Under the physiological state of systemic hypotension (systolic blood pressure <90 mmHg), the interventional strategies have limited effect (Gerhardt et al., 2011; Tavakolian et al., 2014). Pulse pressure and heart rate are recommended or considered for triage decisions (Convertino et al., 2006; Brasel et al., 2007), however, pertinent literature reveals heart rate alone is an unreliable marker of blood loss due its dependency on numerous factors such as, pain, vagal and sympathetic tone, and hormones, as well as showing late changes in response to central hypovolemia (Victorino et al., 2003; Cooke et al., 2006; Brasel et al., 2007; Soller et al., 2008), while high intersubject variability has been demonstrated with pulse pressure as a marker of central blood loss (Cote et al., 2012). The quantification of compensatory mechanisms, corresponding to the maintenance of blood pressure homeostasis, therefore, could dispense paramount information pertaining to the progression of central hypovolemia (Nadler et al., 2014; Convertino et al., 2015, 2016; Janak et al., 2015).

Baroreceptors, the stretch receptors localized in the carotid sinus and the aortic arch play a central role toward maintenance of blood pressure homeostasis (Heesch, 1999). The perturbation to circulatory homeostasis via redistribution of central blood volume away from the heart and the associated decline in the arterial blood pressure is sensed by the stretch receptors on the blood vessels, which leads to a reduction in baroreceptor discharge to the brain. Rapid withdrawal of vagal and activation of sympathetic nerve activity, which causes an elevation in systemic vascular resistance (SVR) and heart rate, is a consequent baroreceptor mediated reflex response to an external perturbation to regulate arterial blood pressure (Rowell, 1993; Ricci et al., 2015). The baroreflex activity quantified via muscle sympathetic nerve activity has been shown to track central hypovolemia (Cooke et al., 2009; Ryan et al., 2012), however, given the requirement of sophisticated instrumentation and invasive nature; it has limited application toward surgical triage. The non-invasive quantification of sympathetic tone via heart rate variability has also been explored to track central hypovolemia, nonetheless, high inter-subject variability has been reported with such an approach (Cooke et al., 2006; Ryan et al., 2010).

In response to an external perturbation to the hemodynamic homeostasis, heart rate, and blood pressure are known to act in conjunction (closed loop) toward maintenance of arterial blood pressure homeostasis (Porta et al., 2002, 2011; Faes et al., 2013). This causal interaction has feedforward (non-baroreflex) and feedback (baroreflex) controls, governed by the Frank-Starling mechanism on blood pressure and the baroreflex mediated control of heart rate, respectively (Porta et al., 2011; Faes et al., 2013; Javorka et al., 2017). The imperative role of causal heart rate and blood pressure interaction toward regulation of arterial blood pressure is well documented in the literature under external disruption to the hemodynamic homeostasis, induced by head-up tilt and stand tests (Nollo et al., 2005; Porta et al., 2011, 2017; Javorka et al., 2017). The physiological response of the cardiovascular system to such stressors can be hypothesized to be analogous to blood loss simulated via the lower-body negative pressure (LBNP), as all experimental protocols aim to translocate central blood volume downwards to the peripheral regions, as a consequence of gravity in a head-up tilt or stand test and lower body suction in the LBNP protocol.

LBNP is an acclaimed experimental tool for replicating controlled central hypovolemia of variable degree in a laboratory setting. The response of the cardiovascular system to LBNP has been shown to be analogous to hemorrhage (Cooke et al., 2004; Blaber et al., 2013; Hinojosa-Laborde et al., 2014b; Johnson et al., 2014; Janak et al., 2015). Therefore, LBNP provides a practical, reproducible, and a safe physiological surrogate to study the hemodynamic response manifesting hemorrhage (Cooke et al., 2004). The continuous application of LBNP sequesters central blood volume in the compliant venous system of lower peripheral regions (Goswami et al., 2008; Blaber et al., 2013), causing a substantial decline in the central venous return and preload. Consequently, a reduction in stroke volume and cardiac output is observed (Cooke et al., 2004; Blaber et al., 2013; Hinojosa-Laborde et al., 2014b).

Depending on the amount of blood volume lost from the circulation, a hemorrhage is categorized as a mild, moderate, or severe (Cooke et al., 2004). The LBNP stage of −20 to −40 mmHg simulates the spectrum of a moderate category hemorrhage, displacing 10–20% of total blood volume or approximately up to 1,000 ml of blood loss (Cooke et al., 2004). Therefore, a successful differentiation of compensatory mechanisms regulating arterial blood pressure at −30 mmHg LBNP from rest can serve as a potential tool toward effective triage, to impede hemorrhage advancement to a severe category.

In the current research, we systematically studied the causal behavior of the heart rate and blood pressure interaction at rest and under graded LBNP to −40 mmHg. Since LBNP is known to reduce stroke volume (Cooke et al., 2004; Hinojosa-Laborde et al., 2014b; Tavakolian et al., 2014), the directional behavior was studied with respect to LBNP, to gain inference regarding the dynamics of closed loop heart rate and blood pressure interaction in response to simulated blood loss. The interaction was quantified using a non-linear convergent cross mapping (CCM) method. The potential of CCM method in outlining the directional information flow between physiological signals has been demonstrated in the literature (Heskamp et al., 2014; Schiecke et al., 2015, 2016; Verma et al., 2016, 2017; Xu et al., 2017). Additionally, given the inherent non-linear nature of heart rate and blood pressure signals, the CCM method holds advantage toward unraveling underlying directional information flow compared to traditional Granger causality, which has been limited in effect with signals of non-linear nature (Schiecke et al., 2016).

The non-linear causal relationship between heart rate and blood pressure in response to LBNP induced central hypovolemia remains to be generalized. The current work is an attempt toward studying this relationship in response to simulated hemorrhage with a hypothesis that such non-linear relationship can assist early detection of blood loss. As LBNP decreases venous return, and heart rate and blood pressure show causal interaction, we hypothesize a significant contribution from the feedback arm of such an interaction; that is, the baroreflex mediated heart rate control in response to LBNP.

# METHODS

# Experimental Protocol

The lower body of each participant was placed in the LBNP chamber and sealed at the level of the iliac crest. The participants lay supine inside the chamber for 5 min after which, the pressure inside the chamber was gradually reduced to −20 mmHg, from this point the chamber pressure was reduced in steps of 10 mmHg up to −40 mmHg. Five minutes of negative pressure was applied at each LBNP stage. A straddling bicycle seat inside the chamber prevented participants from getting further pulled inside the chamber. The chamber pressure was immediately terminated if a participant exhibited (1) pre-syncopal symptoms, (2) sudden drop in blood pressure and/or heart rate (3) any discomfort, or (4) upon request.

The experimentation was conducted in the Aerospace Physiology Laboratory in the Department of Biomedical Physiology and Kinesiology, Simon Fraser University (SFU), Canada. A participant in the age range of 18–40 years and without a history of cardiovascular disease was eligible to participate in the study. The experimental protocol was approved to be of minimal risk and complied with the rules and regulations set forth by the research ethics board of SFU. Written informed consent for participation was obtained from each participant prior to any experimentation. A registered nurse was present during experimentation for the safety of participants.

# Data Acquisition

Simultaneous electrocardiogram (ECG) and blood pressure were acquired from 27 young, healthy participants (15 males and 12 females, age: 27 ± 1 years, weight: 66 ± 2 kg, height: 169 ± 2 cm, mean ± SE) who underwent graded LBNP. The ECG signal was acquired in a lead II configuration using LifePak8 (Medtronic Inc., MN, USA) and the blood pressure signal was acquired using a finger photoplethysmograph cuff (FMS, Amsterdam, The Netherlands) applied on the mid phalanx of the middle finger (left hand). Five minutes of data were acquired during rest i.e., baseline and each LBNP stages using an NI 9205 analog input module (National Instruments Inc., TX, USA) at a sampling rate of 1,000 Hz.

# Convergent Cross Mapping

CCM proposed by Sugihara et al. is a non-linear approach of studying cause and effect relationship based on the state space reconstruction of a time series (Sugihara et al., 2012). The causality between two time series is inferred by studying the correspondence between the manifolds constructed using the optimal dimension of reconstruction and a delay (Krakovská et al., 2015; Ye et al., 2015). The history of response is used to estimate the states of the driver, contrary to the concept of causality proposed by Granger (Schiecke et al., 2015). If there exists a causal information flowing from X to Y (X→Y), then the states of X can be successfully estimated using the states of Y (Sugihara et al., 2012; McCracken, 2016). The strength of this causality is quantified by computing Pearson correlation coefficient between estimated and original variable. The causal information flowing from Y to X is represented as Y → X = <sup>ρ</sup>(Y, <sup>Y</sup><sup>ˆ</sup> <sup>|</sup>MX), similarly, the causal information flowing from X to Y is represented as <sup>X</sup> <sup>→</sup> <sup>Y</sup> <sup>=</sup> <sup>ρ</sup>(X, <sup>X</sup><sup>ˆ</sup> <sup>|</sup>MY).

Where M<sup>X</sup> and M<sup>Y</sup> represent the manifold of X and Y, respectively constructed using time delay and dimension of reconstruction, <sup>X</sup><sup>ˆ</sup> <sup>|</sup>M<sup>Y</sup> and <sup>Y</sup><sup>ˆ</sup> <sup>|</sup>M<sup>X</sup> are estimates of variable X (using Y manifold) and Y (using X manifold), respectively, and ρ is the Pearson correlation coefficient signifying the strength of coupling. Further details of the methodology to infer a causal relationship between two time series is presented in the supplementary material of Sugihara et al. (2012) and in a book on time series analysis by McCracken (2016).

# Data Processing

From the acquired signals, beat-to-beat R-R intervals, SBP, and diastolic blood pressure (DBP) were obtained using Beatscope software (Finapres, FMS, The Netherlands). The mean arterial pressure (MAP) was obtained from the systolic and DBP as; MAP = 2 <sup>3</sup> <sup>×</sup> DBP <sup>+</sup> 1 <sup>3</sup> <sup>×</sup> SBP. An evenly sampled signal was generated using spline interpolation from beat-to-beat signals and was resampled to 10 Hz prior to causality analysis. Only the last 3 min from each LBNP stage was considered for analysis to allow for effective blood pooling induced by LBNP and its effect on the cardiovascular parameters. The optimal dimension of reconstruction (M) was determined according to the false nearest neighbor algorithm in MATLAB using CRP toolbox (Kennel and Brown, 1992; Marwan, 2017), while the delay (τ ) was chosen to be 10 samples to account for changes within a heartbeat range. The false nearest neighbor minimization of approximately 95% was achieved for the last 3 min of SBP, MAP, and RR at M = 3. To limit the effect of noise, which may otherwise lead to a determination of higher dimension of reconstruction, all causality values reported were performed at M = 3 and τ = 10. For each stage, the correlation coefficient value at which each causal event (i.e., SBP→RR, MAP→RR, RR→SBP, and RR→MAP) converged was considered as a degree of directional information flow from one variable to another. The significance of correlation was set at α = 0.05. The MATLAB (Mathworks Inc., MA) implementation of CCM algorithm demonstrated in an application with non-linear signals in the study conducted by Krakovská et al. (2015) was considered for analysis.

# Statistical Analysis

The group mean of RR, SBP, DBP, MAP, pulse pressure (PP = SBP-DBP), the non-baroreflex i.e., feedforward (RR→SBP and RR→MAP), and the baroreflex i.e., feedback (SBP→RR and MAP→RR) causality values for the last 3 min of each LBNP stage was obtained. Test for normality of the data was conducted using the Shapiro-Wilk test at α = 0.05. A one-way test of ANOVA (normally distributed data) or Wilcoxon rank sum test (data failed the normality test) was conducted to test the significance of the difference. The group mean of baroreflex causality (SBP→RR and MAP→RR) was compared with the group mean of nonbaroreflex causality (RR→SBP and RR→MAP) under baseline using one-way ANOVA. A multiple comparison test, to account for the significance of the difference in the cardiovascular parameters and the SBP↔RR and MAP↔RR causality, inflicted by different LBNP stages, was conducted using one-way ANOVA followed by post-hoc analysis using the Tukey-HSD method. All tests for significance were conducted using a statistical toolbox of MATLAB (Mathworks Inc., MA, USA). The test result at α = 0.05 was considered as significant. All tabular results are presented as mean ± SD while all graphical results are presented as mean ± SE unless mentioned otherwise.

# RESULTS

The cardiovascular parameters (RR, SBP, DBP, MAP, PP), as well as the non-baroreflex and the baroreflex causal events, passed the test of normality (p > 0.05). The behavior of cardiovascular parameters with progressive LBNP is summarized in **Figure 1**. The effect of different LBNP stages on the cardiovascular parameters was compared using one-way ANOVA followed by post-hoc analysis using the Tukey-HSD method, the resulting p-value of the comparison is summarized in **Table 1**. A significant reduction in pulse pressure (p = 0.001) at −40 mmHg, and RR interval at −30 mmHg (p = 0.001), and −40 mmHg (p < 0.001) was observed compared to the resting stage. Additionally, a significant difference was observed between −20 and −40 mmHg LBNP in PP (p = 0.02) and RR interval (p < 0.001), **Figure 1**. In response to different LBNP stages, we observed no change in SBP (p = 0.50), DBP (p = 0.79), and MAP (p = 0.99), **Figure 1**.

The compensatory behavior, responsible for maintaining blood pressure equilibrium via SBP↔RR and MAP↔RR interaction was quantified in response to progressing LBNP. This behavior is summarized in **Figure 2**. During the resting stage, a significantly stronger causal drive from heart rate toward SBP (RR→SBP) and MAP (RR→MAP) in comparison to causal drive in the reverse direction (SBP→RR and MAP→RR; **Figure 3**) was observed. With the increase in the lower-body negative pressurization, the causal strength of the baroreflex arm (MAP→RR) increased and achieved a statistical significance as early as −30 mmHg LBNP compared to rest while the baroreflex causality via SBP→RR interaction did not change until −40 mmHg LBNP (**Figure 2**). In response to LBNP, the RR→SBP (p = 0.76) and RR→MAP (p = 0.60) causality did not change (**Figure 2**). The SBP→RR (p = 0.04) and MAP→RR (p = 0.01) successfully differentiated −40 mmHg LBNP from −20 mmHg LBNP. **Table 2** lists the value of all variables studied in this research (mean ± SD) in response to LBNP.

**Figure 4** details the percent change in baroreflex causality (MAP→RR) at −30 mmHg LBNP compared to resting stage for individual study participants. Eighty two percentage participants showed an increase in MAP→RR causality during −30 mmHg LBNP compared to baseline. The group mean of baroreflex causalities (SBP→RR and MAP→RR), RR, and PP other than able to differentiate moderate category (−30 or −40 mmHg LBNP) hemorrhage from resting baseline, showed a significant correlation (p < 0.05) with the LBNP stages. Also, the group mean of SBP, DBP, and RR→SBP was significantly (p < 0.05) correlated with LBNP stages while the group mean of MAP (p = 0.26) and RR→MAP (p = 0.14) showed no relation with the LBNP stages.

# DISCUSSION

The major finding of the current research was the feasibility of the causal heart rate and blood pressure interaction toward differentiating the onset of moderate intensity of simulated hemorrhage from the resting stage. A significant elevation was observed in the strength of baroreflex arm (MAP→RR) of the heart rate blood pressure interaction in response to the LBNP induced redistribution of central blood volume to the lower peripheral parts. This elevation was evident as early as −30 mmHg of LBNP. On the contrary, the conventional indicators of blood loss i.e., blood pressure (SBP, DBP, MAP, or PP) failed to differentiate −30 mmHg LBNP intensity from the resting stage, thus, highlighting the potential of compensatory baroreflex arm of the heart rate and blood pressure interaction toward early detection of progressing blood loss.

Early detection of hemorrhage progression has eluded the research and clinical communities for years. Stroke volume is shown to be an early indicator of mild hypovolemia or blood loss (Westphal et al., 2007; Elstad et al., 2009; Holme et al., 2016), however, an accurate measurement of stroke volume entails sophisticated instrumentation and an expert operator (Scherhag et al., 2005), thus, limiting its application in the environment where majority of hemorrhagic shock occurs. Therefore, non-invasively acquired arterial blood pressure is often relied on for surgical triage. Nevertheless, numerous physiological mechanisms responsible for arterial blood pressure regulation have limited the efficacy of blood pressure from exhibiting early symptoms of blood loss from the circulation. To limit mortality, corresponding to hemorrhagic shock via effective triage, there is a profound interest toward early detection of hemorrhage non-invasively. Pulse pressure has been proposed in the literature to obtain early insights regarding hemorrhage progression, for its association with stroke volume (Convertino

−40 (*p* < 0.001) mmHg LBNP compared to rest. Pulse pressure (B) reduced significantly (*p* = 0.001) at −40 mmHg compared to rest. Additionally, both R-R intervals (*p* < 0.001) and pulse pressure (*p* = 0.02) reduced significantly at −40 mmHg LBNP compared to −20 mmHg LBNP. The systolic blood pressure (B), diastolic blood pressure (B), and the mean arterial pressure (B) did not change (*p* = 0.50, *p* = 0.79, and *p* = 0.99, respectively) in response to graded lower-body negative pressure, \* and † represents significant change (*p* < 0.05, *post-hoc* result) compared to rest and −20 mmHg, respectively.

et al., 2006). In an alignment with existing findings in the literature, the current research found a strong and significant correlation between PP and LBNP stages. However, PP (p = 0.10) failed to show a significant change in its dynamics compared to resting stage until the tail end of moderate category hemorrhage (−40 mmHg), which raises concern regarding its potential to provide an early trace of central blood loss.

Furthermore, cardiovascular parameters such as, SBP and DBP were significantly (p < 0.05) correlated with LBNP stages, however, neither of them showed a significant change in their dynamics with the application of LBNP simulating moderate category hemorrhage (**Figure 1**). This observation further highlighted the ineffectiveness of the cardiovascular parameters alone as an early indicator of blood loss. Out of all cardiovascular parameters studied in this research, we found heart rate to be most sensitive to LBNP, showing significant change (p = 0.001) as early as −30 mmHg compared to resting baseline (**Figure 1**). However, making a triage decision based on heart rate alone has been a subject of controversy due to the association of heart rate with several physiological mechanisms (Victorino et al., 2003; Brasel et al., 2007) as well as owing to existing discrepancies in the literature whether heart rate is an early or late marker of central hypovolemia (Convertino et al., 2006, 2016; Cooke et al., 2006; Soller et al., 2008). In addition to heart rate, pulse pressure was observed to be superior to SBP, MAP, and DBP when tracking simulated blood loss via LBNP. Even though PP failed to differentiate −30 mmHg from baseline, it successfully differentiated tail end of moderate category hemorrhage (−40 mmHg) from resting stage and −40 mmHg LBNP from −20 mmHg LBNP, while SBP, DBP, and MAP showed no change with the application of LBNP (**Figure 1**).

To impede progressing hemorrhage, it is central to quantify the compensatory mechanisms regulating arterial blood pressure. To this end, the current research investigated the sensitivity of causal heart rate and blood pressure interaction for monitoring simulated central hypovolemia. Furthermore, owing to the robustness of the non-linear methodology, we successfully highlighted the contribution of the non-baroreflex (RR→SBP and RR→MAP) and the baroreflex (SBP→RR and MAP→RR) mechanisms responsible for blood pressure regulation under a variable degree of LBNP induced physiological stressor (**Figure 2**). The directional information flow, mediated by both baroreflex and non-baroreflex arms of the interaction was observed, with the non-baroreflex (RR→SBP and RR→MAP) arm being dominant (p < 0.001) during the resting stage compared to the baroreflex arm (SBP→RR and MAP→RR), **Figure 3**.

With the application of external perturbation to the hemodynamic homeostasis in the form of lower body negative pressure, an elevation in the causal activity, in the direction of blood pressure to heart rate (SBP→RR and MAP→RR) was observed, representing activation of baroreflex mediated control of heart rate toward maintenance of arterial blood pressure homeostasis. Progression of central hypovolemia to moderate intensity (−30 mmHg) was accompanied by no significant elevation in the SBP→RR (p = 0.07) causality but significant elevation in the causal drive from MAP→RR (p = 0.004) while no change was observed in the strength of reverse drive, which is a representative of heart rate mediated blood pressure changes i.e., RR→SBP (p = 0.76) and RR→MAP (p = 0.60). This observation indicated that under resting condition blood pressure is primarily maintained through blood pressure changes mediated by heart rate, contrarily, under physiologically perturbed cardiovascular system due to a decline in venous return, the baroreflex mediated heart rate control acts as a compensatory mechanism leading to arterial blood pressure homeostasis. Thus, the two blood pressure regulatory mechanisms interact in closed loop at any given time in order to maintain blood pressure homeostasis. Furthermore, the baroreflex causality (SBP→RR and MAP→RR)

Verma et al. Blood Pressure Regulation during Simulated Hemorrhage

TABLE 1 | Comparison of the response of variables to different LBNP stages.


*The table lists post-hoc analysis (Tukey-HSD method) p-value for different comparisons.* \* *Represents significant difference (p* < *0.05). Only the variables with significant (p* < *0.05) one-way ANOVA results are listed.*

*R-R, RR intervals; PP, Pulse pressure; SBP*→*RR, systolic blood pressure to heart rate causality (baroreflex); MAP*→*RR, mean arterial pressure to heart rate causality (baroreflex); LBNP, Lower-body negative pressure.*

TABLE 2 | Response of different variables to graded lower-body negative pressure.


*The table lists the value of each variable for respective LBNP stages.* \**Represents significant (Tukey-HSD post-hoc analysis, p* < *0.05) difference compared to rest and † represents significant difference compared to* −*20 mmHg.*

*LBNP, Lower-body negative pressure; R-R, RR intervals; SBP, systolic blood pressure; DBP, diastolic blood pressure; MAP, mean arterial pressure; PP, pulse pressure; RR*→*SBP, heart rate to SBP causality (non-baroreflex); SBP*→*RR, SBP to heart rate causality (baroreflex); RR*→*MAP, heart rate to MAP causality (non-baroreflex); and MAP*→*RR, MAP to heart rate causality (baroreflex).*

was able to differentiate −40 mmHg from −20 mmHg, therefore highlighting its capability to track and differentiate varying intensity of hemorrhage (**Table 1**).

This behavior of closed loop heart rate and blood pressure interaction under varying physiological conditions ascertained the contribution of either arm of the blood pressure regulation mechanism. The findings of the current research corroborated with the previous findings regarding heart rate and blood pressure interaction highlighted under head-up tilt and the stand test, which demonstrated an elevation in the baroreflex activity during orthostatic challenge compared to baseline with SBP being an indicator of blood pressure functioning (Czippelova et al., 2014; Javorka et al., 2017; Silvani et al., 2017). However, the literature is limited in terms of comprehensive quantified knowledge of such non-linear behavior with respect to progressing blood loss simulated by LBNP.

The report by Dorantes-Mendez et al. (2013) and Silvani et al. (2017) investigated heart rate and blood pressure coupling with respect to LBNP and actual blood loss, respectively. However, a linear methodological approach was considered for such quantification. Moreover, the heart rate and blood pressure interaction is known to be of non-linear nature, therefore, a more robust approach would be a prerequisite, for accurately underpinning the continuous dynamics of the non-baroreflex and the baroreflex arms of such interaction. With the application of non-linear methodology and higher sample size compared to previous two works, in the current research, we systematically demonstrated the degree of statistical alteration in both the nonbaroreflex and the baroreflex mechanisms of blood pressure regulation in response to the simulated progressing hemorrhage (LBNP). Our study, therefore, provided comprehensive insights regarding the feasibility of compensatory directional interaction for monitoring progression of hemorrhage and for surgical triage.

Additionally, we compared the use of SBP and MAP as a marker of baroreflex mediated heart rate control (baroreflex causality) in response to LBNP. As such, we found MAP to be a more sensitive marker of baroreflex mediated heart rate control in response to moderate category (−30 mmHg) LBNP compared to SBP (**Figure 2**). The MAP→RR causality achieved statistical significance at −30 mmHg LBNP compared to baseline while the SBP→RR causality did not show a significant change in its dynamics until −40 mmHg LBNP. However, SBP→RR and MAP→RR both showed a significant change in their dynamics at −40 mmHg compared to baseline and −20 mmHg (**Figure 2**). This observation leads us to conclude that MAP→RR causality is more sensitive to the early phase of central blood loss simulated by LBNP, thus, a better marker of baroreflex activity when aimed to gain early information regarding progressing blood loss from the circulation. MAP perhaps is a better indicator of early phase of central blood pooling in the lower limbs compared to SBP due its relationship with cardiac output (CO), SVR, and central venous pressure (CVP); MAP = (CO × SVR) + CVP (Klabunde, 2011), validation of this hypothesis under gravityinduced orthostatic stress (head-up tilt or stand test) requires future work.

The behavior of non-linear causal heart rate and blood pressure interaction in response to simulated hemorrhage via LBNP is not extensively explored in the literature. The references that exist regarding such behavior have been outlined under orthostatic challenge induced via head-up tilt and stand test, which have considered SPB as a marker of arterial blood pressure. The LBNP is hypothesized to exert orthostatic challenge on the human body analogous to head-up tilt (Taneja et al., 2007; Goswami et al., 2008). Therefore, the physiological response to LBNP is expected to be analogous to that of head-up tilt and quiet standing. Nevertheless, the outcomes of some studies have highlighted the differences in the cardiovascular, cerebrovascular, and the hormonal responses when using LBNP to evoke orthostatic challenge compared to head-up tilt (Hinghhofer-Szalkay et al., 1996; Kitano et al., 2005; Taneja et al., 2007; Bronzwaer et al., 2017).

The absence of gravity induced hydrostatic gradient during lower-body negative suction is shown to be the major contributor toward the existence of such difference. The application of LBNP empties splanchnic blood volume (analogous to hemorrhage)

while the gravity induced orthostatic stress (such as, headup tilt) increases the blood volume in the splanchnic bed (Taneja et al., 2007). Recent work by Silvani et al. (2017) further highlighted such discrepancy, where a significant change in baroreflex response (using SBP) during head-up tilt but not during 1,000 ml of blood loss was observed, similarly, the baroreflex response (SBP→RR) in our analysis failed to differentiate −30 mmHg LBNP (equivalent to approximately 1,000 ml of blood loss) from baseline. These observations, besides highlighting the fact that baroreflex response differs between head-up tilt and central blood loss, raise concern regarding SBP

\*Represents significantly (*p* < 0.05, one-way ANOVA) stronger causality.

as baroreflex marker when aimed to gain early insights regarding blood loss.

The discrepancies that might exist in the blood pressure regulation via causal heart rate and blood pressure interaction during orthostatic challenge induced by the application of LBNP compared to head-up tilt or stand test is not the scope of this article, and future work shall follow to address such concerns to further our understanding regarding underlying physiology. The current article aimed at investigating the capability of causal heart rate and blood pressure interaction in tracking progressing simulated hemorrhage. In such context, the findings of the study

are promising and underscored capability of MAP→RR causality to differentiate moderate category hemorrhage (−30 mmHg LBNP) from resting baseline.

Although a significant (p = 0.004) increase in the baroreflex arm (MAP→RR) of the interaction was observed in response to early phase of moderate category hemorrhage (−30 mmHg LBNP; **Figure 2**), to gain clinical applicability toward early detection of central hypovolemia, it is necessary to outline the behavior of such compensatory mechanism for individual participants. To this end, the MAP→RR causal activity at resting stage (baseline) was compared with that of −30 mmHg for each participant. **Figure 4** details the percent increase or decrease in the behavior of MAP→RR causal activity at −30 mmHg compared to rest for all study participants.

In 22 out of 27 participants, there was an increase in the baroreflex causality (MAP→RR), in 2 participants the causality decreased at −30 mmHg, while in three participants the causality did not change (increase/decrease <1%; **Figure 4**). Additionally, an abrupt decline in the strength of MAP→RR causality in four participants was observed at −40 mmHg compared to −30 mmHg, which could be symptomatic of pre-syncopal feeling or autonomic decompensation. As such, the relationship between the strength of MAP→RR causality and pre-syncopal feeling shall be explored in the future. The percent increase or decrease behavior of non-baroreflex arms (RR→SBP and RR→MAP) were highly variable, moreover, the non-baroreflex causal activities did not achieve statistical significance (p > 0.05) in response to LBNP stages. Accordingly, the outcome of the analysis highlights the sensitivity of the baroreflex mediated heart rate control (MAP→RR) toward monitoring progression of central hypovolemia, ascertained by its significant increase in the strength as early as the onset of moderate category hemorrhage (−30 mmHg LBNP).

# Limitations and Future Directions

Despite the fact that the current study highlighted the potential of quantified compensatory mechanism toward early detection of hemorrhage progression, there exist certain limitations that need to be discussed. (1) In the case of hemorrhage simulated by LBNP, blood volume is not lost from the circulatory system compared to actual hemorrhage but sequestered in the lower periphery of the body, which reduces venous return and therefore, preload. Since the blood is not lost from the circulatory system, inter-subject variability is expected, due to variable tolerance level of individual participants (Hinojosa-Laborde et al., 2014a). (2) In the current study, the respiration and calf electromyography signals were not acquired, therefore the role of respiration and skeletal muscle pump toward facilitating venous return to the heart was not considered (Rowell, 1993; Miller et al., 2005; van Dijk et al., 2005), and it could also have contributed to the variability in the baroreflex causality discussed in **Figure 4**. Additionally, the respiration is known to affect both the arterial blood pressure and heart rate and should be incorporated in the causality analysis along with heart rate and arterial blood pressure (Faes et al., 2011; Porta et al., 2012). Thus, the generalization of the behavior of baroreflex mediated heart rate elevation in response to blood loss and potential clinical application toward early detection of hemorrhage requires further study.

(3) The degree of blood pooling achieved as a consequence of LBNP was unknown. To accurately quantify the redistribution of blood volume to the peripheral regions, future work could utilize near-infrared spectroscopy (NIRS) in the experimental protocol, analogous to Blaber et al. (2013). The investigation of the strength of SBP→RR and MAP→RR causalities in relation to the degree of blood pooling, quantified via NIRS will shed further light on the effectiveness of SBP→RR or MAP→RR as a choice of baroreflex marker in response to central hypovolemia.

(4) The signals considered for quantifying directional interaction are commonly employed vital sign monitors, therefore, clinical application in a hospital setting is feasible. For application in a battlefield, home, or rural settings, a less sophisticated way of blood pressure estimation, e.g., pulse transit time (Mukkamala et al. (2015) should be considered, which would require a sensor placement on the xiphoid process and on a finger (Verma et al., 2015a,b; Yang and Tavassolian, 2017). Feasibility of such system should be investigated in the future.

# CONCLUSION

The current work studied the compensatory behavior responsible for blood pressure regulation with the progressing LBNP, of which the findings showcase the potential of causal heart rate and blood pressure interaction toward differentiating the severity of simulated hemorrhage. We demonstrated that nonbaroreflex causality dominates during resting stage followed by no significant (p > 0.05) change in response to the LBNP. On the contrary, the baroreflex causality increased with the progression of LBNP and can be differentiated from rest at the onset of a moderate level of simulated hemorrhage (−30 mmHg LBNP).

Additionally, elevation in the baroreflex arm of the interaction was observed in 82% of the study participants, which further

# REFERENCES


accentuates its potential toward early identification of blood loss. The inter-subject variability in the baroreflex causality in response to progressing hypovolemia due to several physiologic factors such as, tolerance level, presyncope symptom, and autonomic decompensation should be addressed in the future work to improve clinical applicability. Pulse pressure, traditionally considered an early indicator of blood loss was unsuccessful in differentiating progression of central hypovolemia until its advancement to the tail end of the moderate category of hemorrhage (−40 mmHg). Therefore, the compensatory mechanisms shall also be considered in addition to the traditional non-invasive markers of central blood loss (i.e., heart rate and arterial pulse pressure) to gain early and reliable insights regarding the progression of hemorrhage for effective surgical triage.

# AUTHOR CONTRIBUTIONS

AV and KT conceived research. AB and KT designed experiment. AB and KT performed data acquisition. DX and KT preprocessed data. AV analyzed data, performed statistical analysis, created figures and tables, and drafted the manuscript. AV, DX, AG, AC, NG, AB, and KT, interpreted results. All authors read, edited, and approved the final manuscript for publication.


transit time: theory and practice. IEEE Trans. Biomed. Eng. 62, 1879–1901. doi: 10.1109/TBME.2015.2441951


between heart period and arterial pressure in response to postural changes in humans. Front. Physiol. 8:163. doi: 10.3389/fphys.2017.00163


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2017 Verma, Xu, Garg, Cote, Goswami, Blaber and Tavakolian. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Human Locomotion in Hypogravity: From Basic Research to Clinical Applications

Francesco Lacquaniti 1, 2, 3 \*, Yury P. Ivanenko<sup>3</sup> , Francesca Sylos-Labini 2, 3 , Valentina La Scaleia2, 3, Barbara La Scaleia<sup>3</sup> , Patrick A. Willems <sup>4</sup> and Myrka Zago<sup>3</sup>

<sup>1</sup> Department of Systems Medicine, University of Rome Tor Vergata, Rome, Italy, <sup>2</sup> Center of Space BioMedicine, University of Rome Tor Vergata, Rome, Italy, <sup>3</sup> Laboratory of Neuromotor Physiology, IRCCS Santa Lucia Foundation, Rome, Italy, <sup>4</sup> Laboratory of Biomechanics and Physiology of Locomotion, Institute of NeuroScience, Université catholique de Louvain, Louvain-la-Neuve, Belgium

We have considerable knowledge about the mechanisms underlying compensation of Earth gravity during locomotion, a knowledge obtained from physiological, biomechanical, modeling, developmental, comparative, and paleoanthropological studies. By contrast, we know much less about locomotion and movement in general under sustained hypogravity. This lack of information poses a serious problem for human space exploration. In a near future humans will walk again on the Moon and for the first time on Mars. It would be important to predict how they will move around, since we know that locomotion and mobility in general may be jeopardized in hypogravity, especially when landing after a prolonged weightlessness of the space flight. The combination of muscle weakness, of wearing a cumbersome spacesuit, and of maladaptive patterns of locomotion in hypogravity significantly increase the risk of falls and injuries. Much of what we currently know about locomotion in hypogravity derives from the video archives of the Apollo missions on the Moon, the experiments performed with parabolic flight or with body weight support on Earth, and the theoretical models. These are the topics of our review, along with the issue of the application of simulated hypogravity in rehabilitation to help patients with deambulation problems. We consider several issues that are common to the field of space science and clinical rehabilitation: the general principles governing locomotion in hypogravity, the methods used to reduce gravity effects on locomotion, the extent to which the resulting behavior is comparable across different methods, the important non-linearities of several locomotor parameters as a function of the gravity reduction, the need to use multiple methods to obtain reliable results, and the need to tailor the methods individually based on the physiology and medical history of each person.

Keywords: human locomotion, body weight support, hypogravity simulators, moon walk, parabolic flight, locomotion rehabilitation, robotic gravity-assist

# INTRODUCTION

Human missions are considered vital for harvesting the maximum benefits from space exploration (White and Averner, 2001). However, space travelers are exposed to several challenges and risks, ranging from radiation to isolation and altered gravity effects. Here we are concerned specifically with the effects of reduced gravity (hypogravity, 0 < g < 1) on locomotion, such as it would be

#### Edited by:

Olivier White, INSERM U1093, Université de Bourgogne Franche Comté, France

#### Reviewed by:

Christopher E. Carr, Massachusetts Institute of Technology, United States Marcel Egli, Lucerne University of Applied Sciences and Arts, Switzerland

> \*Correspondence: Francesco Lacquaniti lacquaniti@med.uniroma2.it

#### Specialty section:

This article was submitted to Integrative Physiology, a section of the journal Frontiers in Physiology

Received: 07 September 2017 Accepted: 24 October 2017 Published: 07 November 2017

#### Citation:

Lacquaniti F, Ivanenko YP, Sylos-Labini F, La Scaleia V, La Scaleia B, Willems PA and Zago M (2017) Human Locomotion in Hypogravity: From Basic Research to Clinical Applications. Front. Physiol. 8:893. doi: 10.3389/fphys.2017.00893

**129**

experienced on the Moon or Mars. In this regards, a recent survey (White et al., 2016) has remarked that, while the effects of sustained weightlessness (0 g) on sensory and motor functions have been rather extensively investigated, much less is known about the effects of sustained hypogravity on these functions. White et al. (2016) further remarked that the transient and adaptive effects on sensory-motor functions have been wellstudied following transitions from 0 to 1 g (and, to a lesser extent, from 1 to 0 g), whereas little is known about the transitions from 1 g to hypo-g and vice versa.

Examination of the records from previous space flights, especially at low-Earth orbits, revealed that astronauts often suffer from several problems related to the locomotor system, from osteoporosis to muscle atrophy, changes of tendons elasticity and altered neural control of posture and movement (Bloomberg et al., 2016; White et al., 2016; Lang et al., 2017). These problems are exacerbated after long-duration missions. Moreover, landing on the Moon or Mars will further challenge the motor control system because of the sudden transition from the weightlessness of the space travel to the restored albeit reduced gravity of the Moon (0.16 g) or Mars (0.38 g). In such circumstances, motor problems may lead to disastrous outcomes in an environment with limited or no medical assistance. A simple fall due to muscle weakness or impaired control of posture and movement can lead to muscle strains, ligament sprains, bone fractures, head traumas, or other more or less severe injuries that would be very difficult to treat in an alien environment. Damages to the spacesuit or the portable life support system resulting from a fall may also be life-threatening. These and related issues have been brought to the attention of the space biomedical community by NASA (see NASA Roadmap for Human Health Risks), ESA, and EC (see THESEUS roadmap). It has been remarked that no existing countermeasure is deemed sufficient to reduce some of these risks to an acceptable level.

The main current countermeasure for preventing disorders of the locomotor apparatus in space is represented by daily, intensive exercises, especially walking and running on a treadmill. Unfortunately, these exercises prevent the motor problems only to a limited extent. Thus, upon return to Earth, most astronauts exhibit significant loss of bone and muscle tissue, as well as a significant reduction of mobility (Mulavara et al., 2010). We believe that the limited success of current countermeasures is due to the lack of sufficient knowledge of the impact of reduced gravity on the motor system, and therefore to the inadequate design of countermeasures including exercise devices and protocols for the astronauts.

A different but related topic is represented by the application of space research to medicine on Earth. There are two aspects to be considered. First, understanding how astronauts adapt to and recover from micro- or hypo-gravity could also benefit patients on Earth who exhibit a decline in motor performance due to a disabling disease or prolonged bed-rest that limited the effects of gravity on their movements. Second, a simulation of reduced gravity—such as with body weight unloading—is used more and more in rehabilitation to help patients with motor disorders. Therefore, a better understanding of the physiological adaptation of locomotor control to hypogravity may help designing more effective simulators for rehabilitation.

Here we first review the basic principles underlying locomotion under gravity, including some developmental and evolutionary issues, then we describe the locomotor behavior that has been monitored in actual hypogravity (Moon, parabolic flight) and simulated hypogravity (Earth laboratory). Finally, we examine current assessments of the therapeutical efficacy of body weight unloading in several pathologies. Specific consideration will be given to the methodologies for simulating hypogravity in the laboratory for research and in clinical environments for rehabilitation.

The present article aims at providing an update relative to the previous reviews on similar topics (e.g., Davis and Cavanagh, 1993; Lackner and DiZio, 1996, 2000; Reschke et al., 1998; Newman, 2000; Bloomberg and Mulavara, 2003; Clément et al., 2005; Clément and Reschke, 2008; Sylos-Labini et al., 2014b; White et al., 2016). Because of space limitations, we consider a number of hypogravity conditions relevant to locomotion but we exclude water immersion, which allows only limited locomotion in humans (e.g., Duddy, 1969; Watenpaugh, 2016). Moreover, we focus on the motion of the center of body mass (COM) and limbs, while for gaze and postural control in hypogravity the reader may want to consult Bloomberg et al. (1997), Lackner and DiZio (2000), Mulavara and Bloomberg (2002), Clément and Reschke (2008), Mulavara et al. (2012).

# GRAVITY EFFECTS ON LOCOMOTION

Gravity effects are essential for locomotion in contact with a support surface (**Figure 1A**). The downward force of gravity is resisted by the vertical ground reaction force (GRF), which a scale measures as body weight (mg) under static posture. When walking, the vertical GRF is time-varying, due to inertia (**Figure 1B**). Because of gravity, we must perform work on the COM at each step, even on a level surface (Cavagna et al., 2000). A full description of the mechanics of human locomotion is complex, given the great number of kinematic degrees of freedom and the even greater number of muscles that are involved at each step. Comprehensive analytical models defining the contributions of individual muscles, joints, and tendons are very difficult to be set up, and they may hide rather than reveal the general principles underlying the energetics of locomotion. By elaborating on the simplest possible mechanical model, instead, we can discern how the control of human locomotion takes advantage of the physical dynamics of bipedal gait (Kuo, 2002).

The simplest mechanical models of human locomotion involve two different oscillatory modes as a function of speed: the pendular mechanism of walking at low speeds and the bouncing mechanism of running at high speeds (Cavagna and Margaria, 1966; Mochon and McMahon, 1980; Full and Koditschek, 1999; Cavagna et al., 2000). Both mechanisms can be predicted theoretically by assuming that the specific gait mode is chosen based on energy optimization, and that energy cost is proportional to muscle work. In fact, these two mechanisms are selected—out of an infinite variety of different gait styles—by a

(±SD, 20 subjects, normalized to individual body weight BW) of the time profile of the vertical ground reaction force (GRF, recorded with a force plate) plotted vs. normalized stance duration. Data replotted from Martino et al. (2014). (C) Changes of the vertical component of in-shoe reaction forces plotted as a function of the spatial coordinates of the foot (outline in red corresponding to the outer elements of the pressure-sensitive matrix interposed between foot and shoe) at three different levels of gravity obtained in the laboratory with a vertical body-weight supporting (BWS) system: 1 g (0 BWS), 0.5 g (50% BWS), and 0.25 g (75% BWS). F1 and F2 denote the two main peaks of force. (D) Mean values of the first (F1) and second (F2) peaks derived from force records such as those of C plotted as a function of gravity level. Data of (C,D) are replotted from Ivanenko et al. (2002).

computer model minimizing the mechanical cost of transport (Srinivasan and Ruina, 2006). At the top speeds of running, however, performance (push-average-power) rather than energy cost is minimized (Cavagna et al., 1991).

# Pendulum-Like Mechanism in Walking

In an idealized fashion, walking is represented by an inverted pendulum under gravity (**Figure 1A**). The COM vaults along a circular arc about the supporting foot, slowing down as it rises, and speeding up as it falls. As a result, the changes of gravitational potential energy of the COM tend to be opposite to those of the corresponding forward kinetic energy, and mechanical energy is saved. A more realistic version of this model takes into account the fact that COM motion is only pendulum-like, and work needs to be performed to raise and lower the COM during single support, and to redirect the velocity vector of COM at the transition from one stance leg to the next (Cavagna et al., 1976; Neptune et al., 2004; Kuo et al., 2005). The maximum energy recovery due to the exchange of potential and kinetic energies is about 65% at the optimal speed of about 5.5 km/h at Earth gravity (Cavagna et al., 1983). Energy recovery is defined as the ratio between the external work saved by the pendulum-like mechanism and the maximum work that would occur without any energy exchange at the COM due to the pendulum mechanism. Recovery is maximum when the changes of gravitational potential energy are about the same amplitude and 180◦ out-of-phase relative to the changes of forward kinetic energy. Deviating substantially from a pendulum-like behavior can lead to significant increases of energy expenditure, as when people try to walk as level as possible by reducing vertical COM displacements (Massaad et al., 2007; Gordon et al., 2009). In addition to the inverted pendulum behavior of the stance limb, the contralateral limb simultaneously swings about the hip as an upright pendulum.

The pendulum-like features of walking are common to all individuals, but subjects differ in their ability to minimize energy oscillations of their body segments and to transfer mechanical energy between the trunk and the limbs (Bianchi et al., 1998a), based on a different tuning of intersegmental coordination (Bianchi et al., 1998b). Moreover, people in specific environments show an improved efficiency of the pendulum mechanism. Thus, African women from the Kikuyu and Luo tribes can carry heavy loads on their head without increasing the rate of metabolic energy consumption (Maloiy et al., 1986), and they do so by means of an improved pendulum-like transfer between gravitational potential energy and kinetic energy at the COM (Heglund et al., 1995; Cavagna et al., 2002).

# Pogo-Stick Mechanism in Running

During running, the oscillations of the COM can be equated to those of a spring-mass system bouncing on the ground, as a bouncing ball or pogo-stick. In this case, gravitational potential

energy and forward kinetic energy at the COM change inphase (Cavagna et al., 1964). At each step the muscle–tendon units absorb and restore both the kinetic energy change of forward motion, due to the braking action of the ground, and the gravitational potential energy change, associated with the fall and the lift of the COM. This results in a large amount of negative and positive work and the chemical energy cost per unit distance is twice as large as that spent in walking at the optimal speed (Margaria, 1938). The metabolic energy expenditure is reduced in running, however, by an elastic storage and recovery of mechanical energy. In fact, when the leg strikes the ground, mechanical energy is temporarily stored as elastic strain energy in muscles, tendons, and ligaments and then it is partially recovered during the propulsive second half of the stance phase. Mechanical energy recovery in running has been estimated to be about 0.55 near the walk to run transition speed, and declining at higher speeds (Kaneko, 1990; Carr and Newman, 2007, 2017).

# Principle of Dynamic Similarity

The principle of dynamic similarity (Alexander, 1989) states that dynamically similar bodies have the same gait when the horizontal speed of COM v is normalized as the dimensionless Froude number: Froude = v 2 /gL, where L is the leg length and g is the acceleration of gravity. The Froude number is proportional to the ratio between kinetic energy and gravitational potential energy. It was originally proposed by William Froude (1874) to compare the hydrodynamic behavior of ships of very different sizes, and it was later applied to locomotion by Thompson (1917) and Alexander (1976), as well as others (see Vaughan and O'Malley, 2005). Thompson (1917) applied the Froude number to compare the theoretical stride length of the inhabitants of Lilliput and Brobdingnag, based on the heights of these people reported in Gulliver's Travels. Alexander (1976) used the Froude number to estimate the speeds of dinosaurs based on their footprints, and then generalized the regression of relative stride length and Froude number across several living quadrupedal mammals (Alexander and Jayes, 1983).

In theory, walking becomes impossible when the centrifugal force (mv<sup>2</sup> /L) exceeds the centripetal force due gravity (mg); above this limit, the body would take off in the aerial phase of running (Alexander, 1989). This corresponds to a Froude number greater than 1 (about 10 km/h for an average human at Earth gravity). The ratio of centrifugal to centripetal force constrains the walk-to-run transition to values of Froude number equal or less than 1, but does not specify a unique value. However, the theory of dynamic similarity (Alexander, 1989; Bullimore and Donelan, 2008) predicts a constant value of Froude number for the walk-to-run transition despite changes in g and L. In practice, on Earth, running becomes energetically more efficient (lower oxygen consumption) than walking at much lower speeds than the upper limit, namely speeds of about 7–8 km/h corresponding to a Froude number of about 0.5. It has also been found that a Froude number of about 0.25 corresponds to the optimal speed of walking (about 5 km/h), that is, the speed associated with the maximal recovery of mechanical energy (Cavagna et al., 1976; Saibene and Minetti, 2003).

Dynamic similarity implies that the recovery of mechanical energy in subjects of short height, such as children (Cavagna et al., 1983; DeJaeger et al., 2001; Ivanenko et al., 2004a), pygmies (Minetti et al., 1994), and people with dwarfism (Minetti et al., 2000), is not different from that of normal-sized adults when compared at the same Froude number. With respect to the present topic, dynamic similarity also implies that both the optimal walking speed and the walk-to-run transition speed depend on the gravitational level (Margaria and Cavagna, 1964). Indeed, the specific value of the acceleration of gravity g appears in the denominator of the ratio that defines the Froude number. Thus, the lower the gravity level, the lower will be the optimal walking speed and the walk-to-run transition speed (Saibene and Minetti, 2003). Moreover, we note that the limb geometry and the musculoskeletal apparatus contribute to the oscillatory behavior of the COM, and that general anatomy and individualized limb segment proportions are optimized in such a way that the Froude number can explain optimal walking speed. Indeed, when the relative proportion of lower limb segments is artificially changed by wearing stilts, one finds that the minimum metabolic cost per unit distance occurs at a lower Froude number than it does when walking without stilts (Leurs et al., 2011).

# DEVELOPMENTAL CONSIDERATIONS

We face gravity effects on the body as soon as we are born. Newborn babies exhibit a number of antigravity responses, such as the righting and the parachute reflexes. In addition, most human neonates can step automatically if supported under the armpits and placed in contact with a table. When stepping, they are able to support up to 40% of their body weight. They do so by generating alternating contractions of flexor and extensor muscles, which are controlled by spinal central pattern generators (CPGs) (Dominici et al., 2011; Ivanenko et al., 2013). Stepping mainly reflects spinal and brainstem control (Forssberg, 1985; Yang and Gorassini, 2006), as shown by the presence of the stepping reflex in anencephalic infants and infants with cervical spinal lesions (Peiper, 1961). Reflex stepping normally disappears around 1–2 months of age, but it can still be evoked during that period with daily practice (Zelazo et al., 1972; Yang et al., 1998) or supporting the weight of the legs by means of water immersion (Thelen and Cooke, 1987). Voluntary unsupported walking develops at 11–19 months, when infants have an improved postural control and weight support. However, many idiosyncratic features of infant stepping are not a simple result of postural instability (Forssberg, 1985; Ivanenko et al., 2005), and toddlers are still unable to fully compensate for body weight unloading (Dominici et al., 2007).

# Maturation of Pendulum-Like Mechanism

Gravity-related signals during the first unsupported steps in toddlers contribute as a functional trigger for gait maturation (Ivanenko et al., 2004a, 2007). The pendulum-like mechanism of walking (**Figure 2A**) matures progressively in children, becoming established at a recovery percentage comparable to that of adults only around 2 years of age (Cavagna et al., 1983; Ivanenko et al., 2004a). In particular, in walking adults, the hip vaults

2004a). (B) stick diagrams of 1 cycle and foot trajectory spatial density plots when the toddler walked unsupported (left) and unloaded (right). Spatial density of foot path was integrated over swing phase (across 10–15 steps) and depicted graphically by means of a color scale (adapted from Dominici et al., 2007): the lower the density (toward blue in color-cued scale), the greater the variability. Plots are anisotropic, vertical scale being expanded relative to horizontal scale. Bottom plots—the same child recorded 3 years later. Note a characteristic (one peak) foot path during normal walking in the toddler and changes in the shape of the foot trajectory when the child was unloaded.

over the stance leg as an inverted pendulum (**Figure 1A**) twice in each gait cycle, with two corresponding peaks in the temporal profile of vertical hip position. Instead, toddlers at the onset of unsupported locomotion show irregular and variable vertical trunk oscillations (**Figure 2A**, left panel). Accordingly, the changes in the potential and kinetic energies of the COM are also irregular, and the percentage of mechanical energy recovery is much lower than in adults (**Figure 2A**, right panels). After few months of walking experience, the vertical trunk oscillations become more regular and the recovery of mechanical energy of the COM approaches that of the adults (**Figure 2A**).

As noticed above, there are two pendular mechanisms during walking, the inverted pendulum of the stance limb pivoting around the foot and the upright pendulum of the contralateral limb that swings about the hip. Body weight support (or partially reduced gravity) noticeably affects the foot trajectory in toddlers, in contrast to adults and older children who show only limited changes in kinematic coordination under such conditions. **Figure 2B** illustrates the effects of the body weight support on the kinematics of the swing phase. When walking unsupported, toddlers demonstrate a characteristic one-peak foot path, while when they are unloaded the shape of the foot trajectory changes significantly. Toddlers tend to make a high foot lift and forward overshoot during the swing phase (**Figure 2B**). These findings suggest that, at the onset of independent walking in children, changes in gravitational loads on the body are not compensated accurately by the kinematic controllers. Instead, compensation and development of pendulum-like mechanism require several months of walking experience (Ivanenko et al., 2004a; Dominici et al., 2007). Thus, not only maturation of the neuromuscular system but also learning by experience represents a powerful optimizing process for a proper integration of the gravity-related limb and body dynamics of walking.

# EVOLUTIONARY AND COMPARATIVE CONSIDERATIONS

# Some Paleoanthropology

The pendulum-like behavior of walking is shared by many terrestrial animals with variable efficiency (Cavagna et al., 1977), but humans are the only extant animals who walk habitually with an erect, bipedal plantigrade posture. This style of locomotion represents a key adaptation of bipeds to gravity effects, since it yields the best alignment of the contact force vector with the lower limb joints and results in limited joint torques during stance (Biewener et al., 2004).

Further insights on its functional significance might be gained by considering the evolution of human bipedal locomotion (Bramble and Lieberman, 2004). Bipedal plantigrade walking appeared at least 4 million years ago (Leakey and Walker, 1997; Haile-Selassie, 2001; Vaughan, 2003; Zollikofer et al., 2005). It probably predated the evolution of large brains, and it may have acted as a co-factor favoring the expansion of the brain. Lucy's (A.L. 288-1) brain volume was similar to that of our closest cousins, the chimpanzees, amounting to only 1/3 the volume of the brain of modern humans, and with a larger cheek-tooth size than that of modern humans (McHenry, 1984). The earliest fossils that can be safely attributed to the hominin clade, the australopithecines like Lucy, resemble extant chimpanzees in numerous aspects of their locomotor skeleton, in addition to the skull (Tardieu et al., 2013). However, Lucy, as a member of Australopitecus afarensis species, probably walked as a habitual biped, whereas chimps typically walk quadrupedally. Lucy was much smaller (about 1–1.2 m high) than modern humans. Therefore, if she relied on pendulum-like walking (which we do not know), her optimal speed was correspondingly slower (as in children or pygmies). The evolution of bipedal locomotion can also be tracked by observing the transition in the footprints from australopithecus to homo erectus: only in the latter (about 1 My ago) do we find the modern features of adducted hallux, medial longitudinal arch, and medial weight transfer before pushoff, while Australopitecus footprints are more similar to those of chimps (Bennett et al., 2009).

# Non-human Primates

Chimpanzees are facultative bipeds in the wild, when they sporadically raise themselves on the hindlimbs. When they walk bipedally, their posture is much more flexed than that of modern humans. Despite being a bent-hip, bent-knee gait, bipedal chimpanzees walk with an inverted pendulum style of locomotion, with vertical oscillations of the COM that are similar in pattern but with a much lower efficiency than that of humans, the recovery percentage amounting to only about 15% (Demes et al., 2015). Notice that another primate closely related to chimpanzees, the bonobo, does not use an inverted pendulum (D'Août et al., 2004). When humans voluntarily adopt a knee- and hip-flexed posture while walking, the pendulumlike behavior is disrupted (Grasso et al., 2000), gravitational potential and kinetic energies fluctuating in-phase rather than out-of-phase as in normal erect walking (Li et al., 1996).

# LOCOMOTION IN ACTUAL HYPOGRAVITY

# Moon Walks

Five years before the first Moon walk, Margaria and Cavagna (1964) made the following theoretical predictions based on the inverted pendulum model: (1) walking would be possible only at very low speeds, (2) there should be a quick transition to running, (3) maximum running speed should be about 5 km/h, (4) jumping would be preferred at higher speeds. Indeed, assuming a Froude number of 0.5 for the walk-to-run transition, one would expect that it should occur around 3 km/h on the Moon (Minetti, 2001).

Anectodical reports and videos from Apollo 11-17 (1969-72) expeditions on the Moon roughly confirm Margaria and Cavagna predictions (Carr and McGee, 2009; Jones and Glover, 2013). There were a total of 14 Apollo expeditions with Moon walks from Apollo 11 (1969) to Apollo 17 (1972) involving 12 different astronauts, the first and the last astronaut to walk on the Moon being Neil Armstrong and Eugene Cernan, respectively. All EVAs (extravehicular activities, such as Moon walks) totaled about 78 h. Although there was considerable inter-subject variability in the preferred gait style (see Jones and Glover, 2013), the astronauts neither walked nor ran most of the time (also because of low frictional contact forces, which are proportional to gravity), but they mainly used one of three styles of locomotion: (1) loping, (2) skipping, and (3) hopping (see for instance https://www.youtube. com/watch?v=x2adl6LszcE). Loping is a kind of slow running with a high, long, prolonged aerial phase, and it is energetically optimal under reduced gravity conditions (Rader et al., 2007).

# Skipping

Skipping is especially interesting (Minetti, 1998), because it is a gait mode intermediate between walking and running, showing features of both the former (double support phase) and the latter (flight phase). Each foot undergoes two lift-off events in each cycle to produce a syncopated stepping. Skipping can be unilateral (right or left, depending on the last foot in contact with the ground prior to the flight), but is most commonly bilateral, with alternating right and left unilateral strides. The metabolic and biomechanical analyses of skipping (performed in the laboratory on Earth) show that the two basic strategies for mechanical energy saving of walking (pendulum-like) and running (elastic bouncing) are operating at different phases of the gait cycle (Minetti, 1998). Moreover, the timing of the major peaks of motoneuronal activity in the spinal cord resembles that of walking and running (Ivanenko et al., 2008a). Although little if ever used by human adults on Earth, skipping belongs to the repertoire of human gaits since childhood. Indeed it is a typically—though sporadically—performed by children at about 5 years of age. The most likely reason why skipping is later abandoned is that it is metabolically inefficient on Earth (Minetti, 1998).

# Role of Space Suit

When considering the Moon walks of astronauts, one should take into account that their weight was reduced by 1/6 (the ratio of Moon gravity to Earth gravity), but inertia was the same as on Earth. It is known that gravity interacts with inertia in complex ways, especially during running (Chang et al., 2000). Also the Apollo spacesuit worn by the astronauts affected locomotion. Although the mass of the space suit was about 80 kg, almost all of it was supported by the internal pressure forces (Carr, 2016). However, the space suit constrained the astronaut's movements considerably and involved significant metabolic costs, limiting the intensity and duration of EVAs.

Carr and McGee (2009) reviewed audio transcripts and video clips of lunar EVAs available from the Apollo Lunar Surface Journal (Jones and Glover, 2013), as well as from several NASA technical reports (cited in Carr and McGee, 2009). They identified gait events that could be classified as walk/lope/run and for which speed could be estimated. In their classification, the lope mode included skipping and hopping, in addition to loping. Using estimated speeds and individual anthropometric characteristics of each astronaut, they computed the Froude number for each such event. They found that the transition between walk/lope and run occurred at an average Froude number of 0.37, close to the value predicted by Margaria and Cavagna (1964) and by the dynamic similarity principle.

# Falls

Another element that characterizes locomotion on the Moon is the relative instability and propensity to fall. Indeed, falls or saved falls were frequent during the EVAs of Apollo missions (Kubis et al., 1972a,b; Bloomberg et al., 2016). In addition to external causes (e.g., slippery, loose or uneven terrain, low visibility, mobility constraints imposed by the spacesuit, see Scott-Pandorf et al., 2009), instability was often caused by inappropriate automatic postural reactions in response to tripping or stumbling. These reactions were often too fast and tended to facilitate rather than impede the fall (Kubis et al., 1972a,b; Bloomberg et al., 2016).

# Treadmill Exercise on ISS

Nowadays, to mitigate the negative effects of prolonged weightlessness on the osteo-articular, muscular and cardiovascular systems, astronauts perform several exercises in the ISS, including walk and run on a treadmill (TVIS, Treadmill with Vibration Isolation Stabilization System). To this end, they wear a harness attached at the waist and shoulders with bungee cords that keep them in contact with the treadmill belt. The cords restore about 60% of the weight of a typical astronaut, and their pulling direction and force change with the movements. Therefore, the restoring forces are variable and are not tailored to each individual.

Recently, a novel Subject Loading System (SLS) has been designed for potential use on ISS (Gosseye et al., 2010). It consists of two pneumatic pistons attached at one end to the trunk harness and at the other end on a trolley sliding with the subject's movements. The resulting traction force is roughly equal to the weight of each subject and remains nearly constant, whatever the position of the subject on the treadmill. Laboratory tests on Earth found that the biomechanics of subjects running with the SLS was reasonably similar to that of normal running (Gosseye et al., 2010). The SLS might therefore represent a better tool for exercise on ISS.

# POSTFLIGHT LOCOMOTION

Transitioning from one gravity level to another one is often troublesome. In particular, prolonged weightlessness affects postflight human performance in several ways, and locomotion is no exception (Bloomberg et al., 2016). Upon return to Earth, crewmembers often experience ataxia, postural instability and navigation difficulties. One day after returning from an average flight duration of 185 days, astronauts completed the functional mobility test (an obstacle course over an unstable, compliant surface) in about twice the time used during the preflight test (Mulavara et al., 2010). Several specific parameters of gait were altered soon after long-duration space flight, and recovered to preflight values in a few days. Thus, knee flexion during stance was significantly increased (Bloomberg and Mulavara, 2003), angular motion at the knee and ankle and vertical accelerations of the COM were increased (Hernandez-Korwo et al., 1983), timing and magnitude of lower limb muscles were slightly but significantly altered, especially at heel strike and toe-off (Layne et al., 1997, 2004) in parallel with changes in toe-clearance (Miller et al., 2010).

# SIMULATED HYPOGRAVITY WITH PARABOLIC FLIGHT

Experiments in parabolic flight represent a viable alternative to spaceflight experiments; they are much less expensive, less demanding, and allow a larger sample of participants (Pletser, 2004; Shelhamer, 2016). Parabolic flight can simulate 0 g, 0.16 g (Moon), 0.4 g (Mars), or hyper-gravity (e.g., 1.5 or 2 g). The main drawback is represented by the limited time available to reproduce hypogravity conditions (∼20, 30, and 40 s for 0, 0.16, and 0.4 g, respectively), which only allows for acute investigations, even though several parabolas are performed during a flight campaign. A few experiments on locomotion were carried out in parabolic flight, and the results depended on the level of simulated hypo-gravity.

# Simulating Martian Gravity

Cavagna et al. (1998, 2000) performed parabolic flights at 0.38 g. Their results were generally in agreement with the predictions of the inverted pendulum model and the principle of dynamic similarity. They found that the amplitude of energy changes (both gravitational potential energy and kinetic energy) was smaller at 0.38 g than at 1 g, and the step period was longer. A decreased work of walking at 0.38 g relative to 1 g is consistent with the decreased metabolic energy consumption reported during walking in reduced gravity Earth simulators (Newman et al., 1994). Maximum pendulum-like recovery of mechanical energy at COM was smaller than at 1 g (about 55 vs. 65%) and occurred at lower speeds (about 3.5 vs. 5.5 km/h). Thus, the range of walking speeds was about half as wide as on Earth, walk-to-run transition occurred at about the optimal speed of walking on Earth, and the external work performed to walk a given distance was about half as much as on Earth.

# Simulating Moon Gravity

De Witt et al. (2014) studied the walk-to-run transition at 0.16 g during parabolic flight. They found that participants preferred to run instead of walking at much higher Froude numbers than predicted by the inverted pendulum model and the dynamic similarity principle (on average, 1.4 instead of 0.5 Froude). They argued that deviations from the predicted Froude number are due to the accelerations induced by the swinging limbs that increase the downward force applied to the body by gravity, as previously suggested by Kram et al. (1997) based on simulated hypogravity on Earth (see below). Thus, if one replaced the term g in the Froude equation with the summed contribution of gravity and downward acceleration due to swinging limbs, one would account for the experimental values of walk-to-run transition speed in hypogravity (Kram et al., 1997; Raichlen et al., 2013). The protocol used by De Witt et al. (2014) to assess walk-to-run transitions during parabolic flight differs substantially from the protocols typically used on Earth. Due to the short parabola durations, participants judged the most comfortable speed among several constant treadmill speeds, instead of switching automatically as it occurs on Earth or Moon. However, these authors also performed the switching protocol in the Earth laboratory using a body weight unloading system and confirmed the results obtained with parabolic flights.

# Jumping and Landing from a Jump

Due to altered sensory information, altered anticipatory postural adjustments (Clément et al., 1984) and body reference configuration (Massion et al., 1997), reduced gravity conditions may also affect locomotor-related tasks. For instance, recent work on landing from a jump during parabolic flight demonstrated clear modifications in the preparatory adjustments and the loading phase, and suggested that otolithic information plays an important role in the control of landing from a jump (Gambelli et al., 2016a,b). In this work, pulldown forces were continuously generated by the SLS described above, simulating 1, 0.6, 0.4, and 0.2 g. Another recent paper studied jumping up on site with knee and hip joints almost extended during Lunar and Martian parabolas (Ritzmann et al., 2016). It found that the new gravitational load was anticipated correctly by the participants being compensated for by gravity-adjusted muscle activities.

# SIMULATED HYPOGRAVITY ON EARTH

# Gravity Compensation Devices

Obviously gravity cannot be changed on Earth, but there exist several simulators that allow low-cost, long duration trials of reduced gravity conditions. Each of these simulators has advantages and disadvantages, but all of them assist posture and locomotion by countering the force of gravity on the body. One of the more frequently used simulators is the body weight support (BWS) system for walking on a treadmill. In vertical BWS, the subject is supported in a harness that applies a controlled upward force to the trunk (**Figure 3A**). Mignardot et al. (2017) modified a ceiling-mounted system to apply multidirectional forces to the trunk (**Figure 3B**), based on the observation that vertically restricted trunk support alters gait dynamics and that the addition of well-calibrated forward forces alleviates these effects. Most available BWS systems constrain locomotion to a treadmill, or a restricted path for vertical unloading systems that are suspended on a shifting cart. Awai et al. (2017) developed a ceiling-mounted rail and deflector system that allows unrestricted walking on level ground or on stairs.

A different approach to counteract the force of gravity in the vertical position consists in applying lower-body positive pressure (LBPP), increasing air pressure around the lower body to create a lifting force approximately at the COM (Cutuk et al., 2006; Grabowski and Kram, 2008; Ruckstuhl et al., 2009; Grabowski, 2010; Schlabs et al., 2013). LBPP systems allow ambulation in the normal erect posture, but high pressure levels may affect systemic blood pressure, head perfusion, and vascular flow, thus requiring caution (Cutuk et al., 2006). Moreover, LBPP provides the desired vertical weight support, but may also generate unwanted horizontal assistance due to the interface between the chamber and the subject (Grabowski and Kram, 2008).

The main limitations of all these simulators are given by the high pressure localized to specific regions of the skin (at the harness attachment for vertical BWS or the body parts within the pressurized chamber for LBPP), and the application of the vertical unloading force only to the trunk, so that during stance the lower limb experiences a reduction of effective gravity force proportional to the unloading force. However, these systems cannot aid the swing movements of the limbs, because they do not pull them in proportion to the simulated gravity level and therefore the limbs remain subject to full gravity during swing.

The latter disadvantage is overcome by the tilted BWS systems (Ivanenko et al., 2011) or horizontal suspension systems (Gurfinkel et al., 1998; De Witt et al., 2010). These simulators do not allow a normal posture, but do not involve major cardiovascular risks. They have been used by both Roscosmos and NASA to train astronauts prior to the mission (Hewes, 1969; Bogdanov et al., 1971; Hansen, 1995). NASA, in particular, has several facilities for horizontal suspension (e.g., the Langley's Reduced Gravity Walking Simulator or the Enhanced Zerogravity Locomotion Simulator). Tilting the subject counteracts a fraction of body weight, the component of the gravity force acting on the body and limbs in the sagittal plane of walking being reduced in proportion to the cosine of the tilt angle relative to the horizontal. One such system (Italian patent number Rm2007A000489, **Figure 3C**) involves a bed supporting the head, trunk, and upper limbs, while the lower limbs are suspended in low-friction, low-mass exoskeletons (Ivanenko et al., 2011; Sylos-Labini et al., 2011, 2013). The subject lies on the side, while the bed and exoskeletons can be tilted by an angle between 0◦ and 40◦ . The subject steps on a treadmill which is tilted by the same angle. In contrast with the vertical BWS, tilted BWS simulators provide a comfortable lying position for the subject and reduce the effects of gravity on both stance and swinging limbs. As a drawback, however, they involve the extra mass of the moving couch and exoskeleton, and limit lateral trunk movements while walking. Another system is represented by the passive gravity balancing system (**Figure 3D**). It is anchored to a wall and compensates the weight of the body by means of a spring-balanced, dual-parallelogram mechanism, and torsosupport assembly, while the weight of each leg is compensated by a leg exoskeleton (Lu et al., 2011; Ma and Wang, 2012). Partial gravity replacement loads can be applied by means of pneumatic pistons with the SLS or similar systems while the subject is in weightlessness (Gambelli et al., 2016a,b) or horizontally tilted in the Earth laboratory (Ivanenko et al., 2011).

Several passive or active gravity compensation devices have been developed in robotics to allow human subjects to step freely over-ground (see Arakelian, 2016). The first exoskeletons were designed to reduce the burden for soldiers and help them carrying heavy objects (e.g., Walsh et al., 2006; Zoss et al., 2006). Subsequently, exoskeletons have been used in rehabilitation, for instance the Indego, EKSO Bionics, ReWalk, Mindwalker (**Figure 3E**) (del-Ama et al., 2012; Esquenazi et al., 2012; Wang et al., 2015). Some exoskeletons do not allow any voluntary contribution from the subject, while others include an assist-as-needed control principle, which is beneficial to learn stepping. The MoonWalker is a passive force balancer that sustains body-weight (Krut et al., 2010). It is controlled using an actuator requiring very low energy on flat terrains to shift the force on the stance leg. The actuator can also

provide part of the energy to climb stairs or slopes. van Dijk et al. (2011) developed a passive exoskeleton to minimize joint work during walking. This exoskeleton uses artificial tendons, acting in parallel with the leg. Artificial tendons are elastic elements that are able to store and redistribute energy over the human leg joints. In contrast with BWS systems, exoskeletons apply the unloading force to the feet rather than the trunk. Therefore, the exoskeleton provides full postural stability and does not necessitate antigravity muscle activity from the subject, but the feet still experience the full subject's weight during walking.

# Biomechanical Effects of Simulated Reduction of Gravity

Simulations of hypogravity on Earth using different BWS techniques generally confirmed the basic principles of gravity effects on locomotion (for a review, see Sylos-Labini et al., 2014b). In particular, they confirmed that reduced gravity involves lower optimal walking speeds and lower preferred walk-to-run transition speeds as compared with normal gravity. However, significant deviations from the Froude numbers predicted by the dynamic similarity principle have been reported (Donelan and Kram, 1997, 2000; Kram et al., 1997; De Witt et al., 2014). Thus, Ivanenko et al. (2011) and Sylos-Labini et al. (2013) compared locomotion at 0.07 g (simulating Pluto), 0.16, 0.38, and 1 g, using the vertical and the tilted BWS, or the 0.16 g replacement load in the recumbent position. Consistent with previous observations (Donelan and Kram, 1997; Kram et al., 1997; De Witt et al., 2014), they found that walk-torun transitions occur at Froude > 0.5 at simulated gravity < 0.2 g. They also found a hysteresis in gait transitions (walkto-run transition speed >> run-to-walk transition speed) at these low gravity levels. More interestingly, they found that gait transitions at reduced gravity were gradual (**Figure 4**), without any noticeable abrupt change as it typically occurs at 1 g, for both kinematics and muscle activity patterns (Ivanenko et al., 2011; Sylos-Labini et al., 2011). **Figure 4** illustrates an example of abrupt gait transition on Earth (1 g) and smooth walk-to-run transition at simulated Moon gravity (0.16 g) in a representative subject. For instance, the percent stance duration, the limb angle at foot contact and the timing of leg muscle EMG activity were all significantly different between the stride immediately before (stride −1) and immediately after (stride +1) the walk-to-run transition at 1 g, but they were not significantly different at reduced gravity (**Figures 4B,C**). It should be noticed that some gait parameters (horizontal foot excursion, maximum horizontal foot speed, relative swing duration) depended on the type of BWS, vertical vs. tilted (Sylos-Labini et al., 2013), suggesting an important contribution of the swing limb dynamics, which is affected differentially by the specific apparatus.

# Kinematics

The more detailed studies with vertical BWS also showed limited changes of the kinematic coordination across a wide range of gravity levels (0, 0.05, 0.25, 0.5, 0.65, 1 g), despite drastic changes

number of the gait cycles (corresponding to the appropriate timing of the trial), y-axis indicates normalized gait cycle, and color indicates EMG amplitude. The white line indicates when toe off occurred. Vertical dashed lines indicate walk-to-run (W-R) and run-to-walk (R-W) transitions. RF, rectus femoris; BF, biceps femoris; TA, tibialis anterior; SOL, soleus. Adapted from Sylos-Labini et al. (2011).

of the kinetic parameters (Ivanenko et al., 2002). Thus, the peak vertical contact forces decreased proportionally to the unloading force (**Figures 1C,D**), so that at 0.05 g they were 20-fold smaller than at 1 g and were applied at the forefoot only instead of showing the classical heel to ball shift. By contrast, the trajectory of the feet in space and the planar covariance of limb segment elevation angles were very similar at all gravity levels except for 0 g, when subjects could only step in air (**Figures 5A,B**). Indeed, the 3D gait loops obtained by plotting the elevation angles one vs. the others lied close to a plane during walking (denoted by grids in **Figure 5A**). Planar co-variation was documented for different walking conditions (Bianchi et al., 1998a; Grasso et al., 2000; Ivanenko et al., 2008b) and it was obeyed at all tested speeds and BWS levels (Ivanenko et al., 2002). However, the specific phase relationships between the elevation angles were different during air-stepping, leading to a different plane orientation at 100% BWS in **Figure 5A**. In addition, there were some non-linear changes in the waveforms of the limb segment elevation angles, for instance, the contribution of the 1st harmonic to the shank and foot elevation angles became more prominent with reduced gravity level (**Figure 6A**).

In the absence of support–related constraints (air-stepping), both inter-stride and inter-subject variability were considerably augmented (Ivanenko et al., 2002, 2008b). On the other hand, minimal contact forces were generated artificially during airstepping by taping a foam-rubber under the subjects' feet that lightly touched the belt of the treadmill during the stance phase or by having the subject stepping on a pillow (**Figure 5C**, left). Such minimal contact forces were sufficient to significantly decrease the variability of the foot path (**Figure 5C**, right), suggesting that the support surface represents an importance reference frame for accurate foot trajectory control (Ivanenko et al., 2002). These results might be relevant for the special conditions of moving around and working on a small asteroid with very low gravity (<10−<sup>3</sup> g, Garrick-Bethell and Carr, 2007).

# EMG Patterns

Also the analysis of the electromyographic patterns (EMG) revealed a non-linear trend with the gravity level (Ivanenko et al., 2002, 2004b, 2013; Van Hedel et al., 2006; Klarner et al., 2010; Sylos-Labini et al., 2014b; Fischer et al., 2015). The temporal components shared by multiple muscles during walking with the vertical BWS were very similar across the tested gravity levels between 1 and 0.05 g (see above), but the weighting coefficient of each component on individual muscles differed considerably as a function of the gravity level (Ivanenko et al.,

vs. the others. Note similar kinematics and covariance plane orientation in all BWS conditions except for air-stepping. Adapted from Ivanenko et al. (2002). (B) shape and variability of endpoint path in 1 subject over 12 consecutive step cycles for trials performed at different BWS levels. (C) effect of minimal contact forces provided by a foam-rubber taped under the subjects' feet or when touching the pillow during air-stepping. Right panel—spatial variability of the foot trajectory (foot path tolerance area, mean ± SD over all subjects). Note that the shape and variability of the foot path is comparable in all conditions except in air-stepping, where variability is much higher (B). Variability decreased substantially in the presence of minimal surrogate contact forces relative to standard air-stepping (C).

2004b). Moreover, the amplitude of net activity of most muscles did not scale proportionally to the percent of body weight loading (**Figure 6B**), as did the amplitude of the peak vertical contact force (Ivanenko et al., 2002). For some muscles, the changes with loading were not even monotonic, and there was a complex reorganization of the pattern of activity which differed across individuals, evidence of variable adaptive adjustments to reduced gravity (Sylos-Labini et al., 2014b). For instance, the mean amplitude of activity in ankle extensors decreased systematically with decreasing simulated gravity, consistent with their antigravity function (lateral gastrocnemius muscle, see LG in **Figure 6C**). By contrast, the activity of rectus femoris muscle showed increments at slow speeds (<3 km/h, **Figure 6C**), while the hamstring muscles demonstrated new activity bursts with body weight unloading (Ivanenko et al., 2002). The latter muscles are those showing the largest variability across subjects (Ivanenko et al., 2002; Sylos-Labini et al., 2014a). Overall, the changes in muscle activity depend on the changed biomechanical requirements with BWS, changed inertial or assistive forces, complex architecture of skeletal muscles, and the dynamic coupling of limb segment motion (Zajac et al., 2003). Even the activity of synergistic muscles, such as gastrocnemius and soleus muscles, may show substantial differential changes during decreased limb loading (Ivanenko et al., 2002; McGowan et al., 2010). In addition, apparently paradoxical elevated EMGs during overground walking assisted by an exoskeleton compared to normal walking have been reported in healthy subjects, consistent with an important contribution of foot loading-related sensory feedback (Sylos-Labini et al., 2014a). These findings indicate that the control algorithms for robotic assistance still

need to be optimized, for instance by being tailored to each individual (Mignardot et al., 2017). There might also be the need for extensive training sessions of the individuals wearing the exoskeleton, so that they become fully adapted to assisted locomotion.

# Body Weight Unloading in Toddlers

There are still few studies on the physiological effects of reduced gravity in children, although this is an important topic given the growing application of BWS techniques in pediatric rehabilitation (see next section). One study (Dominici et al., 2007) compared the locomotion under partial weight unloading in toddlers (about 1 year old) at their first unsupported steps, older children (1.3–5 years), and adults. To simulate various levels of body weight in a manner acceptable by a child, an experimenter held the trunk of the child with both hands and supplied an approximately constant vertical force during stepping on a force platform. In contrast to adults and older children who showed only limited changes in kinematic coordination under reduced-gravity (see above), toddlers lifted the feet too high and forward at the end of swing (**Figure 2B**, Dominici et al., 2007). Intermediate walkers (1.5–5 mo after walking onset) showed only partial improvements in foot trajectory characteristics. Therefore, at the onset of walking, changes in vertical body loads are not compensated fully by the CNS (Forssberg, 1985; Ivanenko et al., 2005; Dominici et al., 2007).

# CLINICAL APPLICATIONS OF SIMULATED HYPOGRAVITY

BWS locomotion training has shown some promise as a tool to facilitate locomotor activity in individuals with neuromotor disorders, such as spinal cord injury SCI (Hubli and Dietz, 2013), stroke (Sale et al., 2012; Mehrholz et al., 2014; Moraru and Onose, 2014), Parkinson disease (Miyai et al., 2000; Picelli et al., 2013), Multiple Sclerosis (Swinnen et al., 2014), Cerebral Palsy, and Down syndrome (Damiano and DeJong, 2009; Valentin-Gudiol et al., 2013). However, in some patients, the efficacy of BWS-treadmill interventions is limited (Morawietz and Moffat, 2013; Picelli et al., 2013). In particular, this intervention seems to have weak or conflicting evidence in children with Cerebral Palsy (Dewar et al., 2015). On the other hand, it may accelerate the development of independent walking in children with Down syndrome (Damiano and DeJong, 2009; Valentin-Gudiol et al., 2013). Stroke patients who are able to walk, but not those unable to walk, benefit from the intervention by increasing walking speed and endurance (Mehrholz et al., 2014), although the superiority of the intervention relative to other control therapies has failed to be established (Charalambous et al., 2013). In SCI patients, locomotor training may induce the reappearance of kinematic regularities (Grasso et al., 2004a), EMG temporal components shared by multiple muscles (Ivanenko et al., 2003), and flexor reflexes (Smith et al., 2014). Robot-assisted training improves mobility-related outcomes to a greater degree than conventional over-ground training for patients with incomplete SCI, particularly during the acute stage (Nam et al., 2017). A low rather than a high testing treadmill speed may be beneficial for an optimal expression of EMG improvements in individual with incomplete chronic SCI (Meyns et al., 2014). A critical combination of sensory cues might be required to generate and improve locomotor patterns after SCI during assisted locomotor training (Hubli and Dietz, 2013). Mignardot et al. (2017) recently developed an adaptive algorithm that personalizes multidirectional forces applied to the trunk based on patient-specific motor deficits. This multidirectional gravityassist enabled natural walking in individuals with SCI or stroke.

In spite of a complex non-linear reorganization of muscle activity patterns with BWS (Ivanenko et al., 2002; Moreno et al., 2013), the basic spatiotemporal structure of the locomotor output tends to be preserved in healthy and SCI subjects (see above), implying that a few oscillating circuits drive the active muscles to produce locomotion (Gerasimenko et al., 2010, 2015). These characteristic spatiotemporal features of spinal motorneuron activation are becoming increasingly important also for the functional assessment and rehabilitation of walking after SCI, both in SCI patients using BWS-treadmill training (Ivanenko et al., 2003; Grasso et al., 2004a), or in animal models of SCI, e.g., in robotic rehabilitation in spinalized rats with epidural stimulations mimicking physiological timings of muscle activations (Capogrosso et al., 2016; Wenger et al., 2016).

Several aspects need to be taken into consideration when using BWS locomotor training to restore the locomotor function. First, the term "normal motor pattern" is somewhat misleading under hypogravity conditions. As noticed above, there are considerable non-linear changes in the muscle activity patterns with changing BWS, especially in proximal leg muscles, even in neurologically intact individuals (Ivanenko et al., 2002; Moreno et al., 2013). It has also been recommended that very low speeds and high levels of BWS, where the EMG differences are most prominent, should be avoided whenever possible in the rehabilitation practice (Van Kammen et al., 2014). Second, the current assessments of therapeutical efficacy of body weight unloading in gait pathologies should consider the complex nature of the control of locomotion, task-dependent features, individual compensatory strategies, and plasticity of neuronal networks. For instance, locomotor training with BWS in SCI patients may not generalize to untrained walking conditions. Thus, SCI patients (ASIA-A, B, and C) were trained to step on a treadmill with BWS for 1.5–3 months (Grasso et al., 2004b). At the end of training, foot motion recovered the shape and the step-by-step reproducibility that characterize normal gait. The patients were then asked to step backward on the treadmill belt moving in the opposite direction. In contrast to healthy subjects who can immediately reverse the direction of walking by time-reversing the kinematic waveforms, all tested patients were unable to step backward initially and they needed specific training in the new direction (Grasso et al., 2004b).

In sum, BWS has contributed importantly to the methodologies that can be used to restore the locomotor function in disabled people. BWS systems are now often used for gait rehabilitation to assist locomotor recovery by performing well-focused and carefully directed repetitive practice. It is also worth noting the beneficial effect of simulated weightlessness on rhythmogenesis and its potential for assessing the state of the spinal pattern generation circuitry and for developing CPGmodulating treatments (Ivanenko et al., 2017). As a final point, an effective strategy to stimulate the spinal cord circuitries and to promote neuroplasticity in disabled people is a combination of locomotor training in simulated hypogravity with other promising experimental approaches, such as epidural electrical stimulation or drug application (Hubli and Dietz, 2013; Angeli et al., 2014; Minassian et al., 2016; Gad et al., 2017; Shah and Lavrov, 2017).

# CONCLUSIONS AND PERSPECTIVES

Our knowledge on the adaptation of locomotion to hypogravity has progressed considerably over the last few years, but remains fragmentary. There are still too few observations in actual hypogravity (Moon and parabolic flight), while most quantitative data come from laboratory simulations. The latter provide only partial replicates of true hypogravity because of the constraints imposed by physics. Indeed, when the locomotor kinematics, EMG, and ground reaction forces are compared between parabolic flight and horizontal suspension on Earth, subtle but systematic differences are revealed (De Witt et al., 2010). Even different simulators on Earth may yield slightly but significantly different results between each other, in terms of both locomotor kinematics and EMG patterns (Ivanenko et al., 2011; Sylos-Labini et al., 2011, 2013, 2014b). The limitations of ground experiments become even more evident when compared with the reports about locomotion on the Moon, where the specific characteristics of the environment (dust, uneven terrain, thin atmosphere, wide range of temperatures, heavy, and cumbersome spacesuits, etc) play an important role. Laboratory experiments with locomotion on soft (Lejeune et al., 1998) or complex terrain (Matthis et al., 2017) are therefore highly relevant.

Training crew members prior to space missions is quite challenging, and it remains particularly difficult to simulate the effects of reduced forces on all physiological systems and apparatus. Thus, prior to their missions, the Apollo crews had limited parabolic flight exposure and mainly trained in the Lunar Landing Training Vehicle, which did not simulate the vestibular effects of 0.16 g. Indeed, upon return after the mission, most Apollo astronauts reported having felt "wobbly" on the lunar surface initially, but that their coordination improved steadily during the first few hours of lunar EVA (Bloomberg et al., 2016). Unsteadiness is even more severe upon re-entry to Earth from a long-term space mission, because the gravity replacement systems used on the ISS to exercise do not act on the whole body, and in particular do not act on the vestibular system. It thus seems imperative to be able to design appropriate training protocols and exercise countermeasures with available technologies, to train space travelers to become more adaptable to transitions in gravity levels, such as from a higher to a lower gravity or vice versa. In the future, rotating habitat units

capable of generating artificial partial gravity might become a valid countermeasure to physiological deconditioning during long duration space missions (such as in the Mars direct project, Clément et al., 2016).

One important lesson we learnt from a wealth of studies, including those reviewed here, is that a reduction of gravity does not lead necessarily to linear, continuous changes of locomotor parameters. Departures from linearity may take the form of non-proportional changes of a given locomotor parameter as a function of the gravity level, or we may observe a discontinuity with the transition to a qualitatively distinct behavior. Some such behaviors are predictable based on theory. For instance, according to the dynamic similarity principle, both the optimal walking speed and the walk-torun transition speed scale with the square root of gravity (Alexander, 1989). Also, the transition from walking to skipping at relatively low speed at Moon gravity can be predicted based on energy optimization criteria (Minetti, 1998; Ackermann and Van den Bogert, 2012; Pavei et al., 2015). Finally, at reduced gravities (Moon gravity or lower), humans could theoretically run on water, an impossible task on Earth. This latter performance was predicted by hydrodynamic modeling, and demonstrated using vertical BWS in the laboratory (Minetti et al., 2012).

However, other non-linear behaviors are discovered empirically, without necessarily being predicted by available theories. Thus, the activity of several limb muscles was shown to scale non-linearly with simulated hypogravity and differentially as a function of the individual muscles (Ivanenko et al., 2002; Sylos-Labini et al., 2011, 2014b). Moreover, several gait parameters change abruptly in the transition from walk to run at 1 g, but they change smoothly, gradually at lower gravities (Ivanenko et al., 2011). Also, below a given (still undefined) gravity threshold, foot contact forces become insufficient to drive

# REFERENCES


the normal kinematic and EMG patterns of locomotion, and the CPGs switch to a different control mode (Ivanenko et al., 2002).

Another lesson stemming from recent studies is the potential usefulness of reduced gravity simulators for clinical applications on Earth. Our brief review has shown that the results are still mixed, possibly because of our incomplete understanding of the physiology of locomotor adaptation to reduced gravity and/or the inadequate methodologies employed to reduce gravity effects on the patient. One take-home message that is shared by the field of space travel and clinical rehabilitation is that individual data about the physiology and medical history of each person will likely contribute to reducing health and mission risks in space travelers and improving the outcomes of rehabilitation in patients. In particular, both reduced gravity simulators and training protocols should be tailored to each individual (White and Averner, 2001; Mignardot et al., 2017), rather than being stereotyped as it still is in the majority of cases.

# AUTHOR CONTRIBUTIONS

All authors listed have made a substantial, direct and intellectual contribution to the work, and approved it for publication.

# ACKNOWLEDGMENTS

This work was supported by the Italian Ministry of Health (IRCCS Ricerca corrente), the Italian Space Agency (grants I/006/06/0 and 2014-008-R.0), the Italian University Ministry (PRIN grant 2015HFWRYY\_002), Lazio Region (INNOVA.1 FILAS - RU 2014\_1033 - RIABILITA), and Horizon 2020 Robotics Program from the European Commission (ICT-23-2014 under Grant Agreement 644727-CogIMon). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.


varying body unloading and Froude number. Gait Posture 30, 446–451. doi: 10.1016/j.gaitpost.2009.07.001


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2017 Lacquaniti, Ivanenko, Sylos-Labini, La Scaleia, La Scaleia, Willems and Zago. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Orthostatic Intolerance in Older Persons: Etiology and Countermeasures

Nandu Goswami <sup>1</sup> \*, Andrew P. Blaber <sup>2</sup> , Helmut Hinghofer-Szalkay <sup>1</sup> and Jean-Pierre Montani <sup>3</sup>

*<sup>1</sup> Gravitational Physiology and Medicine Research Unit, Institute of Physiology, Medical University of Graz, Graz, Austria, <sup>2</sup> Department of Biomedical Physiology and Kinesiology, Simon Fraser University, Burnaby, BC, Canada, <sup>3</sup> Department of Medicine/Physiology, University of Fribourg, Fribourg, Switzerland*

Orthostatic challenge produced by upright posture may lead to syncope if the cardiovascular system is unable to maintain adequate brain perfusion. This review outlines orthostatic intolerance related to the aging process, long-term bedrest confinement, drugs, and disease. Aging-associated illness or injury due to falls often leads to hospitalization. Older patients spend up to 83% of hospital admission lying in bed and thus the consequences of bedrest confinement such as physiological deconditioning, functional decline, and orthostatic intolerance represent a central challenge in the care of the vulnerable older population. This review examines current scientific knowledge regarding orthostatic intolerance and how it comes about and provides a framework for understanding of (patho-) physiological concepts of cardiovascular (in-) stability in ambulatory and bedrest confined senior citizens as well as in individuals with disease conditions [e.g., orthostatic intolerance in patients with diabetes mellitus, multiple sclerosis, Parkinson's, spinal cord injury (SCI)] or those on multiple medications (polypharmacy). Understanding these aspects, along with cardiopostural interactions, is particularly important as blood pressure destabilization leading to orthostatic intolerance affects 3–4% of the general population, and in 4 out of 10 cases the exact cause remains elusive. Reviewed also are countermeasures to orthostatic intolerance such as exercise, water drinking, mental arithmetic, cognitive training, and respiration training in SCI patients. We speculate that optimally applied countermeasures such as mental challenge maintain sympathetic activity, and improve venous return, stroke volume, and consequently, blood pressure during upright standing. Finally, this paper emphasizes the importance of an active life style in old age and why early remobilization following bedrest confinement or bedrest is crucial in preventing orthostatic intolerance, falls and falls-related injuries in older persons.

Keywords: syncope, exercise, mental arithmetic, water drinking, aging, falls

# INTRODUCTION

In most persons, the hemodynamic and neurovascular responses to orthostatic challenge produced by standing up are adequate to stabilize arterial blood pressure and to maintain cerebral blood flow after standing. However, numerous patients come to clinics with the complaint of dizziness upon standing (Rapp et al., 2012). Syncope, defined as a transient self-limiting loss of consciousness, is

Edited by:

*Ovidiu Constantin Baltatu, Anhembi Morumbi University, Brazil*

#### Reviewed by:

*Noman Naseer, Air University, Pakistan Alexander V. Ovechkin, University of Louisville, United States*

> \*Correspondence: *Nandu Goswami nandu.goswami@medunigraz.at*

#### Specialty section:

*This article was submitted to Integrative Physiology, a section of the journal Frontiers in Physiology*

Received: *03 May 2017* Accepted: *29 September 2017* Published: *09 November 2017*

#### Citation:

*Goswami N, Blaber AP, Hinghofer-Szalkay H and Montani J-P (2017) Orthostatic Intolerance in Older Persons: Etiology and Countermeasures. Front. Physiol. 8:803. doi: 10.3389/fphys.2017.00803*

**147**

the end-point of cardiovascular stability. Clinical symptoms preceding syncope can include: nausea, blurring of vision, dizziness, and a sudden decrease in mean arterial pressure (MAP; Grasser et al., 2009a,b) due to a reduction in stroke volume, heart rate, or peripheral resistance. These responses elicited by upright posture characterize the condition of orthostatic intolerance (OI).

The etiology of cardiovascular instability that leads to orthostatic intolerance is multifactorial (Custaud et al., 2002; Weimer and Williams, 2003). Among several causes, orthostatic intolerance can arise due to alterations in cerebral blood flow (Blaber et al., 2011) and/or the control of cardiovascular regulation. Etiology of cardiovascular dysregulation contributing to OI could be centrally located (e.g., arising due to changes in cardiac stretch receptors) or peripherally located (e.g., arising due to limitations in increasing peripheral vascular resistance, venoconstriction and/or sympathetic activity loss, or functional changes in arterial baroreflex).

# Aging, Orthostatic Intolerance, and Countermeasures

There is a steep global trend toward the demographic aging of populations. It is projected that people aged 65+ years in the EU will almost double over the next 50 years, reaching up to 151 million in 2060 (URL1) leading to increasing public health costs as well as costs for older people and their families. The negative consequences of aging such as illness or injury often require admission to hospital. However, the bedrest confinement that occurs during hospitalization is considered a major factor in physiological deconditioning and functional decline and can contribute to a downhill spiral of increasing risk of frailty, and orthostatic intolerance, leading to falls (Mühlberg and Sieber, 2004). A systematic review by Heinrich et al. (2010) indicates that 0.85–1.5% of all health care costs are dedicated solely to the consequences of falls. Therefore, understanding the mechanisms that contribute to falls as well as how risk of falls and falls-related injuries can be reduced is an important aspect in geriatric healthcare delivery (Blain et al., 2016; Bousquet et al., 2017; Broadbent et al., 2017).

As more than 80% of the time during hospital admission is spent lying in bed in older persons (Pedersen et al., 2013), bedrest confinement during hospitalization and its negative effects pose particular challenges in the care of the senior citizens in the acute and chronic care setting. Prolonged periods of physical inactivity or bedrest are associated with negative metabolic and functional effects (Agostini et al., 2010; Pisot et al., 2016; Soavi et al., 2016). Some of these include deconditioning in the cardiovascular, skeletal and neuromuscular systems as well as potential deficits in brain function and structure (Grogorieva and Kozlovskaia, 1987; Leblanc et al., 1990; Traon et al., 1998; Perhonen et al., 2001; Pisot et al., 2008; Lipnicki and Gunga, 2009; Rittweger et al., 2009; Dolenc and Petric, 2013; Marusic et al., 2014, 2016; Li et al., 2015; Cassady et al., 2016).

Further complicating this situation is the fact that orthostatic intolerance incidence is markedly accelerated with intravascular instrumentation (Stevens, 1966) or with increased heat stress (Crandall, 2000), and is greater in taller persons (Ludwig and Convertino, 1994) and in anxiety states (Smith et al., 1994).

In addition to reviewing the state of the art knowledge regarding orthostatic intolerance, this review also examines specific countermeasures that can be used against orthostatic intolerance. For instance, orthostatic tolerance has been shown to be improved by exercise (Howden et al., 2004), water drinking (Schroeder et al., 2002), and mental arithmetic (Goswami et al., 2012a). Other countermeasures that could potentially prevent or mitigate orthostatic intolerance can include simple measures such as standing up slowly (de Bruïne et al., 2017) and/or compression of abdominal region (Figueroa et al., 2010). For bed confined persons, countermeasures can include cognitive training (Goswami et al., 2015) and nutritional supplementation (Muscaritoli et al., 2017)—with and without physical activity—in mitigating orthostatic intolerance.

# ORTHOSTATIC CHALLENGE AND ORTHOSTATIC INTOLERANCE

Orthostatic challenge can lead to problems such as dizziness when the cardiovascular response to this challenge is inadequate. In the upright (orthostatic) position, the hemodynamic responses may not be enough to sustain the arterial blood pressure, and syncope may ensue. In a healthy person, upon standing up, 10–15% (∼700 mL) of blood volume can pool due to gravitational effects in the legs (Naschitz and Rosner, 2007), with 50% of this shift occurring within the first 20–30 s (Toska and Walloe, 2002). The increased hydrostatic pressure induced by standing leads to: increases in transmural pressure in lower extremities; fluid movement from the vascular compartment to the interstitial space; and a decrease in plasma volume (Hinghofer-Szalkay and Greenleaf, 1987). This could potentially cause reductions in venous return, cardiac pre-load and cardiac output. In persons with normal cardiovascular regulatory capability, however, mean arterial blood pressure is maintained by compensatory increases in heart rate and total peripheral resistance (Rowell, 1993).

The autonomic nervous responses to standing are based on a balance between parasympathetic and sympathetic activity: the reflex tachycardia indicates vagal withdrawal and involves increases in sympathetic activity (Ramirez-Marrero et al., 2007). Hormonal responses, in comparison to autonomic neurogenic reflexes, play only a minor role in the early, and rapid hemodynamic responses during standing up.

**Abbreviations:** ADL, Activity of daily life; BP, Blood pressure; BP-MCA, Blood pressure corrected to the level of middle cerebral artery; CCT, Computerized cognitive training; CO, Cardiac output; CVRi, Cerebrovascular resistance index; DBP, Diastolic blood pressure; ETCO2, End tidal CO<sup>2</sup> pressure; HF, High frequency; HR, Heart rate; HUT, Head up tilt; JNC, Joint National Committee; LF, Low frequency; MA, Mental arithmetic; MA + HUT, Mental arithmetic in HUT subjects; MAP, Mean arterial (blood) pressure; MFV, Mean flow velocity (Cerebral artery); MSNA, Muscle sympathetic nerve activity; NTS, Nucleus tractus solitaries; OH, Orthostatic hypotension; OI, Orthostatic intolerance; PP, Pulse pressure; SBP, Systolic blood pressure; SCI, Spinal cord injury; SGLT-2, Sodium glucose co-transporter 2; SV, Stroke volume; TPR, Total peripheral resistance.

# Age-Related Changes in Orthostatic Challenge Induced Responses

It is known that older persons are particularly prone to falls. The increase in falls due to postural hypotension and loss of postural stability is of major concern with aging. Age-associated changes in baroreflexes may also play a role in older subjects: advancing age is associated with increasing barosensory vessel wall stiffness and decreasing cardio-vagal autonomic control effectiveness (Monahan et al., 2001). There are additional important factors outlined below which can lead to orthostatic intolerance in older persons. However, before discussing these aspects it is necessary to outline the cardiovascular responses—as well the role skeletal muscles play in enhancing venous return—that occur during postural changes from supine to upright.

### Cardiovascular Responses to Standing Upright

Standing up decreases aortic and carotid blood pressure thus leading to unloading of the aortic arch and carotid sinus baroreceptors. This results in a rapid increase of heart rate via vagal withdrawal and slower elevation of heart rate and peripheral vascular resistance through sympathetic activation to ensure that arterial blood pressure is maintained (Rowell, 1993). During standing, the baroreflex-mediated sympathetic activity has minimal effect on venous tone in the lower limbs due to scarce sympathetic innervation in the veins within limb muscles (Fuxe and Sedvall, 1965; Samueloff et al., 1966). As a consequence, quiet standing results in extensive venous pooling in the legs due to absence of the action of the skeletal (muscle) pump. Contraction of lower limb muscles (along with one-way venous valves) pumps the pooled blood in the veins back to the heart, increases cardiac pre-load, and consequently cardiac output (Guyton et al., 1962; Ten Harkel et al., 1994).

Musculoskeletal activities of the lower limbs during standing have primarily been studied in relation to the central postural control system which generates proper balance corrections in the upright stance by coordinating sensory-motor control components (somatosensory, visual ocular, vestibular) to naturally optimize body positioning (Dichgans and Diener, 1989). However, the resultant slight postural sway during standing also serves as an important contributor in promoting venous return (Inamura et al., 1996; Murata et al., 2012; Blaber et al., 2013). Surprisingly, despite it being long known that the skeletal muscle pump helps in the maintenance of blood pressure (Guyton et al., 1962), cardiovascular and postural reflexes have been investigated as independent control mechanisms (Winter, 1995; Fadel, 2008). In addition, the influence of the skeletal muscle on the cardiovascular hemodynamic responses have been considered to be mechanical rather than reflex-mediated (Guyton et al., 1962; Ten Harkel et al., 1994).

# Postural Instability with Aging

Postural disturbances, which can arise due to external or internal causes in daily life, can lead to problems with balance during standing. Standing is maintained by a fine balance between the ability to detect postural disturbances and the generation of proper responses. However, these abilities worsen as age increases, and in older persons can cause imbalance and greater falls risk (Blaszczyk et al., 1994; MacKey and Robinovitch, 2006; Hsiao-Wecksler and Robinovitch, 2007). In addition, aging is associated with worsening of the functioning of the somatosensory and motor systems, which in turn can lead to poor standing balance (Lord et al., 1991; Hurley et al., 1998).

The alteration in standing balance is not only a problem of older persons. Changes in reactive postural responses have been seen in healthy young persons with reductions in somatosensory perception in laboratory experiments as well as in patients with peripheral neuropathy or alterations in vestibular function. An understanding of how the cardiovascular, postural and skeletal muscle systems interact to maintain upright posture is important in the development of treatment and training strategies for individuals for whom one or both systems are compromised.

## The Cardio-Postural Model

As introduced above, the control relationships between muskuloskeletal, postural and cardiovacular systems is important and should be investigated. Studies conducted by Claydon and Hainsworth (2006) showed a link between postural sway and prevention of syncope. They reported that participants who had poor tilt table orthostatic tolerance but never fainted during normal standing showed greater postural sway than patients who experienced frequent syncopal episodes (Claydon and Hainsworth, 2006). In addition, Novak et al. (2007) proposed a cardio-locomotion coupling conceptual model in which muscle traction forces generated during walking pump venous blood and propel it to the heart with a step synchronized rhythm. These observations demonstrate the role of the skeletal (muscle) pump in cardiovascular control in conditions of insufficient vascular control.

Based on some studies in which a correlation between postural sway and blood pressure (BP) was observed, a new physiological model—cardio-postural interactions—which includes the interactions between cardiovascular control and postural changes has been developed (Souvestre et al., 2008; Blaber et al., 2009; Goswami et al., 2012b; **Figure 1**). Using this model, a relationship between calf muscle activation, measured with electromyography (EMG) and blood pressure during passive standing has been observed: during passive standing increased EMG activity was associated with subsequent increases in BP and decreased EMG activity followed by decreases in BP (Souvestre et al., 2008; Blaber et al., 2009). Additionally, Garg et al. (2014a) have also reported that older individuals have significant alterations in the relationship between BP and EMG activity. These investigators observed that in non-fainting older subjects the cardio-postural relationship was similar to young subjects; however, the percentage of time of significant coherence between EMG and BP variation increased with decreasing frequency. In the young age group, a similar frequency-based difference was not observed, which could indicate a cardiopostural behavior shift toward larger time scales (i.e., lower frequencies) with increasing age (Garg et al., 2014b). Verma et al. (2017) also showed that causal influence of skeletal muscle pump activity on blood pressure was significantly reduced with aging. Further understanding of the degree and mode of interaction

between the cardiovascular control and centrally-regulated sensory-motor controls and how cardio-postural interactions are affected by bedrest confinement as well as by disease states is required.

Recently, Rodriguez et al. (2017) reported that there is no difference in spontaneous baroreflex sensitivity between stroke patients and healthy controls during standing up from a sitting posture. However, the gain values from SBP to EMG impulse were attenuated in patients, which suggests a post-stroke impairment of muscle-pump and baroreflex. This correlates with the finding that the BP drop upon standing in stroke patients approached the criteria for OH (orthostatic hypotension) (see later sections) and that the stroke patients showed a slightly longer BP recovery time. The impairment of muscle pump and baroreflex could arise due to muscle atrophy after bedrest in patients with stroke and/or by affecting nerve pathways to the muscular system. Further research into recording and analyzing motor unit activation and recruitment should be carried out.

# Orthostatic Hypotension (OH) and Aging

Orthostatic hypotension is common in older adults. The prevalence of asymptomatic orthostatic hypotension [which is defined as a reduction in diastolic BP ≥ 10 mmHg or systolic BP ≥ 20 mmHg, with BP measured supine and at 3-min standing] in 5,201 randomly selected persons aged 65 and older with natural history of—or with risk factors for—cerebrovascular or cardiovascular diseases was found to be 16.2% (Rutan et al., 1992). This prevalence increased to 18.2% when subjects in whom the sit to stand test was aborted due to dizziness upon standing were included (Rutan et al., 1992). Based on different studies, the prevalence of OH varies from 5 to 50% among elderly subjects (reviewed in Weiss et al., 2004).

It has been reported that OH is significantly associated with problems in walking and/ or falls, and histories of transient ischemic attacks and even myocardial infarction (Rutan et al., 1992). Indeed, OH is one of the main risk factors for falls (Tinetti et al., 1988). In older persons, falls are among the top five causes of death (Kannus et al., 2005).

As blood pressure is generally known to be lower in summer than in winter, the question whether seasonal variations in blood pressure affect OH incidence needs also to be further examined. Weiss et al. (2006) studied OH variation between seasons in 502 older male and female in-patients (mean age 81.6 years). Each older person performed orthostatic tests, 30 min after meals, three times per day. In their study, OH was documented in 35% of the population. While the researchers did not observe baseline BP differences between seasons, in older persons the drop in blood pressure upon standing up in the morning was greater in summer. Furthermore, orthostatic hypotension was more common in summer vs. winter months (38 vs. 27%; p = 0.02), and after taking into account all the confounders, orthostatic hypotension risk was found to be 64% higher in summer (Weiss et al., 2006).

# Orthostatic Hypotension vs. Intermediate BP Drop

Due to the limited physiological reserve, older persons are more susceptible to changes in environment or pathological states. Hence, it is not surprising that some older patients complain of light-headedness/dizziness when standing up even though their blood pressure does not fall into the OH range. Intermediate BP drop, defined as reduction of 5–9 mmHg in diastolic BP and/or 10–19 mmHg in systolic BP, can act as an OH predictor. Intermediate BP drop is also associated with higher mortality rates (discussed in Weiss et al., 2004). The current Joint National Committee (JNC) recommendations suggest that a 10 mmHg BP reduction during orthostatic loading—when associated with symptoms—should be considered as clinically relevant (see Weiss et al., 2004)

# Postprandial Hypotension and Aging

Aging is accompanied by an increase in the tendency for blood pressure to decrease following a meal. Classically, hypotension following a meal (postprandial hypotension) is the arbitrary decrease in systolic BP ≥ 20 mmHg or postprandial reduction in systolic BP to <90 mmHg within 2 h following a meal (Jansen and Lipsitz, 1995). The maximal decrease in BP occurs usually between 30 and 60 min post-meal, but it may occur later in some individuals (Luciano et al., 2010). This decrease in BP is associated with a number of pathological events, including greater falls incidence (Le Couteur et al., 2003), syncope due to failure to maintain compensatory tachycardia and normal noradrenaline levels (Lipsitz et al., 1986), coronary events, stroke, and overall mortality (Aronow and Ahn, 1997).

The constitution of a meal seems to contribute toward postprandial hypotension. Carbohydrates (e.g., glucose) show blood pressure reducing effect (Potter et al., 1989). Indeed, older persons with no signs and symptoms of postprandial hypotension show blood pressure decreases following ingestion of a meal rich in carbohydrates (Lipsitz and Fullerton, 1986). Meal ingestion leads to blood pooling in the abdominal region/ vasculature but blood pressure is maintained due to the appropriate homeostatic responses (Lipsitz et al., 1993). In patients with dysautonomia, however, meal ingestion may be accompanied with a decrease in peripheral resistance and reductions in blood pressure. This could potentially lead to the development of presyncopal signs, symptoms or even orthostatic intolerance (Heseltine et al., 1990; Lipsitz et al., 1993).

Post-prandial hypotension has been reported in a large proportion of healthy individuals (Jones et al., 1998; Vloet et al., 2005; Van Orshoven et al., 2010). Additional risk factors for postprandial hypotension include specific co-morbid conditions, such as non-insulin-dependent diabetes mellitus (Jones et al., 1998), especially if accompanied by some degree of autonomic neuropathy (Sasaki et al., 1992), autonomic dysfunction (Shannon et al., 2002), hypertension (Grodzicki et al., 1998), Alzheimer's disease (Idiaquez et al., 1997), and Parkinson's disease (Loew et al., 1995). In addition, older patients with polypharmacy, especially those who use diuretics and psychotropic medications, are more likely to be affected by hypotension following a meal ingestion (Luciano et al., 2010).

Postprandial hypotension is often under-recognized among older persons (Luciano et al., 2010). In a study evaluating 85 older persons (mean age 80 ± 7 years) admitted to geriatric departments in Dutch hospitals, 67% presented a significant post-meal decrease in SBP of 34 ± 4 mmHg. Interestingly, orthostatic hypotension accompanied 52% of patients with an average systolic blood pressure decrease of 44 ± 4 mmHg following standing, and 37% of patients had both orthostatic and postprandial hypotension (Vloet et al., 2005). Indeed, a prospective study in 499 nursing home dwellers (>62 years, mean age 80) showed that the maximal decrease in systolic BP after a meal was a risk factor for syncope, falls, new coronary events or new stroke, and contributed significantly to greater mortality (Aronow and Ahn, 1997). In a study of 179 randomly selected low-level-care older residents (65 years and older, mean age 83 years) in long-term health facilities, postprandial hypotension was shown to be a powerful predictor of deaths arising due to all causes (Fisher et al., 2005).

Post-meal drop in blood pressure is also concomitant with asymptomatic cerebrovascular damage in patients with essential hypertension (Kohara et al., 1999). This observation was extended to residents of the general community (65 ± 9 years old, n = 1308), who were free from coronary heart disease or heart failure and participating in a general health checkup. A higher prevalence of cerebral lacunar infarctions as evidenced by MRI was shown in subjects with postprandial hypotension compared with controls (Tabara et al., 2014).

# Effects of Skin Temperature Changes on Orthostatic Intolerance

Temperature changes have been shown to affect orthostatic tolerance: heat stress enhances and cold stress reduces orthostatic challenge-induced syncope (Crandall, 2000). In addition, cardiovascular responses during changes in posture are influenced by temperature changes: standing up/working in upright positions for longer periods in hot weather is associated with poor orthostatic tolerance and increased incidence of syncope (Wilson and Crandall, 2011). These investigators have further suggested that alterations in orthostatic tolerance and cerebral blood flow seen in thermal stress are not due to the neural-induced postural reflexes but rather due to changes in cardiac mechanics and function. What effect heat has on orthostatic tolerance is particularly important for workers exposed to hot temperatures while standing. While heat may predispose such workers to collapse and orthostatic intolerance, it could also lay the foundation for proposing countermeasures such as skin cooling that could improve orthostatic tolerance. Except for the Weiss study (Weiss et al., 2006), in which older persons in nursing homes showed greater incidence of orthostatic hypotension in summer, there is no systematic study which has investigated the effects of temperature changes in older persons. More research is needed in this area.

# Orthostatic Intolerance in Disease Conditions

There are many disease conditions that appear to affect both postural and cardiovascular control that might be better explained and treated with the integration of the systems into a single predictive model; these include peripheral neuropathy (e.g., diabetic neuropathy), concussion and long term bedrest itself. Orthostatic intolerance can also arise in patients with stroke (Verma and Eltawansy, 2016), diabetes mellitus (Eguchi et al., 2006), multiple sclerosis (Pintér et al., 2015), Alzheimer's and Parkinson's disease (Bae et al., 2014), and traumatic brain injury (Kanjwal et al., 2010). The use of medications in patients such as the new SGLT-2 inhibitor dapagliflozin could also lead to hypotension and dizziness upon standing up (Chao and Henry, 2010). Similarly, patients with autonomic failure have difficulties elevating vascular resistance (Smit et al., 1999) during standing up, which could predispose them to orthostatic intolerance. While the data from aforementioned studies are from mixed age groups (young adults and middleaged persons), to what extent these disease conditions affect orthostatic tolerance in older persons needs to be systematically studied.

We are only aware of one study that has investigated in older persons the post-stroke relationship in heart rate variability (HRV; Rodriguez et al., 2017) at rest and during changes in posture. A greater magnitude of decrease in LF HRV modulation upon standing up for the stroke group—indicating a transition to less sympathetic modulation—on average compared to the control subjects was observed. This suggests a paradoxical decrease in sympathetic modulation of the heart rate during orthostatic challenge, which could potentially arise either due to ischemic damage affecting nucleus tractus solitarius (NTS) signaling, resulting in constant sympathetic activity, which over time can increase resistance at adrenergic beta-receptors to stimulation or there may be a greater inhibition of sympathetic catecholamine release. Patients with hypertension, stroke, and chronic heart disease often display this type of reactivity and the Rodriguez et al. (2017) observation of resting heart rate in the stroke group being lower before and after standing are consistent with either of these theories. While the exact mechanisms for these differences are still unknown, Tang et al. (2012) attributed autonomic dysfunction as a major contributor to the high prevalence of orthostatic hypotension in their stroke patients. Furthermore, the extent of DBP or SBP reduction after standing between the stroke and healthy control groups was not different (Rodriguez et al., 2017). However, the mean BP drop for the stroke group in this study approached the criteria for OH. Earlier studies have reported that a greater risk of syncope occurs when cerebral perfusion reduces in response to a drop in central BP, and that this mechanism is related to a sudden withdrawal of sympathetic activity (Cooke et al., 2004). The greater baseline sympathetic modulation (that is, LF values) in the stroke group could lead to an earlier withdrawal of sympathetic vascular tone, with an increased risk of syncope upon standing up, particularly if beta-adrenergic responses are blunted (Overgaard and Dzavik, 2008).

Another condition that has been shown to increase the susceptibility to orthostatic hypotension is spinal cord injury (SCI; Claydon and Krassioukov, 2006). Indeed, orthostatic maneuvers carried out during mobilization and physiotherapy are associated with hypotension in up to 70% of the patients with SCI (Claydon and Krassioukov, 2006; Popa et al., 2010; Sahota et al., 2012). Although SCI is mostly found in the younger population, older people are at risk of traumatic SCI from falls (Ahn et al., 2015), justifying more studies on the repercussion of SCI on OH in older individuals.

OH that accompanies SCI can also complicate a patient's participation in rehabilitation, especially when carrying out exercises (Claydon and Krassioukov, 2006; Popa et al., 2010). For instance, post-exercise hypotension can occur for up to 12 h following exercise (Claydon et al., 2006). These investigators investigated the effects of graded arm cycling until exhaustion in 29 chronic SCI patients (cervical-level: n = 19; thoraciclevel: n = 8). Following exercise it was observed that MAP decreased in patients with cervical SCI (−9.3 ± 2 mmHg) but increased in thoracic SCI (8.4 ± 5 mmHg; P < 0.001). They suggested that these effects might partially arise due to the loss of descending sympathetic nervous control of the heart and vasculature, which accompanies SCI of higher levels. However, these data were obtained from a sample of younger persons. Therefore, interpretations in terms of effects of exercise in older people with SCI must be applied with caution.

# COUNTERMEASURES

The search for countermeasures against orthostatic intolerance continues. Even after several decades of investigation, the problem of orthostatic intolerance in older persons remains. As the aging population increases worldwide, there an urgent need to develop innovative countermeasures against orthosatic intolerance. It is important, for example, when intervening in the process in which bedrest confinement leads to orthostatic intolerance and falls, that a holistic multifactorial approach which takes into account key factors such as nutrition, (de)conditioning, muscle loss, cardiovascular and vestibular effects, is followed. Such an approach will lead to more effective management of bedrest confinement, reduced orthostatic intolerance and falls and falls-related injury, and consequently decrease public health costs, costs for older people and their families as well as decrease secondary care dependency.

Outlined below are some of the possible countermeasures that are currently in use or being examined. While not all of these interventions have been tested in older persons, we present here some evidence that these countermeasures can reduce orthostatic intolerance in young and middle aged persons.

# Physical Activity/Exercise in Ambulatory Persons and in Immobilized Persons

Physical exercise is therapeutic for many medical disorders (Jolliffe et al., 2001), and it can also counter muscle wasting in acute illnesses. Muscle wasting during bedrest confinement is highly pronounced at old age (Kortebein et al., 2007), and compromised muscle function is the major obstacle to remobilization in elderly patients (Creditor, 1993). Therefore, physical exercise, applied as early as possible during a hospital stay is now seen as the gold standard in acute hospital care for 75 year + people.

# Aging, Bedrest Confinement and the Need for Early Remobilization: Role of Physical Exercise

Mobility, especially ambulatory mobility, is an essential aspect of quality of life and independence. As the aging process proceeds, ambulatory ability begins to deteriorate as a result of sarcopenia (aging-associated degenerative loss of skeletal muscle quality, mass and strength; Phillips, 2015) and dynapenia (agingassociated loss of muscle strength not arising due to muscular and/or neurologic diseases; Clark and Manini, 2008). Sarcopenia is a multiple etiological syndrome, thought to partly arise by cell death of α-motoneurons, i.e., the nerve cells that supply muscle fibers with information from the central nervous system. Skeletal muscle fibers, and α-motor neurons of the faster type are likely to be more affected than those of the slow type, which offers an explanation why speed and power are particularly eroded during aging (Vandervoort, 2002; Deschenes, 2004). Muscular atrophy is different from this—it affects slow fibers more than fast fibers, and one best understands it as muscle's self-adaptation to reduced demands (Li et al., 2013).

In addition to aging, illness and injury can impact ambulatory ability by depriving muscles of normal stimulation related to ambulation. Such disuse atrophy combined with sacropenia and dynapenia can result in significant muscle wasting (Kortebein et al., 2007; Suetta et al., 2009) contributing to deconditioning that can have many negative medical consequences (Brown et al., 2004) including risk of falls and falls injury. Almost immediately after hospitalization admission, skeletal muscle deconditioning begins and deficits in activity of daily life (ADL) can be observed as early as day two of hospitalization (Hirsch et al., 1990). If deconditioning severely limits or blocks ambulatory activity then reduction in muscle strength and power can become self-perpetuating. Such muscle deficits are often accompanied by generalized inflammation and metabolic disorders which complicate muscle function recovery. To avoid such a vicious cycle linking bedrest confinement to muscle wasting and further reduced ambulation requires immediate intervention after hospitalization admission to remobilize the patient as early as possible during and after hospitalization (Singh et al., 2008; Goswami, 2017).

When prescribing physical activity as a countermeasure to muscle deconditioning, aspects such as metabolic state of the muscles as well as amount of mechanical stress and myo-electrical activation to be applied must be addressed. Resistance/"strength" training for instance involves large mechanical stress and large myo-electrical activity, and it leads to enlargement mostly of the faster fiber types. Accordingly, resistive training is recommended by many practioners to combat sarcopenia, in the absence of any causative cure. However, resistive exercise regimens that would normally lead to muscle enlargement (hypertrophy) fail to even maintain calf muscle strength and size under bedrest conditions (Alkner and Tesch, 2004; Rittweger et al., 2005)—like many other countermeasures tested against bedrest-induced deconditioning (Pavy-Le Traon et al., 2007). The available literature thus underlines the difficulty and complexity involved in adequately training muscles in bed confined persons. Therefore, one has to be doubtful with regards to the effectiveness of the general care that is currently being carried out in hospitals, as elastic bands or static exercise using gravity as a resistor are bound to be sub-optimal from a muscle physiologist's point of view, while gait and balance training may perhaps help only to improve balance (Mulder et al., 2015).

To be effective, physical interventions need to be not only multidimensional, standardized, and interdisciplinary, but individualized as well. However, such evidence-based intervention programs (let alone individualized) are still lacking for hospitalized older people. As a result, in many hospitals, the acute care of older persons does not incorporate procedures to promote remobilization early enough. Physical exercise interventions are often delayed and the procedures implemented may be based on individual therapist experience leading to nonstandardized interventions which are often only applied after substantial loss of muscle mass and function has occurred. Thus patients may be discharged without recovering sufficient physical function leading to a vicious circle of hospital deconditioning leading to injury and further hospitalization and further dependency (Singh et al., 2008; Goswami, 2017).

# Ingestion of Water to Attenuate OI or Postprandial Hypotension

The responses to water consumption in young healthy persons are complex with a simultaneous increase in vagal tone, as evidenced by bradycardia and an increase in the high-frequency component of the HRV, and some vasoconstriction, as evidenced by a rise in peripheral resistance (Brown et al., 2005), but no real changes in blood pressure (Jordan et al., 2000; Brown et al., 2005). These changes occur despite an increase in muscle sympathetic nerve activity (MSNA) and higher levels of norepinephrine in plasma. Interestingly, the water-induced vagal stimulation is enhanced by drinking cold water (Girona et al., 2014). In subjects showing some degree of baroreflex impairment, the lack of increased vagal tone may no longer attenuate the vasoconstrictor effects of sympathetic activation and may thus unmask hypertension. Murakami et al. (1993) showed that borderline hypertensives had reduced parasympathetic tone. Parasympathetic withdrawal has been shown to be an important precursor to blood pressure elevations in women with primary hypertension (Dabrowska et al., 1996). Furthermore, a recent paper using HRV (Goit and Ansari, 2016) showed that 120 newly diagnosed hypertensives (37–42 yrs) had lower parasympathetic tone compared to controls. Indeed, in severe autonomic failure patients, water drinking substantially increases blood pressure (Jordan et al., 2000; Cariga and Mathias, 2001) and is associated with increases in plasma norepinephrine (Jordan et al., 2000).

In older persons, the effects of water drinking has been sparsely studied. However, one study has clearly shown that drinking half a liter of water leads to acute increases in blood pressure also in older subjects (Jordan et al., 2000). Therefore, it is no surprise that water ingestion has been proposed to attenuate orthostatic or postprandial hypotension (Shannon et al., 2002). Indeed, drinking half a liter of water in healthy subjects improved orthostatic tolerance (assessed using head-up tilt test followed by graded lower body negative pressure; Schroeder et al., 2002). Intake of approximately half a liter of tap water in <5 min in patients suffering from primary autonomic failure has been shown to improve both the postprandial drop in blood pressure and orthostatic hypotension (Shannon et al., 2002). Similarly, water drinking also decreased the orthostatic tachycardia in patients with idiopathic orthostatic intolerance (Shannon et al., 2002). Grobéty et al. (2015) have also shown recently that ingestion of a light breakfast of about 400 Kcal in older people (67 ± 1 years) lead to reductions in both diastolic and systolic BP, decreases that were not found in younger control subjects (25 ± 1 years), and that the prior ingestion of 500 mL water cut by half the decrease in systolic BP.

# Mental Challenge as a Countermeasure Effects of Mental Challenge

Mental challenge increases heart rate, cardiac output, force of cardiac contraction, and blood pressure (Lackner et al., 2010). It is accompanied by venoconstriction and arterial vasoconstriction in the splanchnic, renal, and cutaneous circulations (Callister et al., 1992). The mental arithmetic task has also been shown to elevate epinephrine, norepinephrine and plasma renin activity in venous blood.

## Beneficial Role of Mental Challenge as a Potential Countermeasure against Orthostatic Intolerance

In a pilot study, Goswami et al. (2012a) have shown that the orthostatic tolerance of young persons can be improved by the application of mental arithmetic. Mental arithmetic (MA) done during head up tilt (HUT) leads to larger increases in heart rate which can maintain cardiac output—than heart rate changes caused by MA or HUT alone (Lackner et al., 2010).

Mental challenge-induced increases in sympathetic activity also affect the venous system. As 70% of total blood volume is approximately in the venous system, and veins are strongly controlled by the sympathetic system (reviewed in Pang, 2001), the venous system can influence the amount of blood returning to the heart—and consequently the cardiac preload—when its capacity is changed. As discussed above, mental challengeincreases the sympathetic activity, which in turn increases cardiac output; this could be sustained by larger increases in venous tone, greater mean circulatory filling pressure and thus greater venous return (see Guyton's analysis, Montani and Van Vliet, 2009). The results of Goswami et al. (2012a) are supported by studies which have reported that mental challenge is accompanied by splanchnic vasoconstriction (Callister et al., 1992).

The responses to mental challenge conducted in upright position are unsurprising, as mental challenge induced central drive modulates physiological responses (e.g., cardiovascular reflexes, Ross and Steptoe, 1980). For instance, mental loading can affect baroreflex function via it's actions at the hypothalamus, pons or medulla (Stephenson, 1984). It also appears that the application of mental stress modifies the input to the baroreceptors, which occurs during orthostatic loading (Goldstein and Shapiro, 1988).

## Using Mental Stressors: Advantages of Mental Arithmetic

Mental arithmetic, Stroop color-word conflict, and public speaking are commonly used mental stressors (Steptoe and Vogele, 1991). Mental arithmetic is a calculations task that involves several cognitive mechanisms, including a consciously executed calculation method or a direct retrieval of the answer from memory (McCloskey et al., 1985). Overall, the frontal lobe is involved in calculations that are conscious but temporocentro-parietal activity is involved in direct automatic retrieval of results (Pauli et al., 1996). Mental arithmetic can be performed in all circumstances, including emergency ones (that is, at the bedside or roadside). However, caution must be observed when extrapolating the effects of mental arithmetic on orthostatic intolerance in older persons, as the data from Goswami et al. (2012a) were obtained mainly from younger subjects.

# Timing of Mental Challenge Application

Goswami et al. (2012a) propose that mental challenge could be used as a countermeasure to prevent orthostatic intolerance, particularly in persons who feel dizzy upon standing up, or to counteract hypotension that often occurs during hemodialysis. However, as the cardiovascular responses to mental arithmetic decrease over time, an important aspect that must be considered when using mental challenge as a countermeasure is the exact timing of the application. Lackner et al. (2010) have reported that a reduction in the hemodynamic responses to orthostatic and mental challenges occurs over time when each of these stressors is applied alone (Lackner et al., 2010). For instance, mental challenge induced cardiovascular effects occur maximally within the first two to three min of application (Lackner et al., 2010). As longer applications of mental challenge can lead to habituation/adaptation of responses (thus decreasing the effectiveness of mental challenge), Goswami et al. (2011) recommended that mental arithmetic should be applied only 2– 3 min preceding the occurrence of orthostatic intolerance (i.e., if a person is known to become dizzy upon standing up, they should start their mental arithmetic calculations 2–3 min prior to standing up).

# Other Potential Countermeasures in Older Persons

Some other countermeasures that could potentially prevent orthostatic intolerance include standing up slowly (de Bruïne et al., 2017). Standing up slowly in older persons with histories of orthostatic hypotension has been shown to antagonize the blood pressure decreases within the first 15 sec of changes in posture/ standing (de Bruïne et al., 2017).

Recent evidence also suggests that cognitive training (Goswami et al., 2015) and nutritional supplementation (Muscaritoli et al., 2017)—with and without physical activityaffect vascular function and, therefore, could potentially improve orthostatic intolerance. Aspects of these interventions are now examined.

## Cognitive Training

The concept of using cognitive training as a countermeasure in preventing functional decline originates from the observations that cognitive interventions promote functional outcomes, particularly improvements in mobility in older sedentary persons (Verghese et al., 2010). With a novel computerized cognitive training (CCT) intervention, developed from an underlying brain-based model, Marusic et al. (2014) and Goswami et al. (2015) recently assessed whether CCT (provided as simulation of walking through a labyrinth) could improve cognitive functioning—and at the same time mitigate possible bedrestassociated decline from a functional perspective. The bedrest confinement, carried out in healthy persons aged between 18– 30 yrs (younger persons) and 55–65 yrs (middle aged to older persons), lasted up to 14 days**,** during which the participants were not allowed to leave the bed. While bedrest triggered various functional and metabolic adaptations in older and younger individuals (Pisot et al., 2016; Soavi et al., 2016), it had no remarkable effects on cognition (Dolenc and Petric, 2013)—which is in agreement with some of the previous bedrest studies (for review see Lipnicki and Gunga, 2009; Marusic et al., 2014). Indeed, CCT intervention during bedrest confinement was effective in enhancing cognitive functioning at the end of bedrest and had sustained/long-term positive effect on cognition (Marusic et al., 2016). Improved cognitive functioning was shown to be further effective in positively affecting functional performance parameters (e.g., improved dual-task walking condition and reduced gait variability, Marusic et al., 2015), which could potentially reduce number of falls following prolonged bedrest, especially in older persons. Recently, a neuroprotective mechanism of CCT has been proposed by Passaro et al. (2017), showing as an unaltered plasma brainderived neurotrophic factor (BDNF) only in the CCT group during bedrest. The control group of older adults (who did not do any CCT during bedrest), on the other hand, showed a significant increase in BDNF level at the end of bedrest (Passaro et al., 2017), which was further interpreted as protective overshooting of the brain to counteract bedrest-related negative effects (Soavi et al., 2016; Passaro et al., 2017). Additionally, CCT provided during bedrest confinement also led to a prevention of decreases in vascular function changes in the older persons group during the bedrest confinement (Goswami et al., 2015). As the changes in vascular function and/vascular reactivity have been shown to contribute to orthostatic intolerance following bedrest confinement as well as to cardiovascular diseases Goswami et al. (2015) propose that cognitive challenge during bedrest confinement may prevent bedrest-induced pathological effects on the vasculature. A limitation of the Goswami et al. (2015) study was that, due to ethical constraints, much older/elderly persons could not be included into the study. However, the pilot data generated from the bedrest study can be used as a basis for motivating the ethics committees to allow recruitment of persons with higher ages than the one that were used in the Goswami et al. (2015) study.

Finally, as mental challenge causes increased blood flow in skeletal muscles (Kuipers et al., 2008), and in addition cognitive training (emphasizing virtual movement) could have effects on brain functions related to walking and mobility (thus potentially affecting peripheral blood flow), further research is needed to assess whether cognitive training during bedrest confinement could prevent post-bedrest induced orthostatic intolerance in older adults.

## Nutritional Therapy

Nutritional therapy, along with resistance training affects muscle mass in healthy older adults (Strandberg et al., 1985). Specifically, a protein-enriched diet, approximately equal to 1.3 g · kg−<sup>1</sup> · d −1 provided by red lean meat has been shown to improve the effects of progressive resistant training on lean tissue mass and muscle strength in older women (Daly et al., 2014). However, a Cochrane review has recently concluded that nutritional therapy can reduce healthcare costs but overall the evidence from the studies is too heterogeneous and of limited quality for concluding whether malnutrition or its treatment helps in reducing re-admissions (Muscaritoli et al., 2017).

### Respiratory Training

Inspiratory-expiratory pressure threshold respiratory training has recently been shown to reduce the incidence of orthostatic hypotension in more than 50% of patients with SCI (Aslan et al., 2016; Legg Ditterline et al., in press). Respiratory training induced increases in respiratory capacity leads to increased sympathetic activation and baroreflex effectiveness as well as improvements in respiratory-cardiovascular interactions during central hypovolemia induced by changes in posture (sit to stand) in patients with SCI. However, the potential benefit of respiratory training in older patients suffering from OH has never been tested.

# SUMMARY

This review outlines orthostatic intolerance connected to age, bedrest confinement, and diseases. Aging-associated illness or injury due to falls often leads to hospitalization. As older patients spend significant amounts of time during hospital admission lying in bed, the consequences of bedrest confinement such as physiological deconditioning, functional decline, and orthostatic intolerance represent a central challenge in the care of the vulnerable older population.

The review also discusses contributing mechanisms of orthostatic intolerance and examines countermeasures such as exercise, water drinking, and mental arithmetic. The timing of the countermeasure application is also considered. Finally, this paper emphasizes the importance of an active life style in old age and why early re-mobilization following bedrest confinement is crucial in preventing orthostatic intolerance, falls and fallsrelated injuries in older persons.

# CLINICAL IMPLICATIONS AND FUTURE DIRECTIONS

There is a need to integrate cerebrovascular and cardiovascular variables into a combined analysis to allow for the development of improved orthostatic intolerance and fall risk-screening monitoring in older persons. Hand held devices for the analysis of global HRV (a key aspect for monitoring autonomic function (Grote et al., 2013; Osborne et al., 2014; Patel et al., 2016; Moser et al., 2017; Rodriguez et al., 2017) for instance, are currently available, which helps in streamlining data collection at homes or in mobile settings. Assessing the autonomic system using a simple method that does not utilize the baroreceptor reflex (e.g., oculocardiac reflex) could also help in separating vessel disease and central autonomic dysfunction at the bedside (McLaren et al., 2005). Assessment of HRV and blood pressure variability is also important as HRV has been associated with mortality following myocardial infarction while greater blood pressure variability is associated with greater disability following stroke.

Additionally, rehabilitation post-stroke could benefit from combination of mobile monitoring of the autonomic and cardiovascular systems with effective physical therapy interventions to ensure that older persons are protected from muscle wasting and muscle loss after stroke. Future sit-tostand protocols that include MSNA evaluation, by which

# REFERENCES


the magnitude and timing of sympathetic responses could be measured, could be carried out to clarify the relationship between MSNA and HRV. All these aspects are important in the asssement of orthostatic intolerance risk in patients during hospitalization and upon discharge.

Finally, there is a need to implement more integrated screening procedures during hospital admission to personalize countermeasures against orthostatic intolerance, which can arise due to bedrest confinement.

# AUTHOR CONTRIBUTIONS

NG conceived the idea and wrote the manuscript. AB supported in editing the manuscript and contributed to sections related to cardio-postural interactions. HH-S helped in writing the manuscript. J-PM supported in editing the manuscript and contributed to sections related to water drinking as well as post-prandial hypotension.


in men versus women after 7 days of head-down bed rest. Auton. Neurosci. 100, 66–76. doi: 10.1016/S1566-0702(02)00132-7


have an effect on the orthostatic cardiovascular and hormonal responses? Acta Astronaut. 68, 1589–1594. doi: 10.1016/j.actaastro.2009.09.033


hypotension in non-insulin-dependent diabetes mellitus. Clin. Sci. 94, 65–70. doi: 10.1042/cs0940065


prevalence related to patient selection and diagnostic criteria. J. Aging Res. 2010:243752. doi: 10.4061/2010/243752


**Conflict of Interest Statement:** The section related to mental stress is based on the Doctoral work of NG.

The other authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2017 Goswami, Blaber, Hinghofer-Szalkay and Montani. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Spinal Health during Unloading and Reloading Associated with Spaceflight

David A. Green1,2,3 \* and Jonathan P. R. Scott 1,2

<sup>1</sup> KBRwyle GmbH, Cologne, Germany, <sup>2</sup> Space Medicine Office, European Astronaut Centre, European Space Agency, Cologne, Germany, <sup>3</sup> Centre of Human and Aerospace Physiological Sciences, King's College London, London, United Kingdom

Spinal elongation and back pain are recognized effects of exposure to microgravity, however, spinal health has received relatively little attention. This changed with the report of an increased risk of post-flight intervertebral disc (IVD) herniation and subsequent identification of spinal pathophysiology in some astronauts post-flight. Ground-based analogs, particularly bed rest, suggest that a loss of spinal curvature and IVD swelling may be factors contributing to unloading-induced spinal elongation. In flight, trunk muscle atrophy, in particular multifidus, may precipitate lumbar curvature loss and reduced spinal stability, but in-flight (ultrasound) and pre- and post-flight (MRI) imaging have yet to detect significant IVD changes. Current International Space Station missions involve short periods of moderate-to-high spinal (axial) loading during running and resistance exercise, superimposed upon a background of prolonged unloading (microgravity). Axial loading acting on a dysfunctional spine, weakened by anatomical changes and local muscle atrophy, might increase the risk of damage/injury. Alternatively, regular loading may be beneficial. Spinal pathology has been identified in-flight, but there are few contemporary reports of in-flight back injury and no recent studies of post-flight back injury incidence. Accurate routine in-flight stature measurements, in- and post-flight imaging, and tracking of pain and injury (herniation) for at least 2 years post-flight is thus warranted. These should be complemented by ground-based studies, in particular hyper buoyancy floatation (HBF) a novel analog of spinal unloading, in order to elucidate the mechanisms and risk of spinal injury, and to evaluate countermeasures for exploration where injury could be mission critical.

Keywords: back pain, spine, microgravity, axial loading, countermeasures, IVD herniation risk

# INTRODUCTION

Insertion into microgravity (µG) is associated with fluid redistribution (Norsk et al., 2015), space adaptation syndrome (Thornton and Bonato, 2013) and increases in stature of up to 7 cm (Brown, 1977; Thornton et al., 1977) or 1–3% (Stoycos and Klute, 1993). Such increases are in excess of those (1% or 1–1.5 cm) observed after 8 h sleep on Earth (Tyrrell et al., 1985) and may be associated with back pain (Kerstman et al., 2012).

Increments in stature can present operational issues, such as astronauts being unable to fit into their extra vehicular activity (EVA) suit (e.g., NASA's EVA Mobility Unit; EMU;

### Edited by:

Andreas Roessler, Medical University of Graz, Austria

#### Reviewed by:

Renée Morris, University of New South Wales, Australia Martino V. Franchi, Universitätsklinik Balgrist, Switzerland

> \*Correspondence: David A. Green david.green@esa.int

#### Specialty section:

This article was submitted to Integrative Physiology, a section of the journal Frontiers in Physiology

Received: 25 October 2017 Accepted: 20 December 2017 Published: 18 January 2018

#### Citation:

Green DA and Scott JPR (2018) Spinal Health during Unloading and Reloading Associated with Spaceflight. Front. Physiol. 8:1126. doi: 10.3389/fphys.2017.01126 Nicogossian, 1989) that are assembled with a ∼2.5 cm tolerance from pre-flight stature (Rajulu and Benson, 2009). Furthermore, presently all astronauts travel to the International Space Station (ISS) in the Russian Soyuz capsule, wearing the Sokol suit and fitted into a pre-molded "Kazbek" seat pan liner which, as a result of on-orbit stature increases, can be problematic.

Prior to EVA suit donning, stature is measured following the "On-orbit Growth Measurement protocol," where an astronaut "stands" against a module wall whilst attempting to stabilize themselves by holding handrails whilst a second astronaut marks their height. In the Shuttle era, stature changes were assessed from seated height (Young and Rajulu, 2011) prior to re-entry (Nicogossian, 1977; Thornton and Moore, 1987). Except before an EVA, stature is no longer routinely measured on the ISS. However, evidence of spinal column changes (e.g., Chang et al., 2016) and a potential increased post-flight intervertebral disc injury risk (Johnston et al., 2010) has created a renewed interest in the effect of the space environment on spinal health. Due to the complexity of conducting human ISS experiments the majority of space spinal research is limited to pre- and post-flight which must be complemented by ground-based µG analogs.

# GROUND-BASED ANALOGS OF MICROGRAVITY FOR SPINAL RESEARCH

The most commonly used ground-based analog of µG is longduration, head-down tilt bed rest (HDTBR). Whilst HDTBR has significant utility in evaluating the effect of disuse and thus countermeasures for musculoskeletal de-conditioning (e.g., Rittweger et al., 2005), it is neither a true representation of the space gravitational environment [i.e., Earth's gravity acts "chest-to-back" (+Gx)] with a headward hydrostatic pressure gradient, nor is reflective of the ISS operational environment. Furthermore, up to 15 mins/day can be spent out of the headdown position, and spinal flexion (up to 30◦ ), plus twisting and turning is permitted when head-down. This may explain why spinal elongation induced after 3 days of HDTBR is no greater than with 8 h sleep on Earth (Styf et al., 1997). It may also explain why both cervical muscle hypertrophy and thoracic intervertebral disc (IVD) compression is observed (Belavý et al., 2013) leading to questioning of its validity as an analog for spaceflight-induced spinal changes (Hargens and Vico, 2016).

Dry immersion, where individuals "float" in a partially "flexed" posture on an impermeable membrane via water buoyancy (Navasiolava et al., 2011) has also been used (Watenpaugh, 2016). Dry immersion induces significant back pain after only 1 day, but spinal elongation is moderate (1.5 cm) (Treffel et al., 2017). Furthermore, dry immersion is poorly tolerated, and in addition the head is supported out of the water which may result in cervical loading and neck afferent activation.

A novel ground-based analog of spinal unloading has been developed at King's College London, termed hyper-buoyancy flotation (HBF) (Green et al., 2015). In HBF, subjects lay supine upon a water bed encased within a frame, partially filled (50%) with a super-saturated and hence dense salt (magnesium sulfate) solution. Thus, subjects are buoyant, sinking into the bed in proportion to segmental body mass in a passive relaxed, supine position (negating the hydrostatic pressure gradient) with little or no requirement for stabilizing muscle activation. Four hours of HBF induces a stature increase of 1.8 ± 0.2 cm (Carvil et al., 2017b), comparable to that with 8 h of normal sleep (Tyrrell et al., 1985). However, greater elongation is observed after 8 h (2.4 ± 0.1 cm), as well as the development of moderate, reversible lower back pain that presents after 5–6 h (Green et al., 2015).

# BACK PAIN: FINDINGS FROM MICROGRAVITY AND GROUND-BASED ANALOGS

In fact, back pain features frequently in astronaut memoirs, with one stating that it "comes with the job" (Mullane, 2006) with back pain incidence ranging from 52 to 68% (Wing et al., 1991; Kerstman et al., 2012; Pool-Goudzwaard et al., 2015). Back pain was the fifth most common reason given for medication use in the Shuttle era (Putcha et al., 1999) and remains an issue on the ISS (Wotring, 2015). Lumbar pain is predominantly reported, typically presenting shortly after insertion into µG. Severity is most commonly reported as mild-to-moderate, although 25% is moderate-to-severe (Wing et al., 1991). Pain typically resolves after 2–3 days, although it can persist for more than a week. Astronauts are reported to adopt a "foetal" tuck in an attempt to reduce pain (Thornton et al., 1977), but this may risk spinal damage (Sayson et al., 2013). Whilst moderate back pain has also been reported after HDTBR (Hutchinson et al., 1995), dry immersion (Treffel et al., 2017), and HBF (Green et al., 2015), the underlying mechanisms remain unclear, although spinal lengthening may exaggerate intrathecal ligament tension (Kershner and Binhammer, 2004).

# MECHANISMS UNDERLYING SPINAL CHANGES AND BACK PAIN WITH UNLOADING

Intervertebral disc expansion, and spinal thoracic and lumbar curvature flattening have been proposed to explain µG-induced statue increases (Young and Rajulu, 2011). Loss of spinal curvature and IVD swelling have been observed post-HDTBR (Belavy et al., 2011a,b) in excess of that following 8 h' sleep (Ledsome et al., 1996). 60-day HDTBR has also been shown to reduce lumbar IVD signal intensity, indicative of reduced glycosaminoglycans concentration (Kordi et al., 2015). Three-day dry immersion is associated with increased lumbar IVD volume and water content (LeBlanc et al., 1994; Treffel et al., 2016), whereas the effect of HBF on IVDs is currently being evaluated.

**Abbreviations:** ARED, Advanced Resistive Exercise Device; Gz, Axial acceleration; ESA, European Space Agency; EMU, EVA Mobility Unit; EVA, Extra vehicular activity; FRED, Functional Re-adaptive Exercise Device; GLCS, Gravity Loading Countermeasure SkinSuit; HDTBR, Head-down tilt bed rest; HBF, Hyperbuoyancy floatation; iRED, Interim Resistive Exercise Device; ISS, International Space Station; IVD, Intervertebral disc; MRI, Magnetic reasonance imaging; µG, Microgravity; NASA, National Aeronautics and Space Administration; ROMFE, Range of motion - flexion-extension; TVIS, Treadmill with Vibration Isolation and Stabilisation; T2, 2nd generation treadmill.

Disc unloading is a critical feature of the daily IVD loadunload cycle, regulating composition and structure (Malko et al., 2002). As IVDs are largely avascular, and thus dependent upon membrane diffusion (Holm et al., 1981), the cycle promotes fluid/molecular exchange (Schmidt et al., 2016). Thus, IVD swelling is hypothesized to reduce diffusion and modify both osmotic and hydrostatic pressures (Humzah and Soames, 1988). Indeed, reduced protoglyocan and annuli fibrosus collagen (markers of disc degeneration) has been observed in rodent hindlimb suspension (Holguin and Judex, 2010) and µG (Maynard, 1994; Jin et al., 2013), although this remains to be confirmed in humans (Belavy et al., 2016a).

Utilisation of a new in-flight ultrasound procedure (Marshburn et al., 2014) with seven long-duration ISS astronauts, revealed 14 spinal changes from pre-flight, including disk desiccation and osteophytes, but no significant changes in IVD height or angle (Garcia et al., 2017). A recent MRI study (Chang et al., 2016) also suggests lumbar IVD swelling is minimal and comparable with that from an 8-day Shuttle mission (LeBlanc et al., 1994). However, post-flight images were recorded 24 and 48 h after landing, following exposure to Gz during re-entry and re-ambulation. Supine MRI images in the same astronauts showed decreased (11%) lumbar lordosis, and active lumbar flexion-extension range of motion (ROMFE) (Bailey et al., 2018), which are associated with impaired spine biomechanics and chronic low back pain on Earth (Hides et al., 1996; Freeman et al., 2010). In contrast, IVD water content and passive range of motion were unaffected. Thus, reduced lumbar lordosis may be a significant factor in spinal elongation, back pain, and potential herniation risk. However, the Bailey et al. (2018) study possessed no lumbar curvature, stature or back-pain in-flight measures, although only astronauts with significant pre-flight endplate irregularities (taskforce: Fardon et al., 2014) reported post-flight chronic low back pain and/or disc herniation (Bailey et al., 2018).

Back pain, the loss of lumbar lordosis and reduced ROMFE with µG exposure may be related to intrinsic (rotatores, multifidus, semispinalis, spinalis, longissimus, iliocostalis) spinal muscle atrophy, observed after 8 days (LeBlanc et al., 1995), and 17 days and 16–28 weeks (LeBlanc et al., 2000). Atrophy of the lumbar paraspinal muscles (multifidus, erector spinae, quadratus lumborum, and psoas at the level of L3/4 has recently been observed in NASA astronauts returning from ISS (Chang et al., 2016), consistent with routine ultrasonic observation of lumbar multifidus and transversus abdominis in European Space Agency (ESA) astronauts (Hides et al., 2016). Multifidus contributes to active sagittal and frontal plane stiffness (Panjabi et al., 1989), proprioception (Brumagne et al., 2008), and supports lordosis (Claus et al., 2009), consistent with limiting the forces acting upon IVDs and facet joints (Adams and Hutton, 1985) during bipedal gait (Sparrey et al., 2014).

Greater increases in multifidus signal intensity post-HDTBR (i.e., recovery) are associated with loss of lumbar (L4/L5) lordosis and the incidence of back pain (Belavy et al., 2011b). In astronauts, pre-post-flight changes in multifidus and erector spinae functional and anatomical cross-sectional area correlate with lordosis and active ROMFE (Bailey et al., 2018). Comparable trunk muscle atrophy has been seen after 60 days of HDTBR (Miokovic et al., 2012). Post-flight multifidus atrophy appears to reduce spinal joint stabilization, increase stiffness (Bailey et al., 2018) and directly affect IVDs (Adams, 2015). Thus, spinal stability may be a factor in determining spinal elongation, back pain, and potentially risk of IVD herniation (Belavý et al., 2016b), which may be elevated post-flight (Johnston et al., 2010). As a result, the Functional Re-adaptive Exercise Device (FRED), which engages multifidus and transversus abdominis (Weber et al., 2017) and is suitable even for those with back pain (Winnard et al., 2017), is currently being evaluated by ESA for post-flight rehabilitation.

# SPINAL LOADING IN-FLIGHT

Whilst long duration ISS missions might be described as "approximately six months of uninterrupted spinal unloading," the current in-flight exercise countermeasure programme followed by ESA (Petersen et al., 2016) and other ISS astronauts (Loehr et al., 2015) involves both resistance and aerobic exercise (cycle ergometry or treadmill running), 6 days per week. When running (∼30 min per session) on the 2nd generation treadmill (T2) in order to be comparable to running on Earth (Genc et al., 2010), astronauts are restrained by an "over-the-shoulder" body harness that typically provides up to 70–80% body weight (Petersen et al., 2016). In addition, resistance exercises, such as squat, deadlift, and heel-raise, are performed on the Advanced Resistive Exercise Device (ARED), which provides axial loading up to 272 kg. The loads used in these exercises are typically in excess of those on Earth to compensate for the fact that astronauts are not working against their own body weight.

As such, only the spine above the level of the shoulders may be considered to be unloaded for an entire space mission. Below this level, a more appropriate description might be "short periods of moderate to very high loading superimposed on a background of spinal unloading." What effect brief loading periods have on the spine is, as yet, unknown (Somers et al., 2015). For instance, on Earth spinal length reduces rapidly with loading induced by standing (Tyrrell et al., 1985), weight-training (Bourne and Reilly, 1991), and running (Dowzer et al., 1998), with additional upper limb loading causing further shortening (Fowler et al., 2006). Whether this is the case for treadmill and resistance exercise in µG is unknown, although astronauts have been reported to perform squats with ARED in an attempt to reduce stature.

Provision of static, axial loading may be a potential countermeasure to spinal elongation and/or back pain in µG. The Russian Pingvin (Penguin) suit was developed as a musculoskeletal countermeasure by imposing axial loading (reported to be ∼40 kg) from the shoulder to foot (Gz) via bungee cords tethered to a waist belt (Sevrin and Svertshek, 1991; Kozlovskaya et al., 1995, 2015). Anecdotal reports suggest that it can transiently reverse stature elongation, although, due to the unnaturally high shoulder loading and poor thermal conductivity (and thus discomfort and skin hygiene issues), the suit is poorly tolerated.

To address the Pengvin Suit's limitations, the Gravity Loading Countermeasure SkinSuit (GLCS) was conceived to produce "1 Gz" using elastic fibers to generate multi-stage tension (that accumulates according to the proportion of body mass) in the vertical axis toward the feet (Waldie and Newman, 2011). Following various prototypes, the Mk III GLCS was found to provide ∼0.7 Gz (measured at the feet) and shown to be compatible with acute strength (Carvil et al., 2017a) and aerobic exercise (Attias et al., 2017). Following several critical design and material innovations, the Mk VI SkinSuit was developed by ESA's Space Medicine Office and King's College London to specifically address whether the modified SkinSuit could reduce in-flight spinal elongation, without being uncomfortable or interfering with nominal ISS spaceflight activities. Ground-based studies using the HBF analog show that the Mk VI SkinSuit which provides axial loading equivalent to 20% of bodyweight both attenuates (Green et al., 2015) and reverses induced spinal elongation (Carvil et al., 2017b).

Following successful parabolic flight tests, the Mk VI SkinSuit was also evaluated during ESA's 2015 short duration "IRISS" and, more recently, during the 2016–17 "PROXIMA" long duration mission, where it partially reversed an increase in stature using a novel inflight stature measurement procedure that showed good within-session repeatability. However, whether the SkinSuit induces stature reductions in flight via lumbar IVD compression and/or induction of lordosis, as is observed with brief 10– 30% bodyweight loading (Neuschwander et al., 2010; Shymon et al., 2014), is unknown. This and the effect on intervertebral motion/laxity (Du Rose and Breen, 2016) is currently the subject of further HBF studies.

# LOOKING FORWARD: SPINAL HEALTH ON LONG-DURATION EXPLORATION MISSIONS

Questions remain concerning the significance of the post-flight spinal changes (Chang et al., 2016; Garcia et al., 2017; Bailey et al., 2018) and their relationship to the apparent increased risk of post-flight IVD herniation (Johnston et al., 2010). In addition, whether the in-flight loading currently experienced by astronauts is protective or provocative with respect to spinal health is unclear. Might regular, albeit brief, axial loading be in some way protective of the spine and serve to attenuate spinal muscle atrophy? Or might applying loads, particularly the large loads used in some resistance exercises, on a spine already weakened muscle atrophy and/or anatomical changes increase the risk of damage and injury?

Of the 44 herniations reported by Johnston et al. (2010), only one occurred in an ISS astronaut and only four following long duration missions (Skylab and Mir). Whilst this might suggest a reduced IVD herniation risk with current ISS operations compared with earlier missions, the study of Johnston et al. (2010), by covering the period April 1959 to December 2006, includes only the first 6 years of ISS long duration missions. Since that time, several important changes have occurred. Firstly, compared with earlier, shorter missions, immediate postlanding ambulation and activity (and thus axial loading) is more carefully managed following ISS missions, which may positively influence the risk of post-flight injury (Johnston et al., 2010).

The second important change is the in-flight loading environment. From December 2000, ISS crew utilized the Treadmill with Vibration Isolation and Stabilisation (TVIS) and performed resistance exercise on the Interim Resistive Exercise Device (iRED) (Korth, 2015). However, because of technical issues, in the first 4–5 years, TVIS was frequently operated with reduced harness loading, whilst the maximum iRED resistance was just 136 kg. Therefore, stronger crew easily reached the maximum load resulting in exercise specialists prescribing onelegged exercises and significant increases in exercise volume (repetitions and sets; Loehr et al., 2015). Exercise was the most frequent (12 of 14) cause of musculoskeletal injuries in ISS astronauts between 2000 and 2006 (Scheuring et al., 2009). Of those 12 injuries, nine were to the back, with the majority involving muscle strains sustained using iRED. In comparison, only 2 of 17 documented injuries from all the Gemini, Shuttle and NASA/Mir missions involved the back/spine.

In 2008, the ARED was installed on the ISS, with the T2 treadmill one year later, facilitating provision of frequent and higher spinal loading during exercise. Heavy squats, which are now possible with ARED, may produce high "uncontrolled" instantaneous, impulse loads through the spine, and thus may present an injury risk (Jennings and Bagian, 1996; Scheuring et al., 2009), particularly as IVDs are relatively uncompliant (Maquer et al., 2014). However, there are no published inflight injury data from 2006 onwards, during which time ARED and T2 have been fully operational. That said, since their introduction, no ESA astronaut has experienced a back injury resulting from in-flight exercise that has required modification of their countermeasure programme. Thus, whilst the effect of this enhanced loading environment on spinal health has yet to be determined, evidently spinal muscle atrophy and structural changes remain (Chang et al., 2016; Hides et al., 2016; Garcia et al., 2017; Bailey et al., 2018).

# CONCLUSION

Space-related spinal elongation, back pain, and elevated risk of IVD herniation have historically been considered to be related to IVD swelling. However, recent evidence suggests that trunk musculature atrophy, in particular multifidus, may precipitate loss of lumbar curvature (and thus stature increments) and lead to spinal instability and IVD dysfunction. Current ISS missions involve short periods of moderate-to-high axial loading, superimposed upon a background of prolonged unloading, yet it is unknown how these affect spinal health, and the risk of post-flight IVD herniation. As a result, routine accurate inflight stature measurements combined with back pain recording should supplement spinal imaging, both in-flight, and as soon as possible upon landing coupled with extended spinal health tracking post-flight. In addition, evaluation of the effect of acute and repetitive graded axial loading is warranted both in orbit and in appropriate ground-based analogs such as HBF. Data from such investigations will help understand the role of unloading and in-flight loading upon spinal health, and therefore spinal injury risk, which could be critical in future human exploration missions.

# AUTHOR CONTRIBUTIONS

All authors listed have made a substantial, direct and intellectual contribution to the work, and approved it for publication.

# REFERENCES


# ACKNOWLEDGMENTS

The authors wish to thank the numerous people and institutions whom have been involved in developing and evaluating the SkinSuit, the subjects that have repeatedly answered our requests, and the various funders that have supported aspects of the work including the European Space Agency, the EPSRC and the Radiological Research Trust. A special thank you goes to Phillip Carvil and Julia Attias who became involved as Space Physiology & Health MSc students at King's College London and whom will shortly move onto pastures new with their axial loading doctorates in hand.

changes in astronauts after long-duration spaceflight on the international space station. Spine 41, 1917–1924. doi: 10.1097/BRS.0000000000001873


new exercise device for lumbo-pelvic reconditioning. Physiol. Rep. 5:e13188. doi: 10.14814/phy2.13188


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Green and Scott. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# The Effect of Bed Rest and Hypoxic Environment on Postural Balance and Trunk Automatic (Re)Actions in Young Healthy Males

#### Nejc Šarabon1,2, Igor B. Mekjavic´ 3,4, Ola Eiken<sup>5</sup> and Jan Babicˇ 3 \*

<sup>1</sup> Faculty of Health Sciences, University of Primorska, Koper, Slovenia, <sup>2</sup> Laboratory for Motor Control and Motor Behaviour, S2P, Science to Practice, Ltd., Ljubljana, Slovenia, <sup>3</sup> Department for Automation, Biocybernetics and Robotics, Jožef Stefan Institute, Ljubljana, Slovenia, <sup>4</sup> Department of Biomedical Physiology and Kinesiology, Simon Fraser University, Burnaby, BC, Canada, <sup>5</sup> Department of Environmental Physiology, Swedish Aerospace Physiology Centre, Royal Institute of Technology, Stockholm, Sweden

#### Edited by:

Andrew Blaber, Simon Fraser University, Canada

#### Reviewed by:

Melissa L. Bates, University of Iowa, United States Davide Susta, Dublin City University, Ireland

> \*Correspondence: Jan Babicˇ jan.babic@ijs.si

#### Specialty section:

This article was submitted to Integrative Physiology, a section of the journal Frontiers in Physiology

Received: 11 September 2017 Accepted: 09 January 2018 Published: 25 January 2018

#### Citation:

Šarabon N, Mekjavic IB, Eiken O and ´ Babic J (2018) The Effect of Bed Rest ˇ and Hypoxic Environment on Postural Balance and Trunk Automatic (Re)Actions in Young Healthy Males. Front. Physiol. 9:27. doi: 10.3389/fphys.2018.00027 Prolonged inactivity, such as bed rest induces several detrimental changes within a short timeframe. Impaired postural balance and responses of trunk muscles to (un)expected perturbations were both shown to be impaired after bed rest. Certain populations (e.g., astronauts) are exposed to hypoxic environment in addition to inactivity, similar to bed rest. While the isolated negative effects of hypoxia on postural balance have been observed before, no study to date has examined the combined effects of hypoxia and bed rest on postural balance or trunk muscle responses. In this study, we examined the effects of 21-day exposure to three conditions: (i) bed rest in hypoxic environment (HBR), (ii) bed rest in normoxic environment (NBR), and (iii) ambulatory hypoxic environment (HAMB). Fourteen healthy male subjects crossed over between conditions in a randomized order, with a 4-month break between conditions to ensure full recovery. Most body sway parameters indicated a similar deterioration of postural balance following both HBR and NBR. Similarly, both anticipatory and reactive responses of the trunk muscles (m. erector spinae and m. multifidus) were impaired after HBR and NBR to a similar degree and mostly unchanged after HAMB. Certain body sway parameters were impaired after HAMB, confirming that hypoxia alone can undermine postural balance. On the other hand, some trunk responses were improved after HAMB. In conclusion, the results of our study confirmed previous findings on negative effects of bed rest, but showed little or no additional effect of hypoxia during bed rest. Physical activity during bed rest is encouraged to preserve neuromuscular functions of the trunk. While the HBR condition in our study resembled conditions during space missions, our results could be relevant to other populations, such as patients with pulmonary diseases exposed to bed rest.

Keywords: inactivity, bed rest, hypoxia, balance, trunk, function

# INTRODUCTION

Prolonged bed rest has been shown to induce several physiological and morphological changes, such as muscle atrophy, decrease in bone mineral density (Parry and Puthucheary, 2015) and impaired cognitive abilities (Lipnicki et al., 2009). Moreover, several studies have demonstrated a debilitation of neuromuscular function following bed rest. Yamanaka and colleagues (Yamanaka et al., 1999) reported an inhibition of H-reflex, but no change in motor evoked potentials after a 20-day bed rest, suggesting that the neuromuscular pathways most affected were the sensory ones. The inactivity/unloading effect of bed rest on the muscle sensory nerves may also be implicated in the delayed postural reflex (i.e., reactive) responses of trunk muscles to sudden mechanical perturbation following bed rest (Panjabi, 1992). Trunk stability functions play a role in preventing spinal injuries and allow for good quality of limb movements (Hoffman and Gabel, 2013). In aging adults confined to bed rest, muscle atrophy is more pronounced in antigravity muscles, including several trunk muscles (Ikezoe et al., 2012). However, bed rest can impair reflex pathways even in limb muscles such as m. biceps brachii (Nakazawa et al., 1997). To our knowledge, the only study that investigated the changes of trunk neuromuscular stabilizing functions following bed rest, found both anticipatory postural adjustments and postural reflex responses of trunk muscles to be impaired, with anticipatory postural adjustments nearly returning to baseline after 14-day rehabilitation period but postural reflex responses remaining substantially deteriorated even after such rehabilitation (Sarabon and Rosker, 2015).

Additionally, several studies have explored the change in postural balance after bed rest. Even a short 5-day bed rest protocol was found to impair postural stability, particularly when the stance task was accompanied with a dynamic head tilting (Mulder et al., 2014). Another study (Kouzaki et al., 2007) found a significant increase in postural sway during quiet stance after a 20-day bed rest; the balance decay taking place regardless of whether the bed rest included a resistance training countermeasure, or not. Furthermore, a 60-day bed rest was shown to impair both static (quiet stance on solid surface) as well as dynamic (standing on an unstable surface) balance in women. In dynamic balance, the anterior-posterior sway was increased more than the medial-lateral sway, suggesting a more pronounced effect of bed rest on distal muscles (Viguier et al., 2009). Interestingly, the static balance measures were restored faster in the group that underwent strength sessions and supine treadmill exercise within lower-body negative pressure box during bed rest, but no difference in the speed of recovery was reported for the dynamic balance measures between this exercise countermeasure group and a control group (no countermeasure). The authors suggested that this observation could have been a result of a specific effect that their intervention had on slow, oxidative muscle fibers, which are responsible for static balance (while faster anaerobic fibers are involved in dynamic balance). Since hypoxia disturbs the oxygen transportation to the muscles, it would be particularly interesting to see, whether it can further aggravate the static balance impairments, caused by bed rest.

Exposure to hypoxia affects postural stability (Fraser et al., 1987), even at relatively low altitudes (∼2,500 m) (Wagner et al., 2011). Body sway changes following hypoxia seem to be more pronounced in the sagittal plane with eyes open (Nordahl et al., 1998). The present study was conducted within the framework of a larger study investigating the separate and combined effects of bed rest and hypoxia on the structure and function of several physiological system (PlanHab study; MEDS-IMPS and I.d.M.e.P.S, 2010; Debevec et al., 2014; Sundblad and Orlov, 2014). Specifically, the study was designed to address the concern that the hypoxic environments anticipated in future habitats on the Moon and Mars (Bodkin et al., 2006) may enhance the effect of reduced loading of the weight-bearing limbs due to the reduced gravity. Reports from the PlanHab study indicated that hypoxia augments the reduction in peak oxygen uptake induced by bed rest (Keramidas et al., 2016), but seems to have no effect on the whole body mass and fat-free mass reductions (Debevec et al., 2014). The focus of the present study was on the combined effect of bed rest and hypoxia on neuromuscular stabilizing functions of the trunk or postural balance control. Our hypothesis was that the changes in postural responses of the trunk muscles and static and dynamic balance will be different following separate and combined exposure to prolonged inactivity/unloading (bed rest) and normobaric hypoxia.

# METHODS

# Participants

Fourteen healthy men (age: 26 ± 5 years, body height: 1.80 ± 0.05 cm, body weight: 76.9 ± 10.8) participated in the study. Each participant underwent careful medical examination prior to enrolment. The exclusion criteria were: history of deep vein thrombosis/pulmonary embolism, active malignancy, uncontrolled hypertension, history of cardiovascular disease, significant hepatic or renal disease, diabetes, chronic inflammatory disease, any significant impairment of the locomotor system, and vestibular or uncorrected visual disturbance. Each participant signed a written informed consent after being thoroughly informed of the study protocol and potential risks. The study protocol was confirmed by the Slovenian National Committee for Medical Ethics.

# Study Design

We used a randomized crossover repeated measures design in this study. The subjects participated in three interventions, with the order of these interventions randomized: (i) bed rest in hypoxic environment (HBR), (ii) bed rest in normoxic environment (NBR), and (iii) ambulatory in hypoxic environment (HAMB). The trials were conducted at the Olympic Sport Centre Planica (Ratece, Slovenia), with the subjects being ˇ confined to one floor of this facility for the duration of each trial. In the HBR and HAMB conditions, hypoxia was established and maintained by a Vacuum Pressure Swing Adsorption System (b-Cat, Tiel, The Netherlands). The system maintained a level of hypoxia equivalent to that at an altitude of ∼4,000 m, accounting for the fact that the facility is situated at 940 m. The partial pressure of inspired oxygen was 90.0 ± 0.4 mmHg in the HAMB and HBR trials, and 133.1 ± 0.3 mmHg in the NBR trials. Since this was a cross-over designed study, the interval between the interventions was a minimum of 3 months to ensure full recovery of the subjects prior to participating in the next intervention. For each trial, subjects arrived at the facility 5 days prior to the onset of the 21-d bed rest, and remained at the facility for 4 days after the conclusion of the bed rest. Baseline measurements before each of the intervention periods were performed during this 4-day period. In the HAMB trial, subjects were confined to the hypoxic facility as in the HBR and NBR trials, with the exception that they were requested to conduct daily 1-h sessions of supervised activity. The activity was designed to mimic the normal daily activity levels of the subjects. During the activity, the subjects' heart rate was monitored with the finger pulse oximeters, and the target heart rate during the activities was 50% of their hypoxic maximum heart rate. The type of exercise was alternated between cycle ergometry, and aerobics/dancing. In addition, during the HAMB trial, subjects were requested to maintain either a seated or standing position throughout the day. Laying supine on the bed was not permitted. At all times, they had to have their feet on the ground. To simulate upright standing activities, subjects were provided with other activities such as table football and darts. Thus, the main difference between the HBR and HAMB trials was the level of activity and loading on the weight-bearing limbs. This was zero in the HBR condition, and normal in the HAMB condition.

By and large, the bed rest protocol was conducted according to the guidelines of the European Space Agency (Heer et al., 2009; MEDS-IMPS and I.d.M.e.P.S, 2010; Sundblad and Orlov, 2014). In the bed rest conditions, participants were confined to bed for 21 days. All activities, including eating, showering, hygiene, etc. were conducted in the horizontal position. The subjects were monitored 24/7 during the bed rest trials, and CCTV cameras ensured compliance with the requirement of the bed rest protocol.

# Measurements

The measurement session included a short-standardized warmup, static balance assessment and assessment of the stabilizing (re)actions of the trunk. The warm-up consisted of spot running with high knees for 2.5 min, 10 squats and 10 push-ups with the subject leaning to the wall at a ∼30◦ angle and hands supported on the wall.

For static balance assessment, the participant was instructed to stand on a force platform (9260AA, Kistler, Winterthur, Switzerland) as still as possible. For each condition, three 30 s trails were performed. The conditions were: parallel stance with eyes open, parallel stance with eyes closed, and semi-tandem stance with eyes open (semi-tandem). The participants were barefoot for all trials and were requested to place their hands on their hips. Three components of the ground reaction force (vertical, anterior-posterior and medial-lateral) were measured using the force plate. The signals were sampled at 1,000 Hz and stored on a personal computer. Average center-of-pressure velocities, amplitudes and frequencies in anterior-posterior and

medial-lateral directions were calculated and stored for further analysis.

Balance assessment was followed by evaluation of trunk stabilizing functions that included ant anticipatory postural adjustments and postural reflex responses measurements. EMG activity was recorded via self-adhesive pairs of electrodes (Blue Sensor N, Ambu A/S, Ballerup, Denmark) placed on designated muscles with 2 cm center-to-center distance, following the SENIAM recommendations (Hermens et al., 2000). The EMG activity of m. multifidus at L5 level m. erector spinae at L1 level and m. deltoideus anterior on the right side of the body was recorded. A reference electrode was placed on the right greater trochanter.

Measurements of anticipatory postural adjustments were performed on random visual cue (LED light) presented at random intervals. The participant stood at hips width, with his arms extended down by his sides, holding a 1.2 kg accelerometer bar with palms facing down. Upon the visual signal, the task was to raise the bar as fast as possible with extended arms from neutral position (i.e., the bar resting on hips) up to the shoulder height and return the bar back down slowly. For postural reflex responses measurements, participants stood relaxed, with their elbows flexed to 90◦ and palms slightly touching the handle of the weight, which was set at 8% of the individual's body mass. After the load release, the participants' task was to return the bar to the initial position, as quickly as possible. For both measurements, all trials were triggered in random manner every 5–12 s. Reliability of the applied test procedures for CoP sway during quiet stance tasks and trunk muscles' stability (re)actions has been tested in previous studies (Markovic et al., 2014; Voglar and Sarabon, 2014a,b; Sarabon and Rosker, 2015; Voglar et al., 2016) and we adopted the suggested reliability optimization guidelines in this study.

The EMG signals were amplified with a factor of 3,000 (Biovision, Wehrheim, Germany), A/D converted, and sampled at 10,000 Hz (USB-6343, National Instruments, Texas, USA). The main outcome parameters were average latencies and amplitudes of the responses, along with the rate of EMG rise in the first 50 ms of the response.

In addition to baseline and post-intervention measurements, daily measurements of capillary oxyhemoglobin saturation were conducted at 7:30 a.m. with a pulse oximeter.

# Statistical Analysis

SPSS 20.0 software (SPSS Inc., Chicago, USA) was used for all statistical analyses. Descriptive statistics was calculated and reported as mean with entitled 95% confidence intervals. The Shapiro–Wilk test was used to test for normality of the distribution. Two-way repeated measures ANOVA [time (2) × condition (3)] was used to test the difference between the sessions. Greenhouse-Geiser correction was applied when Mauchly's sphericity test was statistically significant. Two-tailed t-tests with Bonferroni corrections were used for pairwise comparisons. For all the analyses, the level of statistical significance was set at p < 0.05.

# RESULTS

During the NBR trial, capillary oxyhemoglobin saturation was 98 ± 1% prior to, and throughout the entire period of the bed rest intervention. In contrast, during HBR, capillary oxyhemoglobin saturation decreased from 97 ± 1% prior to the intervention to 83 ± 3% on the first day of the bed rest. Thereafter capillary oxyhemoglobin saturation gradually increased, attaining 98 ± 2% on the last (21st) day of the HBR. A similar response was observed during the HAMB trials, where capillary oxyhemoglobin saturation decreased from 97 ± 1% 1 day prior to the intervention, to 86 ± 1% on the first day of HAMB. As in the HBR condition, capillary oxyhemoglobin saturation gradually recovered, attaining a capillary oxyhemoglobin saturation value of 89 ± 3% on the last day of the 21-day intervention.

The latencies of anticipatory postural adjustments ranged from −19.61 to 24.65 ms for m. multifidus and from −25.30 to 27.75 ms for ES, while postural reflex responses latencies ranged from 106 to 1.384 ms for m. multifidus and from 109 to 1.356 for m. erector spinae. No effect of condition, time or interaction was present for latencies (all p > 0.05). Latencies are presented in **Figures 1**, **2**, respectively, along with the maximal amplitudes of the responses and the rate of EMG rise.

A decrease in anticipatory postural adjustments' maximal amplitudes was observed after HBR and NBR for both muscles, while m. multifidus exhibited increased amplitude after HAMB. Statistically significant interaction was present for both muscles [F(2) = 6.671 and 9.347; p = 0.008 and 0.002; ES = 0.455 and 0.539 for m. multifidus and m. erector spinae, respectively]. However, there were no differences in post-scores between NBR and HBR for neither m. multifidus [t(10) = 0.315; p = 0.760; ES = 0.010] nor m. erector spinae [t(10) = −0.790; p = 0.448; ES = 0.059]. A decrease in rate of EMG rise for m. multifidus was observed after the NBR and HBR conditions [interaction: F(2) = 6.369; p = 0.009; ES = 0.443] with no difference between the two [t(10) = 0.142; p = 0.890; ES = 0.002]. There were no changes in rate of EMG rise for m. erector spinae between prepost-scores, but there was a statistically significant interaction [F(2) = 6.596; p = 0.063; ES = 0.292].

The amplitudes of postural reflex responses were increased after HAMB and decreased after NBR and HBR for m. multifidus [interaction: F(2) = 7.879; p = 0.004; ES = 0.496]. There was no difference between NBR and HBR [t(10) = 1.542; p = 0.154; ES = 0.192]. The trend was similar for m. erector spinae [interaction: F(2) = 4.048; p = 0.038; ES = 0.336], however, a significant difference was present only after NBR. The rate of EMG rise changes for m. multifidus were the same as for the maximal amplitude—an increase was observed after HAMB and a decrease after NBR and HBR [interaction: F(2) = 9.640; p = 0.002; ES = 0.546]. Again, no difference was shown between the bed rest conditions [t(10) = −1.692; p = 0.122; ES = 0.223]. No changes in the rate of EMG rise were observed for m. erector spinae [interaction: F(2) = 3.307; p = 0.063; ES = 0.292].

**Figure 3** presents the changes in antero-posterior and mediolateral body sway velocity. An increase was present after NBR and HBR in both directions for all stances (p = 0.001 – 0.123 for interaction), and additionally for medio-lateral velocity after HAMB in parallel stance with eyes closed. However, this increase was significantly smaller than after NBR and HBR [t(9 and 12) = −3.708 and −4.441; p = 0.005 and 0.001, ES = 0.604 and 0.642, respectively]. No differences were present between NBR and HBR for body sway velocity in any stances or directions [t(12) = 1.812 – 0.597; p = 0.097 – 0.230; ES = 0.029 – 0.230].

A significant interaction was found for antero-posterior body sway amplitude in parallel stance with closed eyes [F(2) = 4.983; p = 0.019; ES = 0.356] and parallel stance with eyes open [F(2) = 4.435; p = 0.025; ES = 0.307]. In both stances, there was a statistically significant increase after NBR (p < 0.001) and HBR (p = 0.001 – 0.003), but not in HAMB. In semi-tandem stance, the increase was observed only after HBR [t(13) = −2.525; p = 0.025; ES = 0.329], but the interaction effect was not statistically significant [F(2) = 2.986; p = 0.073; ES = 0.230]. The mediolateral body sway amplitude in semi-tandem stance increased only after HAMB [t(11) = −2.447; p = 0.032; ES = 0.352], and there was no statistically significant interaction [F(2) = 0.139; p = 0.871; ES = 0.014]. In parallel stance with closed eyes, the increases were observed after all conditions, but were more pronounced after NBR and HBR [interaction: F(2) = 4.382; p = 0.028; ES = 0.327]. Higher medio-lateral amplitude values were also seen after NBR and HBR in parallel stance with eyes open, but there was no statistically significant interaction [F(2) = 2.511; p = 0.106; ES = 0.201].

The mean antero-posterio frequency during parallel stance with closed eyes increased after NBR and HBR (interaction: F = 3.600; p = 0.048; ES = 0.286). In parallel stance with eyes open, the increase was observed after HAMB and HBR, but the interaction was not significant [F(2) = 3.381; p = 0.054; ES = 0.253]. No differences were observed after any condition for mean antero-posterior frequency in semi-tandem stance or mean medio-lateral frequency in either of the parallel stances. In semi-tandem stance, the mean medio-lateral frequency increased after all conditions, with no statistically significant interaction [F(2) = 0.756; p = 0.483; ES = 0.070].

# DISCUSSION

This was the first study to analyze combined and isolated effects of bed rest and hypoxia on neuromuscular responses of the trunk musculature and on postural balance. As expected, NBR and HBR caused greater deteriorations in most of the analyzed measures than HAMB, however, statistically significant differences between NBR and HBR were observed only in a few measures. No changes were observed in anticipatory postural adjustments or postural reflex responses latencies in any of the conditions. The average amplitudes of both responses of m. multifidus were higher in the HAMB trial. The amplitudes of anticipatory postural adjustments were decreased for both muscles after NBR and HBR (with no differences between the two). The postural reflex responses amplitudes of m. multifidus were also lowered following both BR conditions, but only after NBR for m. erector spinae. Additionally, the rate of EMG rise for m. multifidus was decreased after NBR and HBR

for both responses, with no changes observed for m. erector spinae. The above results did not show a potential of hypoxic environment to further augment the unfavorable effects bed rest has on trunk stabilizing functions, whereas the HAMB intervention (most probably its exercise content) even improved certain functions. Previous studies have demonstrated an acute decrease in maximal voluntary contraction force following hypoxia (most likely resulting from decreased activation of highly oxygen-dependent slow motor fibers) (Dousset et al., 2001a). An animal study (Dousset et al., 2001b), investigating neuromuscular responses after chronic exposure to hypoxia showed a decreased response of muscle afferents, while the nerve conduction velocity was higher, which could be the mechanism behind unchanged latencies in our study. However, it has further been demonstrated that both acute and chronic hypoxia have negative effects on muscle afferent activity in humans (Dousset et al., 2003). The improvements following HAMB in our study could have also occurred due to a learning effect, as preand post-intervention measurements within the conditions were conducted 25 days apart, while the interval between exposures to different conditions and corresponding measurements was a minimum of 4 months.

The body sway velocity in both directions was increased after NBR and HBR in all of the stances, but mostly remained the same after HAMB. Similarly, antero-posterior amplitude increased after NBR and HBR in both parallel stances. However, it was only increased following HBR in semi-tandem stance. The medio-lateral amplitude increased after NBR and HBR in parallel stance with eyes closed, after HAMB and HBR in parallel stance with eyes open and interestingly, only after HAMB in semi-tandem stance. Changes in Mean Frequency were also not uniform across stances and directions. The mean medio-lateral frequency was increased after all conditions in the semi-tandem stance, while the mean A-frequency was higher after NBR and HBR in parallel stance with eyes closed, and after HAMB and HBR in parallel stance with eyes open. The antero-posterior amplitude in semi-tandem stance was the only measure that HAMB condition seemed to further increase (more than the bed rest alone). Based on the data from previous studies (Nordahl et al., 1998), demonstrating hypoxia to have a greater effect on postural stability in the sagittal plane, we would expect such results across more than one measure. However, the majority of the postural stability measures deteriorated to the same degree after NBR and HBR. Moreover, changes in frontal plane following HAMB were just as frequent as those in sagittal plane. Our results confirm the findings of previous studies suggesting that hypoxia

alone can impair certain measures of postural balance (Fraser et al., 1987; Nordahl et al., 1998; Wagner et al., 2011; Degache et al., 2012), but do not indicate that hypoxia deteriorates anteroposterior measures in particular, as previously reported (Nordahl et al., 1998). On the other hand, this study did not provide any clear evidence for hypoxia to augment the negative changes of prolonged inactivity. It is important to note that previous studies (Fraser et al., 1987; Nordahl et al., 1998; Wagner et al., 2011; Degache et al., 2012) mostly explored acute effects of hypoxia. In our study, the exposure time was of sufficient length for at least some short-term adaptations (Powell et al., 1998) to occur.

While the results of the present study are non-uniform and equivocal, they do support the notion that minimizing the period of inactivity in populations simultaneously exposed to hypoxia and physical inactivity should be the primary goal when trying to preserve neuromuscular function. With regards to astronauts on missions to the Moon and Mars, who may be exposed to the hypoxic conditions within the habitats, the current practice of daily exercise will minimize deterioration of neuromuscular function. Since the focus of the overall project was the effect of hypoxic inactivity/unloading on astronauts in future missions to the Moon and Mars, the current study was conducted on healthy young male subjects, but the results could be of relevance to specific patient populations rendered either inactive and/or hypoxic. Thus, preservation of neuromuscular function could be jeopardized in patients suffering chronic obstructive pulmonary disease or cardiac insufficiency.

# ETHICS STATEMENT

This study was carried out in accordance with the recommendations of Republic of Slovenia National Medical Ethics Committee with written informed consent from all subjects. All subjects gave written informed consent in accordance with the Declaration of Helsinki. The protocol was approved by the Republic of Slovenia National Medical Ethics Committee (no. 205/02/11).

# AUTHOR CONTRIBUTIONS

NŠ, IM, OE, and JB designed the study and performed the experiments. NŠ analyzed the data. NŠ, IM, OE, and JB wrote the manuscript.

# FUNDING

The study was funded by the European Union FP7 (PlanHab; grant no. 284438), the European Space Agency (ESA) Programme for European Cooperating States (ESTEC/contract no. 40001043721/11/NL/KML: Planetary Habitat Simulation), and the Slovenian Research Agency (contract no. L3-3654: Zero and reduced gravity simulation: the effect on the cardiovascular and musculoskeletal systems). The study was also partially supported by the Slovenian Research Agency through the programs "Automation, robotics and biocybernetics" (P2- 0076) and "Kinesiology of monostructural, polystructural and conventional sports" (P5-0147 (B)).

# REFERENCES


on postural stability and gait. J. Musculoskelet. Neuronal Interact. 14, 359–366.


**Conflict of Interest Statement:** NŠ was employed by company S2P, Science to Practice, Ltd.

The other authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Šarabon, Mekjavi´c, Eiken and Babiˇc. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Myocardial CKIP-1 Overexpression Protects from Simulated Microgravity-Induced Cardiac Remodeling

Shukuan Ling1†, Yuheng Li 1†, Guohui Zhong1†, Yongjun Zheng<sup>2</sup> , Qing Xu<sup>3</sup> , Dingsheng Zhao<sup>1</sup> , Weijia Sun<sup>1</sup> , Xiaoyan Jin<sup>1</sup> , Hongxing Li <sup>4</sup> , Jianwei Li <sup>1</sup> , Huiyuan Sun<sup>5</sup> , Dengchao Cao<sup>6</sup> , Jinping Song<sup>1</sup> , Caizhi Liu<sup>1</sup> , Xinxin Yuan<sup>6</sup> , Xiaorui Wu<sup>1</sup> , Yinlong Zhao<sup>4</sup> , Zizhong Liu<sup>1</sup> , Qi Li <sup>1</sup> and Yingxian Li <sup>1</sup> \*

#### Edited by:

Andreas Roessler, Medical University of Graz, Austria

#### Reviewed by:

Mitchel Tate, Baker Heart and Diabetes Institute, Australia Mónica Isa Moreira-Rodrigues, University of Porto, Portugal

#### \*Correspondence:

Yingxian Li yingxianli@aliyun.com † These authors have contributed equally to this work.

#### Specialty section:

This article was submitted to Integrative Physiology, a section of the journal Frontiers in Physiology

Received: 14 September 2017 Accepted: 11 January 2018 Published: 25 January 2018

#### Citation:

Ling S, Li Y, Zhong G, Zheng Y, Xu Q, Zhao D, Sun W, Jin X, Li H, Li J, Sun H, Cao D, Song J, Liu C, Yuan X, Wu X, Zhao Y, Liu Z, Li Q and Li Y (2018) Myocardial CKIP-1 Overexpression Protects from Simulated Microgravity-Induced Cardiac Remodeling. Front. Physiol. 9:40. doi: 10.3389/fphys.2018.00040 <sup>1</sup> State Key Lab of Space Medicine Fundamentals and Application, China Astronaut Research and Training Center, Beijing, China, <sup>2</sup> Medical Administration Division, The 261th Hospital of PLA, Beijing, China, <sup>3</sup> Core Facility Center, Capital Medical University, Beijing, China, <sup>4</sup> Key Laboratory of Molecular and Cellular Biology of Ministry of Education, College of Life Science, Hebei Normal University, Shijiazhuang, China, <sup>5</sup> Xiyuan Hospital, China Academy of Chinese Medical Sciences, Beijing, China, <sup>6</sup> State Key Laboratory of Agrobiotechnology, College of Life Sciences, China Agricultural University, Beijing, China

Human cardiovascular system has adapted to Earth's gravity of 1G. The microgravity during space flight can induce cardiac remodeling and decline of cardiac function. At present, the mechanism of cardiac remodeling induced by microgravity remains to be disclosed. Casein kinase-2 interacting protein-1 (CKIP-1) is an important inhibitor of pressure-overload induced cardiac remodeling by decreasing the phosphorylation level of HDAC4. However, the role of CKIP-1 in the cardiac remodeling induced by microgravity is unknown. The purpose of this study was to determine whether CKIP-1 was also involved in the regulation of cardiac remodeling induced by microgravity. We first detected the expression of CKIP-1 in the heart from mice and monkey after simulated microgravity using Q-PCR and western blotting. Then, myocardial specific CKIP-1 transgenic (TG) and wild type mice were hindlimb-suspended (HU) to simulate microgravity effect. We estimated the cardiac remodeling in morphology and function by histological analysis and echocardiography. Finally, we detected the phosphorylation of AMPK, ERK1/2, and HDAC4 in the heart from wild type and CKIP-1 transgenic mice after HU. The results revealed the reduced expression of CKIP-1 in the heart both from mice and monkey after simulated microgravity. Myocardial CKIP-1 overexpression protected from simulated microgravity-induced decline of cardiac function and loss of left ventricular mass. Histological analysis demonstrated CKIP-1 TG inhibited the decreases in the size of individual cardiomyocytes of mice after hindlimb unloading. CKIP-1 TG can inhibit the activation of HDAC4 and ERK1/2 and the inactivation of AMPK in heart of mice induced by simulated microgravity. These results demonstrated CKIP-1 was a suppressor of cardiac remodeling induced by simulated microgravity.

Keywords: CKIP-1, simulated microgravity, cardiac remodeling, HDAC4, AMPK, ERK1/2

# INTRODUCTION

Human cardiovascular system has adapted to Earth's gravity of 1G, and cardiac muscle is well regulated in response to changes in loading conditions (Tuday and Berkowitz, 2007; Hill and Olson, 2008). When exposed to microgravity during space flight, there are various changes in cardiac mass and cardiac systolic volume (Hill and Olson, 2008). Microgravity can induce the chronic reduction in metabolic demand and oxygen uptake which reduces the demand on cardiac output, resulting in cardiac atrophy and the decline of cardiac function (Dorfman et al., 2007; Zhong G. et al., 2016; Zhong G. H. et al., 2016). There are many important factors which can regulate cardiac remodeling induced by external or intrinsic stimuli, including AMPK, ERK1/2, and HDAC4 (Ling et al., 2012; Ruppert et al., 2013; Myers et al., 2017). Our previous study demonstrated that pathological cardiac remodeling signals, such as HDAC4 and ERK1/2, were activated, and physiological cardiac remodeling signals, such as AMPK, were inactivated in heart of mice after hindlimb unloading, which might lead to cardiac remodeling and decline of heart function (Zhong G. et al., 2016). Beyond that, we know little about the mechanisms regulating cardiac atrophy induced by microgravity. Also, the prevention of cardiac remodeling induced by simulated microgravity via inhibiting the changes of these signals activity need to be elucidated.

The PH domain-containing casein kinase 2 interacting protein-1 (CKIP-1) was first identified as an interactive protein of casein kinase 2a (Bosc et al., 2000), and functions as a scaffold protein integrating multiple signaling pathways and is highly expressed in the heart under physiological functions (Ling et al., 2012). CKIP-1 was implicated in tumor cell proliferation (Zhu et al., 2017), muscle cell differentiation (Safi et al., 2004), cell apoptosis (Zhang et al., 2005), the regulation of cell morphology (Canton et al., 2006), and the actin cytoskeleton (Canton et al., 2005). We previously showed that CKIP-1 functioned as a novel regulator of pathological cardiac remodeling induced by pressure overload (Ling et al., 2012). We found that CKIP-1-deficent mice displayed age-dependent cardiac hypertrophy and decline of cardiac function. Pressure overload-induced cardiac hypertrophy was exaggerated in CKIP-1-deficient mice. Moreover, myocardial-specific CKIP-1 overexpression protects from cardiac hypertrophy induced by pressure overload. The anti-hypertrophic effects of CKIP-1 are mediated by dephosphorylation and nuclear retention of HDAC4. CKIP-1 is a potential target for inhibiting cardiac remodeling induced by pressure overload (Ling et al., 2012). As with other form of cardiac- remodeling cardiac hypertrophy induced by pressure overload, cardiac atrophy induced by microgravity although linked with different phenotypes (atrophy vs. hypertrophy), generate interesting similar changesupregulation of cardiac remodeling marker genes and decline of cardiac function (Depre et al., 1998). However, the role of CKIP-1 in cardiac remodeling induced by microgravity was unknown.

In this study, we found CKIP-1 mRNA and protein levels in hearts of mice and rhesus monkeys after simulated microgravity were significantly decreased. The results showed myocardial CKIP-1 overexpression protected from the decline of cardiac function, the atrophy of cardiomyocyte and changes in phosphorylation levels of signal factors induced by simulated microgravity. We demonstrated myocardial CKIP-1 overexpression protected from cardiac remodeling induced by simulated microgravity.

# MATERIALS AND METHODS

# Animal Experiments

All animal studies were performed according to approved guidelines for the use and care of live animals (Guideline on Administration of Laboratory Animals released in1988 and 2006 Guideline on Humane Treatment of Laboratory Animals from China, and also referring to European Union guideline 2010/63). The experimental procedures were approved by the Animal Care and Use Committee of China Astronaut Research and Training Center.

The healthy rhesus monkeys with body weight of 5–8 kg and 4–8 years old were purchased from Beijing Xieerxin Biology Resource (Beijing, China). The monkeys were maintained−10 degree head-down tilt position for 42 days. The whole process were supervised and monitored 24 h/day. This 42-day bed rest experiment of rhesus monkeys have been reported previously (Chen et al., 2016).

Myocardial-specific CKIP-1 transgenic mice have been reported previously (Ling et al., 2012). All WT and CKIP-1 transgenic (TG) mice used in this study were bred and housed at the specific-pathogen-free (SPF) Animal Research Center of China Astronaut Research and Training Center (12:12-h light-dark cycle, temperature: 23◦C). All the experiments were repeated three times performed with CKIP-1 mice (2 month) and age-matched WT controls. The hindlimbs of hindlimbunloading procedure were elevated by tail suspension, as described before (Zhong G. et al., 2016). Briefly, the 2 months old mice were maintained in individual cage and suspended with a strip attached the tail and linked a chain hanging from a pulley. The mice were elevated to an angle of 30◦ to the ground, and only the forelimbs of mice can touch the floor, so the hindlimb-unloading mice can move freely to eat and drink. The mice were retained to hindlimb unloading by tail suspension for 28 days, the height of hindlimb suspension was modulated to prevent the hinklimb from touching the ground. This is identified as the "unloaded" state. Similar numbers of control mice of the age-matched littermates and the same strain background were instrumented and monitored in the identical cage conditions without tail suspension.

# Histological Analysis

Sections were generated from paraffin embedded hearts, and were stained with H&E for gross morphology, Masson's trichrome for detection of fibrosis, Frozen sections were used to visualize cardiomyocyte cell membranes by staining with TRITCconjugated wheat-germ agglutinin (Sigma-Aldrich), as described before (Ling et al., 2012).

# RNA Extraction and Real-Time Polymerase Chain Reaction

Total RNA was extracted from heart tissues by using RNAiso Plus reagent (Takara) according to the manufacturer's protocol. The RNA was reverse transcribed into cDNA, and qPCR was performed using a SYBR Green PCR kit (Takara) in a Light Cycler (LightCycler 96, Roche, USA). The mRNA level of each gene was normalized to that of Gapdh, which served as an internal control. Primers (synthesized by Sunbiotech Co, China) for Col1a1, Col3a1, BNP, Ckip1, and Gapdh were as follows:

Mouse:

Col1a1 sense primer: 5′ -CTGACTGGAAGAGCGGAGAGT-3 ′ ,

Col1a1 anti-sense primer: 5′ -AGACGGCTGAGTAGGGAAC AC-3′ ;

Col3a1 sense primer: 5′ -ACGTAAGCACTGGTGGACAG-3′ , Col3a1 anti-sense primer: 5′ -CAGGAGGGCCATAGCTGAA C-3′ ;

BNP sense primer: 5′ -TGTTTCTGCTTTTCCTTTATCTG-3 ′ ,

BNP anti-sense primer: 5′ -TCTTTTTGGGTGTTCTTTTGT GA-3′ ;

CKIP-1 sense primer: 5′ -GCCGTGAGTCCTGAAGAGAAG-3 ′ ;

CKIP-1 anti-sense primer: 5′ -CGAGTAGGGTGGGCAAGAT AG-3′ ;

Gapdh sense primer: 5′ -ACTCCACTCACGGCAAATTCA-3 ′ ;

Gapdh anti-sense primer: 5′ -GGCCTCACCCCATTTGATG-3 ′ .

# Monkey:

Col1a1 sense primer: 5′ -TGACGAGACCAAGAACTGCC-3′ , Col1a1 anti-sense primer: 5′ -CAGGAGATTACCTCGACGC C-3′ ;

Col3a1 sense primer: 5′ -CAAAAGGGGAGCTGGCTACT-3′ , Col3a1 anti-sense primer: 5′ -CAACAGCTTCCTGTTGTGC C-3′ ;

BNP sense primer: 5′ -AATGGTCCTGTACACCCTGC-3′

, BNP anti-sense primer: 5′ -ATCTTCCTCCCAAAGCAGCC-3 ′ ;

CKIP-1 sense primer: 5′ -TCAGGATGGAAACCAGCA-3′ ; CKIP-1 anti-sense primer: 5′ -TTCAGCACCACATAGCGG T-3′ ;

Gapdh sense primer: 5′ -CGAGAGTCAGCCGCATTTTC-3′ ; Gapdh anti-sense primer: 5′ -GACTCCGACCTTCACCTTC C-3′ .

# Echocardiography

Animals were lightly anesthetized with 2,2,2-tribromoethanol (0.2 ml/10 g body weight of a 1.2% solution) and set in a supine position. Two dimensional (2D) guided M-mode echocardiography was performed using a high resolution imaging system (Vevo 2100, Visual-Sonics Inc., Toronto, ON, Canada). Two-dimensional images are recorded in parasternal long- and short-axis projections with guided M-mode recordings at the midventricular level in both views. Left ventricular (LV) cavity size and wall thickness are measured in at least three beats from each projection. Averaged LV wall thickness [anterior wall (AW) and posterior wall (PW) thickness] and internal dimensions at diastole and systole (LVIDd and LVIDs, respectively) are measured. LV fractional shortening ((LVIDd– LVIDs)/LVIDd), relative wall thickness ((IVS thickness + PW thickness)/LVIDd), and LV mass (LV Mass = 1.053 × [(LVID;d + LVPW;d + IVS;d)3 – LVID;d3) are calculated from the M-mode measurements. LV ejection fraction (EF) wascalculated from the LV cross-sectional area (2-D short-axis view) using the equation LV%EF = (LV Vol;d – LV Vol;s)/LV Vol;d × 100%. The studies and analysis were performed blinded as to experimental groups. Structure and function of LV were assessed as described before (Ling et al., 2012).

# Western Blot Analysis

Hearts of mouse and monkey were crushed by homogenizer (Power Gen125, Fisher Scientific) and then lysed in lysis buffer (50 mM Tris, pH 7.5, 250 mM NaCl, 0.1% SDS, 2 mM dithiothreitol, 0.5% NP-40, 1 mM PMSF and protease inhibitor cocktail) on ice for 15 min. Protein fractions were collected by centrifugation for 15 min (12,000 g, 4◦C) and then applied to 10% SDS-PAGE gels, electrophoresed at 80 V for 30 min and 120 V for 90 min. After electrophoresis, protein was transfected to a polyvinylidene fluoride membrane using a Criterion blotter apparatus (Bio Rad). The membrane was then blocked in 5% non-fat dry milk (Becton, Dickinson and Company) in TBST (10 mM Tris–Cl, 150 mM NaCl, 0.05% Tween-20, pH 7.5) for 2 h. After that, the membrane was incubated with primary antibody overnight at 4◦Cfollowed by incubation with a secondary antibody conjugated to horseradish peroxidase (HRP), and visualized using an chemiluminescence kit (Thermo Pierce, No.32 109). Specific antibodies to p-HDAC4 (Cell Signaling Technology, #3443S), HDAC4 (Cell Signaling Technology, #5392S), p-AMPKα (Cell Signaling Technology, #2531S), AMPKα (Cell Signaling Technology, #2532), p-ERK1/2 (Cell Signaling Technology, #4370S), ERK1/2 (Cell Signaling Technology, #4696S), CKIP-1 (proteintech, #24883-1), Gapdh (Santa Cruz Biotechnology, sc-25778) were used to detect protein levels. Gapdh was used as a loading control.

# Statistical Analysis

Data are presented as mean ± SEM per experimental condition. Statistical differences among groups were analyzed by oneway analysis of variance with a post-hoc test to determine group differences in the study parameters. Statistical differences between two groups were determined by the Student's t-test. We use the two-way analysis of variance with unequal variances to account for 2 factors and their interactions. Firstly, we test the equality of variances using a factorial effects analysis of variance on the absolute values of the residuals. If the variances are unequal, we then fit a mixed model with heterogeneous variances. Otherwise, we use the regular linear model. Bonferroni adjustment was used for multiple comparisons. P < 0.05 is considered statistically significant. All the statistical tests are analyzed by Prism software (Graphpad prism for windows, version 5.01) and SPSS (Version 14.0).

# RESULTS

# The Changes of CKIP-1 Expression in the Hearts of Mice and Rhesus Monkeys after Simulated Microgravity

To assess the potential role of CKIP-1 in cardiac remodeling induced by simulated microgravity, hearts from mice after 28 days of hindlimb unloading were assessed for CKIP-1 expression. As shown in **Figures 1A,B**, CKIP-1 mRNA and protein levels were significantly decreased in the hearts of mice after hindlimb unloading. Simulated microgravity can downregulate CKIP-1 expression in mice heart. Moreover, we detected CKIP-1 expression in hearts of rhesus monkeys after 45 days of bed rest. Firstly, real-time polymerase chain reaction analysis showed that the transcripts of the pathological cardiac remodeling genes-Col1a1, Col3a1, BNP, and β-MHC were constantly increased in the hearts of monkey after bed rest, however no obvious changes were showed in α-MHC mRNA level (Supplementary Figure 1). And the changes in phosphorylation levels of HDAC4, AMPK, and ERK1/2 which were proved to be involved in cardiac remodeling of mice after hind limb unloading were detected in hearts of rhesus monkey after 45 days of bed rest. The results showed that following 45 days of bed rest, pathological cardiac remodeling signals-HDAC4 and ERK1/2 were activated, and physiological cardiac remodeling signal-AMPK were inactivated in monkey hearts (Supplementary Figure 2). Meantime, CKIP-1 mRNA and protein levels were also significantly decreased in the hearts of rhesus monkeys after bed rest (**Figures 1C,D**). These results indicated that simulated microgravity reduced the mRNA and protein levels of CKIP-1 which was a suppressor of pathological cardiac remodeling.

# Myocardial CKIP-1 Overexpression Protects from Simulated Microgravity Induced-Decline of Cardiac Function

To further investigate the potential role of CKIP-1 as a suppressor of cardiac remodeling induced by simulated microgravity, we utilized the transgenic (TG) mice with CKIP-1-specific overexpression in cardiomyocytes by using α-MHC promoter. Wildtype (WT) and CKIP-1 TG littermates at 2 months of age were subjected to hindlimb unloading by tail suspension for 28 days. The relevant control groups were treated equally, with the exception of tail suspension. Body weight, heart weight and left ventricular mass were recorded (**Figures 2A–C**), heart rate and cardiac function were calculated by echocardiography (**Figures 2D–F**). All echocardiographic measurements were made while the heart rates of mice were maintained at 400–500 beats per minute (**Figure 2D**). The two way ANOVA was carried out on left ventricular mass, left ventricular fractional shortening (FS), and left ventricular ejection fraction (EF) by condition (control / hindlimb unloading) and genotype (WT/TG). There were all statistically significant interactions between the effects of condition and genotype on left ventricular mass, EF and FS. Compared with the control group, left ventricular mass of WT mice after hindlimb unloading were significantly reduced, meantime left ventricular FS and EF decreased significantly in WT after hindlimb unloading. In comparison with WT mice, CKIP-1 TG mice exhibited a decreased response to hindlimb unload. The changes in left ventricular mass, EF and FS of TG mice after hindlimb were not apparent, and the values were higher than WT-HU group. These results indicated that myocardial CKIP-1 overexpression protected from simulated microgravity induced-decline of cardiac function and loss of left ventricular mass.

# Transthoracic Echocardiography Assessing the Left Ventricular Structure of WT and TG Mice after 28 Days of Hindlimb Unloading

To validate the influence of CKIP-1 TG in the heart structure after hindlimb unloading, we performed transthoracic echocardiography to determine the left ventricular structure of WT and TG mice following hindlimb unloading (**Figure 3A**). A two way ANOVA was carried out on the end-diastolic left ventricular posterior wall thickness (LVPWd) by condition and genotype. There was a statistically significant interaction between the effects of condition and genotype on LVPWd. Compared with control, LVPWd of WT mice was decreased following hindlmb unloading, however LVPWd of TG mice after hindlimb unloading did not change, and the value was higher than that in WT mice after hindlimb unloading (**Figure 3B**). Meantime, the end-systolic left ventricular posterior wall thickness (LVPWs) (**Figure 3C**), the enddiastolic anterior wall thickness (LVAWd) (**Figure 3D**), the end-systolic anterior wall thickness (LVAWs) (**Figure 3E**), the end- diastolic LV internal diameter (LVIDs) (**Figure 3F**), and the end-systolic LV internal diameter (LVIDd) (**Figure 3G**) did not change in the WT and TG mice after hindlimb unloading. These data indicated that CKIP-1 TG protected from the shortening of LVPWd induced by hindlimb unloading.

# Myocardial CKIP-1 Overexpression Protects from Simulated Microgravity Induced-Cardiac Atrophy

To address the effect of myocardial CKIP-1 overexpression on cardiac atrophy induced by hindlimb unloading, hearts from WT and TG mice were assessed for changes in morphology and cardiac remodeling genes expression. As shown in **Figure 4A**, in hematoxylin and eosin-stained (H&E) sections, gross evidence of edema was easily observed by separation of the myofibers in the WT mice after hindlimb, however TG mice after hindlimb had no obvious change. Masson trichrome staining (MTT) showed a deeper staining of collagen in the heart of mice after hindlimb unloading, and CKIP-1 TG could inhibit this change. A two way ANOVA was carried out on myocyte cross sectional area by condition and genotype. There was a statistically significant interaction between the effects of condition and genotype on myocyte cross sectional area. Histological analysis demonstrated decreases in the size of individual cardiomyocytes of WT mice after hindlimb unloading, however, the cardiomyocytes of TG mice after hindlimb unloading had no obvious changes compared with control mice, and higher than WT HU group (**Figure 4B**). The two way ANOVA analysis reports showed there were statistically significant interactions between the effects of condition and genotype on the levels of Col1a1, Col3a1, and BNP mRNAs. The results showed that transcripts for the pathological cardiac remodeling genes-Col1a1, Col3a1, and BNP were significantly increased in the hearts of WT mice after hindlimb unloading, compared with WT control group. And myocardial CKIP-1 overexpression protected from the increases of cardiac remodeling genes expression induced by simulated microgravity (**Figures 4C–E**).

# Myocardial CKIP-1 Overexpression Inhibits the Changes of Signaling Pathway in Mice Heart Induced by Simulated Microgravity

To gain more insights into the effect of CKIP-1 overexpression on the signaling pathways involved in the cardiac remodeling induced by simulated microgravity, we examined the phosphorylation levels of AMPK, ERK1/2, and HDAC4 in heart tissues of WT and TG mice after hindlimb unloading (**Figure 5A**). The two way ANOVA analysis reports showed there were statistically significant interactions between the effects of condition and genotype on the phosphorylation levels of HDAC4 phosphorylation at Ser246 and Ser632, Erk1/2 phosphorylation at Thr202/Tyr204 and AMPK phosphorylation at Thr172. As shown in **Figure 5B**, quantifications of phosphorylation levels normalized to total proteins in the heart revealed HDAC4 phosphorylation at Ser246 and Ser632 and Erk1/2 phosphorylation at Thr202/Tyr204 were increased in WT mice after hindlimb unloading, but it had no obvious changes in TG mice after hindlimb unloading. The phosphorylation level of AMPK at Thr172 was decreased in WT HU group compared with WT control group, but CKIP-1 TG inhibited this change. These results indicated that myocardial CKIP-1 overexpression inhibited the changes in phosphorylation levels of signal factors in mice heart induced by simulated microgravity.

# DISCUSSION

Here, we identified CKIP-1 as a novel regulator of simulated microgravity–induced cardiac remodeling. CKIP-1 mRNA and protein levels were significantly downregulated in the hearts of mice after 28 days of hindlimb unloading and rhesus monkeys after 42 days of head-down bed rest. Myocardial CKIP-1 overexpression protected from simulated microgravity induced-decline of cardiac function and loss of left ventricular mass. Histological analysis demonstrated CKIP-1 TG inhibited the decrease in the size of individual cardiomyocytes of mice after hindlimb unloading. Moreover, the pathological cardiac remodeling signals, such as ERK1/2 and HDAC4, were activated, and physiological cardiac remodeling signals, such as AMPK, were inactivated in heart of WT mice after hindlimb unloading, however CKIP-1 TG mice displayed a different trend. Myocardial CKIP-1 overexpression inhibited the changes of phosphorylation levels of signal factors in mice heart induced by simulated microgravity. So, CKIP-1 manifests important functional significance during cardiac stress response resulting from space flight.

The cardiac muscle is well regulated in response to changes in loading conditions (Hill and Olson, 2008). With prolonged pressure overload, the heart undergoes pathologic hypertrophic remodeling, resulting in dilatation of the failing heart (Hill and Olson, 2008; Ling et al., 2017). Cardiac atrophy was a complication for prolonged microgravity during space flight and

mechanical unloading with a ventricular assist device (Levine et al., 1997; Hill and Olson, 2008; Westby et al., 2016). When exposed to 10 days of spaceflight, left ventricular mass decreased by 12 ± 6.9% (Hill and Olson, 2008). After 6 weeks of bed rest, the left ventricular mass decreased by 8.0 ± 2.2%. Thus, cardiac atrophy can be induced by short-term spaceflight or prolonged bed rest (Perhonen et al., 2001). Hindlimb unloading is widely utilized to study the effects of microgravity in mice or rats (Respress et al., 2014; Zhong G. et al., 2016), and head-down tilt bed rest model for non-human primate- rhesus monkeys or human volunteers is also a classical ground-based model of microgravity (Wang et al., 2012; Chen et al., 2016; Ling et al., 2017). The hindlimb of mice are lifted by tail suspension to generate 30-degree head-down tilt for 28 days, and rhesus monkeys were maintained 10-degree head-down tilt position for 42 days. The tilt and unloading of the hindquarters leads to a shift in body fluids toward the head, cardiac remodeling and other physiological changes, which is similar to what is

FIGURE 4 | Myocardial CKIP-1 overexpression protects from simulated microgravity induced-cardiac atrophy. (A) H&E-stained sections of hearts from WT and CKIP-1 TG mice after 28 days of hindlimb unloading. Sections of hearts are stained with Masson trichrome (MTT) to detect fibrosis (blue). Wheat germ agglutinin (WGA) staining is used to demarcate cell boundaries. Scale bars: 50µm. (B) The cardiomyocyte crosssectional area was measured from 8-µm-thick heart sections that had been stained with WGA by using ImageJ software (NIH). Only myocytes that were round were included in the analysis. The studies and analysis were performed blinded as to experimental. Data represent the means ± SEM (n = 6), \*P < 0.05, \*\*P < 0.01. (C–E) The mRNA levels of Col1a1, Col3a1, and BNP were analyzed by Q-PCR from WT and CKIP-1 TG mice after 28 days of hindlimb unloading. The relative abundance of transcripts were quantified and normalized to GAPDH. Data represent the means ± SEM (n = 6), \*P < 0.05.

founded in humans during space flight (Zhong G. et al., 2016). In a rodent hindlimb unloading model, the decrease of left ventricular mass occurred within 21 days (Bigard et al., 1994). We previously showed simulated microgravity could induce remodeling of the left and right ventricle of mice (Zhong G. et al., 2016). Cardiac atrophy although is associated with distinct phenotypes (atrophy vs. hypertrophy), generate strikingly similar changes-upregulation of cardiac remodeling marker genes and decline of cardiac function (Depre et al., 1998). As with other form of cardiac remodeling, little is known about the specific

Western blots for HDAC4 and phosphorylation at Ser246, AMPKα and phosphorylation at (Thr172), and ERK1/2 and phosphorylation at (Thr202/Tyr204) in hearts from WT and CKIP-1 TG mice after 28 days of hindlimb unloading. Gapdh levels served as a loading control. (B) Quantification of phosphorylation levels normalized to total protein levels of heart, Values are means ± SEM (n = 6), \*P < 0.05.

mechanism governing the microgravity-induced cardiac atrophy. It is critical to understand the mechanisms regulating cardiac remodeling during microgravity-induced myocardial atrophy in addition to those at play during hypertrophy. Most of the researches suggested that hindlimb unloading can simulate the effect of microgravity which caused a systemic stress, such as fluid shift, and lead to cardiac remodeling and the decline of cardiac function (Watenpaugh, 2002; Zhong G. et al., 2016; Zhong G. H. et al., 2016). In this study, LVPWd of WT mice was decreased following hindlmb unloading, however LVPWd of TG mice after hindlimb unloading did not change, and the value was higher than that in WT mice after hindlimb unloading, TG mice also inhibit the decline of cardiac function induced by simulated microgravity.

Well-characterized signaling molecules that regulate cardiac remodeling induced by pressure overload include HDAC4 (Ling et al., 2012), AMPK (Kovacic et al., 2003), and ERK1/2 (Liu and Hofmann, 2004). HDAC4 shuttles between the cytoplasm and the nucleus in a phosphorylation-dependent manner (Kurdi and Booz, 2011). The nuclear export of HDAC4 can inhibit the transcriptional activity of myocyte enhancer factor-1 (MEF2) which is a master positive regulator of cardiac hypertrophy (Zhang et al., 2002; Kong et al., 2006; Ago et al., 2008). AMPK can regulate cardiac homeostasis, and is a key factor of physiological cardiac remodeling (Schisler et al., 2013; Daskalopoulos et al., 2016). ERK1/2 are members of mitogen-activated protein kinase (MAPK) family, and their activation can regulate pathological cardiac remodeling and heart failure (Ogata et al., 2014; Gu et al., 2016). We previously showed that the pathological cardiac remodeling signals, such as HDAC4 and ERK1/2, were activated, and physiological cardiac remodeling signals, such as AMPK, were inactivated in heart of WT mice after hindlimb unloading (Zhong G. et al., 2016). In the present study, we demonstrated that following 45 days of bed rest, pathological cardiac remodeling signals-ERK1/2 and HDAC4 were activated, and physiological cardiac remodeling signal-AMPK were inactivated in monkey hearts. And the levels of CKIP-1 expression were significantly decreased in the hearts of mice and rhesus monkeys after simulated microgravity. We previously demonstrated that CKIP-1 was a novel regulator of cardiac hypertrophy induced by pressure overload. CKIP-1 TG mice exhibited a strong reduction of pathological cardiac hypertrophy. CKIP-1 functioned as a suppressor of pressure overload-induced cardiac remodeling by upregulating the dephosphorylation and nucleus retention of HDAC4 (Ling et al., 2012). CKIP-1 can interact with HDAC4 and PP2A, increases the interaction between HDAC4 and PP2A, and enhances the activity of PP2A, thus promoting the dephosphorylation of HDAC4 (Ling et al., 2012). Moreover, PP2A can also interact with ERK1/2 in cardiomyocytes, PP2A decreases H2O2-induced ERK1/2 activation (Liu and Hofmann, 2004). CKIP-1 may also inhibit the phosphorylation of ERK1/2 via PP2A. The phosphorylation of AMPK can be negatively regulated by Akt in heart (Kovacic et al., 2003), furthermore, CKIP-1 plays a critical role in the regulation of macrophage homeostasis by inhibiting Akt activation (Zhang et al., 2014). We speculate that CKIP-1 can enhance the phosphorylation of AMPK via inhibiting Akt activity. In this study, CKIP-1 TG prevented from simulated microgravity-induced cardiac atrophy, as evidenced by gravimetric, echocardiographic, and cell size analysis. CKIP-1 TG can inhibit the activation of HDAC4 and ERK1/2 and the inactivation of AMPK in heart of mice induced by simulated microgravity.

Taken together, our present work has demonstrated that CKIP-1 was a suppressor of cardiac remodeling induced by simulated microgravity. This study provides evidence of the important role of CKIP-1 in cardiac atrophy. These results suggest a novel strategy for positively affecting cardiac

# REFERENCES


remodeling induced by space flight through the upregulation of CKIP-1 in the heart. Nutritional and/or pharmacological interventions may be exploited to prevent cardiac remodeling induced by microgravity via upregulation of CKIP-1 expression in the heart.

# AUTHOR CONTRIBUTIONS

YL and SL: conceived the study; YL and GZ: performed the experiment with support from YZ and HL; QX: performed the transthoracic echocardiography; WS, JL, HS, and DC: analyzed and interpreted the results; DZ, JS, XJ, CL, and XY: provided intellectual contribution; XW, YZ, ZL, and QL: performed the statistical analysis; SL: wrote the manuscript; YL: revised the manuscript and gave final approval of the submitted manuscript; All authors have reviewed and approved the final manuscript.

# FUNDING

This work was supported by the National Natural Science Foundation of China (No. 31670865, 31300698, and 31325012), Beijing Nova Program (No. Z161100004916111), and 1226 project (No. AWS16J018).

# SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fphys. 2018.00040/full#supplementary-material


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Ling, Li, Zhong, Zheng, Xu, Zhao, Sun, Jin, Li, Li, Sun, Cao, Song, Liu, Yuan, Wu, Zhao, Liu, Li and Li. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Relationship between Aortic Compliance and Impact of Cerebral Blood Flow Fluctuation to Dynamic Orthostatic Challenge in Endurance Athletes

#### Tsubasa Tomoto<sup>1</sup> , Tomoko Imai <sup>2</sup> , Shigehiko Ogoh<sup>3</sup> , Seiji Maeda<sup>4</sup> and Jun Sugawara<sup>1</sup> \*

*<sup>1</sup> Human Informatics Research Institute, National Institute of Advanced Industrial Science and Technology, Ibaraki, Japan, <sup>2</sup> Center for General Education, Aichi Institute of Technology, Toyota, Japan, <sup>3</sup> Department of Biomedical Engineering, Toyo University, Kawagoe, Japan, <sup>4</sup> Faculty of Health and Sport Sciences, University of Tsukuba, Tsukuba, Japan*

#### Edited by:

*Andrew Blaber, Simon Fraser University, Canada*

#### Reviewed by:

*Mark Butlin, Macquarie University, Australia Jie Liu, Department of Ultrasound Diagnostics, Tangdu Hospital, Fourth Military Medical University, China*

> \*Correspondence: *Jun Sugawara*

*jun.sugawara@aist.go.jp*

#### Specialty section:

*This article was submitted to Integrative Physiology, a section of the journal Frontiers in Physiology*

Received: *28 September 2017* Accepted: *09 January 2018* Published: *25 January 2018*

#### Citation:

*Tomoto T, Imai T, Ogoh S, Maeda S and Sugawara J (2018) Relationship between Aortic Compliance and Impact of Cerebral Blood Flow Fluctuation to Dynamic Orthostatic Challenge in Endurance Athletes. Front. Physiol. 9:25. doi: 10.3389/fphys.2018.00025* Aorta effectively buffers cardiac pulsatile fluctuation generated from the left ventricular (LV) which could be a mechanical force to high blood flow and low-resistance end-organs such as the brain. A dynamic orthostatic challenge may evoke substantial cardiac pulsatile fluctuation via the transient increases in venous return and stroke volume (SV). Particularly, this response may be greater in endurance-trained athletes (ET) who exhibit LV eccentric remodeling. The aim of this study was to determine the contribution of aortic compliance to the response of cerebral blood flow fluctuation to dynamic orthostatic challenge in ET and age-matched sedentary (SED) young healthy men. ET (*n* = 10) and SED (*n* = 10) underwent lower body negative pressure (LBNP) (−30 mmHg for 4 min) stimulation and release the pressure that initiates a rapid regain of limited venous return and consequent increase in SV. The recovery responses of central and middle cerebral arterial (MCA) hemodynamics from the release of LBNP (∼15 s) were evaluated. SV (via Modeflow method) and pulsatile and systolic MCA (via transcranial Doppler) normalized by mean MCA velocity (MCAv) significantly increased after the cessation of LBNP in both groups. ET exhibited the higher ratio of SV to aortic pulse pressure (SV/AoPP), an index of aortic compliance, at the baseline compared with SED (*P* < 0.01). Following the LBNP release, SV was significantly increased in SED by 14 ± 7% (mean ± SD) and more in ET by 30 ± 15%; nevertheless, normalized pulsatile, systolic, and diastolic MCAv remained constant in both groups. These results might be attributed to the concomitant with the increase in aortic compliance assessed by SV/AoPP. Importantly, the increase in SV/AoPP following the LBNP release was greater in ET than in SED (*P* < 0.01), and significantly correlated with the baseline SV/AoPP (*r* = 0.636, *P* < 0.01). These results suggest that the aortic compliance in the endurance athletes is able to accommodate the additional SV and buffer the potential increase in pulsatility at end-organs such as the brain.

Keywords: cerebral hemodynamics, aortic compliance, endurance training, pulsatile blood flow, lower body negative pressure stimulation

# INTRODUCTION

Exaggerated hemodynamic fluctuation would be a profound mechanical force to high blood flow and lowresistance end-organs (e.g., the brain and kidneys) (O'rourke and Safar, 2005; Mitchell, 2008). Normally, central elastic arteries (e.g., aorta and carotid artery) effectively buffer cardiac pulsatile fluctuation generated from the left ventricular (LV) (Nichols and McDonald, 2011). Conversely, the impaired buffering ability of central elastic arteries leads to chronic exposure to mechanical stress which could be a strong risk factor for cerebrovascular disease (e.g., white-matter damage and lacuna stroke) (Bateman, 2004; London and Pannier, 2010; Tuttolomondo et al., 2010; Tarumi et al., 2014; Wahlin et al., 2014). In addition, smaller cardiac ejection (e.g., SV) is likely to be associated with smaller cerebrovascular hemodynamic fluctuation and vice versa (Sugawara et al., 2017).

It is well-known that chronic endurance training induces the eccentric remodeling (e.g., increased chamber size) and superior compliance characteristics of the LV (Levine et al., 1991; Pluim et al., 2000; Scharhag et al., 2002; Caselli et al., 2011; Tomoto et al., 2015). Previously, these cardiac adaptations with superior aortic compliance in endurance athletes were reported (Dupont et al., 2017). As stroke volume (SV) increases in response to an increase in the end-diastolic volume which synchronized with the venous return (Guyton and Hall, 2011), such individuals exhibit a greater increase in SV after volume loading. Postural alteration evokes drastic gravitational hemodynamic changes such as greater pulsatile fluctuations of blood flow and blood pressure. In this context, postural alteration-related volume loading (e.g., increased venous return) may cause the substantial change in SV in endurance-trained athletes. However, it is unknown the impact of rapid postural change on central and cerebral hemodynamics in endurance-trained individuals.

The aim of this study was to test the hypothesis that, in endurance-trained athletes who have greater arterial compliance of the proximal aorta, postural alteration-induced substantial increase in SV does not cause the augmented pulsatile fluctuation of cerebral blood flow (CBF). To test our hypothesis, we applied lower body negative pressure (LBNP) release technique as previously used to elicit an immediate increase in central blood volume and blood pressure (Zhang et al., 1998; Ogoh et al., 2015).

# METHODS

# Subjects

Twenty young healthy men voluntarily participated in this investigation. Each subject received a verbal and written explanation of the associated objectives, techniques of measurement and risks. In accordance with the Declaration of Helsinki, each subject provided written informed consent for participation and all protocols were approved by the Ethics Committee of the Human Research Institutional Committee of the National Institute of Advanced Industrial Science and Technology (No. 2013-434). Group of endurance-trained (ET, n = 10) and age-matched sedentary (SED, n = 10) subject were stratified according to their recent history of exercise training. ET had been participating in regular aerobic endurance training for 12 ± 4 h/week (mainly endurance exercise training) and engaging long distance running and/or swimming for 10 ± 2 years; while, untrained subjects did not regularly partake in regular aerobic and resistance exercise (1 ± 1 h/week). All subjects were healthy, normotensive (<140/90 mmHg), nonobese (body mass index, BMI < 25 kg/m<sup>2</sup> ), nonsmokers, free of medication, overt chronic heart and lung diseases, head trauma as assessed by medical history. None of the subjects were taking cardiovascular-acting medication. Prior to the experimental sessions, each subject was familiarized with the equipment and the experimental protocol. Subjects were requested to abstain from food intake for 3 h, caffeinated beverages for at least 12 h, and strenuous physical activity and alcohol intake for at least 24 h before the testing. The subjects were allowed to drink water to keep well hydration.

# Experimental Protocol

Experimental measures including body composition, cardiac echocardiography at rest, and the response of cerebral blood flow velocity (CBFV) and arterial blood pressure at rest, during 30 mmHg negative pressure (LBNP −30 mmHg) and after release up to 15 s. All measurements were conducted in an environmentally controlled laboratory with a quiet and airconditioned room (24–25◦C).

After arrival at the laboratory, the subjects underwent body composition assessment (e.g., weight and height), and supine rest at least 20 min. After cardiac data acquirement and instrumentation for LBNP testing, each participant was placed in a supine position inside the LBNP testing chamber. Once inside, they straddled a wood seat with their feet clear of the base of the chamber and sealed at the level the iliac crest. Following 6 min rest at ambient barometric pressure, negative pressure was gradually induced using a commercially available vacuum and quantified with pressure transducer was invoked. The LBNP test was terminated when the participant completed 4 min at LBNP −30 mmHg. During this experimental protocol, negative pressure was gradually increased and carefully monitored the signs of impending presyncope include dizziness, nausea, profuse sweating, or a rapid change in blood pressure defined as either a decrease in systolic blood pressure by 25 mmHg or a decrease in diastolic blood pressure by 15 mmHg within 1 min. When these signs were reported or observed, the experimental was stopped. These experimental procedures have been demonstrated in our previous research (Sugawara et al., 2017).

### Anthropometric Parameter

For the anthropometric parameters, BMI measurement was calculated by the formula weight in weight/height. The body surface area (BSA) was obtained using the formula: √ (height×weight)/3,600 (Mosteller, 1987).

### Cardiovascular Measurements

Echocardiographic examinations were performed in the left lateral decubitus position using a CX50 xMATRIX (Philips Ultrasound., Bothell, WA) equipped with a multifrequency probe (2.5-MHz transducer). The end-diastolic interventricular septum (IVST) and LV posterior wall thicknesses (PWT) and LV end-diastolic dimension (LVEDd) and end-systolic dimensions (LVESd) were measured on M-mode images in the long parasternal view. LV ejection fraction (EF) and LV fractional shortening (FS) were calculated by Teichholz's method (Teichholz et al., 1976). Doppler velocity time integral (VTI) was obtained from LV ejection velocity wave form recorded via an apical three-chamber view. SV was computed from multiplying the VTI by cross sectional area of LV outflow tract (LVOT) via parasternal long axis view, as previously reported (Lewis et al., 1984). LVOT diameter was measured from the distance between the base of aortic valve right after the valve was fully open. Left ventricular mass (LVM) was calculated following equation: LVM (g) = 0.8 × [1.04 × (IVST + LVEDd + PWT)<sup>3</sup> − (LVEDd)<sup>3</sup> ]+ 0.6. (Devereux et al., 1986). LVM index (LVMi) was normalized LVM by BSA. Electrocardiographic (ECG) data were recorded using a three-lead system for calculation of heart rate (ML 132 Bio Amp, ADInstruments Inc., Colorado Springs, CO). Radial arterial pressure waveforms were continuously recorded at the left wrist by a validated applanation tonometry-based automated radial blood pressure waveform measurement device (Jentow, Nihon Colin Co, Komaki, Japan), and those were calibrated with oscillometry-derived brachial mean and diastolic blood pressures. Beat-by-beat SV was derived from radial arterial pressure waveforms via the Modelflow method (BeatScope 1.1a, Finapres Medical System BV, Amsterdam, the Netherlands), and which was calibrated by the reference value of SV by Doppler echocardiography technique so that the baseline Modelflowderived SV was made equal to the reference value in each subject. Cardiac output was calculated by multiplying SV by heart rate. Systemic vascular conductance was computed by dividing cardiac output by mean arterial pressure. Beat-to-beat aortic (Ao) blood pressure waveforms were computed from radial arterial pressure waveforms by a validated generalized transfer function technique (SCOR-Mx, SphygmoCor, AtCor Medical, Sydney, Australia), as we previously reported (Sugawara et al., 2017). To determine the aortic compliance, SV/AoPP was calculated where AoPP is pulse pressure at proximal aorta.

## Cerebral Vascular Hemodynamics

CBFV was continuously measured at least 6 min at middle cerebral artery (MCA) over the temporal window ipsilateral using 2-MHz transcranial Doppler (TCD) probe (EZ Dop; DWL, Sipplingen, Germany.) The sampling depth was set from 42 to 55 mm, and the angle of the Doppler probe and the sampling depth were adjusted to optimize the signal quality for each subject according to standard procedures (Alexandrov et al., 2012). During data collection, subjects were instructed to breathe normally. The partial pressure of endtidal carbon dioxide (PETCO2) was monitored by a metabolic cart equipped with a respiratory gas analyzing system (AE280S; Minato Medical Science, Tokyo, Japan). Doppler signal was stored at 200 Hz with an acquisition system (PowerLab 8/30, ADInstruments, Colorado Springs, CO, USA) interfaced with a personal computer equipped with data acquisition software (LabChart 7.1, ADInstruments, Colorado Springs, CO, USA). Beat-to-beat time average velocity (e.g., mean velocity), systolic MCA velocity (MCAv), diastolic MCAv, and pulsatile (=systolic − diastolic) MCAv were obtained from 6 min for baseline, 4 min for LBNP testing, and 15 s for releasing LBNP stimulation by offline analysis, and the averaged values of each parameter were reported. Because normalized values of MCA systolic, diastolic, and pulsatile for mean (i.e., time-averaged) velocity (%) (%systolic MCAv, %diastolic MCAv, and %pulsatile MCAv, respectively) have reported with strong correlation with carotid compliance (Tomoto et al., 2015) and brain structures (Tarumi et al., 2014) in previous research, we also obtained these MCAv% parameters. The pulsatile indices of MCA were calculated from the following equations:

Systolic MCAv% = (Absolute systolic MCA velocity /time averaged MCA velocity) ×100 Diastolic MCAv% = (Absolute diastolic MCA velocity /time averaged MCA velocity) ×100 Pulsatile MCAv% = (Absolute pulsatile MCA velocity /time averaged MCA velocity) ×100

Cerebrovascular resistance index was calculated as a ratio of mean radial arterial pressure to mean MCA velocity (MCAv).

# Statistics

Students' independent t-test by groups was performed to determine the impact of continuous endurance training habits on variables of interest and the magnitude difference of recovery responses in central and cerebral hemodynamics from the release of LBNP. Simple correlation analysis was performed to identify the effect of aortic compliance at baseline to the response to the LBNP release. Statistical comparisons of variables were made utilizing a repeated-measures two-way analysis of variance (ANOVA) with a 3 × 2 design (LBNP × group). A Student-Newman-Keuls test was employed post-hoc when interactions were significant. Statistical significance was set at P < 0.05 and results are present as mean ± SD.

# RESULTS

# General Observations

Subjects' anthropometric parameters, systemic hemodynamics, and LV parameters are shown in **Table 1**. Anthropometric parameters did not significantly differ between the SED and ET. In systemic hemodynamics, arterial pressures did not differ between groups. Whereas heart rate was significantly lower and SV was significantly larger in ET, thereby, cardiac output did not differ between groups. In LV characteristics, LVEDd, LVESd, LVM, LVMi, PWT, and IVST were significantly greater in ET than SED, as expected. Whereas, LV fractional shortening and ejection fraction showed no differences between each group. In cerebrovascular parameters, mean and diastolic MCAv did not differ between the groups; whereas, the systolic and pulsatile MCAv in ET were significantly lower than SED (**Table 2**) with higher SV/AoPP (2.2 ± 0.4 vs 2.7 ± 0.3 ml/mmHg, P < 0.01).

# LBNP Testing

No subject showed the signs of impending presyncope during the LBNP testing. Before SV convert from value recorded via



*Values are means* ± *SD. BP, blood pressure;* \**P* < *0.05,* \*\**P* < *0.01 vs. Sedentary.*

modelflow to value recorded via echocardiogram, we performed two-way ANOVA to verify the presence of significant response difference in SV to LBNP between groups. There was a significant difference between groups (F = 6.5, P < 0.01) and SV after the LBNP release in ET was significantly larger compared with SED (P < 0.01). Reported SV values in all figure and **Table 3** were SV after the adjusted value to echocardiogram (calculated SV).

**Table 3** presents the responses of hemodynamics from proximal aorta to cerebral artery to LBNP stimulation and release. HR and cerebrovascular resistance gradually increase associated with increased LBNP pressure in both the groups. Whereas, calculated SV, cardiac output, SV/AoPP, mean and pulsatile MCAv, and PETCO2, gradually decrease associated with increased LBNP in both group. At LBNP-30 mmHg, calculated SV does not differ between the groups, whereas cardiac output was maintained. Diastolic MCAv% was significantly elevated at LBNP-30 mmHg in endurance-trained athletes.

**Figure 1** shows a typical response of radial arterial blood pressure, MCAv via TCD, and LBNP testing at baseline, at LBNP-30 mmHg and at release up to 15 s. **Figure 2** shows the change in calculated SV and SV/AoPP right after the LBNP release. Following the LBNP release, calculated SV was significantly increased in SED by 14 ± 7% and greater in ET by 30 ± 15%, and the change in calculated SV/AoPP TABLE 2 | Subjects' cerebrovascular hemodynamics.


*Value are means* ± *SD. MCA, middle cerebral artery; PETCO2, partial pressure of end-tidal carbon dioxide.* \**P* < *0.05 vs. Sedentary.*

was also significantly increased. Importantly, the increase in SV/AoPP following the LBNP release was greater in ET than in SED (P < 0.001, **Figure 2**), and such response significantly correlated with the baseline SV/AoPP (r = 0.636, P < 0.01, **Figure 3**). **Figure 4** shows the change in normalized systolic, diastolic, and pulsatile MCAv in each group. Although calculated SV significantly increased in ET, no significant differences were observed in normalized MCAv parameters.

# DISCUSSION

The primary findings from the present study are as follows. First, systolic and pulsatile CBF velocity at rest were significantly lower in ET with greater aortic compliance compared with SED. Secondly, SV and pulsatility index of MCA blood flow velocity significantly increased during the release of LBNP in both groups. The increase in SV was significantly greater in ET compared with SED, however, we did not observe a significant groupdifference in responses of pulsatile CBF. These results might be attributed to the concomitant increase in aortic compliance assessed by SV/AoPP. Importantly, the increase in SV/AoPP following the LBNP release was greater in ET than in SED and significantly correlated with the baseline SV/AoPP. These results suggest that the aortic compliance in the endurance athletes is able to accommodate the additional SV and buffer the potential increase in pulsatility at end-organs such as the brain.

In this study, SV was significantly increased after the LBNP release (via an acute increase in central blood volume) in both groups. This finding is consistent with previous observation (Zhang et al., 1998; Ogoh et al., 2015). Interestingly, the change in SV was greater in the ET compared with the SED group. This phenomenon might be explained by LV characteristics in the endurance-trained athletes such as exercise-induced cardiac remodeling and myocardial compliance. Increase in the left and right ventricular end-diastolic volume commonly observed in endurance-trained populations (Pluim et al., 2000; Serrador et al., 2000). Additionally, this cardiac remodeling also accommodated with the elimination of LV compliance which reflect to expand LV ventricular with less pressure (Levine et al., 1991; Bhella et al., 2014). According to the Frank-Starling law of the heart, SV increases in response to an



*Values are mean* ± *SD. Bold values represent P* < *0.05. BP, blood pressure; SV, stroke volume; PP, pulse pressure; MCAv, cerebral blood flow velocity; PETCO2, partial pressure of end-tidal carbon dioxide; TVC, total vascular conductance.* \**P* < *0.05 vs. baseline, † P* < *0.05 vs. ET.*

increase in the end-diastolic volume which synchronized with the venous return (Guyton and Hall, 2011). Taken together, the LBNP release evoked substantial increased venous return in both groups; however, greater LV capacity with expandable myocardium in ET induced the greater change in SV compared with SED.

The acute increase in blood pulsatile fluctuation at the brain may cause of cerebrovascular damage. Also, physiologically, an acute increase in SV would be a cause of elevation in systolic and diastolic blood pressure and mean arterial pressure, which is mechanical stress at the peripheral organs. Similarly with a change in SV, following the release of LBNP up to 15 s, pulsatility index of MCA velocity significantly increased in both groups. However, there was no significant group-difference in responses of pulsatile CBF despite the greater increase in SV among ET compared with SED. In addition, after the LBNP release, SV/AoPP significantly elevated in ET but not in SED, and the increase in SV/AoPP following the LBNP release was significantly correlated with the baseline SV/AoPP. Taken together, in ET individuals with greater aortic compliance, postural change-related rapid augmentation of cardiac ejection does not cause substantial augmentation of cerebral pulsatile fluctuation.

Cerebral autoregulation may partially take a role of explaining this pulsatile attenuation in cerebral arteries. Cerebral autoregulation reflects the constriction or dilation of cerebrovascular within normal arterial pressure. Dynamic cerebrovascular autoregulation refers to the latency from the start of stimulus to the onset of cerebrovascular conuterregulation. Lind-Holst et al. have reported that endurance-trained individuals appear to weaken dynamic cerebral autoregulation, and this adaptation may increase the risk of symptomatic cerebral hypoperfusion during marked and rapid reductions in blood pressure (Lind-Holst et al., 2011). This hyporeactivity of cerebrovascular may affect to the substantial fluctuation of CBF. Whereas, Ichikawa et al. have reported the no associated with an attenuation in dynamic cerebral autoregulation in endurance training (Ichikawa et al., 2013). According to these inconsistency findings, cerebrovascular autoregulation

may have minimal effect on the blood flow fluctuation at CBF.

There is increasing recognition that not only the adequate cerebral perfusion (evaluating by MCAv) but also the attenuated pulsatile fluctuation at cerebrovascular hemodynamics is an important determinant of cerebrovascular health. The change in pulsatile fluctuation may partially accommodate with compliant aorta. The greater amount of evidence has shown the age-related cerebral hypoperfusion and cerebrovascular disease (Davis et al., 1983; Postiglione et al., 1993). Recently, in addition, pulsatile CBF fluctuation was induced by increased in systolic and decrease in diastolic blood flow velocity (Tarumi et al., 2014). Whereas, Zhu et al (Zhu et al., 2013) demonstrated in the cross-sectional study that the lower in systolic and diastolic MCA velocity with lower aortic stiffness were observed in the Masters' athletes compared with age-matched sedentary peers. Our finding provides the evidence of the findings in these previous studies, and the mechanism of endurance exercise preventing pulsatile CBF stress could be explained partly by the greater Windkessl function among endurance trained athletes.

In the present, there are technical limitations that warrant mention. First, TCD measures blood flow velocity in the MCA rather than CBF. Blood flow velocity reflects blood flow only if the diameter of the blood vessel has measured. Early evidence suggested that the diameter of MCA appears to slightly change with LBNP −40 mmHg for 5 min (Serrador et al., 2000). The MCA diameter change would be no greater than in the study (Serrador et al., 2000), and this diameter change would not

FIGURE 2 | The changes in stroke volume and aortic compliance (SV/AoPP) after the release of LBNP up to 15 s.

significantly impact the findings in this research. Secondary, cerebrovascular resistance index was calculated using the mean arterial pressure calibrated by radial signal, not local blood pressure. Most of the invasive human research has this issue. Thirdly, aerobic capacity has not been measured in this study.

We focused more cardiac adaptation rather than aerobic capacity; thus, athletic history and recent endurance training status were recorded from each participant to clarify the cardiac adaptation

# REFERENCES


occurred due to prolonged exercise. Fourth, a previous research has reported that there is different response in cardiac output on release of LBNP (Ogoh et al., 2015). Accodring to this, it appears likely that there is a different response in cardiac output between two groups; however, we did not detect the change due to small sample size. Lastly, we studied only apparent small number of healthy young men. To gain generalizability of these findings and a better understanding of the pathophysiology, further studies are needed on the larger number of subjects and other populations such as women, middle-aged and elderly individuals, and patients with impaired Windkessel function or cerebrovascular disease.

In conclusion, we found that systolic and pulsatile CBFV at rest were significantly lower in ET with superior proximal aortic compliance compared with SED group. SV and pulsatility index of MCA blood flow velocity significantly increased after the release of LBNP in both groups. The increase in SV was significantly greater in ET than in SED, whereas there was no significant group-difference in responses of pulsatile CBF. These results might be attributed to the concomitant with the increase in aortic compliance assessed by SV/AoPP. Additionally, SV/AoPP significantly elevated after the LBNP release in ET but not in SED, and the increase in SV/AoPP following a LBNP release was significantly correlated with the baseline SV/AoPP. These results suggest that the aortic compliance in the endurance athletes is able to accommodate the additional SV and buffer the potential increase in pulsatility at end-organs such as the brain.

# AUTHOR CONTRIBUTIONS

TT, TI, SO, and JS conception and design of research; TT, TI, and JS performed experiments; TT, TI, and JS analyzed data; TT and JS interpreted results of experiments; TT and JS prepared figures; TT, SO, SM, and JS drafted manuscripts; TT, TI, SO, SM, and JS approved final version of manuscripts.

# FUNDING

This study was supported by special coordination funds of the Japanese Ministry of Education, Culture, Sports, Science, and Technology (25702045, 26670116, JS).

# ACKNOWLEDGMENTS

We also thank all subjects of this study and Ms. Mitsuho Watanabe for her technical assistance.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Tomoto, Imai, Ogoh, Maeda and Sugawara. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Hypoxia Exacerbates Negative Emotional State during Inactivity: The Effect of 21 Days Hypoxic Bed Rest and Confinement

#### Nektarios A. M. Stavrou1,2, Tadej Debevec3,4, Ola Eiken<sup>5</sup> and Igor B. Mekjavic3,6 \*

*<sup>1</sup> School of Physical Education and Sport Science, National and Kapodistrian University of Athens, Athens, Greece, <sup>2</sup> Sport Psychology Department, Hellenic Sports Research Institute, Olympic Athletic Center of Athens "Spyros Louis", Athens, Greece, <sup>3</sup> Department of Automation, Biocybernetics and Robotics, Jozef Stefan Institute, Ljubljana, Slovenia, <sup>4</sup> Faculty of Sport, University of Ljubljana, Ljubljana, Slovenia, <sup>5</sup> Department of Environmental Physiology, Swedish Aerospace Physiology Centre, Royal Institute of Technology, Stockholm, Sweden, <sup>6</sup> Department of Biomedical Physiology and Kinesiology, Simon Fraser University, Burnaby, BC, Canada*

### Edited by:

*Andrew Blaber, Simon Fraser University, Canada*

#### Reviewed by:

*Beth J. Allison, Hudson Institute of Medical Research, Australia Simona Mrakic-Sposta, Istituto di Bioimmagini e Fisiologia Molecolare (CNR), Italy*

> \*Correspondence: *Igor B. Mekjavic igor.mekjavic@ijs.si*

#### Specialty section:

*This article was submitted to Integrative Physiology, a section of the journal Frontiers in Physiology*

Received: *19 September 2017* Accepted: *09 January 2018* Published: *08 February 2018*

#### Citation:

*Stavrou NAM, Debevec T, Eiken O and Mekjavic IB (2018) Hypoxia Exacerbates Negative Emotional State during Inactivity: The Effect of 21 Days Hypoxic Bed Rest and Confinement. Front. Physiol. 9:26. doi: 10.3389/fphys.2018.00026* Hypoxia and confinement have both been shown to influence emotional state. It is envisaged that the inhabitants of future planetary habitats will be exposed to concomitant confinement, reduced gravity and hypoxia. We examined the independent and combined effects of a 21-day inactivity/unloading and normobaric hypoxia under confined conditions on various psychological factors. Eleven healthy men participated in three 21-day experimental campaigns designed in a cross-over manner: (1) Normobaric hypoxic ambulatory confinement, (2) Normobaric hypoxic bed rest and (3) Normobaric normoxic bed rest. The Profile of Mood States, and the Positive and Negative Affect Schedule were employed to assess the participants' psychological responses before (Pre), during (Day 7, Day 14, and Day 21) and after (Post) the confinements. The most negative psychological profile appeared on days 14 and 21 of the hypoxic bed rest campaign. A significant increase in depression, tension, and confusion was noted on days 14 and 21 of the hypoxic bed rest condition. Concomitantly, a decrease, albeit not statistically significant, in positive psychological responses was observed. The psychological profile returned to the initial level at Post following all confinements. These data suggest that the combined effect of hypoxia and bed rest induced the most negative effects on an individual's mood. However, significant intra- and inter-individual differences in psychological responses were noted and should be taken into consideration.

Keywords: psychological responses, positive affect, negative affect, hypoxia, bed rest, simulated microgravity

# INTRODUCTION

Research in aid of human Space exploration has to date focused primarily on the microgravity-induced consequences of relative inactivity and unloading of the weight-bearing limbs, on integrative physiological responses. These studies have been performed during missions on the International Space Station (ISS), or by using ground-based simulations of microgravity-induced inactivity/unloading, such as the sustained bed rest (BR) model. While the BR studies to-date have focused extensively on investigating the effect of unloading/inactivity on physiological systems (Pavy-Le Traon et al., 2007), the analyses of the effects on psychological and emotional responses have been limited (Ishizaki et al., 2002; Liu et al., 2012). Nevertheless, both reduced gravity (Kanas et al., 2009) and unloading (Lipnicki and Gunga, 2009) have been shown to importantly modulate individuals' psychological responses. For logistical reasons, the environment within future habitats on the Moon and Mars will be hypoxic. Thus, in addition to confinement and reduced gravity (Eiken and Mekjavic, 2015), astronauts/cosmonauts will be exposed to continuous systemic hypoxia within the envisaged habitats. Given the fact that, similar to inactivity/unloading, hypoxia has also been shown to influence psychological status (Hornbein et al., 1989; Virués-Ortega et al., 2004), understanding the independent and combined effects of these factors on psychological indices is crucial for the success and safety of future space exploration.

The first and only study to-date (Stavrou et al., 2015) that investigated the combined effects of inactivity and hypoxia on psychological indices incorporated a cross-over repeated measures design, such that participants took part in three shortterm interventions including normoxic BR (NBR), hypoxic BR (HBR), and hypoxic ambulation (HAMB). We (Stavrou et al., 2015) showed that 10-day NBR and HAMB interventions did not importantly influence the participants' psychological status. Conversely, HBR resulted in significant alterations, reflected in an increase of negative psychological responses such as negative affect, tension, anger, and confusion, and decrease of positive factors such as vigor, and positive affect. Based on Stavrou et al. (2015) results, it seems that exercise has a positive impact on psychological responses, increasing the positive and ameliorating the negative emotions. Previous research (Yeung, 1996; Hillman et al., 2008; Erickson et al., 2015) supports the positive impact of exercise on mood, affect, cognition, and brain activity, although these effects are highly dependent on exercise type/intensity as well as the participants' characteristics (e.g., age, physical condition).

To date, several studies have investigated the effects of altitude exposure on the indices of psychological status (Gallagher and Hackett, 2004; Bärtsch and Saltin, 2008). Reduction of O<sup>2</sup> availability within the central nervous system, secondary to high altitude exposure induces a variety of neuropsychological impairments (Winget and DeRoshia, 1986; Hornbein, 2001) including alterations in cognition, mood, behavior and sleep indices. Other symptoms that are manifest during hypoxic exposure include impairments of coordination, vision, cognitive function, alertness, vigor, etc. (Acevedo and Ekkekakis, 2001; Virués-Ortega et al., 2004). Importantly, these symptoms are commonly complemented by increases of negative psychological and emotional responses and a decrease of the positive psychological profile elements (Yeung, 1996; Ishizaki et al., 2002; Lipnicki and Gunga, 2009). It has to be noted, however, that significant inter-individual differences in neuropsychological changes at high altitude have been reported (Hornbein et al., 1989; Hornbein, 2001). Indeed, large inter-individual differences might lead to different cognitive responses to various situations and consequently modulate an individuals' ability to adapt to adverse environments (Bahrke and Shukitt-Hale, 1993; Acevedo and Ekkekakis, 2001). Moreover, a moderating role of personality characteristics has also been suggested to influence the measured outcomes (Virués-Ortega et al., 2004).

Accordingly, the present study sought to determine the independent and combined effects of prolonged inactivity and hypoxia on select psychological indices and thereby extend our previous findings (Stavrou et al., 2015) obtained during a shorter exposure time (10 days). This is especially important given that the up-to-date body of research on psychological responses to hypoxic or bed rest environments is limited, and moreover the prolonged combined effect of both factors has not yet been investigated. The results of the current study do not only have application for space missions and performance settings but also have an important clinical aspect for many patients rendered both hypoxic and inactive as a consequence of cardiovascular of respiratory diseases.

# METHODS

This study was a sub-project of larger international research investigation into the physiological and psychological effects of simulated planetary habitation on healthy humans. All experimental procedures within the project were conducted according to the European Space Agency (ESA) bed rest standardization recommendations (Standardization of bed rest study conditions 1.5, August 2009). These recommendations have been developed to standardize the methodology of bed rest studies. This allows the comparison of results from studies conducted at different facilities. The outlined recommendations ensure that the main aspects of each study, particularly those related to overall study protocol, participants selection, medical care and data acquisition are uniform to the highest degree possible. The study was approved by the National Medical Ethics Committee of the Republic of Slovenia, registered at ClinicalTrials.gov (NCT02293772) and conducted according to the Declaration of Helsinki guidelines. The current study focused on psychological and emotional responses before, during and after the experimental interventions.

# Participants

Besides the general inclusion/exclusion criteria outlined in the protocol, individuals recently (<2 months) exposed to altitudes above 2,000 m were ineligible to participate. Following comprehensive interviews, fitness tests, and medical examinations, 14 participants were invited to participate in the study. They were provided with an information booklet detailing all experimental protocols, and were also given oral presentations regarding the experimental protocols included in the overall study. All participants gave their written informed consent prior to inclusion in the study. Two participants did not complete the last campaign while one participant had to be withdrawn from the study during the course of the last campaign due to a medical issue that was not related to participation in the study. Of the original 14 participants who entered the study, eleven healthy, sea level residents (age: 27 ± 6 years (mean ± SD); stature: 180 ± 3 cm; body mass: 77 ± 12 kg; BMI: 23.7 ± 3.0 kg·m−<sup>2</sup> ; body fat %: 21 <sup>±</sup> 5%; VO2max: 44.3 <sup>±</sup> 6.1 mL·kg−<sup>1</sup> ·min−<sup>1</sup> ) completed all trials and thus only their data is included in the present analysis.

# Study Outline

The project outline has been extensively described previously (Debevec et al., 2014). Briefly, the participants underwent the following three, separate campaigns in a randomized, counterbalanced manner: (1) normobaric normoxic bed rest (NBR; partial pressure of inspired oxygen, PiO<sup>2</sup> = 133.1 ± 0.3), (2) normobaric hypoxic ambulatory confinement (HAMB; PiO<sup>2</sup> = 90.0 ± 0.4; ∼4,000 m simulated altitude), and (3) normobaric hypoxic bed rest (HBR; PiO<sup>2</sup> = 90.0 ± 0.4; ∼4,000 m simulated altitude). All experimental interventions were performed at the Olympic Sport Centre Planica (Ratece, ˇ Slovenia), located at 940 m above sea level. Two participants per day entered each campaign in a sequential and fixed order. For participants, each of the three campaigns lasted 32 days and comprised the following three phases: (1) The initial testing phase (Pre) lasted 7 days. This phase enabled the participants to familiarize themselves with the experimental facility, and the researchers to obtain the baseline measures (Pre). (2) The confinement phase lasted 21 days (Day 1–Day 21), during which each participant was assigned to their designated environmental condition (NBR, HAMB, & HBR). (3) A four-day recovery phase that enabled cautious re-ambulation of the participants and obtainment of the post-intervention measurements (Post). A 4 month recovery period was implemented between each campaign to enable sufficient wash-out of physiological and psychological intervention effects.

# Bed Rest and Hypoxic Standard Procedures

Throughout all three study campaigns, the participants were accommodated in double bed-rooms with two participants accommodated in each room. The lights were turned on and the participants awakened at 7:00 a.m. daily and room lights were turned off at 11:00 p.m. Napping was not allowed during the day hours. The facility environmental conditions were constantly controlled and remained stable during the three campaigns (ambient temperature = 24.4 ± 0.7◦C; relative humidity = 53.5 ± 5.4% and ambient pressure = 684 ± 4 mm Hg). During the experimental campaigns, participants were able to communicate and socialize with other participants and/or research personnel either in their rooms or within the joint common area of the facility. In the latter case, participants were transported to the common area by a gurney, and maintained a horizontal position on the mattresses provided in the common area. Also, highspeed internet connection and a personal tablet computer were provided to each participant. No targeted cognitive tasks or mental stimulations were applied.

The participants were confined to strict horizontal BR during the NBR and HBR campaigns. Horizontal BR has been identified as a valid ground-based model to simulate microgravity-induced physiological adaptations (Pavy-Le Traon et al., 2007). All habitual activities (e.g., showering, lavatory) were performed in the supine position. For head support, the participants were allowed to employ one head-pillow. No physical activity, besides changing recumbent position, was allowed during the confinements. Compliance to the BR protocol was monitored using continuous CCTV and permanent medical supervision. To alleviate potential backache during the initial days of each BR the participants were provided with passive physiotherapy.

During HAMB the participants were encouraged to move freely in the common hypoxic area (110 m<sup>2</sup> surface) and encouraged to engage in their usual daily routines. The HAMB participants also performed two, 30-min exercise sessions of low intensity each day to mimic their normal habitual activity levels. The exercise mode (stepping, dancing or cycling) was rotated daily.

The normobaric hypoxic environment within the facility was generated by the Vacuum Pressure Swing Adsorption system (b-Cat, Tiel, The Netherlands) that supplied the O2-depleted gas to the designated rooms and the common hypoxic area. The ambient air within the rooms was monitored and analyzed for O<sup>2</sup> and CO<sup>2</sup> content at regular intervals (15-min). In the event of a variation in the O<sup>2</sup> levels the system restored the target O<sup>2</sup> levels immediately. During all hypoxic campaigns, the participants wore, or had in close proximity, a portable ambient O<sup>2</sup> concentration analysers (Rae PGM-1100, California, USA) that activated a safety audible alarm, if the O<sup>2</sup> level decreased below the pre-set level.

# EXPERIMENTAL PROCEDURES

# Daily Physiological Monitoring

In order to monitor the participants general adaptation during the campaigns, heart rate (HR), capillary oxygen saturation (SpO2) and arterial pressures (AP) were measured each morning upon awakening using a HR telemetry device (iBody, Wahoo Fitness, Atlanta, USA), finger oximetry device (3100 WristOx, Nonin Medicals, Minnesota, USA) and manual sphygmomanometer (Diplomat-presameter, Riester, Jungingen, Germany), respectively.

# Acute Mountain Sickness Assessment

To assess the presence and severity, of acute mountain sickness (AMS) during the confinement the participants were requested to complete the Lake Louise AMS questionnaire daily (Roach et al., 1993). The Lake Louise score (LLS; 0-15) was consequently calculated and the score of ≥3 along with a presence of headache was used to define AMS occurrence.

# Profile of Mood States

The participants' mood state was assessed using the Profile of Mood States - Short Form (POMS-SF) (Shacham, 1983). The POMS-SF is a 37-item self-evaluation questionnaire comprised the following six subscales: tension-anxiety, depressiondejection, anger-hostility, vigor-activity, fatigue-inertia, and confusion-bewilderment. The participants' provided their subjective mood states on a 5-point scale ranging from 0 "not at all" to 4 "extremely." Several validation studies (Curran et al., 1995) lent support for the internal consistency and test-retest reliability of the POMS-SF subscales. The reliability of the POMS subscales (Cronbach a; Cronbach, 1951) in the current study was acceptable (tension = 0.67–0.88, depression = 0.76–0.95, vigor = 0.81–0.91, anger = 0.79–0.92, fatigue = 0.79–0.94, confusion = 0.79–0.92).

# Positive and Negative Affect Schedule

The Positive and Negative Affect Schedule (PANAS) (Watson et al., 1988) is also a self-report instrument comprising 20 items. It was initially developed to enable the measurement of respondents' positive affect (PA; 10 individual items) and negative affect (NA; 10 individual items). The participants were asked to rate the extent to which they experienced each of the noted affects outlined in PANAS. Participants provided their answers on a 5-point scale with anchors between "very slightly or not at all" (1) to "extremely" (5). The PANAS indicated acceptable reliability in the current study (Cronbach a: negative affect = 0.91–0.95, positive affect = 0.78–0.96).

The participants completed both of these psychological instruments (POMS &PANAS) before, during and after each confinement period. In particular, the data were obtained two days before the onset of intervention (PRE), on the 7th (D7), 14th (D14), and 21st (D21) days of each intervention and 1 day following the intervention cessation (POST). At each time point, the participants were instructed to complete the questionnaires solely based on their feelings exactly at the time the questionnaire was administered.

# Statistical Analysis

A multivariate analysis of variance (Conditions: NBR, HBR, HAMB) × 5 (Time: PRE, D7, D14, D21, POST), with repeated measures on the second factor (RMANOVA) were performed for subscales of both PANAS and POMS. In addition, separate analyses of variance were, based on the RMANOVA outcomes, performed on each of the affect (PANAS) and mood (POMS) factors to determine the between-participant differences in each experimental condition and for the within-participant repeated measures within the conditions. When the assumptions of sphericity were not met in the within-participants repeated measure analyses, the Green-House Geisser correction and the corresponding degrees of freedom, were used to estimate the F statistic (Tabachnick and Fidell, 2006; Field, 2009). A post hoc- Bonferroni corrected t-tests were employed to determine any significant differences between the pairwise ANOVA comparisons (Tabachnick and Fidell, 2006; Field, 2009). In order to determine the relationship between the measured physiological and psychological indices a pooled Pearson's correlation analysis was employed. The significance level was set a priori at 0.05. The Statistical Package for Social Sciences (SPSS 21.0 Win) was employed for all of the data and statistical analyses.

# RESULTS

# General Adaptation

With the exception of transient head- and back-aches the participants did not experience any significant adverse healthrelated issues. Due to severe hypoxemia (SpO<sup>2</sup> < 75%) and dizziness of one participant during the initial exposure to HBR intervention, he was individually exposed to a simulated altitude of ∼3,000 m and ∼3,500 during the next 2 days. He recommenced his exposure to 4,000 m on Day 3 without any additional adverse effects. The same protocol was thereafter replicated during the HAMB campaign. As noted in **Table 1** the resting HR was significantly higher during the HAMB and HBR compared to NBR (p < 0.05), but remained stable throughout each confinement period. As expected, during both hypoxic confinements (HAMB, HBR) SpO<sup>2</sup> was significantly reduced (p < 0.001) compared to PRE values and values observed during NBR. Both, systolic and diastolic AP values were significantly higher during the HAMB and HBR compared to the NBR campaign (p < 0.01). Compared to PRE values, AP was significantly higher on D7 and D21 during the HAMB and D14 and D21 during the HBR. As reported previously (Debevec et al., 2014), the subjective AMS symptoms were noted in three participants during the HAMB and five participants during the HBR campaign on the first 2 days of the confinements. Moreover, significantly higher values of LLS were observed during the D7 and D21 of the HAMB and HBR campaigns as compared to NBR.

# Baseline Psychological Responses

The baseline responses to the PANAS (PA: positive affect, NA: negative affect) and POMS subscales (tension, vigor, anger, fatigue, confusion, and depression) in the three experimental conditions (HAMB, HBR, and NBR) were obtained in the preintervention period, 2 days before the start of each 21-days intervention. No significant baseline differences were noted on either the PANAS (F = 0.49, ns, η 2 <sup>p</sup> <sup>=</sup> 0.03), or the POMS (F = 0.84, ns, η 2 <sup>p</sup> <sup>=</sup> 0.16) subscales between the experimental conditions.

# Positive and Negative Affect Schedule

The data from the PANAS subscales in each condition is presented in **Figure 1**. Significant Condition × Time interaction was noted (F = 1.91, p < 0.05, η 2 <sup>p</sup> <sup>=</sup> 0.11) for the PANAS subscales. The means and standard deviations of the PANAS subscales are presented in **Figure 1**. The results indicate that there was a significant Condition X Time interaction for NA (F = 3.36, p < 0.01, η 2 <sup>p</sup> <sup>=</sup> 0.18). A significant difference for the NA was observed in the HBR condition (F = 4.24, p < 0.05, η 2 <sup>p</sup> <sup>=</sup> 0.28), demonstrating an increase in NA from PRE to D14 (p < 0.05) and D21 (p < 0.05). NA later decreased from D14 to POST (p < 0.05). No significant changes over Time were noted during NBR (F = 1.30, p = 0.294, η 2 <sup>p</sup> <sup>=</sup> 0.12) or the HAMB conditions (F = 2.42, p = 0.104, η 2 <sup>p</sup> <sup>=</sup> 0.21). The results do not indicate a significant Condition × Time interaction for PA (F = 0.57, p = 0.799, η 2 <sup>p</sup> <sup>=</sup> 0.04; **Figure 1**).

# Profile of Mood States

The results of the POMS subscales are detailed in **Table 2**. An observed significant interaction for the POMS subscales (Condition × Time) (F = 2.08, p < 0.001, η 2 <sup>p</sup> <sup>=</sup> 0.12), indicates that mood-state was changing over time differently during the three experimental conditions.

The analysis of depression showed a significant Condition × Time interaction (F = 2.36, p < 0.05, η 2 <sup>p</sup> <sup>=</sup> 0.13). The depression values in HAMB (F = 4.30, p < 0.01, η 2 <sup>p</sup> <sup>=</sup> 0.30)


TABLE 1 | Means *(M)* and Standard Deviations *(SD)* of the Heart Rate (HR), Capillary Oxyhemoglobin Saturation (SpO2), Arterial Pressure (AP), and Lake Louise Score (LLS) values before during and after experimental conditions.

*PRE, 2 days before the onset; D7, 7th day; D14, 14th day; D21, 21st day; POST, first day after; HAMB, hypoxic ambulatory condition; HBR, hypoxic bed rest condition; NBR, normoxic bed rest condition;* \**Denotes significant difference compared to PRE (p* < *0.05);* #*Denotes significant difference compared to corresponding NBR values (p* < *0.05).*

changed significantly over time. Depression on D7 was higher than PRE (p < 0.05) and D21 (p < 0.01) measures. Compared to D7, the POST values also approached statistical significance (p = 0.057). Additionally, differences were revealed during HBR (F = 6.70, p < 0.01, η 2 <sup>p</sup> <sup>=</sup> 0.38), showing an increase of depression in the D21 measures compared to PRE (p < 0.01) and POST (p < 0.05) measures. Also, the difference between D21 and D14 approached statistical significance (p = 0.059). Finally, no significant depression-level changes appeared over time in NBR trial (F = 1.50, p = 0.248, η 2 <sup>p</sup> <sup>=</sup> 0.13). When we examined the differences between the three conditions, the differences were only significant on D21, with HBR demonstrating higher depression levels compared to both, the NBR (p < 0.05) and HAMB (p < 0.01).

A significant Condition × Time interaction for fatigue (F = 2.36, p < 0.05, η 2 <sup>p</sup> <sup>=</sup> 0.13) was detected. The results indicate that fatigue significantly changed in the HBR condition (F = 5.45, p < 0.05, η 2 <sup>p</sup> <sup>=</sup> 0.33). There were no significant fatigue differences across time during hypoxic BR exposure, except for the difference between D21 to POST, which approached statistical significance (p = 0.06). No significant time-related changes were observed in both, the HAMB (F = 1.19, p = 0.318, η 2 <sup>p</sup> <sup>=</sup> 0.11), and the NBR condition (F = 2.01, p = 0.158, η 2 <sup>p</sup> <sup>=</sup> 0.17). Examining the difference between the experimental conditions across the five time measures, significant differences were only noted on the POST measure (F = 3.42, p < 0.05), during which the HBR condition showed higher fatigue than the HAMB condition (p < 0.05).

The results for the tension subscale indicated a significant Condition × Time interaction (F = 2.75, p < 0.001, η 2 <sup>p</sup> <sup>=</sup> 0.15). A significant fluctuation in tension over time was noted in the HBR (F = 10.90, p < 0.001, η 2 <sup>p</sup> <sup>=</sup> 0.50). Specifically, tension was augmented on D14 (p < 0.05) and on D21 (p < 0.01) compared to PRE and then reduced on POST measure compared to D14 (p < 0.05) and D21 (p < 0.01). Differences were also revealed during NBR (F = 7.55, p < 0.01, η 2 <sup>p</sup> <sup>=</sup> 0.43), showing that the D14 measure of tension increased compared to D7 (p < 0.05), while decreased on POST measure (p < 0.05). On D14 of the HBR condition tension was higher when compared to the HAMB (p < 0.05). Anger was higher in the HBR condition in the D21 and POST measures compared to the NBR (p < 0.05). For tension, no differences were noted across time in the HAMB (F = 1.11, p = 0.353, η 2 <sup>p</sup> <sup>=</sup> 0.10).

A significant Condition × Time interaction was revealed for confusion (F = 2.77, p < 0.05, η 2 <sup>p</sup> <sup>=</sup> 0.15). Significant changes were noted within the HBR condition (F = 9.26, p < 0.001, η 2 <sup>p</sup> <sup>=</sup> 0.46). Post hoc analysis indicated that confusion increased on D14 (p < 0.05) and D21 (p < 0.001) compared to PRE measure. On the other hand, confusion decreased from D21 to POST (p < 0.05) during HBR. The between conditions analysis indicates that on D21 (F = 6.87, p < 0.01) significant differences were revealed between the HBR and NBR (p < 0.01) and HAMB conditions (p < 0.01). No significant Condition × Time interaction for NBR (F = 1.36, p = 0.280, η 2 <sup>p</sup> <sup>=</sup> 0.12) or HAMB (F = 2.23, p = 0.147, η 2 <sup>p</sup> <sup>=</sup> 0.18) was found.

Finally, no significant Condition × Time interaction was noted for vigor (F = 0.92, p = 0.482, η 2 <sup>p</sup> <sup>=</sup> 0.06) and anger (F = 0.89, p = 0.502, η 2 <sup>p</sup> <sup>=</sup> 0.05) was detected. Based on the POMS subscales, we subsequently calculated the total mood disturbance score [TMD = tension + depression + anger + confusion + fatigue + (24 – vigor)] (Curran et al., 1995) and compared it among the three experimental conditions. The results indicate no significant Condition X Time interaction (F = 1.11, p = 0.363, η 2 <sup>p</sup> <sup>=</sup> 0.07) for TMD.

(*p* < 0.05).

# DISCUSSION

The main finding of the present study is that prolonged hypoxic inactivity significantly augmented negative mood. This also lends further support to the notion that hypoxia exacerbates negative emotions usually noted during BR of longer duration. Participants did not exhibit any such psychological changes during the HAMB condition, suggesting that even habitual levels of physical activity might counteract the negative effects of hypoxia on mood and emotional status. Finally, it is also noteworthy that significant variation in the intra- and interindividual psychological responses was observed in both the hypoxic and normoxic BR condition.

# General Adaptation

As noted in **Table 1**, significantly lower SpO<sup>2</sup> values were noted during both hypoxic conditions. This reduction in SpO<sup>2</sup> was a direct consequence of the lower ambient O<sup>2</sup> in the HBR and HAMB as compared to the NBR condition, which also resulted in higher daily heart rate and arterial pressure levels. Based on the performed correlation analysis and given the strictly controlled environmental conditions in the present study the observed differences in the psychological status of subjects between the NBR and HBR conditions can most likely be attributed to systemic hypoxemia and other consequential physiological responses. It is noteworthy, that during HAMB participants were also continuously confined to the hypoxic facility, but did not exhibit any negative responses. This may be, in part, due to the fact that they spent time with the HBR participants in their rooms, which ensured more social interaction among the participants then commonly reported for previous BR studies. Undoubtedly, this social interaction attenuated the psychological effects of confinement per se.

# The Effect of Bed Rest

In the NBR condition, a significant increase appeared in tension on D14 compared to the PRE measure, while a decrease was noted at POST, immediately following the BR condition. Although non-significant changes appeared in the remaining psychological factors it is important to note that depression and TMD increased on D14 and D21, and decreased only after the end of the BR exposure. On the other hand, non-significant changes were revealed in participants' positive psychological state (positive affect, vigor), which is in contrast to previous findings (Ishizaki et al., 2002; Liu et al., 2012; Stavrou et al., 2015). There are two key differences between the present study and our previous study (Stavrou et al., 2015): (1) the duration of the present study (21 days) was double the duration of the previous one (10 days), and (2) as noted above, much more emphasis was placed on the social interaction between participants, and between participants and staff. It is important to note that the BR-related increase in negative emotions can also be related to the confinement per se, as well as to the social isolation and subsequent lack of social interaction (Palinkas, 2001; Manzey, 2004; Kanas et al., 2007).

# Correlations between Physiological and Psychological Indices

condition (HAMB, open circles), hypoxic bed rest condition (HBR, open squares) and normoxic bed rest (NBR, full squares) condition. \*Denotes significant difference within the HBR condition compared to PRE (*p* < 0.05); #Denotes significant difference within the HBR condition compared to POST

The only two physiological indices that were not significantly correlated to any of the reported psychological variables were Vigor and PA. On the other hand, the LLS values were only significantly correlated to the Fatigue subscale (r = 0.75, p < 0.001). A significant correlation with the changes in the SpO<sup>2</sup> were noted for Tension (r = −0.55, p < 0.05) Anger (r = −0.80, p < 0.001), Confusion (r = −0.61, p < 0.05), Depression (r = −0.70, p < 0.001), TMD (r = −0.65, p < 0.001) and NA (r = −0.51, p < 0.05). Similarly, changes in HR in each condition were correlated to Tension (r = 0.70, p < 0.05), Confusion (r = 0.72, p < 0.01), Depression (r = 0.50, p < 0.05), TMD (r = 0.72, p < 0.001) and NA (r = 0.64, p < 0.01). Finally, changes in mean AP also indicated significant correlations to Tension (r = 0.65, p < 0.001), Anger (r = 0.57, p < 0.05), Confusion (r = 0.62, p < 0.01), Depression (r = 0.68, p < 0.05), TMD (r = 0.65, p < 0.01), and NA (r = 0.53, p < 0.05).


*PRE, 2 days before the onset; D7, 7th day; D14, 14th day; D21, 21st day; POST, first day after; HAMB, hypoxic ambulatory condition; HBR, hypoxic bed rest condition; NBR, normoxic bed rest condition; TMD, total mood disturbance.*

Based on the results of the current and previous work, it seems that the effect of BR on emotional responses and cognitive function is currently still unclear. Although most of the studies suggest detrimental effects of BR (Ryback et al., 1971; Edwards et al., 1981; Stavrou et al., 2015), others found either an improvement during BR (DeRoshia and Greenleaf, 1993; Ishizaki et al., 2002) or no changes (Shehab et al., 1998). Such equivocal findings might be due to duration of BR exposure or might be underlined by differences in participants' emotional characteristics, such as the manner in which they individually respond and adapt to adverse environments', and the management of negative emotions. It may be more appropriate to examine participants' individual psychological responses, rather than average or median results, as this approach would take into account the role of personality characteristics in the adaption to adverse environments. Future research should not only focus on psychological responses during bed rest and/or hypoxic exposure, but also examine participants' personality characteristics. This approach would be especially valuable during the process of selecting crew members for space missions. Individuals that have an appropriate personality profile, leadership and coping skills may have an advantage in adapting to space habitat environments ensuring a successful completion of a space mission (Palinkas, 2001). In addition, better understanding of the potential detrimental effects of extremely low levels of physical activity has important implications for clinical populations (Park et al., 2001; McDonald, 2002; Gaul et al., 2006).

# The Effect of Hypoxia

Our results indicate that the HBR intervention elicited the most negative emotional state compared to the NBR and HAMB conditions. HBR participants exhibited higher negative affect, depression, fatigue, and tension compared to the participants of the NBR and the HAMB conditions. In contrast, no significant effect was noted in positive emotions such as positive affect and vigor, which is in accordance with our previous findings (Stavrou et al., 2015).

The combination of BR and hypoxia exerts a detrimental effect on participants' mood, as reflected by the consecutive increase of depression and tension, remaining on a high level on D14 and D21. Participants in the HBR compared to those in the HAMB condition exhibited higher tension and depression on D14 and D21, respectively, indicating the protective role of exercise in the deterioration of mood in adverse environments. In addition to the above, it is important to note that the HAMB participants showed a significant increase of depression on D7 compared to PRE measure. However, the fact that depression levels decreased thereafter is suggestive of participants' successful adaptation to the respective conditions. It is also noteworthy that fatigue in the HBR condition reached the highest value in the POST measure. This increase in fatigue can be attributed to either participants' mood state during the HBR exposure, or it can be linked to personality characteristics as there was a large variability in participants' responses, ranging from 0 to 14.

The current results are in line with previous research findings (Banderet and Burse, 1991; Bahrke and Shukitt-Hale, 1993; Davidson, 2001; Stavrou et al., 2015) supporting the notion that terrestrial altitude exposure-related reduced systemic O<sup>2</sup> supply to the nervous system can provoke significant negative effects on the neuropsychological state and adaptation. In addition, regarding the effect of altitude exposure on psychological responses, future investigations should examine the relation between an individual's cognitive function and related psychological responses in order to further elucidate the key factors that enable efficient adaptation to adverse environments.

In summary, based on the current and previous results (Stavrou et al., 2015) of the combined effect of bed rest and hypoxia on psychological indices, it seems that both factors significantly contribute to detrimental effect on mood, as hypoxia provokes negative psychological responses mainly in the inactive, bed ridden participants. Interestingly, such response is not evident in the hypoxic ambulatory participants, indicating that even low activity levels can be beneficial as has also been previously reported during protocols of short duration (Stavrou et al., 2015).

# The Effect of Activity

Comparison of the psychological responses between the HBR and HAMB clearly shows that the participants in the HBR condition revealed the most negative psychological profiles on D14 and D21, when their psychological profile continuously deteriorated. Specifically, the participants in the HBR compared to those of the HAMB condition showed a more pronounced negative mood, consisting of higher tension, depression, confusion, and fatigue. It seems that the regular daily physical activity as well as other daily activities performed by the participants in the HAMB condition significantly blunted the psychological impairment observed in the HBR condition.

Participants' mood state was consistently compromised during the HBR intervention, as demonstrated by the significant changes in most of the negative psychological factors observed during the hypoxic exposure as compared to the baseline. Conversely, the emotional responses remained fairly stable during the HAMB condition with no significant differences during the hypoxic exposure. This is indicated by the nonsignificant differences for positive or negative emotions during intervention (D7, D14, and D21) compared to PRE or POST measures. The only significant alteration for the HAMB participants was limited to an increase of depression on D7 compared to PRE measure. However, this may be a transient augmentation of depression as a consequence of hypoxic exposure, with values returning to the initial pre-intervention level as a result of the beneficial effect of exercise. This is further supported by recent research findings underlying exercise benefits for depression treatment (Barbour and Blumenthal, 2005; Russo-Neustadt and Chen, 2005).

The daily physical activity of the HAMB participants was meant to mimic participants' normal life daily activity, and was not meant to provide a training stimulus. This daily physical activity during HAMB seemed sufficient to ameliorate the deterioration of participants' psychological responses during HBR. Extensive research has supported the positive effect of exercise on psychological responses and cognitive function (Ekkekakis, 2003; Hillman et al., 2008; Erickson et al., 2015). However, the independent effects of various activity levels on psychological responses during hypoxic/altitude exposures have, to-date been largely ignored. Our results seem to be in line with previous studies supporting the positive effect of exercise on different psychological variables as well as on cognitive function (Hillman et al., 2008; De La Torre et al., 2012; Erickson et al., 2015). Yeung (1996) review on the effects of exercise on mood and mental health, clearly shows that (85%) of the studies report improved mood and, moreover, that this benefit seems to be at least partially dependent on the duration and intensity of the activity (Lind et al., 2005; Ekkekakis and Lind, 2006). Gordon et al. (2008), Hötting and Röder (2013), and Weinstein et al. (2012) support the important role of cardiorespiratory fitness and physical exercise in the facilitation of neuroplasticity and greater prefrontal cortex volume, and the improvement of cognitive functioning. Cognitive functioning seems to have a mediating role in participants' experience of negative emotions providing a better adaptation to stressful and adverse environments (Acevedo and Ekkekakis, 2001). It is however, noteworthy, that contradictory results appeared in BR studies regarding the effect of exercise on mood and emotion. In particular, some studies suggest that short-term exercise interventions are already sufficient to restore mood and brain cortical function to pre-isolation values (Van Baarsen et al., 2009), while others did not reveal any effect of exercise, since it did not limit the deterioration of psychological responses (DeRoshia and Greenleaf, 1993; Ishizaki et al., 2002).

# Implications for Clinical Populations

Patients with various clinical conditions (e.g., chronic obstructive pulmonary disease, heart failure), rendering them less-active also have systemic hypoxia-related elevated levels of negative psychological responses and reduced exercise capacity (Ritz et al., 2000; Mikkelsen et al., 2004). This exerts a detrimental effect on their quality of life, and establishes a vicious cycle of declining physical activity and increasingly unpleasant perception of exercise, affecting them both physically and emotionally (Troosters et al., 2013). Wilson (2006) revealed that increased levels of depression and anxiety can be noted in severe COPD patients, and moreover that the psychological effects can consequently contribute to COPD worsening. Based on the aforementioned results, the BR model combined with hypoxia might be a useful simulation providing fruitful information on the treatment of chronically ill patients who are rendered inactive and hypoxic. While, it is currently speculative whether implementing appropriate activity levels in this population would contribute to reduction of their negative psychological responses, it is certainly an area worthy of further consideration, having in mind the positive or protective effect of exercise in mood during hypoxia.

# Implications for Planetary Habitats

Bed rest studies are normally conducted to examine the effects of unloading/inactivity and not to mimic the exact activity of the astronauts in Space habitats. As such they cannot be considered a high-fidelity analog of a Space mission, and are also not designed for such simulations. Since the boundary conditions of future Lunar habitats are as yet unresolved, the present study somewhat exaggerated the hypoxic and inactivity/unloading conditions, that are expected in the Lunar habitats. Nevertheless, our results clearly demonstrate that combining BR and hypoxia detrimentally affects psychological responses, despite a large inter- and intra-individual variability. Research findings emphasize the importance of cognitive function, psychological responses, and social interactions between crewmembers to ensure a successful space mission (Palinkas, 2001; Manzey, 2004; Kanas et al., 2009). Individuals' psychological profile, personality and cognitive characteristics, as well as, interpersonal and psychosocial issues are closely linked to astronauts' behavior and performance.

# Implications for Hypoxic/Altitude Training

Hypoxic training is becoming increasingly popular in competitive sports. Athletes live and train at altitudes, to take advantage of the altitude-related hematological and non-hematological adaptations. Some others prepare for competitions at altitude and sea level, by living in hypoxic conditions and training at low altitudes or also employ different intermittent hypoxic protocols (Millet et al., 2010). Regardless of which training regimen is adopted, the focus to date has been on the benefits of such training in terms of increased physiological indices, such as erythropoiesis and the subsequent increase in erythrocytes and hemoglobin. The deleterious effects that hypoxia may have on athletes have not been equally scrutinized. The hypophagia and hypodipsia associated with altitude exposure will certainly affect the nutritional and hydration state of the athlete and may consequently affect performance. Similarly, hypoxia-induced central sleep apnoea will affect the sleep architecture and may also have a negative influence on performance. The present study demonstrates that hypoxia increases the negative psychological responses in conditions of extreme physical inactivity. On the other hand, the results demonstrate that the observed hypoxic effect can be counteracted by daily physical activity. However, the activity levels in the present study were designed to mimic the activity during the participants' normal daily activity. Whether hypoxic exposure in combination with strenuous exercise training has a similar effect on psychological responses requires further investigation.

# Study Limitations and Methodological Considerations

While the present study provides an important insight into the independent and combined effects of hypoxia and inactivity on psychological status, there are certain limitations that should be taken into account when interpreting the obtained outcomes. Firstly, although we focused on the analysis of the group/condition average values a significant inter- and intraindividual variability was noted during all hypoxic and/or bed rest exposures indicating that this approach might be limited to effectively and comprehensively capture participants' psychological responses. Secondly, participants' personality characteristics should be taken into account by future prospective studies as they constitute an important variable that can modify participants' mental and cognitive ability in successfully adapting to adverse environments. Finally, due to the project's logistical and financial constraints the sample size employed in this study is limited and should be considered when interpreting the results. Indeed, this limitation is common to all such bed rest studies. Although difficult, prospective projects in this area should aim for larger sample sizes that would enable improved and more accurate estimations of the psychological changes induced by various environmental factors. Based on the above, future, controlled studies are clearly needed and should aim to account for participants' personality characteristics and also consider significant between-individual variability in responses to the tested stressors (i.e., confinement, (in)activity, and hypoxia).

# CONCLUSION

We conclude that hypoxia enhances the negative psychological characteristics in inactive, bed ridden participants, but not in active, ambulatory participants. In line with our previous results (Stavrou et al., 2015), the present outcomes further demonstrate that the hypoxic effect on mood is transient, is only present during the hypoxic stimulus; and is not sustained once the hypoxic stimulus is removed, nor does it appear to be dependent on the duration of the hypoxic inactivity. Our results also suggest that a small volume of daily physical activity is sufficient to negate the negative emotional responses induced by a hypoxic environment. Thus, it seems warranted to further explore the effect of exercise as a means of improving the psychological responses during prolonged hypoxic/altitude exposures, for space settings as well as for sport/clinical applications. Finally, the present study also provides an impetus for further investigations into the manner by which increased activity may improve the quality of life of patients rendered hypoxic and inactive by various clinical conditions.

# AUTHOR CONTRIBUTIONS

IM and OE initiated and designed the study. Together with TD and NS they performed the measurements, analysis, and interpretation of the data. All authors participated in the preparation of the manuscript.

# FUNDING

This study was funded by the European Commission Framework Programme 7 (PlanHab project; Grant no. 284438), the European Space Agency (ESA) Programme for European Cooperating States (ESTEC/Contract No. 40001043721/11/NL/KML: Planetary Habitat Simulation), and the Slovene Research Agency (ARRS Grant No. L3-3654, P3-0338).

# REFERENCES


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

The handling editor declared a shared affiliation, though no other collaboration, with one of the authors IBM.

Copyright © 2018 Stavrou, Debevec, Eiken and Mekjavic. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Aging Decreases Hand Volume Expansion with Water Immersion

Jamila H. Siamwala, Davina G. Moossazadeh, Timothy R. Macaulay, Rachel L. Becker, Rekha H. Hargens and Alan R. Hargens\*

*Department of Orthopedic Surgery, University of California, San Diego, San Diego, CA, United States*

Hands may show early signs of aging with altered skin texture, skin permeability and vascular properties. In clinics, a hand volumeter is used to measure swelling of hands due to edema, carpal tunnel syndrome or drug interventions. The hand volume measurements are generally taken without taking age into consideration. We hypothesized that age affects hand volumeter measurements and that the younger age group (≤40 years) records a greater change in hand volume as compared to the older group (>40 years). Four volumetric measurements were taken at 5 min intervals during 20 min of water immersion using a clinically-approved hand volumeter. After 20 min of immersion, the hand volume changes of the younger age group were significantly higher than the older age group (*p* < 0.001). Specifically, the right-hand volume of the younger age group (≤40 years, *n* = 30) increased by 4.3 ± 2%, and the left hand increased by 3.4 ± 2.1%. Conversely, the right-hand volume of the older age group (>40 years, *n* = 10) increased by 2.2 ± 2.0%, and the left hand decreased by 0.6 ± 2.4% after 20 min of water immersion. The data are presented as Mean ± SD. Hand volume changes were not correlated with body mass index (BMI) or gender, and furthermore, neither of these two variables affected the relationship between age and hand volume changes with water immersion. We conclude that the younger age group has a higher increase in hand volume with water immersion as compared to the older age group.

#### Keywords: hand volumetry, age, gender, BMI, water immersion, time

# INTRODUCTION

Hand volume measurements by water immersion are used extensively in the clinical setting for measuring swelling of limbs due to post-mastectomy lymphedema (Karges et al., 2003), carpal tunnel syndrome (Braun et al., 1989), rheumatoid arthritis (Rembe, 1970) and post-traumatic conditions (Griffin et al., 1990). The degree of swelling is graded based on the weight of water displaced after hand immersion. Hand volume is typically measured using a hand volumeter, by weighing the water displaced after a limb is immersed in a water-filled hand volume chamber. The volume of the hand is directly proportional to the weight of the water displaced, according to Archimedes' principle. Due to consistent results and ease of performance with limited resources, water displacement is considered a gold standard for hand volume measurements in several diseases. Moreover, this technique is popular in less developed countries with scarce resources and poor clinical facilities. One issue with this technique is that when hand volume measurements are made in the clinic, patient age is generally not taken into account while interpreting the differences in hand volume pre and post-water immersion.

#### Edited by:

*Jack van Loon, VU University Amsterdam, Netherlands*

#### Reviewed by:

*Chris Bolter, University of Otago, New Zealand Abhijit Patwardhan, University of Kentucky, United States*

> \*Correspondence: *Alan R. Hargens ahargens@ucsd.edu*

#### Specialty section:

*This article was submitted to Integrative Physiology, a section of the journal Frontiers in Physiology*

Received: *30 September 2017* Accepted: *19 January 2018* Published: *14 February 2018*

#### Citation:

*Siamwala JH, Moossazadeh DG, Macaulay TR, Becker RL, Hargens RH and Hargens AR (2018) Aging Decreases Hand Volume Expansion with Water Immersion. Front. Physiol. 9:72. doi: 10.3389/fphys.2018.00072*

**206**

Although the hands of most people are very similar anatomically and physiologically, they are susceptible to the vagaries of aging, as they are one of the most frequently used body parts and exposed to environmental insults on a daily basis. With time and injury, hand functions begin to decline and manifest the effects of aging. Dietary, behavioral, genetic, metabolic and environmental factors are associated with agerelated loss of hand dexterity (Flatt, 1960). Dietary factors, including malnutrition, lead to loss of homeostasis of minerals, especially alterations in calcium metabolism. Behavioral factors such as sedentary lifestyle and lack of exercise may also lead to accelerated decline in hand function (Flatt, 1960). Certain genetic factors involving mitochondrial activity, fatty acid metabolism and cancer may influence aging of hands (Reed et al., 1991; Glass et al., 2013). Frequent exposure to ultraviolet light (Akiba et al., 1999) and airborne particulate matter (Kim et al., 2016) may lead to greater aging effects and breakdown of collagen, especially in exposed skin. The skin of the hand is also exposed to other kinds of environmental stresses such as abrasions, lacerations and burns, thus accumulating these insults over a lifetime. Dry skin, which is characteristic of aging, has poor skin barrier functions (Epstein and Maibach, 1965; Plewig, 1970; Rougier et al., 1988), which may affect skin permeability. Thus, the decline in skin permeability with age may be measurable by hand volume changes during water immersion. Anatomical aging of hands affects joints, muscle, tendon, bone, nerve, blood supply, skin, and fingernails. The effects of aging on hand volume after immersion are unknown. In this study, we determined if such age-related effects exist. The hypothesis of the study was that an older age group will have less hand volume change with water immersion as compared to a younger age group.

# METHODS

# Participants

Forty participants between the ages of 13 and 65 years participated in this study. Twenty three were males and 18 were females. Thirty participants were in the 13–40 age group, and 10 were in the 41–65 age group. Thirty subjects had a relatively low body mass index (BMI < 25), and 10 subjects were overweight (BMI ≥ 25). None of the subjects had cardiovascular problems nor were they taking any medication at the time of this study. The protocol was approved by the Institutional Review Board of University of California, San Diego. Prior to the study, the experimental protocol was explained to the subjects and their written informed consents were obtained. The subjects also signed HIPAA consent forms to grant the researchers access to their medical histories, which they provided as a part of the study. For subjects under 18 years of age, both the subject and the subject's parent or legal guardian signed the consent form.

### Equipment Hand Volumeter

All hand volume water displacement measurements were taken using a commercially-available volumeter designed specifically to measure hand volume (Hand Volumeter, Volumeters United, Pheonix, AZ). The device is a clear acrylic box with a stopper in the form of a horizontal bar located at the base to maintain consistency of the hand immersion between subjects. For each experiment, the volumeter was filled with a 3% ethanol 97% water solution to the level of an extended lip through which the displaced liquid flows. The water solution contained ethanol so as to reduce surface tension and improve accuracy and reproducibility of measurements (Hargens et al., 2014).

# Water Bath

The subjects' hands were immersed in a large acrylic tub filled with tap water. The water in the tub was kept at constant temperature of 37◦C by a heating coil submerged in the water. A thermometer was placed in the water at the beginning and end of each 5 min interval to ensure that the water was maintained at 37◦C. Paper outlines in the shapes of typical adult right and left hands were taped on the outside of the acrylic tub on the wall closest to the subject to act as a guide and maintain consistencies in hand position. On the right outline was written "Place right hand here" and on the left outline "Place left hand here" in order to ensure that the subject's hand was in the same position for all four immersion periods of each hand. The subjects were seated for the hand immersion protocol and the chair adjusted according to the height of the individuals. The hands were positioned at the same levels as the heart. This was consistent between the subjects.

# Experimental Protocol

Subjects were asked to remove all rings, watches and bands from their hands to eliminate extraneous variables from hand volume measurements. The subjects were also asked to wash their hands with soap solution, rinse and dry them thoroughly before the experiment. The hand volumeter was over-filled with the 3% ethanol solution and maintained at 37◦C. A measuring cup was placed under the extended lip of the hand volumeter to collect the excess solution. The filled apparatus was allowed to sit and drain the excess fluid until no more solution flowed out of the extended lip. At this point, a plastic container, which was previously zeroed on a balance (OHAUS Triple Beam Balance 700 Series, accurate to 0.01 g), was placed under the extended lip. The subject's right hand was slowly immersed in the apparatus such that the third web (intermetacarpal) space of the subject's hand placed slight pressure on the stopper rod. The subject's hand was kept in that position until the ethanol solution stopped flowing. At this point, the plastic container containing the displaced ethanol was removed and placed on the weighing balance. The measurement was recorded to 0.01 g after each time the subject's hand was immersed in the hand volumeter. This mass measurement was later converted to a volume measurement (ml) by dividing it by the density of the 3% ethanol solution (0.99367 g/ml). After every measurement, the subject was provided with tissue paper to dry their hand in order to prevent systematic errors.

After an initial baseline volume measurement, the subject's right hand was immersed for 5 min in the 37◦C water bath. The subject's hand was relaxed but reproducibly submerged in the water, as guided by the paper outlines. After 5 min, the subject was asked to remove and dry their hand. The steps specified above (measurement of volume displaced and 5 min immersion in 37◦C water) were repeated in this order 3 times for a total of 4 rounds and 20 min of immersion in the 37◦C water. Following the final immersion in the water, the subject's right hand volume was measured one last time and recorded for a total of 5 measurements of volume displaced. This entire process was then repeated with the subject's left hand in random order.

# Statistics

For statistical evaluation, subjects were divided into two groups: ≤40 years and >40 years. The experiments were randomized with either the right hand first or the left hand. The data are expressed as means ± standard deviations of changes in the hand volume with respect to age, time, gender and BMI. The statistical significance between the hand volume changes between age groups (≤40 years and >40 years) was computed using unpaired two-tailed t-tests and Mann-Whitney tests. The differences between left and right hand volume changes over time (0, 5, 10, 15, and 20 min), and comparisons by BMI and gender were performed using multiple t-tests on the raw values. All statistics were performed in the graphpad prism software. Significance was set at p<0.05.

# RESULTS

# Hand Volume Changes with Age

Volumetric assessments showed significant differences in hand volume measurements between the younger (≤40 years) and older (>40 years) age groups. The younger age group had significantly higher changes in hand volume after 20 min of water immersion compared to the older age group (**Figure 1**). The right-hand volume change of the younger age group was 4.3 ± 2.0% from baseline compared to 2.2 ± 2.0% in the older age group (**Figure 1A**). The left-hand volume differences between the younger and older age group were more apparent. The lefthand volume of the younger age group increased 3.5 ± 2.1% from baseline compared to minimal change of 0.6 ± 2.4% from baseline in the older age group (**Figure 1B**). The hand volume measurements of younger age group were significantly higher than the older age group.

# Association of Hand Volume Changes with Age and Time

Temporal changes from baseline in the right and left hand volumes after different times of immersion (5, 10, 15 and 20 min) were plotted for each age group (**Figure 2**). Overall, the rightand left-hand volumes increased with time (**Figures 2A,B**). In the younger age group, the right-hand volume changes with time were 2.7 ± 1.9% after 5 min, 3.3 ± 1.6% after 10 min, 4.3 ± 1.7% after 15 min and 4.5 ± 1.9% after 20 min of water immersion. In the older age group, the right-hand volume changes with time were 2.4 ± 1.4% after 5 min, 3.0 ± 1.9% after 10 min, 2.5 ± 3.0% after 15 min and 2.2 ± 2.0% after 20 min of water immersion (**Figure 2A**). The left-hand volume increased incrementally with time for the younger age group: 2.6 ± 1.9% after 5 min, 2.9 ± 2.3% after 10 min, 3.0 ± 2.4% after 15 min and 3.2 ± 2.4% after 20 min of water immersion. However, the left-hand volume changes decreased with time for the older age group. The left hand volumes were 0.6 ± 0.9% after 5 min, 0.2 ± 1.1% after 10 min, 0.5 ± 0.9% after 15 min and 0.1 ± 1.9% after 20 min of water immersion (**Figure 2B**). Consistently the older age group showed less change in hand volume over time compared to the younger age group.

# Association of Hand Volume Changes with Age and Gender

To evaluate if hand volume changes were associated with gender, the hand volume changes after 20 min immersion for male and female participants of each age group were compared. The mean for the right-hand volume of males was 3.4 ± 1.5% compared to 3.8 ± 1.2% for females in the younger age group (**Figure 3A**). The mean for the right-hand volume of older age group males was 2.2 ± 1.8% compared to 4.8 ± 3.5% for females. The mean for the left-hand volume of males was 3.2 ± 1.6% compared to 4.1 ± 2.7% for females in the younger age group (**Figure 3B**). The mean for the right-hand volume of older age group males was 0.2 ± 2.2% compared to 1.1 ± 2.9% for females. The gender differences are not statistically significant suggesting that gender may not play the primary role in the age-related difference in hand volumeter measurements.

# Association of Hand Volume Changes with Age and BMI

The results suggested that BMI may not be associated with age and hand volume changes. Although BMI had no significant effect on the hand volume change, larger hands displace more water compared to smaller hands and the change from baseline was higher in overweight individuals compared to normal weight individuals. However, the effects of BMI on hand volume were independent of age. The right-hand volume change of overweight individuals (BMI ≥ 25) was 2.6 ± 2.2% compared to 2.2 ± 3.7% in normal weight (BMI < 25) individuals in the younger age group (**Figure 4A**). The left hand volume change of overweight individuals is 2.8 ± 2.5% compared to 0.4 ± 1.8% in normal weight individuals in the younger age group (**Figure 4B**). This documented that hand volume changes depend more on age and are not associated with the corresponding BMI.

# DISCUSSION

Measurement of hand volume using the hand volumeter is a useful tool in clinics with limited infra-structure to measure edema in disease conditions or after limb swelling surgery. The technique is more sensitive and consistent compared to tape measures used more commonly in the clinics. Our study supports our hypothesis that after 20 min of immersion in water, hand volume of the younger age group (≤40 years) increases more than that of the older age group (>40 years). Gender and BMI have no significant effect on the age-related changes in hand volumes measured using the hand volumeter. Our data have important clinical implications in measurement of hand volume changes with immersion in younger vs. older patients. In other words, patient age must be considered along with disease state.

FIGURE 1 | Age-related effects on hand volume measurements. The data represent percentage changes in hand volume measured using a hand volumeter after 20 min of hand immersion in water for the younger age group (≤40 years) and older age group (>40 years) in the right hand (A) and left hand (B).Values are means ± SD. \**p* < 0.05 by two-tailed unpaired *t*-test, *n* = 40.

Certain diseases such as post-mastectomy lymphedema (Karges et al., 2003), carpal tunnel syndrome (Braun et al., 1989), rheumatoid arthritis (Rembe, 1970), and post-traumatic conditions (Griffin et al., 1990) result in the swelling of hands. Apart from tape measurement, a more accurate reading of hand swelling is obtained using a hand volumeter in the clinical setting. The volumeter measurement is considered the most accurate "gold standard" as it can detect small changes in hand volume (DeVore and Hamilton, 1968; Holbrook and Odland, 1974; King, 1993; Hargens et al., 2014). However, age is generally not taken into consideration while measuring swelling of the hand using a hand volumeter.

Our results document that age affects hand volumeter data and therefore should be considered while interpreting the results for patients. Exactly how aging affects changes in hand volume during immersion is not known. Decreased skin permeability and accumulation of dead skin cells with age may be factors that contribute to reduced water uptake during hand immersion with age (Epstein and Maibach, 1965; Plewig, 1970; Rougier et al., 1988).

The skin of the hands mirrors the first signs of human aging, and is reflected by an increase in skin permeability, slow replacement of lipids leading to disturbed barrier function, reduction of cutaneous vascular responsiveness and thinning of the epidermis by 10–50% (Zouboulis and Makrantonaki, 2011). Skin permeability and barrier function are measured by percutaneous absorption and transepidermal water loss through the skin of the hand (Rougier et al., 1988). The size of corneocytes is the main contributing factor for skin permeability. In the elderly, the inward and outward movement of water is correlated with corneocyte size. However, in younger people, corneocyte surface area is low. Thus, percutaneous water absorption and transepidermal water loss increases, but only to a certain point (Rougier et al., 1987). At a surface area of

FIGURE 3 | Comparison of the right hand volume changes (A) and left hand volume changes (B) by gender (male vs. female) after 20 min of hand immersion in water. Values are means ± SD.

1,000 µm<sup>2</sup> , there is a limit to the percutaneous absorption or transepidermal water loss. In previous studies of 65 to 80 years olds, the surface area of corneocytes is found to be 20–25% greater than for ages 45–55 years (Plewig, 1970; Rougier et al., 1988).

Skin permeability, changes in corneocyte size and epidermal cell turnover with age may be responsible for the left hand volume changes seen in subjects above the age of 40 years. Apart from corneocytes, the number of sebaceous glands at a particular anatomic site may also influence absorption of water (Blank and Scheuplein, 1969). Another route of water absorption is the transfollicular route in the hands (Scheuplein, 1967). There is an inverse relationship between corneocyte size and epidermal cell turnover. Epidermal cell proliferation decreases with age, which could be another factor responsible for the changes in permeability (Epstein and Maibach, 1965). Specifically, epidermal cell turnover decreases by 30–50% in the third and eighth decade of life (Baker and Blair, 1968; Kligman, 1979). The change in epidermal cell proliferation and keratinization manifests as altered structure and function of the stratum corneum.

Changes in gravitational effects may also affect hand volume changes. Gravity's effects on blood circulation are less in water compared to on land. On land, gravity pulls the blood flow toward the hands and feet in the upright position. However, underwater, the buoyant force of water takes over the functions of gravity and the position of the body is no longer significant. Kraus and colleagues find that by altering local blood pressure by changing the position of the arm, hand volume is affected (Kraus et al., 2013). The question remains if age-related changes in the hands affect hand volume changes with water immersion.

Apart from the changes in skin structure, certain underlying vascular changes also occur with age (Laurent, 2012). Arterial stiffness may affect vascular compliance and decrease absorption of water during hand immersion (Arnett et al., 1994). As early as 40 years, vascular changes may begin to manifest in the form of progressive thickening of the arterial wall layers (intima/media complex), leading to changes in vascular compliance (Arnett et al., 1994). Vascular changes may also include reduction in the number of cutaneous blood vessels, reduced vessel size, increased leakiness of blood vessels and reduced forearm blood flow.

Reduction in the number of microcirculatory vessels and increased arterial stiffness could prevent clearance of transdermally absorbed water from the dermis with aging (Montagna and Carlisle, 1979). Aging is associated with reduction in cutaneous vessel size and loss in vessel density and surface area for exchange. Photoaging, characterized by UV-induced erythema, also has a vascular response similar to that of aged skin. Hence, although differences in hand volume changes are likely related to changes in skin permeability and slow vascular responsiveness, UV-associated aging of the exposed hand should also be taken into consideration. On the other hand, UV exposure also induces upregulation of VEGF and downregulation of angiogenesis inhibitor thrombospondin-1 through MAPK activation in human skin and cultured keratinocytes, leading to leaky vessels and hyperpermeability (Kim et al., 2006). UV exposure third together with chemical aging may be responsible for negative hand volume changes seen by us in the older subjects.

The leakiness of the blood vessels increases with age, and this may contribute to the difference in hand volume change seen in subjects ≤40 and >40 years. Vascular endothelial dysfunction is considered as the primary expression of normal human aging. The capacity of endothelial cells to generate nitric oxide (NO) reduces with age (Toda, 2012). Baseline flow mediated, brachial arterial dilatation is also reduced in older subjects compared to younger subjects (Gates et al., 2007). Although both endothelium-dependent and -independent forearm vasodilation are reduced with aging, diseases such as artherosclerosis induce global vascular dysfunction. Reduction in NO with aging may result in vasoconstriction and reduced uptake of water, although we did not measure NO levels in our subjects. Thus, normal and progressive vascular aging may be primarily responsible for the difference in hand volume changes seen in the subjects of different age groups. Hence, all of these factors should be taken into consideration before establishing a standard hand volumeter reading index for different age groups during the routine clinical procedure of hand volume measurements.

# REFERENCES


## Limitations

Although variables such as water temperature, equipment setup and experiment time were diligently controlled, there are some sources of variability that may limit interpretation of our results. For example, the spill-over time, or drip rate, requires standardization for consistency before measurements. Adding 3% ethanol solution reduces this issue and provides a more consistent measurement of hand volume (Hargens et al., 2014). In addition, the hands of participants must be relatively still to get consistent results. Therefore, this method cannot be used in people with paralysis or Parkinson's disease or diseases in which the hands tremble. Furthermore, skin lesions and open wounds on hands can interfere with the hand volume results.

# ETHICS STATEMENT

The study was carried out in accordance with the recommendations of the Institutional Review Board (IRB) of the University of California, San Diego and carried out according to the ethical guidelines of the Declaration of Helsinki. Written informed consent was obtained from all participants and their parents prior to study enrollment.

# AUTHOR CONTRIBUTIONS

JS, AH discussed, designed the study and obtained the IRB approvals. RH, JS, and DM performed the experiments. JS, DM, and RH analyzed the results. JS interpreted the results. JS wrote the manuscript. JS prepared the figures. AH revised the manuscript. All the authors read the manuscript for submission.

# ACKNOWLEDGMENTS

This study was supported by a Wood Whelan travel fellowship (JS) and NASA grant NNX13AJ12G (AH). We are also immensely grateful to Robert Healey and Dr. Brandon R. Macias for help with the statistics.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Siamwala, Moossazadeh, Macaulay, Becker, Hargens and Hargens. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Grip Force Adjustments Reflect Prediction of Dynamic Consequences in Varying Gravitoinertial Fields

#### Olivier White<sup>1</sup> \*, Jean-Louis Thonnard2,3, Philippe Lefèvre2,4 and Joachim Hermsdörfer <sup>5</sup> \*

1 INSERM UMR1093-CAPS, Université Bourgogne Franche-Comté, UFR des Sciences du Sport, Dijon, France, <sup>2</sup> Institute of Neuroscience, Université Catholique de Louvain, Louvain-la-Neuve, Belgium, <sup>3</sup> Physical and Rehabilitation Medicine Department, Cliniques Universitaires Saint-Luc, Université Catholique de Louvain, Louvain-la-Neuve, Belgium, <sup>4</sup> Institute of Information and Communication Technologies, Electronics and Applied Mathematics, Université Catholique de Louvain, Louvain-la-Neuve, Belgium, <sup>5</sup> Department of Sport and Health Sciences, Institute of Human Movement Science, Technische Universität München, Munich, Germany

#### *Edited by:*

Yury Ivanenko, Fondazione Santa Lucia (IRCCS), Italy

#### *Reviewed by:*

Alessandro Moscatelli, Università degli Studi di Roma Tor Vergata, Italy Ke Li, Shandong University, China Bing Chen, University of Miami, United States

#### *\*Correspondence:*

Olivier White olivier.white@u-bourgogne.fr Joachim Hermsdörfer joachim.hermsdoerfer@tum.de

#### *Specialty section:*

This article was submitted to Integrative Physiology, a section of the journal Frontiers in Physiology

*Received:* 27 October 2017 *Accepted:* 08 February 2018 *Published:* 23 February 2018

#### *Citation:*

White O, Thonnard J-L, Lefèvre P and Hermsdörfer J (2018) Grip Force Adjustments Reflect Prediction of Dynamic Consequences in Varying Gravitoinertial Fields. Front. Physiol. 9:131. doi: 10.3389/fphys.2018.00131 Humans have a remarkable ability to adjust the way they manipulate tools through a genuine regulation of grip force according to the task. However, rapid changes in the dynamical context may challenge this skill, as shown in many experimental approaches. Most experiments adopt perturbation paradigms that affect only one sensory modality. We hypothesize that very fast adaptation can occur if coherent information from multiple sensory modalities is provided to the central nervous system. Here, we test whether participants can switch between different and never experienced dynamical environments induced by centrifugation of the body. Seven participants lifted an object four times in a row successively in 1, 1.5, 2, 2.5, 2, 1.5, and 1 g. We continuously measured grip force, load force and the gravitoinertial acceleration that was aligned with body axis (perceived gravity). Participants adopted stereotyped grasping movements immediately upon entry in a new environment and needed only one trial to adapt grip forces to a stable performance in each new gravity environment. This result was underlined by good correlations between grip and load forces in the first trial. Participants predictively applied larger grip forces when they expected increasing gravity steps. They also decreased grip force when they expected decreasing gravity steps, but not as much as they could, indicating imperfect anticipation in that condition. The participants' performance could rather be explained by a combination of successful scaling of grip force according to gravity changes and a separate safety factor. The data suggest that in highly unfamiliar dynamic environments, grip force regulation is characterized by a combination of a successful anticipation of the experienced environmental condition, a safety factor reflecting strategic response to uncertainties about the environment and rapid feedback mechanisms to optimize performance under constant conditions.

Keywords: motor control, grip force, switching, gravity sensing, uncertainty, hypergravity

# INTRODUCTION

We can easily lift a bulb and subsequently handle a hammer with appropriate grip forces. Motor adaptation and context switching often occur when we interact with the environment. However, robot-based experiments demonstrate some limitations of the brain to concurrently learn different task dynamics (Gandolfo et al., 1996; Conditt et al., 1997; Karniel and Mussa-Ivaldi, 2002), even when explicit cues inform about the expected dynamics (Krakauer et al., 1999; Osu et al., 2004). Some other contexts, however, allow the motor system to learn different dynamics if these are associated with distinct tools (Kluzik et al., 2008), objects (Ahmed et al., 2008), control policies (White and Diedrichsen, 2013) or effectors (Nozaki et al., 2006). Hence, participants' ability to switch between contexts critically depends on experimental details.

Quite surprisingly, many examples of successful switching were also shown between altered gravity environments despite the fact these environments affect the human body in its entirety including many physiological parameters. Parabolic flights and human centrifuges provide unique means to change the gravitoinertial environment. In parabolic flights, the participant is exposed to a repeated gravitational profile (e.g., 1, 1.8, 0, 1.8, and back to 1 g, where 1 g is Earth gravity). Similarly, in long arm human centrifuges arbitrary gravitoinertial environment can be generated (e.g., gradual or step functions from 1 to 3 g). In contrast to robotic experiments, where only the end-effector (e.g., the hand) is perturbed, parabolic flights and rotating-room environments plunge the subject into an unexplored setting. Nearly perfect and surprisingly quick adaptation of motor responses in those challenging environments were nevertheless observed in dexterous manipulation (Hermsdörfer et al., 1999; Nowak et al., 2000; Augurelle et al., 2003; White et al., 2005; Göbel et al., 2006; Mierau et al., 2008; Crevecoeur et al., 2009; Barbiero et al., 2017) and arm movement tasks (Papaxanthis et al., 1998; White et al., 2008).

A question arises as to why switching is facilitated in radically new contexts whilst it is much more difficult in some laboratory robot-based experiments? Adaptation is a hallmark of successful tuning of internal models. In other words, our brain develops strategies to anticipate and counteract expected perturbations. To do so, it needs information and time. Visual inflows provide key information to refine our priors about an upcoming action. For instance, before lifting an object, our brain analyses different features such as size (Gordon et al., 1991a,b), shape (Jenmalm and Johansson, 1997) and weight distribution (Johansson et al., 1999). All these factors influence predictive scaling of fingertip forces in dextrous manipulation. Anticipatory grip force adjustments are reflected through complementary temporal and dynamic variables. For instance, temporal variables include the duration of the preload phase (i.e., period of contact of the fingers with the object before lift-off) or the synchronization between peaks of load force and grip force. In predictive manipulation tasks, the preload phase is short and force peaks are perfectly synchronized, whatever the profile of destabilizing load force. Good predictability is also reflected by high correlation between time series of grip and load forces and, in particular, a linear relationship between peaks of grip force and load force (or their first derivative). The lack of grip force adjustment is observed either through accidental slips or abnormally high safety margins. One such situation can be generated by the well-known size-weight illusion paradigm used in cognitive psychology. When participants were asked to lift a large and a small object, which seemed to be of the same material but were designed to have equal weight, peak grip and load force rates were initially scaled to object size, whereas after four trials, these signals were similar for the two objects and appropriately scaled to object weight (Flanagan and Beltzner, 2000). Recently, the perception of heaviness and hence the anticipatory grip force adjustment—has been shown to report more on a mass-volume relationship than on visual cues (Platkiewicz and Hayward, 2014). After only a few practice trials, the central nervous system is capable to build two representations that can be selected on a trial basis upon context. Whether this is the same internal representation but parameterized by external information or two hard coded independent internal models remains controversial (Wolpert and Kawato, 1998).

While the importance of vision is not disputed, other sensory information, such as haptics, are processed to refine representation of internal models underlying object manipulation. In the particular contexts of altered gravitoinertial environments, vestibular signals influence motor control from planning to task execution (Bockisch and Haslwanter, 2007). We hypothesize that however radical and new the environment is, coherent sensory inflows will provide much more useful information to the brain to optimize the behavior. In other words, visual, haptic, proprioceptive and vestibular feedback emerging in a homogeneous environment should yield coherent information to the brain to speed up adaptation between unusual dynamics. This result would contrast with slower adaptation usually observed when local perturbations are applied to a subset of sensory modalities (e.g., haptic perturbation of the hand). Here, we test how participants adapt to and switch between unusual dynamical contexts generated by rotation of a long-arm human centrifuge. We expect that participants will adopt an optimal motor strategy since the first trial in the new environment, and that this will be reflected through temporal and dynamic variables underlying grip force control.

# MATERIALS AND METHODS

# Participants

Seven right handed male participants (42.1 years old, SD = 9.3) participated in this experiment. A medical flight doctor checked their health status before the experiment. The protocol was reviewed and approved by the Facility Engineer from the Swedish Defence Material Administration (FMV) and an independent medical officer. The experiment was overseen by a qualified medical officer. The study was conducted in accordance with the Declaration of Helsinki (1964). All participants gave informed and written consent prior to the study.

# Centrifuge Facility and Instrumented Object

Centrifugation took place at QinetiQ's Flight Physiological Centre in Linköping, Sweden. The centrifuge has a controllable swinging gondola at the end of a 9.1 m long arm (**Figure 1A** and see Levin and Kiefer (2002) for technical details). Preprogrammed G-profiles could be specified and the closed loop control of the gondola ensured that the gravitoinertial force was always aligned with body axis (Gz). Participants were strapped while seated and cushioning was provided for comfort. Their electrocardiogram was continuously monitored during the entire centrifuge run for safety reasons. One-way video and two-way audio contacts with the control room were available at all time. In order to minimize nauseogenic tumbling sensations during acceleration and deceleration, participants were instructed to avoid head movements. Furthermore, G-transitions between stable phases were operated below 0.32 g/s until the desired level was reached.

The wireless test object (mass = 0.13 kg) incorporated a strain gauge force sensor, which measured the grip force applied against the grip surfaces (MAK 177, range 0–100N, Rieger, Rheinmünster, see **Figure 1B**). The design of the sensor guaranteed an accuracy < ±0.1N, even if the location of the center of force application was off-axis. The load force was measured along the axis of a stand until object lift-off with the same accuracy (MAK 177, range ±50N, ±0.1N). An accelerometer that measured combined gravitational and kinematic accelerations along the object's long axis was mounted inside the test object (AIS326DQ, range ±30 m/s<sup>2</sup> , accuracy ±0.2 m/s<sup>2</sup> ). After lift-off, load force was calculated from the product of object mass and the gravitational and inertial accelerations (Hermsdörfer et al., 2003). Force sensors were calibrated off-line with calibration weights and acceleration sensors with use of the gravitation vector. All signals were A/D-converted and sampled at a frequency of 120 Hz. The digitized signals were then transmitted to a Palm device through a Bluetooth connection. Data were downloaded after the recordings to a standard PC for analysis.

# Procedure: Lift Task during Centrifugation

The centrifuge was programmed to deliver the same ramp up/ramp down Gz-profile for 180 s (**Figure 1C**). Participants

Frontiers in Physiology | www.frontiersin.org

gravitoinertial environments (1.6 s).

were aware of the profile and could get prepared. The initial 1 g phases (idle) lasted for 27.4 s. Then, the system was controlled to generate 1.5, 2, 2.5, 2, and 1.5 g. Each phase lasted 18.4 s and transitions lasted 1.6 s (0.31 g/s). After a last transition, the system reached its final 1 g phase and recording stopped after another 27.4 s period. Note that transitions between 1 and 1.5 g were longer (13.4 s, 0.04 g/s) as they were more likely to induce motion sickness. **Table 1** reports mean and standard deviations of accelerations recorded during each trial and in each stable phase of the centrifugation profile and shows that the environments were very stable.

The operator was provided with feedback about real time gravity and was in continuous verbal contact with the participant. At each GO signal ("LIFT!"), the participant adopted a precision grip configuration to grasp and lift the manipulandum with the thumb on one side and the other fingers or only the index on the other side at a comfortable speed (see **Figure 1B**). The elbow remained in contact with the support and the upper arm made an angle of ∼30 degrees with the horizontal. When the operator announced the STOP signal ("DOWN!") after about 2 s of stationary holding, the participant gently let the object down on the support. The same task has been extensively used in previous investigations (Westling and Johansson, 1984). Four trials were completed during each stable gravitational phase. Between consecutive trials and during Gz-transitions until the first trial in the new environment, participants adopted a relaxed posture with the hand and forearm resting on the ulnar edge, and the index finger and thumb positioned ∼2 cm apart from the instrument grip surfaces.

# Data Analysis

Grip force, load force and object acceleration along the vertical axis were low-pass filtered at 20 Hz with a zero phase lag autoregressive filter. The derivatives of the force signals (force rates) were then computed with a finite difference algorithm.

**Figure 2** presents load force (red trace) and grip force (blue trace) in a typical trial in 1.5 g that resembled those in earlier studies (Johansson and Westling, 1984; Westling and Johansson, 1984). We first determined peaks of grip force and load force for further analysis (GFMAX, LFMAX, **Figure 2**). Grip force and load force onsets were identified when force rates exceeded 0.4N/s for 125 ms (respectively tGF<sup>o</sup> and tLFo). We identified the time at which load force rate fell below −2 N/s for at least 125 ms and subtracted 250 ms to define the end of the trial. Load force and grip force plateaus were measured as the average load force and grip force during the last second of the trial. We then fitted an exponential function to the grip force profile between its peak and the end of the trial: GF (t) = a+be−ct (**Figure 2**). This allowed us to reliably quantify grip force decay through parameter c and the plateau phase of grip force with the offset parameter a. There was a good correlation between this last parameter and the average grip force during the last second of the trial, r = 0.83, p < 0.001.

Furthermore, two temporal parameters that characterize the grip-lift task were extracted (**Figure 2**): the duration of the preload phase (delay between grip force and load force onsets, tLF<sup>o</sup> − tGFo) and duration of the loading phase (delay between load force onset and the moment load force equals the object's weight, tLF<sup>w</sup> − tLFo). Finally, we calculated the cross correlation between load force rate (reference signal) and grip force rate. We shifted grip force rate between tLFo−150 ms and tLFo+150 ms with respect to load force rate. This procedure yielded the largest coefficient of correlation and the time shift for which this condition was fulfilled. These two values were computed for each individual trial and provided an estimate of the overall synergy of the grip-lift movement. Correlations quantified how well grip and load force profiles matched, which indicated the quality of anticipatory scaling of grip force to load force. Time-shifts

FIGURE 2 | Grip force (blue trace) and load force (red trace) over time for a single lift trial. The first two vertical cursors (tGFo and tLFo) enclose the preload phase. Time 0 ms corresponds to grip force onset (tGFo). The loading phase is the time between tLFo and tLFw (two last vertical cursors, see Methods). The dashed line is the best exponential fit to grip force between its peak (GFmax, black dot) and the end of the trial. Parameter a provides a more reliable estimate of grip force reached in the plateau phase than a classical average during the last portion of the trial. The rate of decrease of grip force following its maximum was quantified by parameter c.

TABLE 1 | Magnitude of the local gravitoinertial acceleration (Gz) in each programmed environment in the centrifuge and during each individual lift (rows).


We first calculated mean and SD of the accelerations during the trial. Cells contain the average and standard deviations of the above values across participants (N = 7). The gondola rotated at a very constant rate during each phase, leading to highly reliable and stable environments.

provided a measure of the asynchrony between the two forces. A positive time-shift indicates that grip force led load force, as it is usually reported in healthy humans in dextrous tasks (Johansson and Westling, 1988; Forssberg et al., 1991).

Participants performed four trials in each phase. We were also interested to quantify the difference of grip force peak between the last trial in one environment and the first trial in the upcoming environment. We defined an index (1GF) by subtracting grip force peak recorded in the last trial in the previous gravity level from grip force peak during the first trial in the next environment, 1GF g = GFnext g trial 1 <sup>−</sup> GFprev g trial 4 . We defined the same index but for load force peaks (1LF).

Quantile-quantile plots were used to assess normality of the data. Repeated-measure ANOVAs were performed on the above variables to test for the effects of gravity (factor GRAVITY = 1, 1.5, 2, or 2.5 g) and trial (factor TRIAL = T1, T2, T3 or T4). In complementary analyses, we compared the first, ascending, 1, 1.5, and 2 g phases with the second, descending, 2, 1.5, and 1 g phases (factor PHASE = ascending or descending). Participants were only faced once to the 2.5 g-phase. Therefore, it was not included in the ANOVA when factor PHASE was considered. Post-hoc Scheffé tests were used for multiple comparisons and paired t-test of individual subject means were used to investigate differences between conditions. Alpha level was set at 0.05. Because the sample size is small (n = 7 participants), partial etasquared are reported for significant results to provide indication on effect sizes. The dataset was visually inspected to ensure these parameters were accurately extracted by custom routines developed in Matlab (The Mathworks, Chicago, IL).

# RESULTS

Participants performed a precision grip lifting task when the gravitoinertial environment was varied with a centrifuge. **Figure 3A** depicts average load force during the plateau phase in each gravitational environment separately for each trial. Since the object was held stationary during that period, the load force reflects the weight of the manipulandum during the respective Glevels. Consistently, a 3-way RM ANOVA shows that load force plateau—or object weight—was only influenced by GRAVITY [F(2, 131) = 143920.04, p < 0.001, η 2 <sup>p</sup> <sup>=</sup>0.99] and not TRIAL [F(3, 131) = 0.8, p = 0.494] or PHASE [F(1, 131) = 0.3, p = 0.619], with no interaction effect (all F < 1.6, all p > 0.191). Participants matched this level of static load force with grip force, as illustrated in **Figure 3B**. Again, the ANOVA only reported a main effect of GRAVITY, F(3, 131) = 28.9, p < 0.001, η 2 <sup>p</sup> = 0.49, with no

FIGURE 3 | Mean and SEM of parameters of the task that characterize the plateau phase (left column, A–C) and when load force reached a maximum (right column, D–F). Data are presented in chronological order, following the successive exposures to 1, 1.5, 2, 2.5, 2, 1.5 g and back to 1 g (same color code as in Figure 1). Data are also shown separately for each trial in a given environment (C, 1 g). The "ascending" (resp. "descending") phase comprised the increasing (resp. decreasing) gravitoinertial environments 1–2.5 g (resp. 2.5–1 g).

significant TRIAL, PHASE or interaction effect (all F < 2.9, p > 0.094). The ratio between grip force and load force was not affected by any of the factors (**Figure 3C**; all F < 0.4, all p > 0.318) except PHASE [F(1, 131) = 5.0, p = 0.025, η 2 <sup>p</sup> <sup>=</sup> 0.03]. We indeed found a small decrease of the safety margin during the plateau phase in the second (descending) phase compared to the first (ascending) phase (ascending: 1.55; descending: 1.37).

Load force depends on gravity (mass x gravitational acceleration) and kinematics through its inertial component (mass x acceleration). Peaks of load force were on average only 12.3% larger than load force plateau (compare **Figures 3A**,D). The statistical analysis again reported a main GRAVITY effect on peak load force [**Figure 3D**, F(3, 131) = 2769.2, p < 0.001, η 2 <sup>p</sup> <sup>=</sup> 0.9]. Furthermore, load force peaks were also 8% smaller in the descending phase PHASE [PHASE: F(1, 131) = 7.4, p = 0.007, η 2 <sup>p</sup> <sup>=</sup> 0.01]. Participants moved the manipulandum in such a way that apart from the somewhat lower acceleration during decent, it was not influenced by any other factor (all F < 0.9, p > 0.495).

**Figure 3E** shows that grip force peaks were adjusted to load force [main effect of GRAVITY, F(2, 131) = 18.8, p < 0.001, η 2 <sup>p</sup> <sup>=</sup> 0.2]. Interestingly, **Figure 3E** also shows two additional effects. On the one hand, grip force peaks decreased across trials during the first and second exposures to 1, 1.5, and 2 g. On the other hand, the average level of peak grip force seemed to be lower in the second, descending, phase. An ANOVA confirmed that grip force peaks decreased with TRIAL [F(3, 131) = 4.7, p = 0.004, η 2 <sup>p</sup> <sup>=</sup> 0.08] and were lower in the descending phase of the profile [PHASE, F(1, 131) = 5.1, p = 0.025, η 2 <sup>p</sup> <sup>=</sup> 0.03], without any interaction (all F < 0.5, all p > 0.643). Post-hoc tests revealed that trial 1 was marginally different from trials 2–4 (p < 0.045). When the ANOVA was conducted only with trials 2, 3, and 4, it yielded no results, F(2, 97) = 0.7, p = 0.496. This effect is also reflected in the ratio between grip and load forces when load force was maximum [**Figure 3F**; TRIAL: F(3,131) = 4.8, p = 0.003, η 2 <sup>p</sup> <sup>=</sup> 0.09 and PHASE: F(1, 131) = 7.1, p = 0.009, η 2 <sup>p</sup> <sup>=</sup> 0.05]. Finally, we also found that, within trials, grip force decreased faster to its value in the plateau in the descending phase [parameter c, exponential decay in ascending vs. descending: 714 ms vs 384 ms, main effect of PHASE F(1, 131) = 4.6, p = 0.035, η 2 <sup>p</sup> <sup>=</sup> 0.04] with no other effects (all F < 0.73, all p > 0.588). Altogether, these results show that participants adopted stereotyped movements from the very first trial in every gravitational environment and decreased grip force across trials and exposure according to the g-level.

The grip force level during the plateau phase was adjusted more than a second after first contact occurred with the object and this regulation was probably influenced by feedback mechanisms. Although grip force peaks occurred rather early after lift-off (mean across participants = 273.2 ms, SD = 58.8 ms), peak grip force rates, which always occur earlier than grip force peaks (mean = 104.2 ms, SD = 37.5 ms) are therefore sometimes considered a reliable measure of feedforward processes. Interestingly, the same analysis as above led to even more significant conclusions: Peak grip force rates were proportional to GRAVITY [F(3, 131) = 17.0, p < 0.001, η 2 <sup>p</sup> <sup>=</sup> 0.18], decreased with TRIAL [F(3, 131) = 4.2, p = 0.007, η 2 <sup>p</sup> <sup>=</sup> 0.07] and were lower in the descending phase of the profile [PHASE, F(1, 131) = 5.1, p = 0.025, η 2 <sup>p</sup> <sup>=</sup> 0.03] with no interaction (all F < 0.5, all p > 0.712).

This anticipatory strategy is compatible with the very short lags observed between load force rate and grip force rate (mean = 1.7 ms, SD =7.3 ms) as well as associated high correlations (mean = 0.91, SD = 0.1). A two-way ANOVA revealed that the correlation increased with TRIAL [**Figure 4A**, F(3, 93) = 4.2, p=0.008, η 2 <sup>p</sup> <sup>=</sup> 0.1], was not affected by GRAVITY [F(3, 93) = 1.69, p = 0.17] and that the lag was not significantly altered [**Figure 4B**, all F < 0.51, all p > 0.123]. The ANOVA did not report any significant result when we excluded trial 1, F(2, 65) = 0.8, p = 0.452.

The preload phase, i.e., the delay between object-finger(s) contact and the first increase in load force, is an important underlying variable that characterizes a grip-lift task because, during this short period of time, physical properties of the grasped object are encoded by mechanoreceptors. **Figure 4C** depicts average preload phases in the four trials. The ANOVA reported a significant effect of TRIAL, F(3,93) = 3.8, p = 0.013, η 2 <sup>p</sup> <sup>=</sup> 0.11, with no other effect (all <sup>F</sup> <sup>&</sup>lt; 0.94, all <sup>p</sup> <sup>&</sup>gt; 0.423). A post-hoc test showed that the preload phase in trial 1 was longer than during trials 2–4 (70.9–46.9 ms, 33.8% drop, p = 0.041).

The most striking finding is that participants could adjust grip force to load force from the very first trial in the new environment. To quantify this ability, we calculated the correlation between load force and grip force peaks in each phase but only for trial 1. We found very good and similar correlations in the ascending and descending phases (**Figure 5**). When we pooled phases together, the correlation reached r = 0.91 and was significant (p < 0.001).

In the previous sections, we showed that although participants moved the object consistently across conditions, grip force was not completely adapted upon entry in the new environment. Indeed, there were genuine differences between trial 1 and the three following trials in the same condition. **Figure 6** illustrates average load force (red) and grip force (blue) profiles in the first trial (T1, solid line) and in the last trial (T4, dashed line). The upper row reports these time series in the ascending phase (**Figures 6A–D**, 1, 1.5, 2, and 2.5 g) and the lower row depicts these data in the descending phase (**Figures 6E–H**, 2.5, 2, 1.5, and 1 g). Note that for the sake of clarity and comparison, panels D and E report data from the same grip and load force profiles in 2.5 g. While **Figure 6** shows that load forces overlapped between trial 1 and trial 4, grip force was always larger in trial 1 compared to trial 4.

We quantified the participants' ability to switch between environments by analyzing the index 1GF that is illustrated between 1 and 1.5 g in the ascending phase in **Figure 6** between panels A and B. We ran an ANOVA with factors PHASE (ascending vs. descending) and a new factor that characterizes the two environments between which 1GF is calculated (SWITCH: 1–1.5, 1.5–2, and 2–2.5 g). The ANOVA reported that 1GF was significantly larger in the ascending phase [F(1, 34) = 19.05, p < 0.001, η 2 <sup>p</sup> <sup>=</sup> 0.32]. These effects are illustrated in **Figure 6I** (see also **Table 2**). Similar results were found when peak grip force rates were used to calculate the index. Furthermore, a t-test

showed that 1GF was significantly larger than 0 in the ascending phase [mean = 2.33N; t(20) = 6.05, p < 0.001, η 2 <sup>p</sup> <sup>=</sup> 0.65], but not in the descending phase (mean = −0.01N; t(18) = −0.02, p = 0.984). This analysis reveals an asymmetric behavior between phases, suggesting that the peak grip force in the current environment is not only planned on the basis of the performance in the last trial in the previous environment combined with the anticipated effects of the upcoming gravitoinertial context. As outlined below, a strategy consisting in adopting a safety margin that is large in the first trial in a new environment but dissipates in the following trials could be relevant.

A question naturally arises as to why this 1GF is asymmetric while step of load forces are symmetric between phases? The ANOVA reported that 1LF, which equals mgt+1−mg<sup>t</sup> , calculated between peaks of load force, were of course different between the ascending and descending phases [F(1, 34) = 15.6, p < 0.001, η 2 <sup>p</sup> <sup>=</sup> 0.97] but did not vary within phase [F(2, 34) <sup>=</sup> 0.43, p = 0.654] and were symmetric [t(18) = −1.4, p = 0.167]. **Table 2** reports these values of 1GF. The prediction of what increment of grip force to apply can be based on the expected increment of load force. These forces have been shown to be reliably linearly correlated with a gain α: 1GF = α1LF = α (LFt+<sup>1</sup> − LFt), where t denotes the last trial in the previous context and t+1 the first lift in the next context. We assume that similar accelerations were produced by participants, which yields 1GF = α mgt+<sup>1</sup> − mg<sup>t</sup> = αm1g. Furthermore, participants often adopt some security margin β that reflects task and environmental uncertainties and risk aversion. The increment of grip force can therefore follow this simple rule: 1GF = αm1 g + β, where the first term quantifies prediction based on experienced and expected information and the last term includes uncertainty.

We can now test two alternative hypotheses to explain why 1GF does not follow the simple model described above (**Figure 6I**). On the one hand, the prediction can be correct and constant but uncertainty can vary. We set the value of the gain β to 1.5, which corresponds to the mean of the grip to load force ratio in all first trials in a new environment (**Figure 3C**). **Table 2** (correct prediction, αm1g) reports values of the predictive term that are proportional to 1LF. In order to match the observed 1GF, the second term had to be adapted (**Table 2**, correct prediction, β). Alternatively, if we set an uncertainty value to the constant 1.44N that corresponds to uncertainty measured in a normal case (1 g), the term αm1g becomes variable, which is caught in α (**Table 2**, α). In the first hypothesis, uncertainty is rather constant in the ascending phase but jumps to a negative value before increasing again in the descending phase. Usually, uncertainty decreases over time, when one gets more confidence in the task. Instead, our data seem to favor the second hypothesis. In that case, the internal model is wrongly adjusted, especially in the first descending step (**Table 2**, bold and italic row).

FIGURE 5 | Correlation between load force peaks (x-axis) and grip force peaks (vertical axis) in the first trial and separately for each phase. Each point corresponds to the average across the seven participants in each of the four gravitoinertial contexts. The linear regressions were significant in both ascending (r = 0.99, p = 0.002, slope = 1.6, offset = 2.2) and descending phases (r = 0.99, p = 0.012, slope = 1.67, offset = 1.68). Vertical and horizontal error bars correspond to STD. Note that the point in the upper right corner was identical in the ascending and descending phases (only one 2.5 g phase).

# DISCUSSION

Humans use many different objects in various situations. For instance, when cooking, one hand can move an egg off the table and shortly after, manipulate a heavy pan. Fortunately, the brain developed strategies that allow to anticipate taskrelevant parameters and adjust the control policy accordingly, from the very first instant we start the task. This ability has been demonstrated in the past in several experimental contexts and is formalized by the concept of internal models. In particular, it was suggested that the brain can store multiple predictive models and select the most appropriate one according to the task at hand (Wolpert and Kawato, 1998).

Importantly, these models are flexible. Most of the time, participants can learn the appropriate dynamics of a new task in a matter of a few trials. Quite surprisingly, this also holds when the environment is radically altered like in parabolic flight (Nowak et al., 2000; Augurelle et al., 2003) or when participants are confronted to artificial new dynamics (Flanagan and Wing, 1997). It seems that, after sufficient training, participants can switch between these models effortlessly. However, forcefield learning experiments show that two dynamic internal models cannot be learned concurrently unless the posture of the arm is changed between conditions (Gandolfo et al., 1996; Karniel and Mussa-Ivaldi, 2002). Interestingly, abstract representations of different objects can be combined in the brain to create a new one, adapted to a new situation. In a nicely designed paradigm, participants trained to lift objects of different masses and were then asked to lift the combined object (Davidson and Wolpert, 2004). Grip force rates were adjusted predictively in the very first trial for the combined object, suggesting they stacked both previously formed models. How do we reconcile experimental contexts in which adaptation needs time and others that do not, or, in other words, that allow switching? We posit that a fundamental difference between these conditions is the availability of different sensory information that allow much more efficient adaptation.

Here, we asked participants to lift a lightweight object but in different gravitoinertial environments generated by a long arm human centrifuge. The dynamics of the system and the visual environment in the gondola were such that the different gravitoinertial levels were felt like pure gravitational increments. Participants are extremely familiar with the employed lifting task and with this kind of object but not at all with the environment. Before the experiment, they were told what gravitoinertial profile (amplitude and time course) was implemented in the system. They were also warned in real time, during the experiment, when a new transition was about to occur. All participants had therefore a cognitive knowledge but not (yet) a multisensory experience of the task.

We found a remarkable ability of participants to scale their grip force to gravity from the outset. How was that possible? First, the brain could use information from all sensory modalities. This task, once the object was contacted by the fingers, relied mostly on tactile and proprioceptive feedback. Initial perfect adaptation underlines the importance of that sensory modality. This is in agreement with the work mentioned above (Davidson and Wolpert, 2004) since in that study, participants were prevented from any visual or auditory cue. However, Davidson and Wolpert's experiment was conducted in a familiar, terrestrial, environment. Second, participants also had a theoretical knowledge of the environment. However, it was also shown that pure cognitive knowledge about a change of context is sometimes insufficient to allow prediction. For instance, when participants decreased the weight of a handheld glass of water by drinking with a straw, they could match the change of weight with grip force which they couldn't when lifting the object after drinking while the object was left on the table (Nowak and Hermsdörfer, 2003). Similarly, the prediction of the effects of gravity of a falling virtual object was only possible when a physical interaction with the object was required (Zago et al., 2004). Third, repeating the same trial many times triggers use-dependent mechanisms (Diedrichsen et al., 2010). This propensity of performing the same action if it was successful during the previous trials is responsible for the appearance of large errors if a contextual parameter is changed unbeknownst to the participant. While this process may have been used within a gravitoinertial phase, it was certainly not the case between phases. Altogether, this suggests that multisensory information is essential to switch between environments. Two learning mechanisms may both contribute to adaptation during this task but their respective importance may be weighted differently. Prediction errors are used by error-based learning processes when switching while use-dependent mechanisms are active within each constant environment.

and T4 in all conditions. The upper row (A–D) corresponds to the ascending phase and the lower row depicts time series in the descending phase (E–H). For clarity, since we had only one 2.5 g environment, D,E present the same data. The index 1GF (illustrated in A) quantifies the switching between environments and is calculated as the first grip force peak in the next environment (at trial 1) minus the last grip force peak reached in the current environment (at trial 4). (I) Average and SEM of 1GF for each of the six transitions. Bar plots are bicolour; left color corresponds to the current environment and right color corresponds to the next environment (refer to the sketch above).

TABLE 2 | Values of different parameters between two consecutive trials in two different environments (Trials).


Are reported: step of LF and GF and the predictive (αm1g) and uncertainty (β) terms under two different hypotheses ("Prediction correct" and "Prediction incorrect"). The bold italic row highlights the first descending step, i.e., between 2.5 and 2 g.

Despite the fact we observed good overall adaptation of grip force to load force in all phases, there were nevertheless more subtle exceptions noticeable at two different timescales. On the one hand, when comparing equivalent environments, grip forces were smaller in the second, descending, phase of the experiment. This was however a weak although significant effect (low effect sizes). This is also reflected by a faster decay of grip force to a smaller plateau value. During a parabolic flight campaign, the static grip force produced to hold an object stationary was massively increased during the first experience of 0 and 1.8 g suggesting a strong effect of stress induced by the novel environmental conditions (Hermsdörfer et al., 1999). This increase in grip force levels resolved however quickly across the subsequent exposures to the new gravitoinertial conditions. This behavior may reflect habituation and not a change in motor prediction. On the other hand, there were subtle adjustments in grip force (not load force) between the first trial and the next trials, particularly in the second 1 g environment. Namely, peak grip force, grip force rate, the grip to load ratio and the preload phase all decreased after the first trial and the synergy between both forces improved. At first sight, this is a counterintuitive result since participants are again back in a 1 g stable and well known environment. Beside the fact participants experienced a very stressful environment, a transition between 2.5 and 2 g and between 1.5 and 1 g are very different in terms of vestibular inputs. Indeed, the central nervous system interprets 1 g as an absence of rotation and a strong sensorimotor conflict arises. We made these transitions much smoother to avoid motion sickness. The suboptimal parameters observed in trial 1 in the second 1 g environment may reflect the fact participants have to readjust grip force. Finally, it is also worth mentioning that two trials are necessary before a decay becomes observable in peak grip forces in the first 1.5 g phase (**Figure 3E**), that is, during the first seconds spent in a hypergravity environment. Therefore, a pure, perfect switch really needs at least one trial to occur.

It is immediately clear in **Figure 6** that the change in grip force directly after a change of g-level does not directly reflect the change in load. It rather seems that the change in grip force is exaggerated since it is reduced substantially in the following trials. In all environments, a safety margin, linked to self-perception of uncertainty, was obviously employed during the first contact with the object in the new gravitoinertial environment. This margin decreased with time and confidence. The only exceptions are the trials in the highest g-level, 2.5 g. In that extreme situation, participants experienced the highest mental and physical stress and may not have relaxed during the duration of the 2.5 g interval.

Interestingly, the shape of the grip force switch was asymmetric between ascending and descending g-changes. One reason may be that the second (descending) phase was not entirely novel for participant. This is particularly true for the transition from 2.5 g down to 2 g, since the ascending 2 g phase was still in the recent sensorimotor history. Furthermore, because inertial fluctuations were weak, predicting the weight could have been sufficient to adjust grip force and vestibular afferents are good candidates to allow such prediction of weight even before the first movement in the new environment. It seems that a combination of grip force prediction according to the change in the gravitoinertial environment and a separate safety margin can predict the data quite accurately. A simple linear model that includes (1) a gain factor which reflects the calculation of the grip force change from the load change and (2) a constant magnitude of grip force increase as safety margin approximate the data well. This factor seems however not constant, but may depend on context like ascending or descending g-levels, time in the experiment, or experience with g-changes.

Finally, our data should also be put in the perspective of more theoretical motor control considerations. Despite the very new context, participants never dropped the object. In the presence of such environmental uncertainty, what strategy does the central nervous system adopt to predict a feedforward grip force command in the new phase condition? One approach consists in minimizing the squared error of potential feedforward predictions (Körding et al., 2004), i.e., penalizing too high grip forces. This can be achieved by averaging previous lifts (Scheidt et al., 2012; Hadjiosif and Smith, 2015) or using a Bayesian framework (Körding and Wolpert, 2004). The latter is more flexible as in addition to estimating physical properties linked to the object, it can also build a representation of environment uncertainty. Once both are integrated, a point estimate can be formed. By a genuine manipulation of probability distribution of object masses, a recent study showed that the sensorimotor system indeed uses a minimal squared error

# REFERENCES


strategy to predict grip force (Cashaback et al., 2017). This view is not quite compatible with our results, as it does not explain the switching we observe. Another approach consists in selecting the feedforward prediction that is most likely to be correct. This strategy has been shown to occur in sequential object lifting. When confronted to lift objects of increasing weights, participants expect the next trial to be even heavier (Mawase and Karniel, 2010). Here, participants that were immersed in these gravitoinertial contexts could have formed a reliable representation of the object dynamics in the environment. This could have provided solid information in order to infer a good prediction and therefore a good switch. This is further supported by the fact grip forces were even smaller during the second descending phase. Overall, this view is compatible with the fact both mechanisms are implemented in parallel (Cashaback et al., 2017). However, psychological factors such as stress, could be responsible for the asymmetry observed in the switching between ascending and descending phases. One way to address this would be to perform the same experiment as Mawase and Karniel (2010) but using decreasing weights in the laboratory environment.

# AUTHOR CONTRIBUTIONS

OW, JH, J-LT, and PL designed the experiment. OW, J-LT and JH recorded the data using the human centrifuge. OW analyzed the data. JH, PL, and J-LT discussed the analyses. OW, JH, J-LT, and PL wrote the manuscript.

# ACKNOWLEDGMENTS

This research was supported by the European Space Agency (ESA) in the framework of the Delta-G Topical Team (4000106291/12/NL/VJ), the "Institut National de la Santé et de la Recherche Médicale" (INSERM) and the "Conseil Général de Bourgogne" (France). This work was supported by the German Research Foundation (DFG) and the Technical University of Munich (TUM) in the framework of the Open Access Publishing Program.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 White, Thonnard, Lefèvre and Hermsdörfer. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Alterations in Leg Extensor Muscle-Tendon Unit Biomechanical Properties With Ageing and Mechanical Loading

#### Christopher McCrum1,2 \*, Pamela Leow<sup>1</sup> , Gaspar Epro<sup>3</sup> , Matthias König<sup>3</sup> , Kenneth Meijer <sup>1</sup> and Kiros Karamanidis <sup>3</sup>

<sup>1</sup> Department of Human Movement Science, NUTRIM School of Nutrition and Translational Research in Metabolism, Maastricht University Medical Centre+, Maastricht, Netherlands, <sup>2</sup> Institute of Movement and Sport Gerontology, German Sport University Cologne, Cologne, Germany, <sup>3</sup> Sport and Exercise Science Research Centre, School of Applied Sciences, London South Bank University, London, United Kingdom

#### Edited by:

Nandu Goswami, Medical University of Graz, Austria

#### Reviewed by:

Davide Susta, Dublin City University, Ireland Falk Mersmann, Humboldt-Universität zu Berlin, Germany

\*Correspondence: Christopher McCrum

chris.mccrum@maastrichtuniversity.nl

#### Specialty section:

This article was submitted to Integrative Physiology, a section of the journal Frontiers in Physiology

Received: 30 September 2017 Accepted: 13 February 2018 Published: 28 February 2018

#### Citation:

McCrum C, Leow P, Epro G, König M, Meijer K and Karamanidis K (2018) Alterations in Leg Extensor Muscle-Tendon Unit Biomechanical Properties With Ageing and Mechanical Loading. Front. Physiol. 9:150. doi: 10.3389/fphys.2018.00150 Tendons transfer forces produced by muscle to the skeletal system and can therefore have a large influence on movement effectiveness and safety. Tendons are mechanosensitive, meaning that they adapt their material, morphological and hence their mechanical properties in response to mechanical loading. Therefore, unloading due to immobilization or inactivity could lead to changes in tendon mechanical properties. Additionally, ageing may influence tendon biomechanical properties directly, as a result of biological changes in the tendon, and indirectly, due to reduced muscle strength and physical activity. This review aimed to examine age-related differences in human leg extensor (triceps surae and quadriceps femoris) muscle-tendon unit biomechanical properties. Additionally, this review aimed to assess if, and to what extent mechanical loading interventions could counteract these changes in older adults. There appear to be consistent reductions in human triceps surae and quadriceps femoris muscle strength, accompanied by similar reductions in tendon stiffness and elastic modulus with ageing, whereas the effect on tendon cross sectional area is unclear. Therefore, the observed age-related changes in tendon stiffness are predominantly due to changes in tendon material rather than size with age. However, human tendons appear to retain their mechanosensitivity with age, as intervention studies report alterations in tendon biomechanical properties in older adults of similar magnitudes to younger adults over 12–14 weeks of training. Interventions should implement tendon strains corresponding to high mechanical loads (i.e., 80–90% MVC) with repetitive loading for up to 3–4 months to successfully counteract age-related changes in leg extensor muscle-tendon unit biomechanical properties.

Keywords: Achilles tendon, aged, bed rest, locomotion, quadriceps femoris, patellar tendon, resistance training, triceps surae

# INTRODUCTION

The leg extensor muscle-tendon units (MTUs) play important roles in locomotion, with the muscles opposing gravity and controlling and generating progression by decelerating and accelerating the center of mass and the tendons storing and returning elastic energy to the musculoskeletal system (Biewener and Roberts, 2000; Roberts, 2002; Pandy and Andriacchi, 2010). As a consequence, the tendons can have a large influence on movement effectiveness (Hof et al., 2002; Lichtwark and Wilson, 2007; Pandy and Andriacchi, 2010; Huang et al., 2015). Specifically, the mechanical properties of the Achilles (AT) and patellar (PT) tendons (e.g., tendon stiffness) can greatly influence the contributions of the triceps surae (TS) and quadriceps femoris (QF) to forward propulsion and energy absorption during gait.

In the literature, it is well established that ageing mammalian tendons experience biochemical, cellular, mechanical and pathological alterations, causing progressive deterioration (Noyes and Grood, 1976; Vogel, 1991; Kjaer, 2004; Komatsu et al., 2004). In vitro, the connective tissues of older adults have a declined failure stress compared to younger adults (Noyes and Grood, 1976). In vitro animal studies have associated ageing with an increase in irreducible collagen cross-linking, a reduction in collagen fibril diameter and its crimp angle, an increase in more extensible elastin content, a reduction in extracellular water and glycosaminglycans content and an increase in collagen type V (Vogel, 1991; Nakagawa et al., 1994; Tuite et al., 1997; Dressler et al., 2002; Kjaer, 2004). These changes may lead to altered biomechanical properties of tendons in vivo, which in turn, could affect the overall function of the MTUs. In addition to biological changes, altered environmental mechanical stress may influence ageing tendon. Unloading of the leg extensor MTUs can occur due to immobilization and inactivity, and can lead to muscle atrophy. In vivo studies have demonstrated that chronic inactivity (20–90 days bed rest) results in a reduction in tendon stiffness (Kubo et al., 2004; Reeves et al., 2005). Collectively, this means that ageing tendon is affected not only by processes of biological ageing per se, but also the reduced habitual loading due to decreased physical activity and muscle strength (Iannuzzi-Sucich et al., 2002; Lauretani et al., 2003).

Since tendon is a mechanosensitive and adaptive tissue, its properties can change depending on its exposure to mechanical loading (Tkaczuk, 1968; Butler et al., 1978; Woo et al., 1980, 1982). Such changes are believed to be regulated through mechanotransduction (Chiquet et al., 2009). Mechanical load generated by the muscle contractions deforms the tendinous tissue, whereby the resultant tendon strain is transferred to its cellular cytoskeleton via the extracellular matrix, causing structural changes (Wang, 2006) and various molecular responses (Robbins and Vogel, 1994; Pins et al., 1997; Arnoczky et al., 2002; Yang et al., 2004; Olesen et al., 2006). These responses have been linked to modifications in tendon mechanical properties following long-term mechanical loading (Kjaer, 2004; Wang, 2006; Heinemeier and Kjaer, 2011; Galloway et al., 2013).

For older adults in particular, the capacities of the legextensor MTUs are highly relevant for locomotion. Karamanidis and Arampatzis (2007) and Karamanidis et al. (2008) found significant associations between leg extensor MTU mechanical properties (i.e., TS and QF muscle strength and PT stiffness) and stability control following sudden release from a forwardinclined body position and Onambele et al. (2006) reported that AT stiffness was a decisive predictor of single leg stance ability. Additionally, Stenroth et al. (2015) found that lower AT stiffness was associated with slower timed "up and go" test and 6 min walk test results among healthy older adults. Interventions to counteract age-related changes in the leg extensor MTU mechanical properties may have the potential to positively influence the safety and effectiveness of human locomotion.

This review aims to examine age-related differences in human leg extensor MTU biomechanical properties and if changes in these properties can be counteracted in older adults. Therefore, we provide an overview of recent literature examining agerelated differences in human leg extensor MTU biomechanical properties in young and older healthy adults, including muscle strength and the mechanical (tendon stiffness), morphological (tendon cross sectional area: CSA) and material (Young's modulus of the tendon) properties of the AT and PT. Secondly, mechanical loading interventions to trigger alterations in these properties in older adults are reviewed, in order to determine if and to what extent age-related changes can be counteracted and if particular criteria for successful interventions exist.

# AGE-RELATED CHANGES IN HUMAN MUSCLE-TENDON UNIT BIOMECHANICAL PROPERTIES

# Muscle Strength

As the leg extensor MTUs are comprised of muscular and tendinous tissue, any alterations in the tendon biomechanical properties must be interpreted in parallel with changes in the muscle. Twelve articles discussed in this review that examined the tendon biomechanical properties also reported muscle strength [determined during maximum voluntary contractions (MVC) and reported in kg, N, Nm or body weight normalized values]. Seven of the 12 articles analyzed the TS and six assessed the QF MTU. The median number of older adults assessed in the studies was 11 (range of 6–67), with mean ages from the studies ranging from 64 to 79 years. Overall, the age-related changes in muscle strength ranged from −52 to −26.4% for the TS and −29.3 to −1.4% for the QF with an overall median of −29% (**Figure 1**).

# Tendon Stiffness

Tendon stiffness describes the force-elongation relationship of the tendon, assessed in the linear region of the tendon forceelongation relationship. Eleven articles that examined age-related differences in tendon stiffness are discussed in this review (**Figure 1**). Seven of the articles analyzed the AT and six assessed the PT. The median number of older adults assessed in the studies was 12 (range of 6–67), with mean ages from the studies ranging from 64 to 79 years. Overall, the age-related differences

in tendon stiffness ranged from −55 to −3.9% for the AT and −31.2 to −2.4% for the PT with an overall median of −20.3% (**Figure 1**). One study (Csapo et al., 2014) used combined MRI and dynamometry to assess tendon stiffness, whereas all other articles employed synchronized ultrasonography and dynamometry (e.g., Mademli and Arampatzis, 2008). Despite the large range in percentage differences, which may have been a result of methodological differences affecting the assessment of tendon elongation and stiffness such as imaging method (i.e., Csapo et al., 2014), contraction protocol (Kösters et al., 2014; McCrum et al., 2017) or technical differences between the studies (Finni et al., 2013; Seynnes et al., 2015), the literature shows a consistent reduction in tendon stiffness with age.

# Tendon Cross Sectional Area

Ten of the discussed studies examined tendon CSA, six of which assessed the AT, with the other four analyzing the PT (**Figure 1**). Four of the six studies that analyzed the AT used ultrasonography (Onambele et al., 2006; Stenroth et al., 2012, 2015; Tweedell et al., 2016), with the other two using MRI (Magnusson et al., 2003; Csapo et al., 2014), whereas all of the studies examining the PT used MRI to assess the CSA (Carroll et al., 2008; Couppé et al., 2009, 2012, 2014). The median number of older adults included in the studies was 15 (range of 6–67), with a range of mean ages from the studies of 65–79 years. Two of the studies found significantly smaller tendon CSA in the older adults (−18.7 and −7.8%; Onambele et al., 2006; Carroll et al., 2008), while four found significantly greater CSA (Magnusson et al., 2003; Stenroth et al., 2012, 2015; Couppé et al., 2014; Tweedell et al., 2016), with an overall median of 9.1% greater tendon CSA in the older adults (**Figure 1**). As the accuracy of ultrasound-based methods for determining both AT and PT CSA has been shown to be insufficient (Ekizos et al., 2013; Bohm et al., 2016), we suggest that more weight should be given to studies which have used MRI to determine tendon CSA and Young's modulus. If only MRI studies are taken into account, the median difference drops to 4.6%. Aside from imaging methodology, there is variation in how CSA was determined. Most AT studies assessed CSA at a specific tendon length (usually where the CSA is assumed to be smallest), which varied between three and four cm proximal to the insertion of the AT to the calcaneus (Magnusson et al., 2003; Stenroth et al., 2012, 2015; Csapo et al., 2014). The remaining AT and PT studies used multiple (usually three) lengths from which the CSA was averaged. As a result, potential region-specific differences in the CSA between younger and older adults may be excluded, as no existing study has compared the AT or PT CSA between older and younger adults along the entire tendon length; a potentially important gap in the literature. Traininginduced regional changes have previously been reported in young adults (Magnusson and Kjaer, 2003; Arampatzis et al., 2007; Kongsgaard et al., 2007; Seynnes et al., 2009). Due to the diversity in methodologies and results, no firm conclusion can be made about the age-effects on tendon CSA.

# Tendon Young's Modulus

The Young's modulus of a material is defined as the slope of the stress-strain relationship, where stress is tendon force relative to CSA and strain is tendon elongation in relation to resting length. Nine of the discussed studies assessed the tendon Young's modulus, with four and five studies analyzing the AT and PT, respectively. The median number of older adults included was 19 (range of 6–67), with a range of mean ages from 64.5 to 76.7 years. One of the four studies of the AT (Csapo et al., 2014) and all but one (Hsiao et al., 2015) of the five studies of the PT used MRI to assess the tendon CSA, with the others using ultrasound to assess tendon CSA. All but one of the studies used synchronized dynamometry and ultrasound to assess the force-elongation behavior of the tendon, with the final study (Csapo et al., 2014) using combined MRI and dynamometry. A median difference of −27.8% in Young's modulus (−23.9% when only including the MRI-based studies) was found, with no studies showing a higher Young's modulus in older, compared to younger adults (**Figure 1**). There is a relatively consistent reduction in Young's modulus with age, although the above described limitations regarding region-specific CSA should also be kept in mind. Thus, we can conclude that the observed changes in tendon stiffness due to ageing are predominantly due to changes in tendon material properties rather than reduced CSA.

# EFFECTS OF INCREASED MECHANICAL LOADING ON MUSCLE-TENDON UNIT BIOMECHANICAL PROPERTIES IN OLDER ADULTS

In this section, we provide an overview and discussion of intervention studies conducted with older adults (mean age of 60 years or older) that analyzed the leg extensor MTUs' biomechanical properties. Nine intervention groups from six articles are discussed (**Figure 2**; Reeves et al., 2003a,b; Onambele-Pearson and Pearson, 2012; Grosset et al., 2014; Karamanidis et al., 2014; Epro et al., 2017). All interventions consisted of predominantly resistance-based exercise and lasted 12–14 weeks in length, with one study also conducting a 1.5 year long intervention (Epro et al., 2017). All conducted two or three sessions per week that were partly or completely supervised. The exercise protocols ranged from highly specific and controlled protocols (i.e., five sets and four repetitions of isometric plantar flexions at 90% MVC held for 3 s guided by visual feedback in Epro et al., 2017) to more mixed ecological training interventions with multiple strength exercises, as well as hopping or running (Onambele-Pearson and Pearson, 2012; Grosset et al., 2014; Karamanidis et al., 2014). The number of contractions per exercise ranged from 16 to 44 spread over a range of two to five sets, and all but one of the interventions in the study of Grosset et al. (2014) aimed to impose high mechanical loads (e.g., 80– 90% MVC or 80% of five repetition maximum for 10 repetitions). Grosset et al. (2014) compared low and high intensity training groups (40% vs. 80% MVC) and Onambele-Pearson and Pearson (2012) compared male and female groups of older adults. One article (two intervention durations) focused exclusively on the AT (Epro et al., 2017), four articles (six intervention groups) focused exclusively on the PT (Reeves et al., 2003a,b; Onambele-Pearson and Pearson, 2012; Grosset et al., 2014) and one article conducted an intervention targeting both the TS and QF MTUs (Karamanidis et al., 2014). One study did not report muscle strength values (Grosset et al., 2014) but all other articles reported significant muscle strength increases (**Figure 2**) (13.4–25.5% for the TS and 9.2–25.4% for the QF), measured either by maximum joint moments, maximum force during isometric contractions or by one or five repetition maximum values (Reeves et al., 2003a,b; Onambele-Pearson and Pearson, 2012; Karamanidis et al., 2014; Epro et al., 2017). All but one intervention resulted in significant increases in AT (19.6–22.5%) or PT (10.1–82.5%) stiffness (**Figure 2**), with the one non-significant result coming from the low intensity (40% MVC) training group of Grosset et al. (2014). Young's modulus of the tendons was assessed by four of the articles, generally showing significant increases in both the AT (19–22%) and PT (9.5–68.4%), with the low intensity group of Grosset et al. (2014) showing no change (**Figure 2**). Tendon CSA was also assessed by four of the articles, three of which assessed the PT and found no differences post-intervention (Reeves et al., 2003a; Onambele-Pearson and Pearson, 2012; Grosset et al., 2014) and one reported significant increases in AT CSA after both 14 weeks and 1.5 years of intervention (Epro et al., 2017). It is noteworthy that Epro et al. (2017) analyzed the CSA over the entire length of the AT using MRI, whereas the other studies used ultrasound and did not assess the entire length of the AT. This may suggest that the ultrasound method is not sensitive enough to consistently detect the usual range of changes in tendon CSA following exercise interventions (a range of 3.7–9.6% from the studies using MRI reported in the review of Bohm et al., 2015 and in Epro et al., 2017).

Older tendons appear to preserve their adaptability to mechanical loading with age (Reeves et al., 2003a,b; Onambele-Pearson and Pearson, 2012; Grosset et al., 2014; Karamanidis et al., 2014; Epro et al., 2017), but a few results are worth noting when considering the effectiveness of the interventions. Firstly, Epro et al. (2017) found that 14 weeks of resistance exercise was a sufficient time period to trigger adaptive changes in the biomechanical properties of the AT and that these adaptations were maintained for 1.5 years by continuing training, suggesting that there is a non-linear time-response relationship of ageing tendons subjected to mechanical loading. However, the lack of further adaptation may have been related to a plateau in plantarflexion MVC force after 11–12 weeks of training (Epro et al., 2017). These long-term adaptation processes should be investigated in future research. Secondly, the intervention with the lowest exercise intensity (Grosset et al., 2014) and therefore, lowest tendon strain magnitudes, was the only intervention that found no significant changes in the tendon biomechanical properties. This finding is in accordance with evidence from young adults, demonstrating that tendon adaptation is triggered only when a specific threshold of strain magnitude is exceeded during the loading exercise (Arampatzis et al., 2007; Bohm et al., 2015). This might also explain the absence of differences in lower limb MTU biomechanical properties in a number of cross-sectional studies of older endurance runners and their agematched sedentary counterparts (Karamanidis and Arampatzis, 2005, 2006). Future research should continue to explore viable activities for stimulating tendon adaptation in older adults.

In young adults, studies have reported increased tendon stiffness, CSA and Young's modulus in response to tendon loading exercise over 12–14 weeks (Kubo et al., 2001; Arampatzis et al., 2007; Kongsgaard et al., 2007; Bohm et al., 2015; Wiesinger et al., 2015). The importance of strain magnitude for tendon adaptation was originally demonstrated by Arampatzis et al. (2007) and two systematic reviews concluded that resistance training can lead to tendon adaptation providing that sufficient tendon strain magnitudes (or intensities greater than 70% MVC)

are applied (Bohm et al., 2015; Wiesinger et al., 2015). This is in agreement with the results found in the current review in the study of Grosset et al. (2014). Considering the results in older adults, the adaptation magnitudes in stiffness and Young's modulus are similar to those observed after 12–14 weeks exercise in younger adults (increases of 16–36% and 15–45% respectively; Arampatzis et al., 2007, 2010; Kubo et al., 2007; Fletcher et al., 2010; Fouré et al., 2013; Bohm et al., 2014). Moreover, the changes in tendon CSA values after exercise are consistent with younger adults (mean AT CSA increases of between 0.5 and 10%; Arampatzis et al., 2007, 2010; Kongsgaard et al., 2007; Bohm et al., 2014). Importantly, the cyclic tendon strain exercise protocol of Epro et al. (2017) in older adults was the same as Arampatzis et al. (2007) in young adults and relative adaptations in the TS MTU biomechanical properties were similar. Overall, it appears that the leg extensor MTUs of older adults respond to increased mechanical loading in a way that involves similar magnitudes of tendon and muscle adaptation.

# CONCLUSION

Based on the available literature, increasing age appears to result in reductions in human TS and QF muscle strength accompanied by reductions in AT and PT stiffness and elastic modulus, whereas the effect on AT and PT CSA is unclear. Therefore, the observed changes in tendon stiffness due to ageing are predominantly due to changes in tendon material properties rather than changes in tendon CSA. However, tendons appear to retain their mechanosensitivity with age, showing similar alterations in their biomechanical properties in older adults compared to younger adults following training interventions. Exercise interventions should implement tendon strains corresponding to high, repetitive mechanical loading (i.e., 80–90% of MVC) for up to 3 or 4 months in order to successfully counteract age-related changes in leg extensor MTU biomechanical properties.

# AUTHOR CONTRIBUTIONS

Conception of the work: CM and PL; literature acquisition: MK, GE, PL, and CM; literature synthesis: PL, CM, and MK; analysis and interpretation: all authors; drafted the manuscript: CM; prepared figures: CM, PL, and MK; revised the manuscript for important intellectual content: all authors; final approval of the version to be published: all authors; agreement to be accountable for the work: all authors.

# FUNDING

CM was funded by the Kootstra Talent Fellowship awarded by the Centre for Research Innovation, Support and Policy (CRISP) and by the NUTRIM Graduate Programme, both of Maastricht University Medical Center+. MK was supported by the German Social Accident Insurance (Deutsche Gesetzliche Unfallversicherung, Postgraduate Scholarship).

# REFERENCES


properties of tendon structures in the lower limb. Scand. J. Med. Sci. Sports 14, 296–302. doi: 10.1046/j.1600-0838.2003.00368.x


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 McCrum, Leow, Epro, König, Meijer and Karamanidis. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# High-Resolution X-Ray Tomography: A 3D Exploration Into the Skeletal Architecture in Mouse Models Submitted to Microgravity Constraints

Alessandra Giuliani <sup>1</sup> \*, Serena Mazzoni <sup>1</sup> , Alessandra Ruggiu<sup>2</sup> , Barbara Canciani <sup>2</sup> , Ranieri Cancedda<sup>2</sup> and Sara Tavella<sup>2</sup>

<sup>1</sup> Sezione di Biochimica, Biologia e Fisica Applicata, Dipartimento di Scienze Cliniche Specialistiche e Odontostomatologiche, Università Politecnica delle Marche, Ancona, Italy, <sup>2</sup> Dipartimento di Medicina Sperimentale, Universita' di Genova and Ospedale Policlinico San Martino, Genova, Italy

#### Edited by:

Jack van Loon, VU University Amsterdam, Netherlands

#### Reviewed by:

Giovanna Valenti, Università degli studi di Bari Aldo Moro, Italy Daniel Donner, Baker Heart and Diabetes Institute, Australia

> \*Correspondence: Alessandra Giuliani a.giuliani@univpm.it

#### Specialty section:

This article was submitted to Integrative Physiology, a section of the journal Frontiers in Physiology

Received: 19 October 2017 Accepted: 20 February 2018 Published: 06 March 2018

#### Citation:

Giuliani A, Mazzoni S, Ruggiu A, Canciani B, Cancedda R and Tavella S (2018) High-Resolution X-Ray Tomography: A 3D Exploration Into the Skeletal Architecture in Mouse Models Submitted to Microgravity Constraints. Front. Physiol. 9:181. doi: 10.3389/fphys.2018.00181 Bone remodeling process consists in a slow building phase and in faster resorption with the objective to maintain a functional skeleton locomotion to counteract the Earth gravity. Thus, during spaceflights, the skeleton does not act against gravity, with a rapid decrease of bone mass and density, favoring bone fracture. Several studies approached the problem by imaging the bone architecture and density of cosmonauts returned by the different spaceflights. However, the weaknesses of the previously reported studies was two-fold: on the one hand the research suffered the small statistical sample size of almost all human spaceflight studies, on the other the results were not fully reliable, mainly due to the fact that the observed bone structures were small compared with the spatial resolution of the available imaging devices. The recent advances in high-resolution X-ray tomography have stimulated the study of weight-bearing skeletal sites by novel approaches, mainly based on the use of the mouse and its various strains as an animal model, and sometimes taking advantage of the synchrotron radiation support to approach studies of 3D bone architecture and mineralization degree mapping at different hierarchical levels. Here we report the first, to our knowledge, systematic review of the recent advances in studying the skeletal bone architecture by high-resolution X-ray tomography after submission of mice models to microgravity constrains.

Keywords: high-resolution tomography, bone microarchitecture, synchrotron radiation, microgravity, animal model, mice

# INTRODUCTION

Bone is considered one of the most complex tissues in the body because of the continuous remodeling process that it undergoes in physiological conditions. This occurs not only in order to bear mechanical loading, but also to inhibit the damage consequent to fatigue demand. Furthermore, remodeling acts in repairing fractures, in achieving the viability of the osteocytes and calcium homeostasis.

Bone cells act synergistically, increasing or decreasing bone mass based on several factors: the

osteoblasts, the bone-forming cells, regulate the deposition of the bone matrix molecules, including type I collagen and a variety of other non-collagenous proteins; the osteoclasts, multinucleated giant cells, are responsible for the mineralized bone matrix resorption (Tavella et al., 2012).

Normally, the remodeling process consists in a slow building phase and faster resorption with the objective to maintain a functional skeleton locomotion to counteract the Earth gravity. Thus, when exercise and movements are reduced, as it happens in bed-rest prolonged conditions or during spaceflights, the skeleton does not act against gravity, with a rapid decrease of bone mass and density, rendering bone fragile (Zhang et al., 2008).

The bone architecture and its remodeling were traditionally studied by X-ray radiography, but this method presents several important limitations, like misleading superimposition of anatomic structures.

In the 1970s, the development of the first equipment for computed tomography (CT), capable of producing threedimensional (3D) virtual reconstructions of objects, in a nondestructive way and with contrast discrimination up to 10<sup>3</sup> times better than conventional radiographs (Claesson, 2001), revolutionized and gave new stimuli to a more in-depth study of the bone structure.

Computed microtomography (microCT) is based on the same physical and methodologic principles of conventional CT currently used for medical imaging diagnostics. While the CT systems typically have a maximum linear resolution of about 500µm, some microCT devices reach a spatial resolution up to 0.3µm (Cancedda et al., 2007; Rominu et al., 2014), with an increase of about three orders of magnitude.

Subsequently, allowing accurate non-destructive 3D examination of objects, the microCT was used to reconstruct the complex architecture of bone tissue at a high resolution. In the field of bone research, different methods have now been developed to extract quantitative architectural parameters from the microCT images. For instance, the 3D mean intercept length (MIL) method is able to measure the trabecular thickness and spacing based on structural geometry assumptions (Hildebrand and Ruegsegger, 1997a). It is also feasible to skip these assumptions, extracting parameters that are model independent (Hildebrand and Ruegsegger, 1997b).

In this scenario, synchrotron radiation (SR) has proved to be of critical relevance in microCT investigations because of its specific features, including the high signal-to-noise ratio, a high photon flux to achieve measurements at high spatial resolution, and the possibility to tune the beam energy, avoiding beam hardening effects.

SR-microCT successfully supported several studies of 3D bone architecture and of mineralization degree mapping at different hierarchical levels (Nuzzo et al., 2002; Bousson et al., 2004; Lane et al., 2005), including also the imaging of the lacuno-canalicular network (Langer et al., 2012; Peyrin et al., 2014). Moreover, SR-based microCT was also employed to reconstruct, at high resolution, the complex architecture of bone tissue in different genetic/environmental conditions (Martín-Badosa et al., 2003; Costa et al., 2013; Hesse et al., 2014), and it is also increasingly becoming a powerful tool for the engineered bone characterization in several areas of skeletal research (Cancedda et al., 2007; Giuliani et al., 2013, 2014; Mazzoni et al., 2017).

However, these recent advances in microCT physics/technology have further stimulated the study of weight-bearing skeletal sites by novel imaging approaches, not only in the validated acquisition of functional images but also to investigate degenerative diseases with induced site-specific bone loss, for instance resulting from weightless environment during spaceflight.

Here we report the first, to our knowledge, systematic review of the recent advances in studying the skeletal bone architecture by high-resolution tomography after submission of mice animal models to microgravity constrains.

# APPLYING MICROGRAVITY CONSTRAINTS TO BONE EVALUATED BY HIGH-RESOLUTION X-RAY TOMOGRAPHY

The growing interest in this specific research started after the Gemini, Apollo, and Skylab missions, when the astronauts experienced a strong bone demineralization coupled to an increased calcium excretion, with negative consequences on bone turnover comparable to those on subjects subjected to long bed-rest (Wronski and Morey, 1983).

Indeed, bed-rest is the most similar condition on Earth to reduced-gravity on the skeleton, with impaired ratio between bone formation and resorption, causing an accelerated bone loss by an increased osteoclast activity that has been demonstrated in different observations (Donaldson et al., 1970; Leblanc et al., 1990, 1995).

Studies on cosmonauts aboard the Russian MIR space station confirmed the previous findings with a significant bone loss in the weight-bearing tibia and unaltered bone mass in the non-weightbearing radius (Vico et al., 2000).

Analogous analysis of crewmembers after 4- to 6-month flights on the International Space Station (ISS) provided morphometric data on cortical and trabecular bone in spine and hip sites using quantitative computed tomography (QCT). It was observed that there was no compartment-specific loss of bone in the spine and that the cortical bone mineral loss in the hip was mainly due to endocortical thinning (Lang et al., 2004).

However, the weaknesses of the previously reported studies was two-fold: on the one hand the research suffered the small sample size of almost all human spaceflight studies, on the other the results were affected by the partial volume averaging on cortical bone measurements, due to the fact that the observed structures were small compared with the spatial resolution of the imaging device (Lang et al., 2004).

These reasons have led to the growing use of animal models, ranging from rats (Cosmi et al., 2009; Keune et al., 2015) to fish (Chatani et al., 2015), to increase the sample size by providing more significant statistical data, and to increasingly perform more informative microCT investigations, being a highresolution 3D imaging technique able to study smaller bone structures.

# Mice in Space

Most studies on microgravity have been carried out using the mouse and its various strains. These experiments have been conducted in space aboard the Space Shuttles and, for each mission, ever more sophisticated animal habitation cages have been developed. Indeed, environmental conditions are expected to affect physiology and behavior of mice both on Earth and in Space. Blottner (Blottner et al., 2009) analyzed the effects of cage confinement on the weight-bearing musculoskeletal system of 24 wild-type C57BL/6JRj mice housed for 25 days in the MIS (Mice In Space) habitat prototype, which was ground-based and fully automated. The system was a portion of the MSRM1 (Mouse Science Reference Module) device produced by Alcatel Alenia Space Inc. (Milano, Italy). As determined by SR-microCT, compared with the mice individually housed in control ventilated cages, the MIS mice revealed no significant changes in either 3D microarchitecture or mineralization degree in any of the investigated bone sites (calvaria, spine, and femur).

Three missions have provided the first documented data on in-flight mice skeletal changes taking advantage of the use of high-resolution X-ray microCT: the 12-day shuttle mission (STS-108) with 2-month-old C57BL6/J female mice, the 15-day shuttle mission (STS-131) with 16-week-old female C57BL/6J mice, and the 91-day mission aboard the ISS with 2-month-old wild-type C57BLJ10 (WT) and pleiotrophin-transgenic (PTN-Tg) male mice.

In the first experiment, Lloyd (Lloyd et al., 2015) tested the ability of Osteoprotegerin-Fc (OPG-Fc) to preserve bone mass during the spaceflight (SF). Twelve mice per group were injected, 24 h prior to launch, with OPG-Fc or an inert vehicle (VEH). Ground control (GC) mice (VEH and OPG-Fc) were kept in environmental conditions mimicking those in the space shuttle, while the age-matched baseline (BL) controls were sacrificed before the launch. MicroCT (µCT20; Scanco Medical AG; Brüttisellen, Switzerland) was used to investigate the trabecular bone architecture with the following experimental parameters: pixel size of 9µm, scan settings of 55 KVp, 145 mA, and 200 ms per projection (pp). The scanning sites were the trabecular portion immediately distal to the growth plate in the tibiae and humerus. The trabecular bone parameters included: trabecular bone volume fraction (BV/TV); connectivity density (Conn.D); trabecular number (Tb. N); trabecular separation (Tb.Sp); and (SMI). In the tibia, the BV/TV of SF/VEH was 26% lower than GC/VEH, while the Conn.D was 27% lower (although not significantly), the Tb.Th was 16% lower, and the SMI was 6% greater. The spaceflight did not produce modification on the same parameters when SF/OPG-Fc mice were compared to GC/OPG-Fc. Both BV/TV and Conn.D were not changed by spaceflight in the humerus site. In synthesis, this microCT study showed that a single treatment with OPG-Fc before the flight efficiently prevented the detrimental effects of microgravity on mouse bone.

In the second experiment, Blaber (Blaber et al., 2013) exposed eight mice to microgravity to test if osteocytic osteolysis, and cell cycle stopping during osteogenesis may contribute to bone resorption in microgravity conditions. MicroCT (SkyScan 1174 microCT scanner, Kontich, Belgium) was used to image and quantify bone morphometry of the ischium region in the right coxa. Images were acquired with the following experimental parameters: pixel size of 6.77µm, scan settings of 50 KV, 800 mA, and 3.5 s per projection (pp). These analysis of the pelvis showed that microgravity induced a decrease in BV/TV of 6.29%, and in bone thickness (B.Th) of 11.91%, without reducing the bone mineral density (BMD).

Afterwards, during the Italian Mice Drawer System (MDS) mission, six mice were exposed for 91 days to microgravity on the ISS (Cancedda et al., 2012). This spaceflight is, to date, the longest one ever experimented: for this reason, it has provided a broad range of results, including insights of the PTN transgene possible protection against bone loss due to microgravity (Tavella et al., 2012). SR-microCT imaging was performed at the SYRMEP Beamline of the ELETTRA Synchrotron Radiation Facility (Trieste, Italy), using a beam energy of 19 keV over 180◦ and with a resulting pixel size of 9µm. Investigations were focused for the weight-bearing sites on the lower third of the left femurs from the patella toward the shaft of the femur (in the trabecular and cortical portions), and onto the vertebral body in the VII lumbar ring. In non-weight-bearing bones, the analysis was restricted to the middle of the parietal bone left portion (transverse direction from the sagittal suture to the border). SR-microCT analysis revealed a bone loss during spaceflight in the weight-bearing bones of both WT and PTN strains, with a decrease of the trabecular number (Tb.Nr) as well as an increase of the trabecular separation (Tb.Sp) after flight (**Figure 1**). Non-weight-bearing bones were shown to be not affected by microgravity constrains.

The Bion-M1 mission offered another opportunity to characterize, by microCT, the skeletal changes in adult (23 weeks-old) male C57/BL6 mice after 30-day spaceflight and an 8 day recovery period (Gerbaix et al., 2017). Like the MDS mission, two ground control groups were included in the protocol: a Habitat Control group, which was kept in the same spacecraft cages; and a Control group, which was kept in standard cages. All left femurs, L3, and T12 vertebrae (5/6 animals per group) were scanned with a high-resolution microCT device (VivaCT40, Scanco Medical, Bassersdorf, Switzerland), using a pixel size of 12.5µm and an analysis protocol previously adopted and described by (David et al., 2003). Comparing the ground control groups, the spacecraft cage confinement was found to negatively affect the femur and lumbar vertebrae (but not the thoracic vertebrae): indeed, the L3 vertebrae and femur trabecular BV/TV and Conn.D were decreased in the Habitat Control vs. the Control group. Moreover, the trabecular BV/TV of the L3 vertebrae decreased in the Flight group vs. both the Habitat Control and Control groups (−35.7 and −56.5%, respectively; p < 0.0033): this was attributed to decreased Tb.N and Tb.Th. Flight group femurs showed a large loss of trabecular BV/TV when compared to both the Control and Habitat Control groups (−85.2%, p < 0.0003; −64.8%, p < 0.017; respectively). Similar trabecular parameters were found in the Flight + Rec and Flight

groups, indicating that 1 week of restored gravity is not sufficient to start bone structure recovery.

In the same study (Gerbaix et al., 2017), five cortical femur sections per group were also investigated by SR-microCT, at the ID19 beamline of the European Synchrotron Radiation Facility (Grenoble, France), with a 0.7µm pixel size, 2000 projections over a total angle of 360◦ and 26 keV beam energy. The 3D analysis was focused on the several thousands of osteocyte lacunae, getting data on total lacunar volume (Lc.V, mm<sup>3</sup> ), lacunar number density (N.Lc/TV) and lacunar volume density Lc.V/TV (%). Being the shape of an osteocyte lacunae approximate to an ellipsoid, the authors used the second-order moments to efficiently measure the lengths of the main axis of the best fitting ellipsoid. The canal volume fraction (Ca.V/TV, %) was also determined. This sophisticated analysis showed that, in the Flight animal group, the osteocyte lacunae had smaller volume with more spherical shape. Furthermore, the number of empty lacunae were significantly (+344%) increased respect to the Habitat Control group. These data demonstrated that microgravity caused osteocyte death, possibly responsible of bone resorption with the consequent bone mass loss.

The 30-day Bion-M1 mission also offered the chance to study six male C57BL/6 mice (19–20 weeks old). The animals were sacrificed 13–15 h after landing. Eight control mice were kept on the ground during the same 30-day period in standard vivarium cages. Caudal motion segments from flight and vivarium mice were loaded to failure in four-point bending (Berg-Johansen et al., 2016). After this test, the same motion segments were imaged by microCT (µCT 50, SCANCO Medical, Brüttisellen, Switzerland) to quantify trabecular microarchitecture and BMD.

published in Figure 3 of (Tavella et al., 2012).


TABLE 1 | Main spaceflight mission experiences: parameters and results on the mice skeletal bone architecture studied by high-resolution tomography (microCT).

\*Lloyd et al., 2015; \*\*Blaber et al., 2013; §Tavella et al., 2012; #Gerbaix et al., 2017; ##Berg-Johansen et al., 2016; <sup>+</sup>Shiba et al., 2017.

Motion segments scanned using a 4µm pixel size and the following scanning parameters: 55 kVp and 109 µA.

It was observed that spaceflight significantly reduced vertebral BV/TV, BMD, and Tb.Th., possibly explaining the tendency of flight specimens to fail within the epiphyseal bone. These microCT results, combined with the mechanical bending tests data, indicate that vertebral bone loss during spaceflight may compromise spine bending capacity, contributing to increased disc herniation risk in astronauts.

However, despite the novelty of the results of the two missions previously described (the 91-day MDS and the 30-day Bion-M1), their experimental protocols appeared deficient in the management of the control mice that were located on the ground instead of in space. Indeed, this choice may have affected the results, preventing reliable conclusions on the impact of microgravity on rodents. More specifically, locating control mice on the ground instead of in space disregarded some factors, like cosmic radiation, microbial environment, flight vibration, shock, and semi-steady acceleration during the launch and return phases inside the space vehicle.

Thus, Shiba (Shiba et al., 2017) developed a novel experimental platform to generate artificial gravity in space. This study was conducted in the framework of a Japan Aerospace Exploration Agency (JAXA) project focused on elucidating the impacts of partial gravity (partial g) and microgravity (µg) on mice using newly developed mouse habitat cage units (HCU) that were implemented in the Centrifuge-equipped Biological Experiment Facility in the ISS. In the first mission of the project, 12 five-week-old C57BL/6 J male mice were housed under µg or artificial earth-gravity (1 g). µg mice floated inside the HCU, whereas artificial 1 g mice were on their feet (on the bottom floor of the HCU, at a centrifugation 77 rpm). After 35 days, all mice were returned to the Earth and investigated. The right femurs of mice were analyzed by microCT, using a ScanXmate-A100S Scanner (Comscantechno,Yokohama, Japan). MicroCT revealed that artificial gravity loading significantly reduced bone loss induced by microgravity conditions. Significant decreases were evident in femur bone density of µg mice, whereas artificial 1 g mice maintained the same bone density as mice in the ground control experiment, in which housing conditions in the flight experiment were replicated. These data further confirmed that the previous experiments' bone loss was specifically due to changes in gravity.

# CONCLUSIONS AND PERSPECTIVES

Most of the data from these mice studies were in agreement with those previously obtained from human studies, confirming the utility of the mouse model in the spaceflight investigations. Despite this, the results should be analyzed with caution and attention, considering the specific experimental conditions, such as animal age, sex, body weight, pregnancy, and other variables, possibly influencing the study output (Tavella et al., 2012).

Indeed, as summarized in **Table 1**, the materials (mouse strain, age, sex, bone site, etc.) and methods (flight duration, animal

# REFERENCES


management, microCT resolution and settings, etc.) reported in literature differed also considerably among themselves. This fact constitutes an incentive to continue research on animal models submitted to spaceflight experience, on the one hand standardizing the survey parameters, on the other by deepening the studies of bone superstructure by nanotomography, possibly based on synchrotron radiation (Langer et al., 2012; Pacureanu et al., 2012; Hesse et al., 2015), as already explored by Maude (Gerbaix et al., 2017) after the 30-day Bion-M1 mission.

Moreover, the possibility to simulate an artificial loading on the musculoskeletal system, as in hypergravity conditions (Canciani et al., 2015), represents an undoubted advantageous tool for scientific research, since it may allow researchers to avoid the complexity of real microgravity studies and makes possible the investigation of a larger number of subjects, thus improving the power of such studies.

# AUTHOR CONTRIBUTIONS

AG: Concept and design; revision of the whole literature; coordination of the work drafting; final version definition and approval. SM: Design (high-resolution tomography); data research in literature on microCT; work drafting; final version approval. AR and BC: Design (physiology concepts); data research in literature on physiological interpretation of data; work drafting; final version approval. RC and ST: Concept and design; data research in literature on spaceflight Projects; work drafting; final version definition and approval. All the authors agreed to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.


tissue-level mechanical properties, osteocyte survival and lacunae volume in mature mice skeletons. Sci. Rep. 7:2659. doi: 10.1038/s41598-017-03014-2


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Giuliani, Mazzoni, Ruggiu, Canciani, Cancedda and Tavella. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Muscle Atrophy Induced by Mechanical Unloading: Mechanisms and Potential Countermeasures

Yunfang Gao<sup>1</sup> \*, Yasir Arfat <sup>1</sup> , Huiping Wang<sup>1</sup> and Nandu Goswami <sup>2</sup>

*<sup>1</sup> Key Laboratory of Resource Biology and Biotechnology in Western China, College of Life Sciences, Ministry of Education, Northwest University, Xi'an, China, <sup>2</sup> Physiology Unit, Otto Loewi Center of Research for Vascular Biology, Immunity and Inflammation, Medical University of Graz, Graz, Austria*

Prolonged periods of skeletal muscle inactivity or mechanical unloading (bed rest, hindlimb unloading, immobilization, spaceflight and reduced step) can result in a significant loss of musculoskeletal mass, size and strength which ultimately lead to muscle atrophy. With advancement in understanding of the molecular and cellular mechanisms involved in disuse skeletal muscle atrophy, several different signaling pathways have been studied to understand their regulatory role in this process. However, substantial gaps exist in our understanding of the regulatory mechanisms involved, as well as their functional significance. This review aims to update the current state of knowledge and the underlying cellular mechanisms related to skeletal muscle loss during a variety of unloading conditions, both in humans and animals. Recent advancements in understanding of cellular and molecular mechanisms, including IGF1-Akt-mTOR, MuRF1/MAFbx, FOXO, and potential triggers of disuse atrophy, such as calcium overload and ROS overproduction, as well as their role in skeletal muscle protein adaptation to disuse is emphasized. We have also elaborated potential therapeutic countermeasures that have shown promising results in preventing and restoring disuse-induced muscle loss. Finally, identified are the key challenges in this field as well as some future prospectives.

Keywords: disuse muscle atrophy, mechanical unloading, protein synthesis, protein degradation, molecular and cellular pathways, therapeutic countermeasures

# INTRODUCTION

Skeletal muscle is composed of muscular fibers and fascicles. It possesses a variety of functions in an organism's body and plays a vital role in the regulation of body metabolism. Skeletal muscles may differ significantly in mass, size, shape, and arrangement, depending upon their location and physical function in the organism. Elevated physical activity, like exercise, leads to increase the muscle mass (Bogdanis, 2012). On the other hand, decreased or limited use of skeletal muscle is one of the greatest contributing factors leading to muscle atrophy. Muscle atrophy may occur in a variety of conditions in unhealthy and healthy individuals, e.g., many common illnesses (Evans, 2010) including diabetes (Bonaldo and Sandri, 2013), cancers (Stephens et al., 2010), renal/heart failure, sepsis (Gordon et al., 2013), muscle genetic diseases (Sandri, 2010) and neurodegenerative disorders (Verdijk et al., 2012) cause significant loss of muscle mass. While in healthy individuals, muscle atrophy can also occur during conditions like spaceflight, bed rest, reduced

#### Edited by:

*Kevin I. Watt, Baker Heart and Diabetes Institute, Australia*

#### Reviewed by:

*Andrew Philp, University of Birmingham, United Kingdom Craig Andrew Goodman, Victoria University, Australia, Australia*

\*Correspondence:

*Yunfang Gao gaoyunf@nwu.edu.cn*

#### Specialty section:

*This article was submitted to Integrative Physiology, a section of the journal Frontiers in Physiology*

Received: *26 October 2017* Accepted: *02 March 2018* Published: *20 March 2018*

#### Citation:

*Gao Y, Arfat Y, Wang H and Goswami N (2018) Muscle Atrophy Induced by Mechanical Unloading: Mechanisms and Potential Countermeasures. Front. Physiol. 9:235. doi: 10.3389/fphys.2018.00235* step, hindlimb unloading (HLU) and immobilization. Furthermore, aging is also associated with muscle loss (Keller and Engelhardt, 2013; Hughes et al., 2017). Over the years, disuse skeletal muscle atrophy has been widely studied in humans as well as in animals.

Low skeletal muscle mass, decreased muscle fiber crosssectional area (mCSA), muscle fiber transition from slow to fast and the change of functional properties have been observed in different muscle types under disuse conditions (Booth and Gollnick, 1983; Baldwin, 1996; Baldwin and Haddad, 2001; Fitts et al., 2001; Ohira et al., 2006). Several studies indicated that bed rest and immobilization cause disturbances in protein turnover (Goldspink, 1977; Janssen et al., 2000; Phillips et al., 2009; Rennie, 2009). It is commonly believed that a disproportion in the rate of protein synthesis and protein degradation is the main cause of muscle loss (Boonyarom and Inui, 2006; Bialek et al., 2011).

This review provides a summary of disuse muscle atrophy. In the first section, the functional and structural adaptations or alterations of skeletal muscle to disuse are discussed in different unloading models such as spaceflight, head down bed rest (HDBR), immobilization and reduced step in humans as well as HLU and immobilization in animals. The second section of the review provides a detailed discussion of anabolic and catabolic pathways and potential triggers involved in muscle protein synthesis and degradation under disuse-induced muscle atrophy. These include insulin-like growth factor-1 protein kinase B-mammalian target of rapamycin (IGF-1- Akt/PKB-mTOR), muscle ring finger 1/muscle atrophy F-box (MuRF1/MAFbx) and forkhead family of transcription factors (FOXO) pathways. The third section discusses the efficiency of potential countermeasures (antioxidants, resistance exercises and protein supplements) to counteract the loss of skeletal muscle.

# MECHANICAL UNLOADING MODELS

Spaceflight, bed rest, immobilization (cast/leg brace), step reduction and HLU are the key models extensively used to study the muscle loss in humans as well as in animals. These models provide deeper insights into the molecular and cellular mechanisms underlying disuse-induced muscle atrophy.

# Human Models of Disuse-Induced Muscle Atrophy

The primary models of disuse in human research include microgravity induced muscle changes during spaceflight (Fitts et al., 2010; Goswami, 2017), bed rest immobilization (Spector et al., 2009), as well as other forms of immobilization (cast or leg brace) (Wall et al., 2013) and step reduction (Breen et al., 2013; McGlory et al., 2017). The majority of the published results of spaceflight experiments indicate that significant declines in skeletal muscle size, volume, CSA and strength occur after exposure to microgravity. It appear that longer sojourns in space lead to further atrophy and weakening of the muscles: 115–197 days in space were associated with significant decreases in muscle volume of gastrocnemius 17%, soleus 17%, and quadriceps 10% (LeBlanc et al., 2000). Another study also reported that prolonged spaceflight (approximately 180 days) significantly reduced force and fibers size in the gastrocnemius and soleus muscles. Specifically, the atrophy order (greatest-least) was: atrophy in soleus type I > soleus type II > gastrocnemius type I > gastrocnemius type II (Fitts et al., 2010). Furthermore, two long term spaceflight studies (140 and 175 days) reported a considerable decrease in characteristics of gastrocnemius muscle strength (Kozlovskaya et al., 1981). Similarly, 6 months Mir-mission in space found that isometric maximal voluntary contractions of the triceps surae muscle and peak tetanic force (Po) decreased by 42 and 25%, respectively (Koryak, 2001). In addition, short term spaceflight studies have also been reported to cause muscle weakness. For instance, volume changes in the knee extensor, knee flexor and plantar flexor muscle ranged from −15.4 to −5.5, −14.1 to −5.6 and −8.8 to −15.9, respectively, after 2 weeks in spaceflight (Akima et al., 2000). Edgerton and colleagues found that the size of all muscle fiber types of the vastus lateralis (VL) muscle decreased after 5–11 days of spaceflight (e.g., type I 16%, IIa 23%, and IIb 36%) and the percentage of type I myofibers decreased 6–8% (Edgerton et al., 1995). For instance, up to 8% decrease in CSA of the knee extensor and the gluteal muscles as well as 10% decrease of strength were observed after 17 days of spaceflight (Tesch et al., 2005).

In addition, differential rates of muscle atrophy have also been observed in response to disuse induced by HDBR in different muscle and fiber types. For example, prolonged bed rest with a 6◦ HDBR is one of the commonly used methods to mimic the effects of microgravity on muscle and bone turnover (Pavy-Le Traon et al., 2007; Spector et al., 2009). It was repeatedly demonstrated that HDBR leads to significant reduction in muscle strength and mass. Using magnetic resonance imaging (MRI), up to 17 and 40% loss in VL muscle volume and function were seen following 84-day HDBR (Trappe et al., 2004). Another 90-day study of HDBR reported up to 26% reduction in mCSA in young, healthy males (Rittweger et al., 2005). Indeed, similar results were obtained following 35 days of bed rest study (e.g., type I fiber CSA of VL showed greater loss than type II fiber) (Brocca et al., 2012). Similarly, a short duration HDBR study of 17–20 days also showed a 10–12% decrease in muscle size (Akima et al., 1997, 2001). On the other hand, Miokovic and associates observed that during prolonged bed rest, intramuscular differential atrophy did occur in most muscles, but some muscles of the lower limb remained unaffected (Miokovic et al., 2012).

In addition to bed rest, various other forms of immobilization have been used in ground-based studies of disuse muscle atrophy. Two week limb immobilization (casting) studies on young males reported reduction in quadriceps muscle volume, mCSA and strength by 9, 5–8, and 23%, respectively (Glover et al., 2008; Suetta et al., 2009). Even some shorter duration disuse studies have also reported significant reductions in muscle size (Wall et al., 2014). A five day study reported a 9% reduction in strength and 4% in quadriceps CSA (Dirks et al., 2014). Furthermore, immobilization appeared to impact the knee extensors to a greater degree than knee flexors (Veldhuizen et al., 1993; Deschenes et al., 2002). In addition to the above models of disuse, reduced step model has been demonstrated to influence muscle functions (Olsen et al., 2008; Knudsen et al., 2012; Breen et al., 2013). For example, reduced daily ambulatory activity has shown to lead to 2.8% loss in lean leg mass (Krogh-Madsen et al., 2010). Further data have shown that loss in muscle mass through lower limb disuse is more pronounced in older individuals (as they show increased vulnerability to loss of muscle size and strength) compared to younger individuals (Trappe, 2009; Degens and Korhonen, 2012; Tanner et al., 2015). It was observed in older participants (aged 68 ± 5 years) that 10-day bed rest leads to 7% reductions in muscle mass (lean tissue) (Kortebein et al., 2006). These findings were consistent with the results from a 7-day bed rest study involving six 60–73 year old participants, which reported 3.0 and 4.1% loss in total lean mass and lean leg mass, respectively (Drummond et al., 2012). Moreover, it has been observed that both contractile rate of force (torque) development and maximal isometric muscle strength are reduced significantly in the elderly as compared to younger men after 2 weeks of unilateral leg casting (Hvid et al., 2010). The discussed differences between young and old subjects are correlated with the effects of age-associated muscle atrophy (sarcopenia). Sarcopenia has been defined specifically as related to a subgroup of older persons with muscle-mass depletion, whose appendicular skeletal muscle mass (kg)/height<sup>2</sup> (m<sup>2</sup> ) is less than two standard deviations below the mean of a young reference group (Baumgartner et al., 1998). The muscle loss of sarcopenia has been attributed to the reduction in muscle fiber size (predominately in type II) and the muscle fiber number (Nilwik et al., 2013). Further detailed characteristics, mechanisms and functional significance of sarcopenia have been discussed in previous reviews (Evans, 2010; Narici and Maffulli, 2010).

From the foregoing discussion, it can be seen that the effects of several models of disuse (spaceflight, HDBR, immobilization and reduced step) on loss of muscle mass and strength have been widely investigated. Not all the studies have, however, reported consistent findings. This could be due to the fact that in most of these studies, either the study was restricted to investigation of only one muscle type or the immobilization was of short duration (≤14 days). Therefore, it is difficult to determine the details regarding differential rate of atrophy from these studies. Additionally, an important limiting factor in human studies is that most of the available data on atrophy and fiber CSA are based on a small biopsy sample taken from a single site.

# Animal Models of Disuse-Induced Muscle Atrophy

Among all models of disuse-induced muscle atrophy, HLU and immobilization are, by far, the most widely employed animal models to study skeletal muscle atrophy in small mammals (Fitts et al., 1989; Morey-Holton and Globus, 2002; Caron et al., 2009). Generally, disuse muscle atrophy in rodents results in a rapid loss of muscle mass as well as in fiber CSA and function (within 14 days of unloading; Bodine, 2013). In the following section, we will attempt to evaluate the effectiveness of these two models with specific regards to the adaptations that are thought to occur in skeletal muscle.

The rodent HLU model is extensively used to simulate the physiological effects of microgravity (Morey-Holton and Globus, 2002; Lawler et al., 2003; Baehr et al., 2016). Most rodent studies focused on the slow soleus muscle, as it showed rapid atrophy (Wang et al., 2006; Sandonà et al., 2012). After 14 days of HLU 34-50% decrease of soleus muscle wet weight and 49% CSA were observed in rats (Ohira et al., 1992; Zhang et al., 2017). Similarly, another study indicated that electromyography of soleus was reduced at the start of unloading but then recovered fully within a week but muscle atrophy continued to increase (Ohira et al., 2015). Furthermore, fast type muscles (including gastrocnemius, plantaris and tibialis anterior but not extensor digitorum longus) also showed significant reduction in muscle mass following HLU (Tsika et al., 1987; Kyparos et al., 2005).

Another key model is limb immobilization, in which the desired part of the animal is covered with a plaster bandage or with a spiral wire and surgical skin staplers, which helps to maintain the joint in a particular position (Caron et al., 2009; Du et al., 2011). It is now well documented in rodents that soleus muscle wet weight and muscle fiber diameters for type I and II significantly decreased after 4 weeks of immobilization (Okita et al., 2001). It should be noted that muscle atrophy varies significantly under different conditions of immobilization. For example, it depends upon the position in which the joint is fixed/immobilized. Additionally, it was reported that muscles immobilized in short positions favor atrophy (Onda et al., 2016). Some studies also reported that the extensor muscles atrophy more than flexor muscles during ankle-joint immobilization (Roy et al., 1991; Adams et al., 2003; Ohira et al., 2006). The degree of skeletal muscles atrophy also showed differential responses in regards to the types of fiber (Jozsa et al., 1988). For example, 4 week hindlimb immobilization studies on male rats reported that type I fibers of the soleus muscle undergo greater reductions than type II fibers (Booth and Kelso, 1973; Thomason and Booth, 1990) and similar results have also been obtained during anklejoint immobilization (Thomason and Booth, 1990; Ohira et al., 2006).

Taken together, varying results among different studies in rodent models of HLU and limb immobilization are related to the muscle type, muscle fiber type and conditions of immobilization. It has been shown that the amount of muscle loss is greater in the extensor muscles of the ankle (soleus and gastrocnemius) as compared to the flexor muscles (tibialis anterior and extensor digitorum longus) (Ohira et al., 2002; Adams et al., 2003; Zhong et al., 2005). Additionally, muscle atrophy appears to be different across muscle types. For example, slow-twitch (type I) fibers are more vulnerable and therefore, show a greater loss in the amount of protein than fast twitch (type II) fibers (Tsika et al., 1987; Thomason and Booth, 1990; Zhang et al., 2015), and the muscles immobilized in short positions showed more sensitivity to disuse (Desaphy et al., 2010). These observations suggest that the rate and extent of muscle loss appear to depend on the degree of unloading, the extent of physical inactivity and the muscle type.

# MECHANISMS OF MUSCLE ATROPHY

Alterations and underlying mechanisms of muscle protein synthesis and degradation have been investigated extensively in different disuse models (Bodine, 2013; Bonaldo and Sandri, 2013; Rudrappa et al., 2016). In addition, some new insights and findings also have been reported recently (Mirzoev et al., 2016; Baehr et al., 2017). In this section, we provide a comprehensive summary of the molecular basis of disuse atrophy including potential triggers to signaling pathways and their ultimate effects on the myofibrillar apparatus.

# Protein Synthesis and Disuse Atrophy

It has been well recognized that decreased protein synthesis in muscle seems to be the major contributor in disuse muscle atrophy (Bodine, 2013). The decline of the basal protein synthesis rate in the early period of unloading has been investigated and confirmed in human and animal models (Booth and Seider, 1979; de Boer et al., 2007; Mirzoev et al., 2016; Baehr et al., 2017). Thus, the underlying mechanisms of decreased protein synthesis in disused skeletal muscle have been a main focus research area the field for the past few decades. Although many problems have not been solved yet, recent research has confirmed that decreased activation of the Akt-mTOR pathway is involved in the mechanisms of the attenuated protein synthesis under disuse conditions. In the following section, we give the details about this pathway both in normal and unloading conditions.

Under normal physiological condition, the IGF-1-Akt-mTOR pathway acts as a key regulator in the translation initiation step of protein synthesis in skeletal muscle (Erbay et al., 2003; Han et al., 2008; Schiaffino and Mammucari, 2011). Firstly, the IGF-1-Akt-mTOR pathway is initiated by binding of the IGF-1 to its specific IGF-1 receptor (IGF-1R), which triggers a signaling cascade mainly stimulating the intrinsic tyrosine kinase activity through insulin receptor substrate-1 (IRS-1). Subsequent activation of phosphatidylinositol 3 kinase (PI3K) may be achieved by binding p85 regulatory subunit with phosphorylated IRS-1 (Jheng et al., 2012). The membrane phospholipidphosphatidylinositol-4,5-bisphosphate (PIP2), is phosphorylated into phosphatidylinositol-3,4,5-triphosphate (PIP3) by PI3K. Then PIP3 recruits phosphoinositide-dependent protein kinase-1 (PDK1) to phosphorylate and activate Akt. Phosphorylated Akt further activates mTORC1 (Inoki et al., 2002; Goodman et al., 2010; Miyazaki et al., 2011). Consequently, activated mTORC1 phosphorylates both 4E-BP1 and S6K1 which finally leads to protein synthesis (Gordon et al., 2013). Besides the indirect pathway by mTORC1, activated Akt can also phosphorylate glycogen synthase kinase 3β (GSK-3β), which directly leads to the increase of global protein synthesis through an increased activity of eukaryotic initiation factor 2B (eIF2B) (Welsh et al., 1998). The simplified version of the regulating mechanism underlying IGF-1-Akt-mTOR pathway on muscle protein synthesis is shown in **Figure 1**.

Under unloading conditions, insulin resistance plays an important role in driving depression of protein synthesis. This has been widely observed in humans subjected to bed rest confinement (Shangraw et al., 1988; Stuart et al., 1988; Hamburg et al., 2007), immobilization (Richter et al., 1989) and HLU in animals (Allen et al., 1997). Research on disuse models of rodents (e.g., HLU, immobilization and denervation) showed that insulin resistance induced attenuation of AktmTORC1 pathway may provide a mechanism for decreased protein synthesis (Gordon et al., 2013). Reduced activation of this pathway which characterized as decreased phosphorylation of Akt, S6K1, and 4E-BP1 has been shown in the soleus and medial gastrocnemius muscles (Bodine et al., 2001b; Hornberger et al., 2001; Sugiura et al., 2005; Haddad et al., 2006; Kelleher et al., 2013). Liu and colleagues also found that the binding of 4E-BP1 and eIF4E altered in rat gastrocnemius muscle during the early period of HLU (Liu et al., 2012). In addition, decreased phosphorylation of GSK3β has been observed in HLU rats (Stevenson et al., 2003; Mirzoev et al., 2016). Recent studies have also reported that an important mTOR signaling repressors such as mRNA expressions of regulated in DNA damage and development 1 and 2 (REDD1/2), significantly elevated following unloading in rats (Kelleher et al., 2013, 2015). Moreover, it has been reported that both Akt-null and mTOR knockout mice exhibited significant skeletal muscle atrophy as well as growth deficiency, which also proved the essential role of AktmTOR pathway on muscle maintenance (Peng et al., 2003; Risson et al., 2009). All the above discussion demonstrates the essential role of Akt-mTOR pathway in animals, but its role in controlling muscle protein synthesis in humans unloading models remains unclear. For example, human immobilization studies reported decrease in protein synthesis rate, but no changes were observed in Akt-mTORC1 signaling pathway (de Boer et al., 2007; Glover et al., 2008; Marimuthu et al., 2011). This suggests that decreased protein synthesis rate could also be regulated through other signaling pathways in human. Among other possible pathways, the most noteworthy one is focal adhesion kinase (FAK), a mechanosensitive non-receptor protein tyrosine kinase, located in the costamere region of skeletal muscle fibers and is sensitive to changes in mechanical loading (Bloch and Gonzalez-Serratos, 2003; Anastasi et al., 2008). Some crosstalk has been reported between FAK and PI3K-Akt-mTOR pathway. On the one hand, phosphorylation of tyrosine 397 of FAK results in the binding of FAK to the SH2 domain of the 85 kDa subunit of PI3K, which can lead to the increase in PI3K activity and subsequently activate Akt-mTOR pathway (Chen et al., 1996). On the other hand, activated FAK may upregulate mTOR through inhibiting TSC2 (a negative regulator of mTOR) by phosphorylation (Graham et al., 2015). Under physical inactivity conditions, reduced phosphorylation of FAK was discovered in humans and animals, which suggests that attenuated activation of FAK-Akt-mTOR is another key contributor to the decreased protein synthesis during atrophy condition (de Boer et al., 2007; Glover et al., 2008; Graham et al., 2015). Collectively, both the declined activation of IGF1–Akt–mTOR pathway induced by impaired IGF-1 signaling/insulin resistance and the decreased activation of FAK-Akt-mTOR pathway caused by reduced FAK phosphorylation play essential roles in regulating skeletal muscle protein synthesis during atrophy conditions. The possible mechanisms are also summarized briefly in **Figure 1**.

In addition to the regulation of translation initiation, it has also been reported that eukaryotic elongation factor 2 (eEF2) plays an important role in the regulation of protein synthesis at the level of elongation of mRNA translation process (Redpath et al., 1996). Recently it was observed that the level of eEF2 phosphorylation (inactive form) in soleus muscle elevated significantly after 14 days of HLU in rats (Lomonosova et al., 2017). Besides, protein synthesis also depends on translational capacity, the main component of which is the number of ribosomes (McCarthy and Esser, 2010; Chaillou et al., 2014). Decreases in the content of both total RNA and 28S rRNA (one of the key markers of ribosome content) were observed after 1, 3, 6, and 7 days of HLU in rat soleus (Bajotto et al., 2011; Mirzoev et al., 2016). Although great progress has been made as described above, there is still a lot of work that needs to be done. For example, studies with frequent biopsy sampling are required to comprehensively understand the role of mTORC1 signaling in regulating the depression in postprandial and postabsorptive muscle protein synthesis, especially in human models. In addition, more research is needed to clarify the relative downstream genes for regulation of ribosome assembly in the Akt-mTOR pathway.

# Protein Degradation and Disuse Atrophy

In contrast to the recognized deficits in muscle protein synthesis during disuse conditions, the role of protein breakdown in disuse-induced muscle atrophy is less clear. This is partly due to the lack of direct measurements of muscle protein degradation in studies. Instead, indirect studies of protein degradation pathways have to be employed to measure molecular markers of muscle proteolysis. The major protein degradation pathways in skeletal muscle include the Ca2+-dependent proteases, lysosomal system, caspases and ubiquitin proteasome pathways (Scicchitano et al., 2015). It was proposed that Ca2+-dependent proteases (calpains) act as a promoter of muscle protein degradation, and might be responsible for the discharge of myofilaments from the surface of myofibrils (Dayton et al., 1976). Subsequently, the myofilaments were ubiquitinated and degraded to amino acids by proteasome intracellular peptidase cathepsins (**Figure 2**; Huang and Zhu, 2016). Emerging evidence suggests that calpains (Huang and Zhu, 2016), caspase-3 (Talbert et al., 2013), autophagy-lysosomal system (Sandri, 2010) and ubiquitin proteasome pathway (Bodine and Baehr, 2014), all are involved in disuse-induced muscle atrophy. However, the ubiquitin proteasome system is often considered as the most important proteolytic system during disuse conditions that promotes muscle wasting (Scicchitano et al., 2015; Baehr et al., 2017). The breakdown of protein via the ubiquitin proteasome system requires three distinct enzymatic components of the ubiquitin proteasome pathway: E1 (ubiquitinactivating enzyme), E2 (ubiquitin-conjugating enzyme) and E3 (ubiquitin ligase, key enzyme which regulates proteolysis as it recognizes multiple target protein substrates). Two muscle specific classes of E3s (MuRF1 and MAFbx/atrogin-1) have been studied widely and play an essential role during skeletal muscle atrophy (Mitch and Goldberg, 1996; Foletta et al., 2011; Bodine and Baehr, 2014). In various animal models of disuse muscle atrophy (HLU, immobilization, spaceflight and denervation), mRNA levels of both genes (MuRF1/MAFbx) were

rapidly increasing and thought to play a crucial role in the initiation of the atrophy process (Bodine et al., 2001a; Lecker et al., 2004; Murton et al., 2008; Allen et al., 2009; Baehr et al., 2017; Gambara et al., 2017). In addition, it has been reported that the degree and the time course of the upregulation of these two genes is not uniform among different muscles. For example, a more significant increase in the expression of MuRF1/MAFbx genes occurred in ankle plantar flexors (soleus and medial gastrocnemius) than dorsi flexors (tibialis anterior), and there was a longer duration of expression in ankle plantar flexors than dorsi flexors in response to unloading (Bodine et al., 2001a; Lecker et al., 2004). Moreover, studies using MAFbx and MuRF1-deficient mouse models further supported a potential contributing role of these two genes in the development of muscle atrophy. For instance, mice deficient in either MAFbx or MuRF1 were found to be resistant to atrophy (Bodine et al., 2001a). Another study also indicated remarkable protection of the soleus muscle in the MuRF1-KO mice during 10 days of HLU (Labeit et al., 2010). It should be noted that various reports have been shown about MuRF1 and MAFbx during disuse-induced atrophy in human models (Murton et al., 2008). For example, both the MuRF1 and MAFbx mRNAs increased significantly after 2 days (Abadi et al., 2009) and 5 days of immobilization (Dirks et al., 2014) or after 3 days of unilateral lower limb suspension (ULLS) (Gustafsson et al., 2010). However, in a 20-day bed rest study, elevated MAFbx but not MuRF1 mRNA were observed in VL (Ogawa et al., 2006), same results were observed during 14 days of leg immobilization (Jones et al., 2004). On the other hand, de Boer and colleagues reported that increased MuRF1, mRNA expression, but not MAFbx were observed during 0–10 days of immobilization (de Boer et al., 2007). In addition, it has been reported that MuRF1 protein expression increased in soleus, but not in VL after 60 days of bed rest (Salanova et al., 2008).

Under disuse conditions, it has been reported that expression of E3 ubiquitin ligases MuRF1 and MAFbx are regulated by various upstream factors or signaling pathways (Bodine and Baehr, 2014). FOXO transcription factors (FOXO1 and FOXO3) have been stated as the major transcription factors regulating both the MuRF1 and MAFbx expressions (Sandri et al., 2004; Stitt et al., 2004). It is noteworthy that the functional aspects of FOXO are determined by their cellular location and predominantly regulated by the IGF-1-PI3K-Akt pathways (Brunet et al., 1999; Zhao et al., 2007). Under normal physiological conditions, Akt phosphorylates FOXO on specific threonine and serine residues, which results in the retention of FOXO in the cytosol instead of translocating to the nucleus (Brunet et al., 1999). Under muscle disuse conditions, dephosphorylated FOXO translocates to the nucleus and up-regulates several different types of atrogenes (E3 ligases) or autophagy related genes as shown in **Figure 2** (Stitt et al., 2004). Several studies reported that mRNA and protein levels of FOXO1 and/or FOXO3 expression in the slow-twitch soleus muscle, mixed fiber type gastrocnemius muscle and fast twitch plantaris muscle are upregulated following different muscle atrophy condition (Giresi et al., 2005; Sacheck et al., 2007; Allen et al., 2009; Levine et al., 2011; Okamoto and Machida, 2017). Sandri and colleagues found that constitutively active FOXO3 acts on the MAFbx promoter to cause MAFbx transcription and dramatic atrophy of myotubes and muscle fiber. It further suggested that stimulation of the two main proteolytic pathways (MuRF1/MAFbx) via overexpression of FOXO1 and FOXO3 leads to a reduction in muscle mass and strength during disuse inactivity (Sandri et al., 2004). Direct or indirect inhibition of FOXO transcriptional activity or suppressed expression of co-factors (MuRF1/MAFbx) and interaction with other transcription factors lead to the attenuation of disuse-induced muscle atrophy (Senf et al., 2008; Reed et al., 2012; Brocca et al., 2017). But these two transcription factors mRNA are not always linked and that changes in the expression levels of both genes (MuRF1/MAFbx) are dependent on the muscle and the time after unloading (Atherton et al., 2016). Interestingly, it was observed that constitutively active FOXO3 controls the stimulation of autophagic/lysosomal proteolysis pathway, thus leading to muscle wasting in fasting and denervation models (Mammucari et al., 2007). Put-together, these findings strongly support the notion that most of these transcription factors and signaling pathways play important roles in the progression of muscle disuse atrophy. Remarkably, increased mRNA levels of both genes (MuRF1/MAFbx) were observed in animals following disuse conditions as shown in **Figure 2**. While, data were inconsistent in human models, a previous review showed that activation of these two genes mainly occurs in muscle wasting caused by inflammation (e.g., cancer, chronic obstructive pulmonary disease, severe head trauma, amyotrophic lateral sclerosis, critical illness, AIDS) with the data in non-inflammatory muscle atrophy is inconsistent (Narici and Maffulli, 2010). Only a few studies have measured the changes in ubiquitinated protein content and proteasomal proteolysis following disuse atrophy (Helliwell et al., 1998; Ogawa et al., 2006; Slimani et al., 2012; Baehr et al., 2016). There is a need for additional studies on human biopsies to explore the activity of ubiquitin-proteosome pathway components.

# Major Triggers of Disuse Atrophy

Besides the above-mentioned pathways regarding muscle protein synthesis and degradation following disuse atrophy, there are two important factors contributing to protein turnover. The following discussion gives a review regarding the possible triggers reported during muscle disuse.

# ROS and Disuse Atrophy

Oxidative stress, characterized by the increased reactive oxygen species (ROS) production and impairment of antioxidant defense systems, has frequently been observed in disuse and other pathological conditions. It is now widely considered as a major trigger of the imbalance between protein synthesis and degradation leading to muscle atrophy (Powers et al., 2005, 2007; Moylan and Reid, 2007; Powers, 2014; Zuo and Pannell, 2015). Production of ROS results in the inhibition of insulin action and acts as a putative mediator in the development of insulin resistance (Bashan et al., 2009; Di Meo et al., 2017). Furthermore, growing evidence indicates that oxidative stress can promote muscle protein breakdown through the following aspects. Firstly, oxidative stress, promotes expression of proteins involved in proteolytic pathways, such as autophagy, calpain and the ubiquitin–proteasome system of proteolysis. Secondly, oxidative stress results in the activation of two important proteases, calpain and caspase-3. Thirdly, increased ROS production in muscle fibers can also promote proteolysis by oxidative modification of myofibrillar proteins, which enhances their susceptibility to proteolytic processing. The details about the three aspects have been discussed in a previous published review (Powers, 2014). ROS also might be the upstream activators of nuclear factor kappa-B (NF-κB) and FOXO pathways in skeletal muscle atrophy (Dodd et al., 2010). The roles played by ROS in protein synthesis and degradation are depicted in **Figures 1**, **2**, respectively.

# Calcium Overload and Disuse Atrophy

Calcium (Ca2+) is necessary to carry out many important body functions such as cell metabolism, cardiac and skeletal muscle contraction, tissue differentiation and neurotransmission (Zhou et al., 2013). Ca2<sup>+</sup> and endogenous inhibitor calpastatin are the two major regulators on calpains activation during disuse conditions (**Figure 2**; Huang and Zhu, 2016). Previous studies reported highly elevated cytosolic free Ca2<sup>+</sup> concentration in soleus and gastrocnemius muscles during disuse conditions (Ingalls et al., 2001; Xu et al., 2012; Hu et al., 2017). Two ubiquitous calpains, calpain1 and calpain2 (also called u- and m-) are activated by elevated intracellular Ca2<sup>+</sup> in HLU rats (Matsumoto et al., 2014; Zhang et al., 2017). In addition, caspase-3-dependent apoptosis, another major signaling pathway involved in disuse muscle atrophy (Talbert et al., 2013), is also activated by intracellular Ca2<sup>+</sup> overload through two distinct pathways. On the one hand, the intracellular Ca2<sup>+</sup> overload leads to the activation of caspase-12 which then activates caspase-3 (Primeau et al., 2002). On the other hand, increasing Ca2<sup>+</sup> levels induces activation of pro-apoptotic protein Bax, which translocated and inserted into the outer membrane of mitochondria via forming Bax/Bax-homo-oligomerization. Bcl-2, another Bcl-2 family protein, could inhibit the formation of Bax/Bax-homo-oligomerization. The decline of the ratio of Bax/Bcl-2 leads to the release of pro-apoptotic factors from the mitochondria, which subsequently activates caspase-9 and caspase-3 (Zha et al., 1996; Antonsson et al., 1997; Chen et al., 2002; Garrido et al., 2006). In one of our previous studies, the increase of Bax/Bcl-2 and cytochrome C release were observed in gastrocnemius in rats following 14-day HLU (Hu et al., 2017). Collectively, both the overproduction of ROS and Ca2<sup>+</sup> overload play an essential role in regulation of protein synthesis and degradation following disuse conditions as shown in **Figure 2**. However, the relationship between ROS and Ca2<sup>+</sup> remains unclear and more research is needed to clarify this relationship.

In addition to the anabolic and catabolic pathways mentioned above, recently emerging evidence indicates some other key factors, such as P53, activating transcription factor 4 (ATF4), growth arrest and DNA damage-inducible 45a protein (Gadd45a) and P21, are significantly elevated under disuse conditions (Ebert et al., 2012; Fox et al., 2014; Bullard et al., 2016). Therefore, we recommend that the reader consult the most recently published reviews on the topic (Brooks and Myburgh, 2014; Adams et al., 2017). Taken together, during disuse conditions, both the protein synthesis and degradation play an essential role during muscle atrophy, and in particular, suppressed protein synthesis has been confirmed. However, more research has to be carried out to clarify the details mechanisms of protein degradation and insulin resistance in driving disuse-induced muscle atrophy.

# THERAPEUTIC COUNTERMEASURES

Various therapeutic interventions, pharmaceutical options and rehabilitation programs have been used to prevent and limit the loss of skeletal muscle. These therapeutic countermeasures can be grouped into three categories: antioxidant and anti-inflammatory compounds, nutritional supplements and physical training and exercise.

# Antioxidant and Anti-inflammatory Compounds

It has been demonstrated that muscle damage, oxidative stress and inflammation have a negative impact on protein turnover of skeletal muscle, predominantly via decreases in protein synthesis (Peterson et al., 2011; Powers et al., 2012). Oxidative stress is thought to be one of the major factors leading to many health-related disorders including skeletal muscle dysfunction (Powers et al., 2012). In addition, increased mitochondria ROS production, as well as endoplasmic reticulum stress and decrease of antioxidant capacity, are three major factors that play key roles in triggering sarcopenia with aging (Drew et al., 2003; Short et al., 2005; Narici and Maffulli, 2010). Thus, antioxidants have been shown to prevent oxidative stress associated damage and have been proven to be effective countermeasures against muscle atrophy to maintain skeletal muscle mass and strength, especially in elders (Servais et al., 2007; Cornetti, 2009; Rieu et al., 2009; Stojiljkovic et al., 2016).

In recent years, polyphenols, a well-recognized antioxidants, have been studied extensively with regard to their roles in the prevention of neurodegenerative diseases/skeletal muscle atrophy. Resveratrol, is one of the naturally occurring polyphenols, well known for its great health benefits, and frequently found in berries, grapes, red wine and some other fruits and vegetables (Brito et al., 2008; Das et al., 2008). It has been suggested that resveratrol plays an important role in the transcription of two important antioxidant enzymes, i.e., Mn superoxide dismutase (SOD) and catalase (Dani et al., 2008; Kode et al., 2008; Robb et al., 2008; Ryan et al., 2010). Six month old adult rats were treated with resveratrol (12.5 mg/kg/day) for 5 weeks (including 2 weeks of muscle immobilization) with results indicating that it reduced the functional decrements and the oxidative stress level (Jackson et al., 2010). In another study, rats were supplemented with resveratrol at a dose (400 mg/kg/day) before unloading and 2 weeks of muscle immobilization. During treatment with resveratrol, muscle disuse atrophy was significantly reduced (Momken et al., 2011). These two studies strongly suggest that metabolic and muscle deconditioning in response to mechanical unloading can be prevented by the use of high dosage of the antioxidant, resveratrol. Similarly, in another study, mice were treated with tea catechins (comprising of up to 81% polyphenols) at a dosage of 46–50 mg/kg. The antioxidant diet was consumed before and during immobilization of 14 and 10 days, respectively. This study showed that the antioxidant (tea catechins) did not suppress muscle atrophy completely, but it helped in the maintenance of tetanic force observed in the soleus muscle in response to immobilization. This study also suggested that tea catechins have positive effects on skeletal muscle function rather than skeletal mass, and they help to improve muscle strength (Ota et al., 2011).

In addition to the above findings, several studies have reported the use of several other antioxidants [e.g., vitamin E or SS-31 (D-Arg-2′ 6 ′dimethylTyr-Lys-Phe-NH2) for the prevention of disuse muscle atrophy, (Servais et al., 2007; Powers, 2014)]. For instance, administration of vitamin E was shown to significantly reduce soleus muscle atrophy during 14-day HLU. The results of this research also demonstrated that the protective role of vitamin E does not depend on its antioxidant activity, but it might be due to alteration in muscle protein degradation (Servais et al., 2007). While, Koesterer and colleagues reported that vitamin E supplementation has no effect on HLU induced soleus and gastrocnemius muscle atrophy (Koesterer et al., 2002). In addition, it has been reported that SS-31, one of the essential mitochondrial-targeted antioxidant, could protect against HLU induced soleus and plantaris muscles atrophy both in rats and mice (Min et al., 2011; Talbert et al., 2013). Additionally, some authors reported that another two widely used antioxidants, either curcumin or N-acetylcysteine treatment could protect the diaphragm against ventilator-induced muscle wasting (Agten et al., 2011; Smuder et al., 2012), but did not prevent against inactivity-induced limb muscle atrophy (Farid et al., 2005).

In addition to the widely used antioxidant to prevent disuse atrophy, some anti-inflammatory compounds and pharmaceutical options were also adopted in this field. Some studies recommended the use of dietary fish oil to prevent immobilization-induced atrophy. Fish oils are well known for their anti-inflammatory properties as they contain different fatty acids (long chain n-3 fatty acid; Fetterman and Zdanowicz, 2009). N-3 fatty acids facilitate insulin-sensitive protein anabolism through the Akt-mTOR-S6K1 pathway, which prevents anabolic resistance and leads to decreased muscle atrophy induced by disuse (Gingras et al., 2007). Furthermore, another study reported that ingestion of 5% fish oil reduces disuse muscle atrophy via Akt pathway through E3 ubiquitin ligases and S6K1 pathway (You et al., 2010). Chromium (Cr) is also believed to preserve muscle mass by inhibiting the elevation of ubiquitin proteasome system pathway and restoring the impaired Akt signal through elevating the Akt phosphorylation (Dong et al., 2009).

Studies carried out in our laboratory show that tetramethylpyrazine is a dietary supplement that could effectively alleviate muscle atrophy in HLU rats (Gao et al., 2005; Zhang et al., 2007; Li et al., 2012). It was demonstrated that attenuating disuse-induced Ca2<sup>+</sup> overload and activation of calpains system might be involved in the underlying mechanism of tetramethylpyrazine to counteract disuse-induced muscle atrophy (Wu et al., 2012; Hu et al., 2017; Zhang et al., 2017). Furthermore, various Chinese herbal medicines have also been used to reduce and prevent muscle loss caused by different unloading models. For example, Sijunzi Decoction (Hu et al., 2009), Angelica Sinensis (Qin et al., 2009; Du and Gao, 2014), Ligusticum (Qin et al., 2009), and Radix Astragali (Gao et al., 2005, 2008; Zhang et al., 2007) have been tested under different muscle atrophy conditions by our laboratory.

The majority of these traditional Chinese medicines have positive effects on blood circulation and/or are blood tonics have been reported to attenuate disuse-induced muscle atrophy (Gao et al., 2005; Zhang et al., 2007, 2017; Wu et al., 2012; Hu et al., 2017). Due to numerous effective components in these herbs, the major effective ingredients and the mechanisms involved against muscle atrophy remain unclear. Thus, future research should be carried out to identify specific and effective components of these herbs, thus leading to wider applications of these herbs to counter muscle atrophy.

# Exercise and Physical Training

Exercise is one of the best countermeasure against disuse atrophy. During physical inactivity conditions such as HDBR and HLU, exercise has been shown to be the most efficient countermeasure to address the deficits in muscle structure and function, as well as for maintenance of balance between muscle protein synthesis and protein breakdown (Herbert et al., 1988; Widrick et al., 1996; Shinohara et al., 2003). Loss in muscle mass during physical inactivity is more challenging in aged people compared to younger persons and, therefore, to maintain muscle protein synthesis rate, older individuals required more resistance exercise than younger individuals (Kumar et al., 2012). For instance, after 2 weeks of ULLS following 4 weeks of resistive exercise employed in younger and older individuals suggested that recovery of strength and muscle size was much more reduced in older adults (Suetta et al., 2009; Hvid et al., 2010). These findings were further supported by a study of older women in which resistance training of 12 weeks was performed (at a frequency of 3 times per week). These investigators observed an increase in quadriceps muscle volume by up to 6% in young women and only 3% increase in older women (Greig et al., 2011). Hvid and co-workers observed marked decrements in knee extensor muscle function in young and old individuals after 4-day lower limb disuse. Following 7 day recovery, knee extensor (isometric or isokinetic) strength was recovered in young individuals, while an impaired ability to fully recover was observed in older individuals (Hvid et al., 2014).

Moreover, it has been documented that different types of resistance exercises play a key role in the maintenance of muscle mass in disuse models (e.g., bed rest, HLU) by improving muscle protein synthesis via activation of the PI3K-Akt-mTOR pathway (Ferrando et al., 1997; Fluckey et al., 2004; Hornberger et al., 2004; Philp et al., 2011). It has been proposed that short period of resistance exercise can activate IGF-1 gene expression in healthy individuals (Chesley et al., 1992). For instance, full restoration of quadriceps muscle mass following 2 weeks of single leg immobilization in humans can be achieved via highresistance training (Oates et al., 2010). Additionally, combination of endurance and resistance exercise is an effective modality to counter the muscle loss associated with disuse or inactivity in HLU mice (Adams et al., 2007). In fact, resistance exercise partly rescued the loss in cytoskeletal and dystrophin-associated glycoprotein in VL and soleus muscles at protein level for the duration of extended bed rest (84 days) in human beings (Chopard et al., 2005). It was also demonstrated that resistance exercises associated with a ∼50% decrease in the stimulation of the ubiquitin-proteasome system (MuRF1/MAFbx) and alleviate muscle atrophy caused by 14 days of HLU (Dupont-Versteegden et al., 2006).

Moreover, a number of other published literatures opine that the increased mechanical load can activate cellular signaling that initiates the protein synthesis independent of a traditionally described functioning IGF-1 receptor (Bickel et al., 2005; Hornberger et al., 2006; Spangenburg et al., 2008; Hamilton et al., 2010; West et al., 2010; Witkowski et al., 2010; Gabriel et al., 2016). Noticeably, these experiments pave the way for future investigations regarding Akt-mediated signaling in response to mechanical loading and other growth stimuli, as well as provide new insights for the prevention of disuse-induced muscle atrophy.

In general, applying exercise and physical training is one of the most widely used, effective and with minimal side effects in all countermeasures, but exercise and physical training cannot always be applied to injured patients with fractures and is often problematic for bed rest patients. In addition, the acceptance of this countermeasure is also difficult for those who do not want to exercise. Thus, exercises that possess minimum load or minimum exercise time against disuse muscle atrophy needs further investigation.

# Nutrition and Protein Supplementation

Exercise alone cannot avert the disuse-induced muscle loss under different unloading conditions, including bed rest confinement during hospitalizations and sedentary lifestyle. Hence, several other approaches, such as nutritional supplementation (essential amino acid and protein), should be used in conjunction with exercise to rescue or counteract better against catabolic processes during chronic disuse. Specifically, muscle mass and strength can be more effectively enhanced by combining the nutritional regulation (dietary carbohydrates and amino acid) with an adaptive exercise regimen than by application of either treatment approach alone (Dreyer et al., 2008; Pasiakos et al., 2011). The use of protein supplement along with heavy resistance training can significantly improve muscle strength and mass in very old individuals (Bechshøft et al., 2017). Generally speaking, it appears that during immobilization and bed rest confinement, protein synthesis can significantly be improved by using amino acid supplementation, but it can only partially avert muscle atrophy (Glover et al., 2008). Amino acids are very well known for their important role in the regulation of protein synthesis and metabolism by accelerating the initiation step of peptide chain formation in skeletal muscle (Biolo et al., 1997; Bohé et al., 2001; Rennie et al., 2004; Stipanuk, 2007). It has consistently been shown that essential amino acid supplementation strengthens the response of muscle protein synthesis and partially rescues skeletal muscle loss experienced during bed rest (Stuart et al., 1990; Paddon-Jones, 2006). Furthermore, ingestion of amino acid supplements has been shown to enhance the fractional synthesis rate of protein in the soleus muscle (Carroll et al., 2005).

Leucine has been widely studied among all nutritional supplementations that were effective to prevent disuse muscle atrophy. It was documented that leucine, as an anabolic factor among all essential amino acids, has potential to affect muscle protein metabolism in several ways and is considered as a strong stimulator of protein synthesis (Katsanos et al., 2006). However, it is still unclear how leucine is sensed by the processes that regulate protein synthesis? Several in vivo and in vitro studies described that leucine plays an important role in improvement of protein synthesis via the IGF1-mTOR-Akt pathway (Kimball et al., 1999; Anthony et al., 2000b; Drummond et al., 2017). Anthony and coworkers reported that leucine is implicated in the stimulation of the eIF4E complex (that is, mRNA binding step in translation initiation; Anthony et al., 2000a). Branched-chain amino acids supplementation was observed to have an anabolic effect on human muscles during 14 days of bed rest, specifically, a slight improvement of protein synthesis in the early recovery period was observed (Stein et al., 1999, 2003). In addition, resistance exercise and essential amino acid supplementation were also shown to lead to preserve skeletal muscle mass and strength during 28-day bed rest (Brooks et al., 2008). Physical inactivity appears to inhibit mTORC1 signaling associated with reduced amino acid transporter protein contents thus suggesting that a blunted response in essential amino acid stimulation could be the underlying basis of muscle loss in older individuals (Hvid et al., 2010; Drummond et al., 2012). However, another study reported that supplementation of protein (0.8–1.5 g/kg/day) throughout 10 days of bed rest in older subjects promoted muscle strength but had no impact on muscle mass loss (Dillon et al., 2009). Discussions related to the efficiency of amino acid countermeasures in preventing protein mass losses during inactivity can be seen in the detailed review of Stein and associates (Stein and Blanc, 2011). In their conclusion, it is mentioned that various nutrition and protein supplements have different degrees of prevention and treatment of muscular atrophy, but the effect of these treatments is limited as a result of the blunted post absorptive and postprandial muscle protein synthesis after disuse atrophy. Therefore, searching for nutritional supplements with less synthetic metabolism resistance is undoubtedly an important research direction for the treatment of muscle atrophy.

# FUTURE PROSPECTIVE

Over the past few decades, our current understanding of the cellular and molecular mechanisms involved in disuse muscle atrophy has significantly increased. However, this understanding remains incomplete with numerous unanswered questions. So far, the importance of protein turnover in driving disuse-induced muscle atrophy has been widely recognized. When the rate of protein synthesis becomes slower than the rate of protein breakdown, muscle atrophy begins. From a growing number of clinical and preclinical experiments, it is now clear that blunted

# REFERENCES


or suppressed muscle protein synthesis seems to be the major drivers of disuse atrophy rather than increased muscle protein breakdown. However, further investigations are still needed to completely understand that how and why the activation of MuRF1/MAFbx, FOXO and other genes affect both protein synthesis and protein breakdown, which strongly suggests that increased proteolysis occurred somewhere. Additionally, it has been suggested frequently that during prolonged periods of disuse, oxidative stress can influence biochemical pathways and gene expression that regulates both muscle protein synthesis and breakdown. There are many factors leading to oxidative stress, such as low temperature, hypoxia, tissue damage, inflammation and calcium overload, etc., but which one among these factors contributing to oxidative stress in disuse-induced muscle atrophy need to be further elucidated.

On the other hand, to date, many studies have demonstrated beneficial effects of therapeutics countermeasure in disuseinduced muscle loss in model organisms. But we still lack of appropriate treatment strategies and have limited pharmaceutical options. The major hindrance, of course, is lack of knowledge regarding the cellular and molecular mechanisms involved in disuse muscle atrophy, which is far more complicated than we have been led to believe. There is much more to learn about how to prevent muscle atrophy as it is evidently not limited to muscle protein synthesis and breakdown only. Even as far as protein turnover is concerned, more accurate and targeted studies are needed to be done to unlock the secrets of abnormal or unbalanced protein metabolism underlying disuse muscle atrophy. More effective therapeutic agents or measures will emerge and develop along with the deepening of the research on the detailed mechanism in future.

# AUTHOR CONTRIBUTIONS

Conceived and designed by: YG and YA; Critical Analysis provided by: YG, YA, HW, and NG; Proof reading and edited the manuscript: YG, YA, HW, and NG; Wrote the paper: YG and YA; Figures drawn by: YA.

# ACKNOWLEDGMENTS

The critical reading of Kashif Majeed is gratefully acknowledged. We apologize to colleagues whose studies were not cited owing to space limitations.

Our work is supported by grants from the National Natural Science Foundation of China (No. 31772459) and the Postdoctoral Science Foundation of China (No. 2016M592831).


contractility associated with controlled mechanical ventilation. Crit. Care Med. 39, 777–782. doi: 10.1097/CCM.0b013e318206cca9


myonuclei, satellite cells and signaling pathways. Front. Physiol. 5:99. doi: 10.3389/fphys.2014.00099


volume, and intensity. J. Gerontol. A Biol. Sci. Med. Sci. 67, 1170–1177. doi: 10.1093/gerona/gls141


activity in mouse brain. Biochem. Biophys. Res. Commun. 372, 254–259. doi: 10.1016/j.bbrc.2008.05.028


load-induced skeletal muscle hypertrophy. J. Physiol. Lond. 586, 283–291. doi: 10.1113/jphysiol.2007.141507


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Gao, Arfat, Wang and Goswami. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Efficacy of Stochastic Vestibular Stimulation to Improve Locomotor Performance During Adaptation to Visuomotor and Somatosensory Distortion

David R. Temple1,2, Yiri E. De Dios <sup>3</sup> , Charles S. Layne1,2,4, Jacob J. Bloomberg<sup>5</sup> and Ajitkumar P. Mulavara<sup>3</sup> \*

*<sup>1</sup> Department of Health and Human Performance, University of Houston, Houston, TX, United States, <sup>2</sup> Center for Neuromotor and Biomechanics Research, University of Houston, Houston, TX, United States, <sup>3</sup> KBRwyle, Houston, TX, United States, <sup>4</sup> Center for Neuro-Engineering and Cognitive Science, University of Houston, Houston, TX, United States, <sup>5</sup> Johnson Space Center, National Aeronautics and Space Administration, Houston, TX, United States*

#### Edited by:

*Andrew Blaber, Simon Fraser University, Canada*

#### Reviewed by:

*Alessandro Tonacci, Istituto di Fisiologia Clinica (CNR), Italy Irene Di Giulio, King's College School, United Kingdom*

> \*Correspondence: *Ajitkumar P. Mulavara ajitkumar.p.mulavara@nasa.gov*

#### Specialty section:

*This article was submitted to Integrative Physiology, a section of the journal Frontiers in Physiology*

Received: *15 August 2017* Accepted: *13 March 2018* Published: *29 March 2018*

#### Citation:

*Temple DR, De Dios YE, Layne CS, Bloomberg JJ and Mulavara AP (2018) Efficacy of Stochastic Vestibular Stimulation to Improve Locomotor Performance During Adaptation to Visuomotor and Somatosensory Distortion. Front. Physiol. 9:301. doi: 10.3389/fphys.2018.00301* Astronauts exposed to microgravity face sensorimotor challenges affecting balance control when readapting to Earth's gravity upon return from spaceflight. Small amounts of electrical noise applied to the vestibular system have been shown to improve balance control during standing and walking under discordant sensory conditions in healthy subjects, likely by enhancing information transfer through the phenomenon of stochastic resonance. The purpose of this study was to test the hypothesis that imperceptible levels of stochastic vestibular stimulation (SVS) could improve short-term adaptation to a locomotor task in a novel sensory discordant environment. Healthy subjects (14 males, 10 females, age = 28.7 ± 5.3 years, height = 167.2 ± 9.6 cm, weight = 71.0 ± 12.8 kg) were tested for perceptual thresholds to sinusoidal currents applied across the mastoids. Subjects were then randomly and blindly assigned to an SVS group receiving a 0–30 Hz Gaussian white noise electrical stimulus at 50% of their perceptual threshold (stim) or a control group receiving zero stimulation during Functional Mobility Tests (FMTs), nine trials of which were done under conditions of visual discordance (wearing up/down vision reversing goggles). Time to complete the course (TCC) was used to test the effect of SVS between the two groups across the trials. Adaptation rates from the normalized TCCs were also compared utilizing exponent values of power fit trendline equations. A one-tailed independent-samples *t*-test indicated these adaptation rates were significantly faster in the stim group (*n* = 12) than the control (*n* = 12) group [*t*(16.18) = 2.00, *p* = 0.031]. When a secondary analysis was performed comparing "responders" (subjects who showed faster adaptation rates) of the stim (*n* = 7) group to the control group (*n* = 12), independent-samples *t*-tests revealed significantly faster trial times for the last five trials with goggles in the stim group "responders" than the controls. The data suggests that SVS may be capable of improving short-term adaptation to a locomotion task done under sensory discordance in a group of responsive subjects.

Keywords: stochastic resonance, vestibular, locomotion, somatosensory, vision

# INTRODUCTION

In astronauts, prolonged exposure to microgravity induces an adaptation to that environment resulting in reinterpretation of visual, vestibular, and somatosensory inputs (Paloski et al., 1992, 1994; Reschke et al., 1994; Bloomberg et al., 2015). Upon return to a gravitational environment, postural control and balance can be severely compromised until the central nervous system (CNS) readapts to correctly process sensory information from a terrestrial environment (Paloski et al., 1992). Walking ability may take as long as 15 days to be fully restored upon returning to Earth (Mulavara et al., 2010). There are two periods in which adaptation is commonly observed in individuals: (1) rapid, within-session trial by trial improvements when completing a task multiple times, and (2) slower evolving incremental performance gain often observed over multiple practice sessions (Mulavara et al., 2010). The processes by which these rapid and slower adaptation curves occur are often referred to as strategic control and adaptive realignment respectively (Redding and Wallace, 2002; Richards et al., 2007; Mulavara et al., 2010). These two processes are considered interdependent (Redding and Wallace, 2002; Richards et al., 2007), and longer-term locomotor adaptive recovery has been shown to be associated with short-term strategic capabilities of astronauts readapting from long-duration spaceflight. Specifically, those astronauts demonstrating faster short-term (strategic) adaptation rates 1 day after their return also show faster overall recovery (Mulavara et al., 2010). If effective countermeasures can be implemented that develop faster short-term improvements in balance and locomotor skills, populations such as astronauts might be able to utilize strategic responses to speed recovery from adaptation to prolonged microgravity exposure after gravitational transitions. One proposed countermeasure, which has been shown to benefit sensory system capabilities and associated performance improvements, is through the addition of small amounts of noise via a phenomenon known as stochastic resonance (SR).

Despite the commonly held belief that noise is a hindrance to signal detection, in more recent years SR has been suggested as a means by which recognition of weak sensory input signals may be enhanced by the addition of an appropriate amount of noise (Moss et al., 1994, 2004; Wiesenfeld and Moss, 1995; McDonnell and Abbott, 2009; Aihara et al., 2010). SR occurs in non-linear systems when addition of noise results in improved signal transmission or detection (Collins et al., 2003; Moss et al., 2004; McDonnell and Abbott, 2009). Recently this phenomenon has been explored with the idea of improving physiological systems through optimizing neuronal noise (McDonnell and Abbott, 2009). Some studies have noted changes in autonomic responses attributed to SR, such as enhanced heart rate and muscle sympathetic nerve activity under conditions of hypovolemic stress, which likely resulted from improved baroreceptor signaling by adding noise directly via carotid sinus baroreceptors (Hidaka et al., 2000, 2001; Yamamoto et al., 2002). Additionally, SR using imperceptible stochastic electrical stimulation of the vestibular system, applied to normal subjects, has been shown to improve the degree of association between the weak input periodic signals introduced via venous blood pressure receptors and the heart rate responses (Soma et al., 2003). When the stochastic current is applied to the vestibular system over 24 h, SR improves the long-term heart rate dynamics and motor responsiveness as indicated by daytime trunk activity measurements in patients with multisystem atrophy, Parkinson's disease, or both, including patients who were unresponsive to standard levodopa therapy (Yamamoto et al., 2005). If detection of weak sensory signals can be improved through SR, then the sensorimotor coordination which accounts for the maintenance of equilibrium also stands to benefit. One proposed hypothesis is that noise generates small changes in receptor transmembrane characteristics allowing the detection of a weak stimulus (Gravelle et al., 2002; Collins et al., 2003; Mulavara et al., 2011). There is evidence that all three sensory systems responsible for balance (i.e., vision, somatosensation, and vestibular) are capable of improved detection of weak sensory signals through SR, thus helping improve performance of balance control (Collins et al., 2003; Priplata et al., 2003, 2006; Sasaki et al., 2006, 2008; Aihara et al., 2008; Pal et al., 2009; Mulavara et al., 2011; Goel et al., 2015). Numerous studies have shown the signaling capacity of somatosensory afferents to be enhanced with the addition of noise stimuli delivered just at or below that of perceptual thresholds (Collins et al., 1996a,b; Dhruv et al., 2002; Liu et al., 2002; Khaodhiar et al., 2003). There is evidence that enhancing signal detection of cutaneous afferents, with the addition of subthreshold noise delivered via a mechanical stimulus such as vibration to the soles of the feet can improve balance (Priplata et al., 2002, 2003, 2006) and locomotion performance (Galica et al., 2009; Stephen et al., 2012). A few studies have noted improvements in balance control during standing and walking when using imperceptible amounts of stochastic electrical current delivered through the vestibular system, which are likely occurring through means of SR both in healthy controls and in patients with a variety of neurological disorders (Pal et al., 2009; Mulavara et al., 2011, 2015; Goel et al., 2015; Samoudi et al., 2015; Wuehr et al., 2016a,b). Stochastic vestibular stimulation (SVS) was found to decrease anterior-posterior (A/P) sway in Parkinson's Disease patients when a small stochastic current (max amplitude of 0.1 mA) was delivered across the vestibular end organs by placing two cathodes on the mastoid processes and an anode over the C7 vertebra. The improvement was very small (4.5%), although a significant enhancement in a normal population was not noted (Pal et al., 2009). More recently, balance performance with SVS delivered in a mediolateral (M/L) fashion (binaural bipolar vestibular stimulation with electrodes placed directly over the mastoid processes) has been shown to improve balance while standing on a foam compliant surface (Mulavara et al., 2011; Goel et al., 2015). Likewise, using similar amplitudes of M/L SVS during locomotion improved walking stability in patients with bilateral vestibulopathy (Wuehr et al., 2016a) as well as normal healthy subjects (Mulavara et al., 2015; Wuehr et al., 2016b).

To determine if SVS can enhance short-term strategic adaptation in a locomotion task, we used subthreshold bipolar binaural SVS in a novel visual and somatosensory discordant environment. We hypothesized that subthreshold electrical vestibular stimulation with white Gaussian distributed noise SVS would significantly improve locomotor adaptation to a novel sensory discordant environment, compared to controls receiving zero vestibular stimulation.

# METHODS

# Participants

A sample of 27 healthy individuals (15 males, 12 females, age = 28.9 ± 5.0 years, height = 167.7 ± 9.2 cm, weight = 72.4 ± 15.0 kg), with no known musculoskeletal or neurological deficits were recruited from the Department of Health and Human Performance at the University of Houston. Written informed consent was obtained from each subject prior to the start of the experimental procedures. Approval to conduct this study was granted by the Committees for the Protection of Human Subjects at the University of Houston, which conforms to the Declaration of Helsinki.

# Procedures

All subjects were required to fill out a physical activity readiness questionnaire (PAR-Q) prior to the study session. Individuals were excluded if the PAR-Q indicated they had any known neurological dysfunction, recent bouts of vertigo, bone or joint issues, prior bad experience with Galvanic vestibular stimulation, poor vision, were pregnant, diabetic, epileptic, had balance or gait problems, or had any major surgeries recently that might impact their balance or locomotion. Height, weight, shoulder height, and shoulder width were measured on all the subjects just prior to preparing them for the electrode placement. Although not a prerequisite for the study, all subjects were right-hand dominant.

# Electrode Placement

All subjects sat in a chair while the skin over the mastoid processes were cleaned and prepared for electrode placement. Two 5 × 10 cm electrodes (Axelgaard Manufacturing, Fallbrook, CA, USA) coated with a thin layer of Signa Gel <sup>R</sup> (Parker Laboratories Corp., Fairfield, NJ, USA) were centered over the two mastoid processes and two soft foam pads were then placed over the electrodes and secured in place by a head strap. Impedance between the electrodes was always confirmed to be less than 1 k.

# Thresholding Task

After electrode placement, a thresholding task designed to identify the level of electrical vestibular stimulation at which subjects could discern head motion induced by the stimulation was conducted, using methods described in previous papers (Goel et al., 2015; Mulavara et al., 2015). Subjects sat on a stool without a backrest, with their feet on the footrest, and held a gamepad (Logitech Gamepad F310, Lausanne, Switzerland). Utilizing the gamepad, they indicated their ability to perceive a sinusoidal bipolar stimulus current (which produces a side-toside head motion sensation) applied between the electrodes. The exact instruction given to each subject was: "Use your dominant hand to push a joystick depending upon the direction of the motion sensation; make sure to do it as long as you feel the sensation". A 1 Hz sinusoidal electrical stimulation signal was chosen for motion threshold determination in this study. In general, the stimulus profile consisted of 15 s periods with the 1 Hz sinusoidal stimulation signals, interspersed with 20–25 s periods of no stimulation. The different current peak amplitudes used were 100, 200, 300, 400, 500, 600, 700, 900, 1100, 1300, 1500 µA. The order of the stimulation levels, within the profile was randomized for all subjects. Total duration for this task was 463 s. For the joystick data, percentage time of "perceived motion reported by the subject" for each stimulation and baseline periods was calculated. Joystick movement was interpreted as "perceived motion reported by the subject," when its output amplitude exceeded 0.05 V (full-scale movement recorded in 0–5 V range). The percentage time at each stimulation and baseline level for perceptual and body motion detection was normalized with respect to the largest value across all levels of stimulation. A binomial distribution function was fit to the data with a generalized linear model and a logit link function, which is very common in psychophysical studies (Treutwein, 1995). Threshold was defined as the amplitude of stimulation at the point of subjective equality, at which there is a 50% chance of motion detection.

Upon finishing the thresholding task and calculating a subject's perceptual threshold, subjects were then randomly assigned to one of two groups: those receiving maximum amplitude SVS at 50% of their calculated perceptual threshold (stim), and those receiving zero stimulation (control). Post analysis two-tailed independent-samples t-tests confirmed that the two groups did not differ significantly in anthropometric measures (height, weight, shoulder height, and shoulder width), perceptual threshold levels, or baseline times for the Functional Mobility Test (FMT) without goggles. Those who received zero stimulation were given zero amplitude of current delivered to their electrodes, while those who received the SVS (stim) were given a zero mean, Gaussian distributed white noise electrical signal in a wideband 0–30 Hz frequency range during all trials of the locomotion task. In both groups, the device was switched on just prior to subject beginning the course. Root mean square (RMS) of the signals was checked to be [(26 µA RMS/100 µA peak) ± 5%]. This type of wide-band vestibular noise has been used in prior studies and found to benefit balance metrics in both static stance (Mulavara et al., 2011; Reschke et al., 2014; Goel et al., 2015) and dynamic locomotor control (Mulavara et al., 2015). We have reasoned with supporting evidence from other studies (Dakin et al., 2007; Songer and Eatock, 2013) that the wideband range of 0–30 Hz can improve postural control performance in standing and walking tasks by stimulating vestibular hair cells (VHCs) that affect posture and evoke vestibulo-myogenic response in the lower limbs (Mulavara et al., 2011, 2015; Goel et al., 2015).

# Locomotion Task/Functional Mobility Test (FMT)

All subjects performed 12 trials of a locomotor assessment consisting of a slightly modified version of the FMT, which has been previously used to assess locomotor capabilities of astronauts returning from long-duration spaceflight (Mulavara et al., 2010; Cohen et al., 2012), bedrest subjects (Reschke et al., 2009), and healthy individuals alike (Moore et al., 2006; Mulavara et al., 2009). All subjects wore the portable current stimulator and electrodes during 12 FMT trials. Neither group reported any sensations of electrical stimulus throughout the FMT trials. Although the stim group did receive SVS during their trials, it was below their perceptual threshold and not perceptible. Thus, subjects were blinded to which group they were in. The 12 FMT trials consisted of three baseline (B<sup>1</sup> – B3) trials without vision distortion and nine goggle (G<sup>1</sup> – G9) trials performed with subjects wearing vision reversing prism (up/down) goggles in order to provide visual discordance (vision is distorted by the goggles). The entire FMT course was performed on 10-cm thick foam to make somatosensory/proprioceptive input unreliable to the subjects and provide a greater postural challenge. In addition to providing a sensory discordance, the foam also served as an added safety benefit. If any subjects were to fall, it created a soft landing surface to prevent injury. Thus, in trials with visual discordance, the most reliable feedback system supporting balance control was the vestibular system. The sensory discordant conditions of the FMT provided an ideal amount of challenge in most subjects, allowing us to observe adaptation curves in our primary metric, time to complete the course TCC in seconds.

Subjects wore socks and were instructed to navigate the FMT course as quickly as they could without running or touching any obstacles (**Figure 1**). Before each FMT trial, subjects were walked to a starting line approximately six inches from the foam surface. Both subjects' feet were moved up to a position just touching this start line prior to beginning. Timing of a trial began on a subject's first movement. Subjects then stepped onto the foam and toward the first "portal" obstacle, which consisted of two successive 31-cm high Styrofoam blocks placed on the foam surface with a horizontal bar hung from the ceiling between the two blocks at a height adjusted to that of the subject's shoulders. The "portal" required subjects to bend at the waist or lower themselves to avoid hitting the bar and balance on a single foot on the compliant surface while they stepped over the Styrofoam barriers. Next a "slalom" section consisted of four foam poles placed vertically from the floor, which made subjects change head directions and challenged their spatial awareness as they navigated around the poles. A larger 46-cm high Styrofoam block was then placed after the first "slalom" section, and it again required subjects to balance with one foot on the compliant surface as they stepped over. After stepping over, subjects then turned and went through a narrow "gate" which consisted of two foam poles hung vertically from the ceiling at a distance set to the subject's shoulder width. Subjects often elected to go through the "gate" sidewise in an attempt to not touch the poles. Once making it through the "gate," subjects then came back through a second "slalom" and "portal" section. Timing of the subjects stopped once both feet touched down on the hard surface located just after the last "portal" section.

After each trial of the FMT was performed, subjects were asked to rate on a scale of 1–5 (with one indicating none and five indicating a great deal) sensations of electrode irritation, nausea, and their degree of difficulty balancing. Additionally, after all FMT trials were performed, subjects were asked to indicate again on the same scale of 1–5 if they had any sensations of pain, tingling, itching, burning, vertigo, fatigue, nervousness, difficulty concentrating, changes in headache perception, general unpleasantness, or visual sensations throughout the FMT trials. Subjects had also been asked to rate these 11 sensations immediately after threshold testing in order to gauge if subjects were experiencing any adverse effects from the vestibular stimulation.

# Data Analysis

There were no significant differences in TCC between the stim and control group for any of the three baseline trials (B<sup>1</sup> – B3), and they were therefore not included in the subsequent analyses described below. TCC data for remaining trials were normalized to each subject's first trial with goggles by the following equation: G<sup>x</sup> = (T<sup>x</sup> / T1) × 100. G refers to the normalized time for a specific goggle trial number (x). T<sup>1</sup> represents the TCC (in seconds) for the first trial with goggles on. The time metrics for G are expressed as a percentage of the time it took during the first trial with goggles on (T1). Thus, G<sup>1</sup> is always equal to 100%.

Microsoft Excel (Microsoft Corp., Redmond, WA, USA) was used to determine the best fit adaptation curves to the normalized TCCs for each subject's nine trials of the FMT with goggles on (G<sup>1</sup> – G9). The fit of four different types of functions were evaluated: exponential, logarithmic, polynomial, and power. Overall means of the R<sup>2</sup> for the power function (mean R <sup>2</sup> = 0.906) indicated it fit the data better than the exponential (mean R <sup>2</sup> = 0.805), logarithmic (mean R <sup>2</sup> = 0.878), and polynomial (mean R 2 = 0.888) functions. Therefore, the power function was used to characterize the short-term, strategic adaptation rates for the nine FMT trials performed with goggles on. The power function used was y = c(x α ), where y is the estimated normalized time, c is a constant, x is the goggle trial number (1–9), and α is an exponent value that can represent how steep the adaptation curve is. Specifically, the more negative values of α become, the steeper the curves are, indicating faster rates of strategic adaptation to the sensory discordant environment.

As the effect of SVS on normal strategic adaptation was paramount to this study, a criterion was established to exclude outliers displaying this lack of adaptation. It has been shown previously that not everyone displays typical strategic adaptation to certain tasks (Bock, 2005). Visual inspection of adaptation suggested that three subjects had extreme difficulty in improving their performance on the locomotor task with goggles, and therefore appeared to be statistical outliers, as the supplementary image shows (Appendix A). We then determined that these subjects' maximum improvement were at least two standard deviations below the mean improvement of all 27 subjects, and they were removed from further analysis. Thus, a total of 24 subjects were analyzed, which consisted of 12 subjects in each group (control, n = 12; stim, n = 12).

Statistical analysis for the data was performed using SPSS 20 (IBM Corp., Armonk, NY, USA). The specific variables assessed in the study were the normalized times from trials with goggles on (G<sup>2</sup> – G9; G<sup>1</sup> was not compared between the groups because all values would be equal to 100%), and the exponent values for the power trendlines indicating the adaptation rates (α) with goggles on. Shapiro-Wilk's tests and evaluations of Q-Q plots were utilized to check for data normality. Homogeneity of variance was assessed using Levene's tests for equality of variances with significance set at p < 0.05. One-tailed independent-samples

t-tests were used to compare differences between the two groups (control vs. stim) with significance also set at p < 0.05.

# RESULTS

**Figure 2** shows each subject's exponent value (α) for the power function [y = c(x<sup>α</sup> )] used to represent the strategic adaptation rates during the nine trials of the FMT done with goggles on. The initial one-tailed independent-samples t-tests revealed significantly faster adaptation rates (α) in the stim (n = 12) group than the control (n = 12) group [t(16.18) = 2.00, p = 0.031], but no significant differences were found between the two groups (control and stim) for any of the normalized TCC times completed during visual discordance. Multiple studies have noted that SVS does not always reveal an effect of improved balance performance in all subjects receiving the stimulation (Mulavara et al., 2011, 2015; Goel et al., 2015). We expected those individuals responding to SVS would display faster adaptation rates than most of the control subjects. Therefore, we identified a subgroup of "responders" as stim individuals whose adaptation rates (α) were faster (more negative) than the lower 95% confidence interval (CI) bound from the control group's mean adaptation rate. Having identified seven responders, we then performed a second analysis with independent-samples t-test comparisons between the control group (n = 12) and the "responders" (n = 7). Consequently, individuals in the stim group who did not have adaptation rates faster than the lower 95% CI bound of the control were defined as "non-responders". Identification of the "responders" and "non-responders" subgroups within the stim group can also be indicated in **Figure 2** by color shading.

In the second analysis when comparing the stim group "responders" (n = 7) to the control group (n = 12) during visual discordance, several significant comparisons emerged by the fifth trial. As **Figure 3** indicates, normalized TCC for goggle trials five through nine (G<sup>5</sup> – G9) were significantly faster with the "responders" of the stim treatment than the control group [G5: t(17) = 1.78, p = 0.047; G6: t(17) = 2.15, p = 0.023; G7: t(16.11) = 4.64, p < 0.001; G8: t(16.18) = 4.56, p < 0.001; G9: t(17) = 3.27, p = 0.003]. As would be expected when we split the stim group by adaptation rates (α) falling above and below the lower bound of the 95% CI for the control group, the "responders" (n = 7) were confirmed to have significantly faster adaptation rates than the control (n = 12) group [t(17) = 5.88, p < 0.001]. Thus, those responding to SVS had significantly faster adaptation rates than controls who received the zero stimulus, and these faster rates of adaptation led to significantly faster trial times by trial number five (G5) under conditions of visual discordance. Once reaching a significant level, these significantly faster trial times in responders continued to be maintained throughout the additional visual discordant trials performed in the experiment (G<sup>6</sup> – G9), as **Figure 3** shows. To assess whether there were differences in adaptation and normalized TCC between the control and our identified "non-responders", we conducted two-tailed independent-samples t-tests. These results revealed no difference in adaptation rate, and at only a single time point (G9) were the "non-responders" slower than the control group.

Rated sensations for degree of difficulty balancing were generally larger for the goggle trails compared to the baseline conditions (B<sup>1</sup> – B3), with G<sup>1</sup> being the greatest reported average (G<sup>1</sup> mean rating = 4.1). These higher ratings indicated that the vision reversing prisms posed a fairly difficult challenge to subjects' balance during locomotion. Perceived nausea and electrode irritation ratings were very low on average (≤ 1.1 for each sensation) and did not change much throughout the trials (mean change ≤ 0.3 units for each sensation).

FIGURE 2 | Shows each subject's exponent value (α) for the power equation [*y* = *c*(*x* <sup>α</sup>)] trendlines used to represent the strategic adaptation rates during the nine trials of the FMT done with goggles on. Values more negative indicate faster adaptation. Squares represent the control group while diamonds depict the stim group adaptation rates. Group means ± the 95% CI are represented by the larger dark filled shapes with error bars. The asterisk denotes significantly faster adaptation rates (\**p* < 0.05) in the stim group (*n* = 12) compared to the control group (*n* = 12). The top five gray shaded diamonds depict five subjects in the stim group whose adaptation rates were not faster than the lower bound of the control group 95% CI and were considered "non-responders" to SVS. Consequently, the bottom seven hollow diamond shapes from the stim group depict adaptation rates of those who were considered "responders" to SVS. The horizontal dotted line represents the cut-off criteria for "responders" and "non-responders" in the stim group (the lower bound of the control group's 95% CI, thus establishing the criteria that "responders" had to have faster adaptation rates than most of the control subjects).

# DISCUSSION

In this study, we investigated if SVS could improve locomotor performance within an adaptation paradigm. We hypothesized that subthreshold levels of electrical broadband white noise delivered to the vestibular system could improve adaptation to a novel sensory discordant environment. The data suggests that adaptation rates were faster in the subjects who received SVS than the controls. Moreover, in a subgroup of subjects who were responsive to SVS, short-term strategic adaptation to the visual and somatosensory discordant environment of the FMT seemed improved relative to controls and non-responders by the fifth trial. We proposed the improvements in short-term strategic adaptation seen in those responsive may have been caused by better detection of vestibular input provided via the SR phenomenon.

The exact mechanism by which SR occurs requires further elucidation. It has been proposed that Galvanic vestibular stimulation acts on spike trigger zones of vestibular afferents (Goldberg et al., 1982, 1984). However, other studies have shown that Type I mammalian vestibular hair cells may have mechanical responses evoked by low frequency electrical current, where rotational mechanical characteristics of the stereocilia may be changed, modifying the hair bundle position, and effecting transduction during head tilt and acceleration (Zenner and Zimmermann, 1991; Zenner et al., 1992). Additionally, it has been reported that subthreshold SVS may improve postural control by facilitating the vestibulo-spinal control system or other non-dopaminergic pathways as has been suggested in Parkinson's disease patients as well as healthy individuals (Pal et al., 2009; Mulavara et al., 2011; Samoudi et al., 2015). All of the above mechanisms may play an important role in the improved behavioral responses observed with SVS. Regardless of the exact mechanism responsible for the SR phenomenon, the improvements to locomotion and postural control when applying subthreshold amounts of a bipolar electrical broadband white noise stimulus to the vestibular system have been well documented (Mulavara et al., 2011, 2015; Goel et al., 2015; Samoudi et al., 2015). Studies have suggested that SVS is capable of improving postural control and locomotor stability in numerous populations with balance deficits, individuals with neuropathies, the elderly, recurrent fallers, and those with Parkinson's disease (Mulavara et al., 2011, 2015; Goel et al., 2015; Samoudi et al., 2015). Furthermore, studies observing these benefits of SVS on balance performance have done so with only a few minor or no adverse effects reported in subjects (Goel et al., 2015; Samoudi et al., 2015). Consistent with this prior research, our subjects did not report any significant adverse effects credited directly to the SVS. The most commonly reported sensations were slight nausea or vertigo, minor headache from the head strap, and difficulty balancing attributed to the sensory discordant conditions with up/down vision reversing prism googles on, experienced during the FMT.

It is worth noting that not all subjects receiving the SVS responded in a manner showing improved strategic adaptation performance beyond that of the controls. **Figure 2** shows that five out of the 12 subjects receiving vestibular stimulation did not have adaptation rates (α) better than the lower bound of the 95% CI from the control mean. We considered these subjects to be "non-responders" to the vestibular stimulation they received. In our previous studies exploring SVS, we have noted rates of "nonresponders" to be around 30% (Mulavara et al., 2011, 2015; Goel et al., 2015; Samoudi et al., 2015). The results in the present study suggests a slightly higher rate of non-responders (5/12 = 41.7%).

The present results suggest that application of low amplitude SVS may be able to assist a countermeasure that has been proposed to improve locomotor capabilities in astronauts after spaceflight. This countermeasure utilizes the phenomenon of adaptive generalization to train the sensorimotor adaptability (SA) capabilities of astronauts (Bloomberg et al., 2015). Adaptive generalization suggests that repeatedly adapting to conflicting sensory environments fosters people's ability to adapt to new, novel displacements (Welch et al., 1993). Practice solving certain classes of motor control problems enables them to adapt faster or "learn to learn." The type of training to make use of this concept of adaptive generalization has been termed SA training.

It has been suggested that SA training programs that expose astronauts to varying sensory input and balance challenges can enhance their ability to assemble appropriate motor patterns in sensory discordant environments, thus improving their ability to quickly adapt. In addition to the lack of postural control and locomotor capabilities post-flight in astronauts which have been well documented (Paloski et al., 1992, 1994; Paloski, 2000; Mulavara et al., 2010), an increased reliance on visual feedback during recovery from spaceflight has also been reported (Reschke et al., 1998). Numerous studies have reported that subjects relying more on vision during locomotion have difficulty adapting walking and postural control strategies when in novel sensorimotor discordant environments (Hodgson et al., 2010; Brady et al., 2012; Eikema et al., 2013). It is suggested however that subjects who are more visually dependent can be trained to better utilize other sensory modalities such as those provided by vestibular and somatosensory inputs (Wood et al., 2011; Mulavara et al., 2015). Future studies in SR should address the issue of sensory bias and evaluate whether or not improved vestibular and somatosensory signal detection can reduce reliance on vision in postural control and locomotion tasks. If those who are more visually dependent can be found to explore other sensory modalities better with the assistance of SR, then effectiveness of SA training may be further enhanced.

As noted, not all subjects responded to SVS, at least not in a measurable way. Reasons for unresponsiveness to the noise stimulus can incur much speculation. Perhaps receiving noise in these individuals at amplitudes of half their perceptual thresholds was not ideal, although we have previously shown ideal amount of SVS provided to improve postural control in a Romberg posture task to be around 46 to 53% of perceptual thresholds (Goel et al., 2015). A recent study found average optimal improvement in locomotor stability to be achieved at vestibular stimulus levels approximately 35% of the perceptual threshold (Mulavara et al., 2015), thus optimal dynamic stability might be achieved at slightly less levels of stimulation. Additionally, the current study did not take into account the potential for various levels of internal noise to exist between subjects. It has been hypothesized that behavioral responses may be optimized by the interaction between the external noise applied and the internal noise already present within the CNS, such that with high levels of internal noise present, less external noise may be needed for optimal performance and vice versa (Aihara et al., 2008; Goel et al., 2015). Adjusting levels of vestibular stimulation received by a metric of internal noise could potentially increase effects of SVS seen on FMT performance. Future studies need to focus on ways to identify these potentially high internal noise individuals that may be unresponsive to the addition of external noise prior to testing, or at least find other objective measures other than task performance which can help to identify potential "responders" and "non-responders" to the noise stimulation. It is also possible that using healthy and relatively young individuals in this study provided us with some subjects whose vestibular systems were already performing at near optimal levels, and thus gaining more sensory input through the additional noise provided may not have been viable (Priplata et al., 2003). Subjects with prior impaired balance capabilities may have shown a greater effect of SVS, had they been used in the present study. Effects of SR on locomotor performance have previously been reported to occur on somewhat of a continuum, where those individuals who display greater gait variability seem to benefit more from SR than younger subjects who show less variability (Galica et al., 2009). In other words, those with greater decrements in their gait ability seem to have a greater chance for SR to improve their performance. Thus, understanding and identifying the limitations of the SR approach via vestibular stimulation is important from a variety of standpoints, relating to the efficacy of using it appropriately in various rehabilitation countermeasures to make it more personalized based on individual characteristics (Seidler et al., 2015; Seidler and Carson, 2017).

A few limitations should be of note for this study. First, the sample size is relatively small. The few number of subjects per group in the analysis could have made it difficult to find significant differences between some comparisons, as well as overinflate the importance of individuals' data when making comparisons that divided them into even smaller groups. As using an adaptation paradigm was paramount to the study though, it was not possible to make within subject comparisons and have subjects serve as their own control. Thus, the total number of subjects collected had to be divided into the between subject comparison groups. Limited funds and time did not make collection of more subjects a viable option. It is suggested that future research observing effects of SVS on adaptation should be conducted with a greater number of participants. Additionally, although the subjects in this study were blinded as to whether or not they were receiving zero or subthreshold SVS, it was not possible to conduct this study in a double-blind fashion. The researcher who assigned the stimulus profile to the current stimulator needed to be present to ensure the device was working correctly and assist in safely spotting the subjects. Great care however, was taken to ensure that the exact same instructions were given to every subject before each trial, and spotting techniques did not change between trials or subjects. Finally, the sole performance metric used in this study was TCC. Although it appeared to be a valid metric to assess postural control during the FMT, as individuals with more difficulty balancing were slower and it has been utilized previously (Reschke et al., 2009; Mulavara et al., 2010), future studies should proceed to collect other forms of kinematic and kinetic data during similar adaptation tasks to more completely characterize the postural control.

# CONCLUSION

Our study indicated that short-term locomotor adaptation to a somatosensory and visually discordant environment may be improved, in some individuals when adding subthreshold amounts of broadband binaural bipolar stochastic electrical stimulation to the vestibular system. The acceleration of

# REFERENCES

Aihara, T., Kitajo, K., Nozaki, D., and Yamamoto, Y. (2008). Internal noise determines external stochastic resonance in visual perception. Vis. Res. 48, 1569–1573. doi: 10.1016/j.visres.2008. 04.022

adaptation as displayed in improved performance is believed possible as a result of SR occurring within the vestibular system. Such improvements in locomotor adaptation could have implications for use as a countermeasure or perhaps in enhancing other countermeasures like SA training programs that strive toward optimizing adaptation capabilities of astronauts who incur sensorimotor challenges during gravitational transition periods (Goel et al., 2015). Results also suggest that SVS may be applicable for improving locomotor performance in populations with balance deficits, such as those with Parkinson's disease (Samoudi et al., 2012, 2015), stroke, diabetic neuropathy, recurrent fallers, or the elderly (Mulavara et al., 2011, 2015). It is suggested that future directions of research in this area should also explore more the possible explanations as to why some subjects appear more responsive to SVS than others, such as observing relationships in sensitivities of various sensory channels (e.g., somatosensory or proprioception, vestibular, or visual). Potential carry-over effects of SVS improving balance should also be studied. Kim noted effects of noisy SVS on brain rhythms were present several seconds after stimulation ceased (Kim et al., 2013), and thus the potential for sustained balance improvements after stimulation warrants further investigation.

# AUTHOR CONTRIBUTIONS

DT helped with study design, collected data, analyzed the data, and wrote the manuscript. YD helped collect the data and edit the manuscript. CL helped with study design and editing the manuscript. JB helped with study design and editing the manuscript. AM helped with study design, analyzing the data, and editing the manuscript.

# ACKNOWLEDGMENTS

This study was supported in part by a NASA-JSC Space Life Sciences Scholarship offered through the Department of Health and Human Performance at the University of Houston and the National Space Biomedical Research Institute through NASA NCC 9-58. The authors would also like to thank Dr. Brian Peters and Dr. Rahul Goel for their collaboration on the topic, as well as Dr. Helen Cohen, Darcy Szecsy, Elisa Allen, and Jan Cook for their contributions to the project.

# SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fphys. 2018.00301/full#supplementary-material

Aihara, T., Kitajo, K., Nozaki, D., and Yamamoto, Y. (2010). How does stochastic resonance work within the human brain? – psychophysics of internal and

external noise. Chem. Phys. 375, 616–624. doi: 10.1016/j.chemphys.2010.04.027 Bloomberg, J. J., Peters, B. T., Cohen, H. S., and Mulavara, A. P. (2015). Enhancing astronaut performance using sensorimotor adaptability training. Front. Syst. Neurosci. 9:129. doi: 10.3389/fnsys.2015.00129


spaceflight with Galvanic vestibular stimulation. Exp. Brain Res. 174, 647–659. doi: 10.1007/s00221-006-0528-1


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Temple, De Dios, Layne, Bloomberg and Mulavara. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Artificial Gravity as a Countermeasure to the Cardiovascular Deconditioning of Spaceflight: Gender Perspectives

#### Joyce M. Evans<sup>1</sup> \*, Charles F. Knapp<sup>1</sup> and Nandu Goswami<sup>2</sup> \*

<sup>1</sup> Department of Biomedical Engineering, University of Kentucky, Lexington, KY, United States, <sup>2</sup> Physiology, Otto Loewi Research Center for Vascular Biology, Immunology and Inflammation, Medical University of Graz, Graz, Austria

Edited by: Lorenza Pratali, Istituto di Fisiologia Clinica (IFC), Italy

# Reviewed by:

Francesca Lanfranconi, Institute of Sport, Exercise and Active Living (ISEAL), Australia Marcel Egli, Lucerne University of Applied Sciences and Arts, Switzerland

#### \*Correspondence:

Joyce M. Evans jevans1@uky.edu Nandu Goswami nandu.goswami@medunigraz.at

#### Specialty section:

This article was submitted to Integrative Physiology, a section of the journal Frontiers in Physiology

Received: 30 January 2018 Accepted: 24 May 2018 Published: 06 July 2018

#### Citation:

Evans JM, Knapp CF and Goswami N (2018) Artificial Gravity as a Countermeasure to the Cardiovascular Deconditioning of Spaceflight: Gender Perspectives. Front. Physiol. 9:716. doi: 10.3389/fphys.2018.00716 Space flight-induced physiological deconditioning resulting from decreased gravitational input, decreased plasma volume, and disruption of regulatory mechanisms is a significant problem in returning astronauts as well as in normal aging. Here we review effects of a promising countermeasure on cardiovascular systems of healthy men and women undergoing Earth-based models of space-flight. This countermeasure is produced by a centrifuge and called artificial gravity (AG). Numerous studies have determined that AG improves orthostatic tolerance (as assessed by various protocols) of healthy ambulatory men, of men deconditioned by bed rest or by immersion (both wet and dry) and, in one case, following spaceflight. Although a few studies of healthy, ambulatory women and one study of women deconditioned by furosemide, have reported improvement of orthostatic tolerance following exposure to AG, studies of bed-rested women exposed to AG have not been conducted. However, in ambulatory, normovolemic subjects, AG training was more effective in men than women and more effective in subjects who exercised during AG than in those who passively rode the centrifuge. Acute exposure to an AG protocol, individualized to provide a common stimulus to each person, also improved orthostatic tolerance of normovolemic men and women and of furosemide-deconditioned men and women. Again, men's tolerance was more improved than women's. In both men and women, exposure to AG increased stroke volume, so greater improvement in men vs. women was due in part to their different vascular responses to AG. Following AG exposure, resting blood pressure (via decreased vascular resistance) decreased in men but not women, indicating an increase in men's vascular reserve. Finally, in addition to counteracting space flight deconditioning, improved orthostatic tolerance through AG-induced improvement of stroke volume could benefit aging men and women on Earth.

Keywords: orthostatic intolerance, microgravity, aging, spaceflight simulations, falls

# CARDIOVASCULAR RESPONSES TO SPACE FLIGHT AND SIMILARITIES TO AGING

A myriad of cardiovascular effects develop during space flight, including an immediate shift of fluid headward and a decrease in central venous pressure (even though transmural central venous pressure increases and therefore venous return of blood to the heart is enhanced) (Buckey et al., 1996a; Levine et al., 2002; Kaderka, 2010; Norsk, 2014). On a longer time scale, astronauts develop a loss of ventricular mass (cardiac atrophy) (Perhonen et al., 2001; Dorfman et al., 2007, 2008), decreased sensitivity of the carotid-cardiac (vagal) baroreflex (Convertino, 2002; Norsk, 2014) and a greater responsiveness of sympathetic neural activity to inflight simulations of standing (Ertl et al., 2002). Overall, the net effect of time spent in space is manifested by decreasing blood pressure and elevation of cardiac output throughout flight implicating peripheral vasodilation as a major body response that may drive the reduction of plasma volume and associated cardiovascular effects. Conundrums exist in the elevation of cardiac output in the face of cardiac atrophy and in the fact that muscle sympathetic nerve activity (MSNA) increases during spaceflight despite, or perhaps to counteract, peripheral vasodilation (Ertl et al., 2002; Levine et al., 2002; Norsk, 2014; Norsk et al., 2015).

Upon return from space missions, cardiovascular effects have been a concern from the time of early Mercury flights when two astronauts were found to have lost tolerance for standing even after a short time spent in weightlessness (Kaderka, 2010). To date, the most persistent post-flight problem has been orthostatic intolerance (OI), as demonstrated in up to 64% of returning astronauts (Buckey et al., 1996b; Platts et al., 2014). Major cardiovascular conditions that present upon return from space flight include increased hematocrit, decreased plasma volume, decreased aerobic capacity, cardiac atrophy, decreased norepinephrine, and decreased vascular responsiveness in response to standing, (even though directly measured MSNA is increased during space flight) (Perhonen et al., 2001; Levine et al., 2002; Kaderka, 2010; Norsk, 2014).

Spaceflight and aging are associated with similar kinds of physiological deconditioning. For example, microgravity during spaceflight has been shown to influence cardiovascular function, cerebral autoregulation, and musculoskeletal function, in a manner that leads to OI upon return to Earth (Blaber et al., 2013; Goswami et al., 2013). Similarly, aging is associated with deterioration of cardiovascular and musculoskeletal systems, which predisposes older persons to dizziness upon standing and/or OI, which can lead to falls and falls-related injuries, and often hospitalization (Blain et al., 2016; Bousquet et al., 2017). Specifically, current scientific knowledge regarding OI and how it comes about provides a framework for understanding (patho-) physiological concepts of cardiovascular (in-) stability in bed rest-confined senior citizens or those on multiple medications (polypharmacy) (Goswami et al., 2017).

Furthermore, since bed rest is used as a model to study effects of spaceflight de-conditioning (Cvirn et al., 2015; Goswami et al., 2015a; O'Shea et al., 2015; Waha et al., 2015) and hospitalized older persons spend a large part of their time in bed, the de-conditioning effects of bed rest confinement on physiological functions and its parallels with spaceflight deconditioning can be exploited to understand and combat both variations of de-conditioning. This knowledge is important as deconditioning due to bed confinement in older persons can lead to a (downward) spiral of increasing frailty, OI, falls, and fall-related injury.

Integration of knowledge regarding deconditioning due to reduced gravitational stress in space, and bed rest-induced deconditioning promotes a comprehensive approach that can incorporate nutritional aspects, muscle strength, and function (Gao et al., 2018), cardiovascular (de-) conditioning, and cardiopostural interactions (Goswami et al., 2013). The impact of such integration can provide new insights and lead to methods of value for both space medicine and geriatrics (Geriatrics meets Spaceflight!) (Goswami, 2017). Finally, as astronauts in space spend substantial amounts of time carrying out exercise training to counteract the microgravity-induced deconditioning – and to counteract OI on return to Earth-, it is logical to suggest some of these interventions for bed-confined older persons.

# Gender Differences

Cardiovascular gender/sex differences noted upon return to Earth, indicate that women are more susceptible to post-flight OI (Fritsch-Yelle et al., 1994; Harm et al., 2001; Waters et al., 2002; Platts et al., 2014) while visual impairment intracranial pressure syndrome appears to be more severe in men (Mark et al., 2014; Platts et al., 2014). In searching for mechanisms responsible for reduced orthostatic tolerance following spaceflight, studies have determined that women demonstrate a greater loss of plasma volume (Waters et al., 2002; Platts et al., 2014), a greater decrease in baroreflex control of heart rate (Fritsch-Yelle et al., 1994; Waters et al., 2002) and an hypoadrenergic responsiveness to orthostatic stress (Waters et al., 2002). Several reports also note that, on Earth, women typically respond to stress with increased heart rate, while men respond with increased vascular resistance (Ludwig et al., 1987; Evans et al., 2001; Arzeno et al., 2013; Mark et al., 2014). The lower vasoconstrictive reserve of women in the Convertino study came in spite of steeper increases in peripheral vascular resistance accompanied by enhanced epinephrine and diminished norepinephrine responses at presyncope, leading the authors to conclude that women clearly demonstrated less effective responsiveness of mechanisms that contribute to blood pressure regulation during orthostatic stress (Convertino, 1998). These results are reinforced by those seen in women's response to both standing (Waters et al., 2002; Meck et al., 2004) and lower body negative pressure (LBNP) stresses (White et al., 1996; Frey and Hoffler, 1998). Finally, coherence between diastolic blood pressure and MSNA is lower in women compared to men at rest and in response to increasing LBNP (Yang et al., 2012).

Recently, a study of post-flight carotid artery stiffness and associated blood biomarkers indicated that, after 6 months in space, carotid artery stiffness and insulin resistance were increased in that group of astronauts with sex differences

noted in pulse transit time, insulin resistance, plasma renin, and aldosterone (Hughson et al., 2016). Overall, the latter study indicated that the significant gender differences from this small group of astronauts who had spent 6 months on the International Space Station, would require additional research to firmly establish gender-specific differences in these important metabolic and vascular remodeling variables (Hughson et al., 2016). Autoregulatory differences in cerebral blood flow may also contribute to greater Ol in women compared to men. Cerebral flow regulation in astronauts who failed a post-flight 10 min stand test was characterized by higher cerebral vasodilation with a significant sex interaction in response to standing: five of eight non-finishers were female while 17 of 19 finishers were male (Blaber et al., 2011). Harm et al. (2001) carried out a compilation of statistics from 140 males and 25 females with respect to occurrence of Ol following 5–16 days of spaceflight and reported that 7% of males and 28% of females developed Ol, similar to that reported by Blaber et al. (2011).

One study established foundations for basic gender differences by using ganglionic blockade to examine reflex responses to vasoactive drugs (Christou et al., 2005). In that study the authors determined that, in the presence of ganglionic blockade, men and women displayed equal increases in blood pressure in response to an alpha 1 agonist. However, before blockade, the women's response was significantly greater than the men's response, clearly establishing that baroreflex buffering of blood pressure by reflex lowering of heart rate was not as robust in reflexive women as it was in reflexive men. The same study also established that women's vasopressin response to the ganglionic blocker was significantly smaller than that of men's, thereby indicating that women's secretion of that hormone to support blood pressure was not as strong as men's (Christou et al., 2005).

# CARDIOVASCULAR RESPONSES TO GROUND BASED SIMULATIONS OF SPACE FLIGHT

# Simulations of Spaceflight

Certain criteria from spaceflight (headward shift of fluid) and return from spaceflight (increased hematocrit and Ol) led to the choice of head down bed rest (HDBR) and immersion (both wet and dry) as the Earth-based protocols that best simulate cardiovascular responses to actual spaceflight. As a general rule, HDBR and dry immersion are used to model long term effects of spaceflight while neck-high immersion in thermo-neutral water is used to model short term effects (Norsk, 2014).

Questions arise concerning the suitability of different models of spaceflight to provoke responses comparable to those seen post-flight. Accompanying this concern are: questions of the appropriate amount of time over which to apply such models in order to induce certain deconditioning effects; the appropriate testing of countermeasures to combat this cardiovascular deconditioning; and the most appropriate model to produce the gender effects seen in actual spaceflight.

# Similarities and Differences Between Spaceflight Simulations

At rest, similarities between HDBR and wet and dry immersion include headward fluid shift, decreased plasma volume, increased venous distensibility, decreased heart muscle strength, and muscle volume (cardiac atrophy) as well as impaired carotidcardiac control of heart rate that occurs on a quicker time scale in response to wet or dry immersion than to HDBR (Fortney, 1991; Levine et al., 1997; Perhonen et al., 2001; Norsk, 2014). Differences between water immersion and HDBR include a larger shift of fluid to the chest accompanied by a greater increase in heart size with immersion and a larger shift of fluid to the head with HDBR. In addition, blood pressure decreases acutely in response to HDBR but not to water immersion (Norsk, 2014).

# Similarities and Differences Between Spaceflight Simulations and Spaceflight

Similarities between immersion to the neck in thermo-neutral water (WI) and spaceflight are the immediate shifting of fluid to the thorax resulting in the acute increase in cardiac preload accompanied by intravascular absorption of interstitial fluid, two mechanisms that seem to dominate the early response to spaceflight (Norsk, 2014).

Similarities of HDBR to spaceflight include a 10–15% decrease in plasma volume accompanied by diminished cardiac performance and baroreflex sensitivity that mirror those of spaceflight (Levine et al., 1997; Dorfman et al., 2007; Pavy-Le Traon et al., 2007). In addition, HDBR subjects experience a rapid decline in aerobic capacity followed by a further decline at a slower rate, impaired vascular reflexes, and altered myocardial mechanics, similar to those observed during spaceflight (Levine et al., 1997; Dorfman et al., 2007; Hargens et al., 2013). In terms of autonomic behavior, HDBR has been shown to result in decreased baroreflex sensitivity in control of heart rate and augmented sensitivity in control of MSNA, also similar to that of spaceflight (Arzeno et al., 2013). Finally, again similar to spaceflight, both cardiac atrophy and MSNA increase across the course of HDBR while the decrease in plasma volume plateaus (Levine et al., 1997; Kamiya et al., 2000; Perhonen et al., 2001; Dorfman et al., 2007; Arzeno et al., 2013; Norsk, 2014).

One critically important difference between HDBR and space flight, is the effect of gravity acting on the intra thoracic pressure of HDBR subjects when compared to the dramatic release of thoracic compression (due to lifting of the weight of the chest walls) in space. Another important difference is MSNA, which, although increasing across both space flight and HDBR, is at higher levels during spaceflight (Norsk, 2014). The effects of MSNA on vascular resistance and thoracic compression on venous return are important and need to be considered when interpreting results of HDBR studies (Norsk, 2014).

# Gender Similarities and Differences in Simulations of Space Flight Similarities

Cardiac atrophy, observed with both echocardiography and magnetic resonance (MR), indicated a loss of left and right

ventricular mass that increased similarly in men and women during 60 days of sedentary, HDBR (Dorfman et al., 2007). Studies have also determined that exercise during HDBR attenuated the loss of cardiac mass or left ventricular compliance to approximately the same degree in both men and women (Arbab-Zadeh et al., 2004; Dorfman et al., 2007). Similarly, a 60 day HDBR study found that some differences in autonomic control (men's tendency toward sympathetic activation and women's parasympathetic dominance in terms of heart rate effect on blood pressure regulation) were preserved even though both parasympathetic modulation and baroreflex sensitivity decreased across bed rest (Arzeno et al., 2013). Greater parasympathetic dominance in healthy, ambulatory women, and greater sympathetic dominance in ambulatory men were previously reported by Frey and Hoffler (1998) and also by our laboratory (Evans et al., 2001). Even short term (4 h) HDBR indicates that both men and women demonstrate similar lowering of blood volume, central venous pressure, forearm vascular resistance, and norepinephrine and higher heart rate and greater loss of stroke volume during LBNP compared to 4 h of seated rest (Edgell et al., 2012).

### Differences

A study comparing orthostatic tolerance before and after 6 h of water immersion with that determined before and after 6 h of HDBR, found that men's orthostatic tolerance limit (OTL) was greater than women's in all cases, and the decrease in tolerance was greater in women than in men after HDBR but not after water immersion (Hordinsky et al., 1981). Two important measures of autonomic function, determined in the Arzeno study, indicated that women's baroreflex sensitivity decreased more than men's in response to HDBR and the group's decrease in systolic and diastolic blood pressure over 60 days of HDBR was due to men, while women maintained blood pressure over the course of the study (Arzeno et al., 2013). Although HDBR-induced decrease in parasympathetic modulation would lead to decreased baroreflex sensitivity, it is unclear what role these changes play in the greater susceptibility to OI and greater loss of plasma volume in female, compared to male, astronauts upon return to Earth (Mark et al., 2014).

A bed rest study by Pavy-Le Trao et al. (2002) showed that, even though there was a slower vasodilatory response to sudden reductions in blood pressure in orthostatically intolerant women, cerebrovascular autoregulation was not impaired in females. However, the large variability in cerebral blood flow responses during HDBR studies, seriously limits the use of HDBR-derived cerebral blood flow data in understanding how cerebral vasculature adapts to microgravity exposure (Blaber et al., 2013).

# CARDIOVASCULAR RESPONSES TO ARTIFICIAL GRAVITY

Artificial gravity (AG) as a countermeasure to physiologic deconditioning of multiple organ systems has long been discussed and proposed (National Research Council, 2011, Chapter 7 of Recapturing a Future for Space Exploration; Shulzhenko, 1992; Vernikos et al., 1996; Vernikos, 1997; Greenleaf et al., 1998; Clement and Pavy-Le Traon, 2004; Evans et al., 2004, 2015; Clement et al., 2016). To date, however, cardiovascular responses to AG applied in the long body axis (to simulate standing on Earth) have come almost exclusively from Earth-based studies in a variety of situations: (1) healthy, ambulatory subjects acutely exposed to hypergravity, (2) healthy, ambulatory subjects before and after a period of AG training, and (3) deconditioned subjects before and after bed rest, HDBR, water immersion, dry immersion, and furosemide-induced simulations of spaceflight. Below, we review what has been learned about the human response to AG in the above environments and seek to determine if gender differences in cardiovascular responses might affect the future of AG as a countermeasure to cardiovascular deconditioning. There are two studies where AG was actually applied to astronauts several times in-flight in order to gather their perception of AG while in space (Benson et al., 1997; Clément et al., 2001). Although those studies did not report actual post-flight cardiovascular responses to gravitational stimulus, later reports (Clement and Pavy-Le Traon, 2004; Clement et al., 2016) stated that none of the four astronauts who underwent AG during that flight exhibited post-flight OI while the three who did not receive AG, did exhibit OI. Although there are a plethora of ground based studies, the need for actual space-based studies focusing on cardiovascular and other system effects of AG is glaring, and has been called out by NASA administrators as a goal for closing critical gaps in the areas of post-flight human performance (National Research Council, 2011, Chapter 7 of Recapturing a Future for Space Exploration; Kaderka, 2010; Mark et al., 2014; Norsk, 2014; Clement et al., 2016).

As glaring as the lack of space-based information from humans who have experienced in-flight AG, is the lack of data from women undergoing AG during bed rest or immersion (wet or dry) studies. Acute responses to AG have been determined from healthy, normovolemic women (Stenger et al., 2007) and from healthy women deconditioned by furosemide (Evans et al., 2015; Zhang et al., 2017). However, there are no AG studies in women deconditioned by HDBR or immersion (wet or dry); the three simulations of space flight considered closest to actual space missions. Therefore, results below will summarize what has been learned from numerous AG studies conducted in healthy men studied in ambulatory conditions and following simulations of space flight, and from the few studies conducted in healthy ambulatory women and the only study of deconditioned (furosemide infusion) women.

# Healthy Ambulatory Subjects Acute Exposure to AG

## **Normovolemic men**

Exposure of healthy, ambulatory men and women to AG is a mainstay of pilot testing and training and such testing has also been used in the general population to collect baseline data for commercial spaceflight (Blue et al., 2012). In that study, data were collected from 65 men and 12 women, 22–88 years old, as they were taken to gray-out with the principal finding being

that gray-out and peak heart rate were inversely related to age. Another study of 22 men and 25 women determined that risk factors for OI to G stress (on NASA Ames' centrifuge) included increased height and reduced plasma volume (Ludwig et al., 1987).

In the last 12 years, centrifugation in the long body axis has been used to select subjects for subsequent bed rest studies where AG was to be tested as a countermeasure to subsequent HDBR (Fong et al., 2007). In that study, 5/6 men were tolerant of a +1 Gz load at the heart (+2.5 Gz at the feet) applied for an hour. A similar study designed to evaluate the role of anthropometric factors in determining cardiovascular stability during two bouts of AG found that tolerance correlated positively with body volume and fat free mass. In that study, 8 of the men were classified as high tolerance and two were classified as low tolerance (Opatz et al., 2014).

### **Normovolemic women**

In the above studies, women were also tested for their tolerance to AG. Being female was a significant risk factor for OI in the Ludwig study (Ludwig et al., 1987). In the Fong study, only 1/5 women were able to withstand a constant +1 Gz load at the heart for an hour, while in the Opatz study, 6/10 women were tolerant of the two bouts of +2 Gz (Fong et al., 2007; Opatz et al., 2014).

Taken together, results of the latter two studies indicate a higher tolerance in men (13/16) than in women (7/15) for matching AG stresses applied at heart level. Further, the intolerance of women for the constant 1 h, +1G protocol of the Fong study was one of the factors that led to the exclusion of women from a subsequent HDBR/AG study (Stenger et al., 2012).

### Chronic Exposure to AG

In examining effects of high G training on orthostatic tolerance of men and women, Convertino et al determined that 4 weeks of AG training (three times a week) at ever increasing orthostatic load, increased calf compliance of both men and women but did not remove the lower orthostatic tolerance of women compared to men (Convertino et al., 1998). A study of effects of chronic AG exposure on subjects deconditioned by dry immersion found that intermittent exposure to 0.8–1 Gz during 7 days of immersion prevented the 28% decrease in orthostatic tolerance seen with immersion alone (Vil-Viliams, 1994).

Due to the sparseness of women's AG studies, the majority of available data in the following section will come from investigations conducted by the authors. Since 1999 we conducted studies of effects of AG training on: ambulatory, acutely deconditioned (furosemide) and bed rest deconditioned men, (Greenleaf et al., 1998; Evans et al., 2004; Stenger et al., 2012; Blaber et al., 2013), and on ambulatory and acutely deconditioned men and women (Greenleaf et al., 1998; Stenger et al., 2007, 2012; Evans et al., 2015; Goswami et al., 2015b; Zhang et al., 2017). In studies conducted before 2007, we looked at effects of AG training (45 min a day, 5 days a week, over a period of 3 weeks) with respect to the ability of short bouts of AG to improve orthostatic tolerance over ambulatory, pretraining tolerance. We found the significant increase in tolerance after AG training to be associated with decreased resting blood pressure and vascular resistance and increased stroke volume (Greenleaf et al., 1998; Evans et al., 2004; Stenger et al., 2007). The improvement in vascular responsiveness was demonstrated through increased low frequency spectral power of blood pressure and heart rate as well as a doubling of the norepinephrine response during tilt (Greenleaf et al., 1998; Evans et al., 2004; Stenger et al., 2007).

## Gender Differences in Ambulatory Responses to Chronic AG Exposure

Similar to tolerance for Earth gravity or matched levels of LBNP, women clearly demonstrate a lower tolerance for orthostatic stress than do men (Ludwig et al., 1987; Frey and Hoffler, 1998; Evans et al., 2004; Fong et al., 2007; Stenger et al., 2007; Opatz et al., 2014). In our studies, 3 weeks of AG training (45 min/day, 5 days a week) improved ambulatory men's orthostatic tolerance more than it improved ambulatory women's tolerance and exercise during AG improved tolerance more than did passive AG. An improvement in orthostatic tolerance was not seen for these ambulatory women unless exercise accompanied the AG sessions (Evans et al., 2004; Stenger et al., 2007). Other investigators determined that a primary source of the increased protection against OI provided by AG training, resulted from an increased ability to mobilize stroke volume and cardiac output during orthostatic stress that was more evident in men than women (Convertino et al., 1998).

In a test of brain cortical activation during stepped increases in G load (to presyncope), beta wave (12.5–35 Hz) activity increased in both men and women, while alpha wave (7.5–12.5 Hz) activity increased only in men (Schneider et al., 2014). This sex difference in cortical activation in response to increasing levels of AG may have implications in the observed sex differences in cardiovascular responses to AG (Stenger et al., 2012).

# Deconditioned Subjects

Recently, Clement and Pavy-Le Traon (2004) and Clement et al. (2016), reviewed studies that explored AG as a countermeasure to deconditioning evoked by bed rest and dry immersion. Results from the 18 studies reviewed, all males, indicated that AG, applied over as little as 30 min twice a day (Sasaki et al., 1999) was successful at mitigating OI in men who were deconditioned by these protocols. Other results included reductions in exaggerated responses to orthostasis, maintenance of autonomic cardiovascular function and attenuation of plasma volume loss. One review also indicated that AG applied intermittently rather than being held at a constant value, was better tolerated by men (Clement et al., 2016). However, the spectrum of AG effects on deconditioned men and women will not be definitive until AG has been tested in deconditioned women.

### Acute Exposure to AG

### **Hypovolemic men**

Recently, we observed that 90 min of AG exposure in a protocol individualized to provide a common stimulus to each person, improved the orthostatic tolerance limit of hypovolemic men compared to a day in which the same men had been mildly deconditioned by hypovolemia plus 90 min of 6 degree

HDBR (Evans et al., 2015). On both study days, subjects had undergone a diet and furosemide protocol to reduce their plasma volume to match the reduction observed during spaceflight. In a companion study, we found that applying a similar, individualized, AG protocol to a separate group of normovolemic men significantly improved the OTL of that group compared to a day in which the same subjects had rested supine for 90 min (Goswami et al., 2015b). Both studies determined that a 90 min exposure to increasing levels of AG (to presyncope) improved a subsequent test of their OTL by up to 30% (hypovolemic men) compared to their OTL on the supine/HDBR day. Mechanisms for improvement in men's orthostatic tolerance appeared to be a result of decreased resting blood pressure accompanied by increased cardiac output at rest and during orthostatic testing following AG exposure (Evans et al., 2015). The Goswami et al. (2015b) study similarly found decreased blood pressure but did not find increased cardiac output in men following their exposure to AG.

In the Clement review of AG as a countermeasure to deconditioning, 18 studies of men undergoing HDBR or dry immersion reported that acute AG applied intermittently was effective to improve orthostatic tolerance, increase MSNA, maintain exercise capacity, and reduce exaggerated responses to orthostatic testing, but was not effective to return plasma volume to normal (Clement et al., 2016).

## **Hypovolemic women**

The hypovolemic study above included women. These women's OTL was improved by the 90 min exposure to AG compared to the day on which they were exposed to 90 min of 6 degree HDBR (Evans et al., 2015). As with normovolemic women and men and hypovolemic men, the mechanism of OTL improvement in hypovolemic women was primarily through increased resting cardiac output (Evans et al., 2015; Goswami et al., 2015b).

An additional report of autonomic responses to the AG exposure of hypovolemic men and women of the Evans study (above) indicated that mechanisms of improved OTL following AG exposure, also included increased responsiveness of the cardiac baroreflex to orthostatic stress (both men and women) (Zhang et al., 2017).

### **Gender perspectives in deconditioned subjects' responses to acute AG**

In addition to the effects of AG to improve baroreflex responsiveness in men and women noted above (Zhang et al., 2017), we determined that AG exposure increased men's low frequency spectral power of systolic blood pressure during the subsequent test of their OTL but did not change women's. In that study we also determined that men's resting blood pressure declined after exposure to AG, but women's blood pressure was not different on the 2 days either at rest or during orthostatic tolerance testing (Evans et al., 2015). Different between the Evans et al. (2015) and Goswami et al. (2015b) studies was the increase in women's cardiac output which was significantly greater on the AG day for women separately in the Goswami study but only as part of the whole group in the Evans study. Similarities between men and women in this study included increased orthostatic tolerance (Evans et al., 2015)and improved baroreflex activity (Zhang et al., 2017) on the day subjects had previously been exposed to AG. Major differences between men and women consisted of decreased blood pressure (Evans et al., 2015) and increased low frequency spectral power of blood pressure in men but not women following exposure to AG (Zhang et al., 2017).

### Chronic Exposure to AG

Studies have been performed to determine effects of AG applied periodically to men during bed rest deconditioning (White et al., 1966; Iwasaki et al., 2001; Pavy-Le Traon et al., 2007; Stenger et al., 2012; Linnarsson et al., 2015). White et al. (1966) used 1.75 Gz (heart level) applied in four, 20 min daily sessions, to prevent the development of OI in men during 10 days of bed rest. Pavy-Le Traon et al.'s (2007) review of 20 years of bed rest studies, included studies in which gravity was used as a countermeasure; it was apparent from early days, that as little as standing 2 h a day would lessen the incidence of OI in men bed rested for 4 days (Vernikos et al., 1996). Iwasaki et al. (2001) study determined that two, 30 min bouts of 2 Gz applied daily, could prevent the development of OI as well as shifts in autonomic balance toward sympathetic dominance during 4 days of HDBR. In 2006, we participated in the first NASA study to survey a comprehensive physiologic response to 3 weeks of HDBR in a group of men who received an hour of AG per day (+1 Gz at the heart) and compared those results to a control group of men who did not undergo the AG exposure (Stenger et al., 2012). In the group of men who received AG, the HDBR-induced decreases in orthostatic tolerance and maximum oxygen consumption were significantly smaller and the profiles of vasoactive hormones in response to head up tilt were improved in the AG group compared to the control group. The recent study by Linnarsson et al. (2015) was able to establish that an intermittent (6 each, 5 min exposures to AG) protocol applied daily to men undergoing 5 days of HDBR resulted in orthostatic tolerance nearer to that of pre-HDBR than did a protocol that delivered 30 min of continuous AG to the same men.

## **Gender perspectives in deconditioned subject's responses to chronic AG**

We could not find any studies that determined women's cardiovascular responses to chronic AG exposure during bed rest, HDBR, or immersion deconditioning. Our group's study of acute exposure to AG following furosemide infusion to induce plasma volume loss similar to that of spaceflight is the only study from which we are able to compare deconditioned men's and women's responses before and after AG (Evans et al., 2015). Similarities between these deconditioned men and women included improved orthostatic tolerance (Evans et al., 2015) and baroreflex function (Zhang et al., 2017) on the day subjects had been exposed to AG. Major differences between men and women consisted of decreased blood pressure (Evans et al., 2015) and increased low frequency spectral power of blood pressure in men but not women following exposure to AG (Zhang et al., 2017).

# CONCLUSION AND PERSPECTIVES

There is emerging evidence that an individual-specific AG training protocol may be a useful tool to assess orthostatic tolerance in both males and females. This has been verified in both normovolemic and hypovolemic men and women. Future studies should consider the usage of individual-specific AG training as highlighted in the Evans et al. (2015) and Goswami et al. (2015b) studies.

Data from the above studies as well as directions from NASA and ESA administrators, indicate that cardiovascular deconditioning as a result of space flight may not be overcome by current countermeasures [Chapter 7 of Recapturing a Future for Space Exploration (National Research Council, 2011)]. Gaps in knowledge include whether AG applied to astronauts will preserve physiologic systems' (including the cardiovascular system) integrity so that astronauts can return safely to live on Earth. There is a cascade of basic studies that will follow the addition of an AG facility to other space-based facilities in order that future investigators be able to ask and answer basic questions as to the protocols that will render space fight safe and will introduce new hardware for future study. For those reasons, the "Crosscutting Issues for Humans in the Space Environment" document recommended that "NASA should

# REFERENCES


reinitiate a vigorous program to. . .develop a simple short-radius human centrifuge for eventual evaluation experiments aboard the ISS."

# AUTHOR CONTRIBUTIONS

JE wrote this review based upon many years of collaborative research into the cardiovascular effects of gravity. CK contributed to this review based upon many years of collaborative research into the cardiovascular effects of gravity. NG contributed to this review based upon his collaborative research into the cardiovascular effects of gravity.

# FUNDING

The research upon which this review is based, was supported by KY NASA EPSCoR Grant NNX07AT58A and the National Center for Research Resources, and the National Center for Advancing Translational Sciences through NIH Grant UL1TR000117. Facilities support was provided by the Gravitational Branch of NASA Ames Research Center and the Human Research Program of NASA Johnson Space Center.



weightlessness. Acta Astronaut. 49, 227–235. doi: 10.1016/S0094-5765(01) 00101-1



**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Evans, Knapp and Goswami. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Indices of Increased Decompression Stress Following Long-Term Bed Rest

#### Mikael Gennser<sup>1</sup> , S. L. Blogg<sup>2</sup> , Ola Eiken<sup>1</sup> and Igor B. Mekjavic3,4 \*

<sup>1</sup> Swedish Aerospace Physiology Centre, Department of Environmental Physiology, CBH, KTH Royal Institute of Technology, Stockholm, Sweden, <sup>2</sup> SLB Consulting, Cumbria, United Kingdom, <sup>3</sup> Department of Automation, Biocybernetics and Robotics, Jožef Stefan Institute, Ljubljana, Slovenia, <sup>4</sup> Department of Biomedical Physiology and Kinesiology, Simon Fraser University, Burnaby, BC, Canada

#### Edited by:

Nandu Goswami, Medizinische Universität Graz, Austria

#### Reviewed by:

Rohit Ramchandra, University of Auckland, New Zealand Johnny Conkin, KBRwyle, United States Matthew Graham White, Defence Science and Technology Laboratory, United Kingdom

\*Correspondence:

Igor B. Mekjavic igor.mekjavic@ijs.si; mikael.gennser@sth.kth.se

#### Specialty section:

This article was submitted to Integrative Physiology, a section of the journal Frontiers in Physiology

Received: 29 September 2017 Accepted: 09 April 2018 Published: 18 July 2018

#### Citation:

Gennser M, Blogg SL, Eiken O and Mekjavic IB (2018) Indices of Increased Decompression Stress Following Long-Term Bed Rest. Front. Physiol. 9:442. doi: 10.3389/fphys.2018.00442 Human extravehicular activity (EVA) is essential to space exploration and involves risk of decompression sickness (DCS). On Earth, the effect of microgravity on physiological systems is simulated in an experimental model where subjects are confined to a 6 ◦ head-down bed rest (HDBR). This model was used to investigate various resting and exercise regimen on the formation of venous gas emboli (VGE), an indicator of decompression stress, post-hyperbaric exposure. Eight healthy male subjects participating in a bed rest regimen also took part in this study, which incorporated five different hyperbaric exposure (HE) interventions made before, during and after the HDBR. Interventions i–iv were all made with the subjects lying in 6◦ HD position. They included (C1) resting control, (C2) knee-bend exercise immediately prior to HE, (T1) HE during the fifth week of the 35-day HDBR period, (C3) supine cycling exercise during the HE. In intervention (C4), subjects remained upright and ambulatory. The HE protocol followed the Royal Navy Table 11 with 100 min spent at 18 m (280 kPa), with decompression stops at 6 m for 5 min, and at 3 m for 15 min. Post-HE, regular precordial Doppler audio measurements were made to evaluate any VGE produced post-dive. VGE were graded according to the Kisman Masurel scale. The number of bubbles produced was low in comparison to previous studies using this profile [Kisman integrated severity score (KISS) ranging from 0–1], and may be because subjects were young, and lay supine during both the HE and the 2 h measurement period post-HE for interventions i–iv. However, the HE during the end of HDBR produced significantly higher maximum bubble grades and KISS score than the supine control conditions (p < 0.01). In contrast to the protective effect of pre-dive exercise on bubble production, a prolonged period of bed rest prior to a HE appears to promote the formation of post-decompression VGE. This is in contrast to the absence of DCS observed during EVA. Whether this is due to a difference between hypo- and hyperbaric decompression stress, or that the HDBR model is a not a good model for decompression sensitivity during microgravity conditions will have to be elucidated in future studies.

Keywords: bed rest, extravehicular activity, decompression sickness, venous gas emboli, space simulation

# INTRODUCTION

fphys-09-00442 July 17, 2018 Time: 16:48 # 2

Human extravehicular activity (EVA) is an essential part of space exploration, having taken place on the moon, and continuing from space vehicles and the International Space Station (ISS). The hostile environment of space means that protective suits have to be worn during EVA to guard against extreme temperatures, radiation and effectively, no atmospheric pressure. Although the ambient pressure in the ISS is the same as that at sea level on earth (1 ata or 101 kPa), the pressure maintained inside the space suit is significantly lower, to allow some flexibility of movement. The Russian 'Orlan' suit has an internal pressure of 38.6 kPa (comparable to 7440 m altitude), while the United States extravehicular mobility unit (EMU) suit maintains 29.6 kPa (9250 m) (Norfleet and Butler, 2001). This means that personnel are subject to decompression when moving from ISS ambient pressure to that of the suit.

Concomitant with decompression is the risk of decompression sickness (DCS). A number of studies have examined DCS risk at altitudes comparable to the corresponding internal pressure of the space suits mentioned above. Overall, the United States Space program found a 20–40% DCS incidence with groundbased simulated EVAs (Conkin, 2001), although these protocols have produced no reported incidence of DCS in space (Conkin et al., 2016).

It has been hypothesized that microgravity might be protective against both the growth of venous gas emboli (VGE) and incidence of DCS. Reasons for this hypothesis include the positive effect of weightlessness, and adynamia (loss of strength). Balldin et al. (2002)showed that simulated weightlessness (supine position) did not provide any protection against DCS incidence over a control ambulatory group after both groups were exposed to hypobaric conditions, even though there was a significantly higher incidence of bubbles in the ambulatory subjects. Conkin et al. (2017) found a higher incidence of VGE and DCS in subjects who exercised before and during hypobaric exposure than in those who did not exercise during the hypobaric exposure, despite both groups undertaking a prebreathe with exercise to denitrogenate themselves. Another study (Conkin and Powell, 2001) compared lower body adynamia (LBA) that included upper body exercise while at altitude with random walking exercise and planned exercise. They found that LBA treatment appeared to protect against DCS and VGE as effectively as random walking and more than following planned exercise.

The nearest we can come to simulating the effects of microgravity on earth is with a 6◦ head down tilt bed rest (HDBR) experimental model, which causes a pooling of fluids in the upper body similar to that seen in microgravity (Charles and Lathers, 1991). In the longer term, it can also simulate the wasting seen in the musculoskeletal system after a period spent in space. In the present study, an extended period of bed rest (5 weeks) was used to investigate various resting and exercise regimen on the formation of VGE post-hyperbaric exposure.

In hypobaric exposures, decompression is not preceded by a period of increased pressure as in diving but is initiated immediately from saturation at ambient pressure. In a study where exercise (150 knee bend squats) was taken immediately before hypobaric decompression to 6706 m, there was a significant increase in VGE production in comparison to the numbers formed when the squats were performed 1 or 2 h prior to the decompression (Dervay et al., 2002). This led the authors to hypothesize that the micronuclei formed by the eccentric exercise had a half-life of around 1 h under the conditions of the study.

Conventionally, exercise at depth is thought to raise the risk of DCS (Francis and Mitchell, 2003), however, in one study no differences in post-dive VGE were found between exercise at depth and no exercise (Jankowski et al., 2004). Exercise during decompression may also have an effect on inert-gas saturation and the amount of VGE produced post-dive. Studies have found that moderate exercise performed during decompression can reduce VGE production and the risk of DCS (Jankowski et al., 1997, 2004). Conversely, cramped body positions in the decompression phase may reduce inert gas washout and so increase DCS risk (Guilliod et al., 1996). In this regard, hypobaric exposures may produce different results; Balldin et al. (2002) found that a group walking around in a chamber during a 4 h hypobaric exposure to 29.6 kPa produced more VGE at altitude than subjects performing occasional mild exercises while maintaining a supine position for the duration of the exposure. However, the incidence of DCS did not differ between the groups. Karlsson et al. (2009) also found that supine inactivity reduced the formation of VGE during acute altitude exposures.

Post-hyperbaric or hypobaric exposure, there is some evidence that exercise, both isometric and isotonic, made in the peridecompression period may increase the risk of DCS occurring (Pollard et al., 1995). Pilmanis et al. (1999) could find no difference (40% occurrence) between the two modes of exercise in eliciting DCS post-hypobaric exposure to 8992 m for 4 h.

The aim of the present study was to investigate whether prolonged bed rest would have an effect on the amount of VGE produced post-hyperbaric exposure, in comparison to a condition of short supine rest prior to the pressure exposure. Also, hyperbaric exposure (HE) after prolonged bed rest was compared with exercise taken immediately before, during and after the HE. We tested the null hypothesis that the bed rest treatment would give rise to less VGE production than the control situations.

# MATERIALS AND METHODS

# Subjects

Ten healthy male subjects gave their written informed consent to participate in the study. Their body mass ranged from 63 to 98 kg (mean ± SD: 75 kg ± 10 kg), their body mass index (BMI) from 21.0 to 27.2 (mean ± SD: 23.4 ± 2.1) and their age from 21 to 28 years (mean ± SD: 23 ± 2) (**Table 1**).

All were participating principally in a long term (35 days) 6◦ head down bed rest study (HDBR) at the Valdoltra Orthopaedic Hospital (Ankaran, Slovenia). Subjects gave their informed consent to participate in all parts of the study and were free to leave the trial at any time. The study protocol and experimental procedures were in accordance with the Declaration of Helsinki and were approved by the Committee for Medical Ethics at the


TABLE 1 | Subjects' physical characteristics.

fphys-09-00442 July 17, 2018 Time: 16:48 # 3

<sup>∗</sup>Not included in statistical analysis.

Ministry of Health of the Republic of Slovenia. For practical reasons, only 10 subjects could be included in the bed-rest regimen. In addition to the 6◦ HDBR, the subjects also consented to take part in the present study, which required that they were subjected to a number of HEs, before, during and after the bed rest. The exposures involved a variety of interventions, with dives made in a dry hyperbaric chamber at the Josef Stefan Institute, Ljubljana, Slovenia.

# General Bed Rest Procedures

The duration of the HDBR was 35 days. Subjects were accommodated in two rooms and remained in the 6◦ head-down position at all times. Subjects were allowed one pillow and could occasionally lean on one elbow to eat or while being moved to a stretcher. Their arms were allowed to move in and above the horizontal plane, but legs were kept in the tilted plane at all times. Muscular exercise was prohibited. Although they were not allowed alcohol, other food and drink was not restricted and subjects were provided three nutritionist-compiled meals per day.

To ensure that the subjects complied with the study regimen and restrictions at all times, and also for subject safety, video cameras provided 24 h surveillance. Each subject received physiotherapy twice a week and also on request, in order to mitigate neck/back pain and stiffness of joints due to bed rest. The therapy was performed in the HDBR position and consisted of massage, assisted (passive) stretching and assisted joint flexion. Each subject was screened for deep vein thrombosis twice a week, using an Ultrasound/Doppler system with a 6.0–11.0 MHz linear array transducer (Aspen, Acuson, Mountain View, CA, United States) to visualize the popliteal veins bilaterally.

Following bed rest, each subject resumed their normal ambulatory lifestyle and also participated in a supervised exercise training program, consisting of 11–12 1 h sessions of cycle ergometry or lower body resistance training. (For further information regarding the bed rest procedure see Eiken et al., 2008.)

# Hyperbaric Exposures

Subjects were split into pairs for their HE. Each HE followed the United Kingdom Royal Navy standard air Table 11, with 100 min spent at 18 m (280 kPa). Decompression commenced at a rate of 15 m/min, with a stop at 6 m for 5 min, and another at 3 m for 15 min. The floor of the hyperbaric chamber was adjusted to tilt to the same angle (6◦ ) as that of the beds in the HDBR study, with the subjects remaining in a head-down, supine position during each pressure exposure. For the HE that was made during the HDBR period (see T1 below) subjects were transported, lying in a 6◦ HD position, to the Josef Stefan Institute in an ambulance.

# Hyperbaric Protocols

Initially the study was designed to compare bubble data collected after the control HE (C1 – see below, **Figure 1**) against that collected after the HE made during the HDBR period (T1). However, after C1, the subjects produced very low numbers or no Doppler detectable bubbles, so further controls (C2–C4) were brought into the study, each adding a facet to the investigation (exercise prior to the HE – C2; exercise during the HE – C3; subjects upright and ambulatory – C4). This was necessary, as if no bubbles could be detected then there would be nothing to provide a comparison for the bubble grades measured during the HDBR, given that our initial hypothesis was that the HDBR would give rise to less VGE than the control situation.

Control 1 (C1) – Performed prior to the HDBR study, to provide baseline control data. The subjects lay down outside the chamber at the Josef Stefan Institute for 1 h prior to the HE. They were carried inside chamber while supine, were subjected to the dive, then lifted out and remained supine for 2 h while post-decompression Doppler measurements were made.

Control 2 (C2) – Also performed prior to the commencement of the HDBR study. Subjects lay at rest for 1 h prior to the dive, but performed 150 knee bends immediately before entering the chamber over a period of 10 min. The subjects then walked inside and lay down during the HE. Following the HE, the subjects were carried out of the chamber and remained supine while the post-decompression measurements were made over a 2 h period.

Treatment 1 (T1) – Performed during the last week of the HDBR study. The subjects were transported to and from the Valdoltra Orthopaedic Hospital and the Josef Stefan

Institute by ambulance, lying in the HDBR position on the stretchers at all times. They remained supine while carried into the chamber, during the HE and during the 2 h of postdive measurements. They were then transported back to the Valdoltra hospital to finish their bed-rest period.

Control 3 (C3) – Performed 6 weeks after the end of bed rest. Again, subjects lay down for 1 h prior to the HE at rest, were then carried into the chamber and remained supine during the HE. They performed some light cycling on their backs while at depth (Monark cycle ergometer; Idass, Glastonbury, United Kingdom), working at 50 W for 10 min, then resting for 10 min and so on throughout the 100 min bottom time. The subjects were carried from the chamber and again remained supine for 2 h during their post-decompression measurements.

Control 4 (C4) – This treatment was also performed 6 weeks after the completion of the bed rest. The subjects lay at rest for 1 h prior to entering the chamber. They were then allowed to walk into the chamber and sat upright during the HE, then walked out at the end of the HE. During the 2 h post-decompression measurement period, subjects were allowed to move around in between the measurements. Doppler measurements were made on supine subjects, but for flex, each subject stood up and did three knee bends, then lay again for measurement to be graded.

C3 and C4 were performed in the same week. On the first dive, one subject in each pair cycled in the chamber, while one sat upright. Three days later, the pairs returned with the one who had cycled now sitting upright and the other cycling.

In all cases, if necessary, the subjects were covered by blankets when lying outside chamber to keep them warm.

# Doppler Protocol

Precordial Doppler audio measurements were made using a Doppler Bubble Monitor (DBM9008; Techno Scientific Inc., Ontario, Canada) to evaluate any VGE produced postdecompression. VGE were evaluated on the Kisman Masurel scale (Kisman et al., 1978) by an experienced operator, with measurements made at 5 min intervals for the first 30 min after decompression, then at 15 min intervals for a total period of 2 h. Measurements were made while the subjects were at rest in the left lateral decubitus position and also after a 'flex', which involved the subjects kicking their feet away from their body vigorously three times then coming back to rest. This protocol was followed for all of the treatment modes bar number five, where the flex measurements were made after the subject stood up and made three knee bends, then lay down again. As Doppler grades are ordinal data, they are represented by Roman numerals in the text (III, IV etc.).

# Statistics

Maximum bubble grades were noted and the Kisman integrated severity scores (KISS) (Jankowski et al., 1997) were calculated for each subject following each treatment. Although an indirect relationship, the higher the bubble load, the more likely DCS is to occur (Francis and Mitchell, 2003), and large numbers of bubbles over a protracted period indicate a high free-gas load so increasing the risk of clinical symptoms (Spencer and Johanson, 1974; Yount and Hoffman, 1986). Therefore, maximum grades are useful to illustrate the highest number of VGE and to infer some idea of DCS risk to the individual at a particular point in time. In addition, the KISS method denotes an 'index of severity' for each protocol, integrating all of the detected bubbles over the measurement period for each subject. KM Doppler grades are ordinal data, therefore statistical testing is usually performed using non-parametric tests such as the Friedman test and Wilcoxon signed rank test. However, in this experiment where several control situations were added, the low power of these tests would make it difficult to discern any differences. Also, there are no generally accepted post hoc tests for these non-parametric tests. Therefore, as suggested by Baguley (2012) the data for the maximum Doppler scores or KISS scores were rank transformed and then a one-way ANOVA for correlated samples performed on the ranks. The differences between pairs of treatments were tested using Tukey HSD test. The data was calculated on the VassarStats website for statistical computation. Comparisons (ANOVA) were made between the supine controls (C1–C3) and the HDBR situation (T1), and between the supine

and the upright control (C4). The significance level was set at p < 0.05.

# RESULTS

Although ten subjects entered the study, the results of eight subjects only were used in the analysis. Of the two excluded, one (subject BB – see **Table 1**) had already experienced substantial weight loss (≈ 20 kg) before being enlisted in the study and this continued during HDBR [change in weight from first measurement (C1) to last (C4) was 11.2 kg; see **Table 1**]; it was felt that the VGE data might be affected by this weight loss and change in BMI. The change in BMI for this subject, −2.9 kg/m<sup>2</sup> , was larger than the average for the remaining nine subjects plus 3.7 standard deviations. The second subject (SB) was excluded as none of the hyperbaric treatment protocols elicited any VGE in him.

Following C1, only one subject (subject three) produced VGE in the 2 h post-dive measurement period, producing a maximum grade of KM III, with a relatively high KISS score of 34.3 indicating a high bubble load. Across all subjects, the median KISS score for C1 was zero. Following C2, where the subjects performed deep knee bends immediately prior to entering the HE, two subjects (two and six) produced bubbles postdecompression and once again the median KISS was zero. All subjects apart from one (subject seven) produced bubbles after T1 (during the last week of the HDBR). Five of the subjects produced a maximum KM Grade of II or above and the median KISS score was 0.8. After C3, which involved supine cycling at depth, only two subjects produced a small number of VGE (subjects two and three; maximum KM I), with a median KISS of zero. After C4, where the subjects were allowed to sit upright during the HE and move around between measurements post-decompression, VGE were produced in five subjects, with maximum KM grades ranging from I–III and a median KISS of 1, the highest across all of the treatments.

Statistical comparison (ANOVA) of the ranked maximum VGE data for the supine control groups (C1–C3) vs. the HDBR (T1) (see **Figure 2**) revealed a significant difference between the groups (p = 0.00123). There was a difference between all the controls and T1 (p < 0.01), but none of the control situations (C1–C3) differed between each other indicating that the HDBR produced significantly higher maximum bubble grades than the supine control treatments, most of which involved exercise before or during the HE (C2 and C3). Similar results were shown for the KISS scores. Statistical comparison (ANOVA) of the ranked KISS scores for the supine control groups (C1–C3) vs. the HDBR (T1) showed a significant difference between the groups (p = 0.00174). There was a difference between all the controls and T1 (p < 0.01), but none of the control situations (C1–C3) differed between each other indicating that the HDBR condition produced significantly more bubbles than the supine control treatments. In short, both bubble indices showed that overall, the number of VGE produced by a decompression challenge post-HDBR was larger than after the supine control situations.

Comparison of the ranked maximum KM grades for C1– C4 showed a significant (between groups) effect (ANOVA, F = 3.18, p = 0.046) (**Figure 3**). However, none of the individual

FIGURE 2 | The maximum KM Doppler grades for each subject after all of the supine controls (C1–C3) plus the HDBR treatment (T1). Statistical testing was performed on ranked data (ANOVA: F = 7.64, p = 0.00123). Only statistically significant pair-wise differences are indicated in the figure. No signs or symptoms of DCS were noted at any point.

comparisons were statistically significant (C1 vs. C4 p = 0.061), although the KM scores for the upright/ambulant were nominally the largest. The KISS results did not differ significantly between these groups.

# DISCUSSION

The present study shows that a 5 week HDBR increases decompression stress after a HE. During space explorations, upon decompression from normal atmospheric pressure to a hypobaric pressure, astronauts are subject to decompression stress. Unfortunately, it was not technically possible to achieve hypobaric decompressions in this study, as a hypobaric chamber was not available in the vicinity. The difference between a hypobaric decompression from normal atmospheric pressure (in essence, a saturation decompression) and a hyperbaric decompression is twofold; firstly, tissue compartments with longer half-times are stressed during the hypobaric decompression, and secondly, the hyperbaric decompression entails two phases, the wash-in phase when the tissue partial pressure of dissolved gas is increased, and the decompression phase.

The fact that the slower compartments were not challenged in this study is a difference that will have to be borne in mind when comparing these results with decompressions experienced during space walks. The difference in the gas wash-in phase is unlikely to have been of any major importance given that the control test, where the subjects exercised lightly in a supine position during the HE, was not different from the other resting supine control tests (C3 vs. C1, C2).

As shown in **Figure 2**, the maximum bubble grades were significantly lower in the supine controls (C1–C3) than after the HDBR (T1). There was no significant difference between the supine control (C1) maximum bubble grades and the upright control (C4), although there was a tendency toward higher grades in the latter (**Figure 3**). It should be noted that the bubble grades observed following the supine control treatments (C1– C3) were surprisingly low. The dive profile chosen for the study (United Kingdom Royal Navy Table 11) has been used to provoke bubbles in a number of trials investigating prophylactic measures to guard against DCS (Blogg et al., 2010, 2017; Jurd et al., 2011; Gennser et al., 2012), as it is known to regularly produce VGE loads across the complete range of the KM grading scale, but with a low incidence of DCS. However, in the present study, hardly any bubbles were produced following the control HE. Although some subjects' maximum grades were toward the high end of the scale on occasion (**Figure 2**), KISS scores, which give an impression of overall bubble load, were quite low across all of the treatments.

The reason for this scarcity of bubbles is not known but could be explained to some extent by the fact that the subjects in the present study were all young and fit (mean age 23 and mean BMI 23.4), which is in contrast to other studies. For example in the study by Gennser et al. (2012), the mean age of the subjects was 40 years, with a mean BMI of 27.7. Further, Conkin et al. (2003) found that age was significantly related to VGE load, with younger subjects having fewer bubbles. In addition to their young age, the subjects were supine during the whole HE and during the post-exposure measurement period. Although Balldin et al. (2002) did not find any protective effect of supine rest on altitude DCS, they did find a significantly lower incidence of bubbles in supine subjects (81% vs. 51%). Similarly, Karlsson et al. (2009)

noted very low bubble grades in subjects exposed to high altitude during supine rest. Whatever the reason for the very low amount of bubbles produced by the subjects after the supine control dives, this low level of bubbling makes the difference between it and the significantly higher maximum grades produced after the HDBR (T1) more obvious.

There are two potential reasons why HDBR may produce more bubbles post-hyperbaric exposure. The first is concerned with peripheral vasoconstriction provoked during HDBR. In a study investigating core temperature during 35 days of bed rest, it was found that skin temperature decreased progressively over the period, with the distal regions being affected the most (Golja et al., 2002). An associated reduction in blood flow to the peripheral areas would reduce wash-out of inert gas upon decompression, so supersaturation of these tissues and bubble production therein would be more likely post-decompression. This is in direct contrast to the positive effect of weightlessness encountered in space walks, which increases peripheral blood flow back to the heart (Arborelius et al., 1972). (Obviously, the fact that a reduced peripheral blood flow may also have been present during the period at pressure would reduce the washin rate of nitrogen. However, given the rather long period at pressure, the reduced wash-in rate would only affect the tissues with longer half-times.)

However, if peripheral vasoconstriction was responsible for the high bubble production seen after the bed rest, then likewise exercise at depth should produce an increased number of bubbles. The period of exercise would serve to increase blood flow, metabolism and inert on-gassing to all areas of the body during the at-depth period, so causing a greater net inert gas balance with the potential to form more VGE post-decompression. Yet it was found that on comparison with the HDBR (T1) data, the maximum bubble grades observed after exercise at depth (C3) were significantly lower (**Figure 2**). Also, as has been mentioned previously, there was no difference between the supine control pressure exposure with exercise during the hyperbaric phase (C3) and the other supine control pressure exposures (**Figure 3**).

The second possible explanation for these results is that with the extended period of rest afforded by the HDBR, bubble micronuclei located in the endothelial walls of the blood vessels were not destroyed as they might be, if high impact exercise had been undertaken. It is assumed that bubble micronuclei form naturally and would continue to do so during bed rest. Exercise is thought to be able to both create nuclei through injury and cavitation (Vann and Thalmann, 1993; Dervay et al., 2002) and to destroy them (Germonpre et al., 2009; Jurd et al., 2011), with the latter studies suggesting that the net balance is toward the reduction of subsequently formed decompression bubbles, so long as the appropriate form of exercise is undertaken.

Historically, exercise before HE (e.g., diving), particularly that strenuous enough to cause muscle soreness, was thought to be a risk factor for DCS. Microscopic muscle tears were considered sites where micronuclei, and later larger bubbles, could form and grow (Vann and Thalmann, 1993). However, recent studies show that certain types of exercise prior to diving and decompression are protective in terms of reducing VGE formed post-dive, though the timing and the mode of any beneficial exercise has proved to be contentious. Initial work indicated that in rats, a bout of high-intensity exercise performed on a treadmill 10– 20 h before a dive, but not thereafter, reduced VGE formation post-dive (Wisloff et al., 2004). Dujic et al. (2004) corroborated these positive findings in a similar study using human subjects performing 280 kPa dry chamber dives, where treadmill running 24 h prior to the dives also reduced post-dive VGE in comparison to no exercise. Further studies investigated the effect of exercise taken closer to the dive time. Recent work in humans involving both medium and high intensity running exercise commencing 2 h prior to a dive, was found to reduce VGE formation (Blatteau et al., 2005, 2007). Similarly, medium or high-intensity cycling exercise commencing 2 h before an open water dive also reduced VGE grades (Pontier and Blatteau, 2007). Endurance running (45 min continuous sub-maximal) exercise immediately before diving was also shown to significantly reduce VGE formation in comparison to control (Castagna et al., 2011). However, submaximal cycling at either 24 or 2 h prior to a dive was not shown to be beneficial in terms of reduction of VGE in another study (Gennser et al., 2012).

These contrasting data indicate that a complex relationship exists between exercise and VGE production. The mechanisms involved may include nitric oxide production, haemodynamics and fluid balance, as well as the mode of exercise undertaken and its effect on the formation of bubble micronuclei. Thus, when the sub-maximal exercise study (Gennser et al., 2012) was repeated, using sub-maximal running/jumping exercise instead of cycling, it was found that replacing the concentric exercise with moderate-intensity impact exercise 2 h prior to a dive caused VGE formation to be significantly reduced post-dive (Jurd et al., 2011). This suggests that high impact exercise might be capable of dislodging gas nuclei in the blood vessels, a hypothesis that was supported further by a study investigating 30 min of wholebody vibration made 1 h before a dive (Germonpre et al., 2009), as VGE formation was again significantly reduced in comparison to non-vibrated control.

In contrast to the positive effect of exercise, inactivity has been shown to have a deleterious effect on the vascular system and among other consequences cause an increase of endothelial microparticles in the blood (Navasiolava et al., 2010; Boyle et al., 2013). Microparticles are small vesicles released from active and injured endothelial cells (Thom et al., 2013). It has been shown that these particles are compressible by an applied pressure of 790 kPa, and pre-pressurization to that pressure abolished the gas phase (Thom et al., 2013). Here it is relevant to consider the second control experiment when subjects were asked to perform squats prior to the compression. A similar experiment prior to hypobaric exposure has shown an increase in circulating bubbles when the squats were performed just prior to the decompression (Dervay et al., 2002). It was estimated that the bubble nuclei apparently created by the squats had a half-time of approximately 1 h. In the present study, the squats were performed just prior to the start of compression, but 100 min prior to the start of decompression. Thus, the finding that the squats did not increase the amount of bubbles post-decompression may be explained partly by the time delay to the decompression (the chamber did not allow the subjects to stand upright and

perform squats) and partly by the pressure increase that would act to compress any pre-existing gas phase. Therefore, for microparticles to be able to act as bubble nuclei during the subsequent decompression they must either be able to withstand the pressure, or form a gas-phase during the period at pressure prior to decompression. It might be hypothesized that gas nuclei formed during prolonged bed rest have more time to stabilize than microbubbles produced during a short bout of exercise.

This idea has gained some support by recent observations that hydrophobic spots on the luminal surface of blood vessels serve to promote stable nanobubbles, which when exposed to gas supersaturation form decompression bubbles. When isolated vessels were exposed to mechanical stresses in vitro, bubbles were released. The released bubbles appeared to deplete the vessel wall of the hydrophobic material, and thus reduce the subsequent propensity for bubble formation (Arieli et al., 2015).

If one accepts that vascular bubbles are indicative of decompression stress and are related to the risk of DCS (Spencer and Johanson, 1974), then it would appear that bed rest is not a good simulation of microgravity for decompression risk. In space flight, astronauts would be active for most of their waking hours, so they would likely be creating and destroying micronuclei constantly. Although it is known that long space flights induce changes in the vasculature (Charles and Lathers, 1991), it is not known whether excessive microparticle production takes place. There were no cases of DCS in the present study, and overall the bubble loads observed in the young, healthy subjects were relatively low, but it was the HDBR treatment that provoked the largest maximum bubble grades. Although HDBR causes a pooling of fluids in the upper body and the wasting of the musculoskeletal system similar to that seen after a period spent in space, it would seem that it does not simulate the potentially positive benefits of mobility in microgravity that might help to balance the equation and reduce the risk of DCS, leading to the low incidence of DCS reported in astronauts.

The only control situation that came close to be as bubble producing as the HDBR situation was the upright control (C4), where the subjects were allowed to sit up inside the chamber and move around during the Doppler measurement period. Although the subjects were not exercising during the postdecompression monitoring period per se, during the upright control treatment, they performed standing knee bends for the flex Doppler measurement, which is a fairly strenuous muscular

# REFERENCES


activity. It is generally accepted that post-dive exercise increases DCS risk, but there are few studies on this topic. Pollard et al. (1995) determined in rats that post-dive exercise (30 min walking) produced a significantly greater occurrence of DCS than did rest after diving. Van Der Aue et al. (1949) exercised human subjects with arm and leg weight lifting after a variety of long no-stop dives that pushed the boundaries of modern dive profile conservatism. The exercise elicited a 47% occurrence of DCS in comparison to a 22% incidence in resting controls.

Neither the level nor the type of activity performed by a diver or subject post-decompression is often considered closely or well described when monitoring post-decompression bubbles and subsequently managing DCS risk. However, the observation of increased amount of bubbles in the upright control situation compared to the supine controls, despite falling short of a statistical significant difference, indicates the need for a close control of the activity during the post-decompression monitoring period in comparative studies of decompression stress.

# CONCLUSION

In contrast to the suggested protective effect of pre-dive exercise on bubble production, a prolonged period of bed rest prior to a HE appears to promote bubbling post-decompression. HDBR does not seem to be a good model with regards to decompression stress in microgravity when the decompression stress is via HE. Whether long-term bed rest has a different effect on hypo- and hyperbaric decompression stress, will have to be clarified in future studies.

# AUTHOR CONTRIBUTIONS

MG, OE, and IB initiated the project. SLB and MG analyzed and interpreted the bubble grade data. All authors conducted the experimental work and contributed to writing and revising the manuscript.

# FUNDING

The research was supported by a grant from the Swedish National Space board no. 151-12.



#### **Conflict of Interest Statement:** SLB was employed by SLB Consulting.

The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Gennser, Blogg, Eiken and Mekjavic. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Gender-Specific Cardiovascular Reactions to +Gz Interval Training on a Short Arm Human Centrifuge

Zeynep Masatli <sup>1</sup> \* † , Michael Nordine1†, Martina A. Maggioni 1,2, Stefan Mendt <sup>1</sup> , Ben Hilmer <sup>1</sup> , Katharina Brauns <sup>1</sup> , Anika Werner <sup>1</sup> , Anton Schwarz <sup>3</sup> , Helmut Habazettl <sup>1</sup> , Hanns-Christian Gunga<sup>1</sup> and Oliver S. Opatz <sup>1</sup>

#### Edited by:

Nandu Goswami, Medizinische Universität Graz, Austria

#### Reviewed by:

Olivier White, INSERM U1093 Cognition, Action et Plasticité Sensomotrice, France Joyce McClendon Evans, University of Kentucky, United States Jie Liu, Rutgers, The State University of New Jersey, Newark, United States

> \*Correspondence: Zeynep Masatli zeynepmasatli@gmail.com

†These authors have contributed equally to this work.

#### Specialty section:

This article was submitted to Integrative Physiology, a section of the journal Frontiers in Physiology

Received: 28 December 2017 Accepted: 11 July 2018 Published: 31 July 2018

#### Citation:

Masatli Z, Nordine M, Maggioni MA, Mendt S, Hilmer B, Brauns K, Werner A, Schwarz A, Habazettl H, Gunga H-C and Opatz OS (2018) Gender-Specific Cardiovascular Reactions to +Gz Interval Training on a Short Arm Human Centrifuge. Front. Physiol. 9:1028. doi: 10.3389/fphys.2018.01028 <sup>1</sup> Center for Space Medicine and Extreme Environments Berlin, Institute of Physiology, Charité-Universitätsmedizin Berlin, Berlin, Germany, <sup>2</sup> Department of Biomedical Sciences for Health, Università degli Studi di Milano, Milan, Italy, <sup>3</sup> Central Medical School, Monash University, Melbourne, VIC, Australia

Cardiovascular deconditioning occurs in astronauts during microgravity exposure, and may lead to post-flight orthostatic intolerance, which is more prevalent in women than men. Intermittent artificial gravity is a potential countermeasure, which can effectively train the cardiovascular mechanisms responsible for maintaining orthostatic integrity. Since cardiovascular responses may differ between women and men during gravitational challenges, information regarding gender specific responses during intermittent artificial gravity exposure plays a crucial role in countermeasure strategies. This study implemented a +Gz interval training protocol using a ground based short arm human centrifuge, in order to assess its effectiveness in stimulating the components of orthostatic integrity, such as diastolic blood pressure, heart rate and vascular resistance amongst both genders. Twenty-eight participants (12 men/16 women) underwent a two-round graded +1/2/1 Gz profile, with each +Gz phase lasting 4 min. Cardiovascular parameters from each phase (averaged last 60 sec) were analyzed for significant changes with respect to baseline values. Twelve men and eleven women completed the session without interruption, while five women experienced an orthostatic event. These women had a significantly greater height and baseline mean arterial pressure than their counterparts. Throughout the +Gz interval session, women who completed the session exhibited significant increases in heart rate and systemic vascular resistance index throughout all +Gz phases, while exhibiting increases in diastolic blood pressure during several +Gz phases. Men expressed significant increases from baseline in diastolic blood pressure throughout the session with heart rate increases during the +2Gz phases, while no significant changes in vascular resistance were recorded. Furthermore, women exhibited non-significantly higher heart rates over men during all phases of +Gz. Based on these findings, this protocol proved to consistently stimulate the cardiovascular systems involved in orthostatic integrity to a larger extent amongst women than men. Thus the +Gz gradients used for this interval protocol may be beneficial for women as a countermeasure against microgravity induced cardiovascular deconditioning, whereas men may require higher +Gz gradients. Lastly, this study indicates that gender specific cardiovascular reactions are apparent during graded +Gz exposure while no significant differences regarding cardiovascular responses were found between women and men during intermittent artificial gravity training.

Keywords: artificial gravity, gender, short arm human centrifuge, cardiovascular deconditioning, artificial gravity training, countermeasure

# INTRODUCTION

Human space exploration inherently leads to micro-gravity (micro-g) exposure, which induces changes in cardiovascular functioning. The sum of physiological adaptations that occur during exposure to micro-g is known as cardiovascular deconditioning (Komorowski et al., 2016). Changes that occur comprise of cephalic fluid shifts, which lead to increases in cardiac output and myocardial atrophy (Norsk et al., 2015) with resulting decreases in blood volume (Agnew et al., 2004). Further changes that have been recorded are reductions in cardiac diastolic function (Convertino and Cooke, 2005), heart rate (Verheyden et al., 2009), and drastic reductions in systemic vascular resistance (Norsk et al., 2015). All of these adaptations contribute to a 50% reduction in baroreflex functioning (Antonutto and di Prampero, 2003; Beckers et al., 2009), which can lead to orthostatic intolerance (OI) upon return to Earth or any other gravitational environment (Lee et al., 2015). This is characterized by the inability of the neurohumoral reflex, namely increases in heart rate and vasoconstriction, to maintain adequate mean arterial pressure. Additional evidence also suggests that women astronauts undergo profound endothelial dysregulation during gravitational unloading (Demiot et al., 2007), which may contribute to the higher incidence of OI in women astronauts upon returning to Earth (Waters et al., 2002; Wenner et al., 2013).

Research in effective cardiovascular deconditioning countermeasures for both genders is imperative and is as a goal of the Human Research Roadmap put forth by NASA, as well as the EU (Aubert et al., 2016; Vernikos et al., 2016). A countermeasure system that can potentially offset micro-g induced cardiovascular deconditioning is the implementation of artificial gravity (AG) exposure via short arm human centrifuge (SAHC). AG via SAHC creates a gravitational vector along the z-axis of the body thereby stimulating the baroreflex system to induce upsurges in cardiac and vascular resistance activity (Moore et al., 2005). Several Earth-based studies have provided evidence that AG training can improve tolerance against OI, and that benefits of short intermittent AG exposure surpass those of continuous static AG exposure (Stenger et al., 2007, 2012; Young and Paloski, 2007; Goswami et al., 2015a; Clément et al., 2016; Zhang et al., 2017). Additionally, there is also evidence to suggest that short intermittent AG exposure improves tissue oxygenation (Marijke et al., 2017). Much like the benefits high-intensity interval training provides for maintaining aerobic fitness (Gibala et al., 2012; Milanovic et al., ´ 2015), AG training could prove to be an effective counter-measure against OI via short duration and intense training intervals. However, more research must be conducted using Earth-based AG protocols prior to testing in microgravity environments.

The limited number of AG studies performed to date, have primarily involved research in men, and have focused on orthostatic stability upon re-introduction of 1G conditions, and not so much on cardiovascular responses during the AG training itself. The cardiovascular responses in women during SAHC AG exposure have not been thoroughly documented and no studies have specifically have recorded and analyzed gender specific cardiovascular reactions active during an intermittent AG exposure. It has yet to be determined whether women and men exhibit similar or diverging cardiovascular responses during an identical AG protocol. A thorough comparison of gender specific cardiovascular responses during AG is important in order to verify its effectiveness on eliciting the cardiovascular responses needed to overcome an OI event. There is also substantial evidence indicating that cardiovascular functioning differs between men and women (Evans et al., 2001; Hart et al., 2009; Hart and Charkoudian, 2014), particularly cardiac and vascular resistance activity, with women tending to exhibit greater cardiac activity, whereas men tend to respond with heightened vascular resistance activity during orthostatic stress (Shoemaker et al., 2001). In addition, women have exhibited a greater gravity-dependent baroreflex sensitivity than men during orthostatic stress, which leads to profound differences in cardiovascular responses during exposure to an equal level of orthostatic stimulus (Drudi and Grenon, 2014). Prior to deployment of any AG protocol for manned space crews, it must be determined what +Gz gradient is required in order to elicit significant increases in cardiac and vascular resistance in both women and men, while minimizing any OI event during the exposure. Therefore, this study implemented a graded +Gz interval training (GIT) via SAHC in order to establish whether the cardiovascular systems involved in maintaining orthostatic integrity can effectively be stimulated amongst men and women, and to examine if any gender specific cardiovascular responses became apparent.

# METHODS

# Subjects

Twenty-eight healthy, Caucasian civilian subjects gave their written informed consent to participate freely in this experiment, which took place at the German Aerospace Institute (DLR), at the European Space Agency (ESA) short arm human centrifuge (SAHC) test facility in Cologne, Germany. The first 13 subjects were tested in the fall of 2012, while the remaining 15 were studied in summer of 2015. The subject pool consisted of 16 women and 12 men who were matched for age (28.4 ± 5.3 years). Screening comprised of a medical questionnaire and a physical examination performed by an independent general physician who was not involved in the study. This screening examination included a resting ECG, a Schellong test to screen for orthostatic susceptibility, and cycle ergometry to determine baseline cardiovascular fitness. Upon completion of the screening, no subjects were excluded, nor had any history of cardiovascular, metabolic, or neurological diseases. Furthermore, it was ensured that none were commercial or military pilots, professional or elite athletes. This study was carried out in accordance with the recommendations of the Medical Ethics Committee of Nordrhein-Westfalen, Germany (Aerztekammer Nordrhein, Düsseldorf, Germany) in accordance with the Declaration of Helsinki. The protocol was approved by the Medical Ethics Committee of Nordrhein-Westfalen, Germany (Aerztekammer Nordrhein, Düsseldorf, Germany).

# Study Protocol

The SAHC (SAHC-TN-007-VE, QinetiQ Company, Antwerp, Belgium) used for this study has a +Gz range of +0.1 to +5.0, an acceleration rate of maximal +0.4Gz/s, a maximum radius of 2.82 m and a maximum RPM of 40. The ESA-SAHC has been extensively used for +Gz training and various studies in the field of gravitational physiology (Zander et al., 2013; Frett et al., 2015).

The +Gz profile (see **Figure 1**) consisted of an initial baseline for 15 min followed by two identical +1/2/1Gz graded acceleration/deceleration rounds, composed of 3 phases each. These 2 rounds were separated by a +0Gz phase. Except for phase 1 (P1, 15 min) and P5 (5 min), each phase lasted 4 min. Acceleration/deceleration between each phase lasted 15 s. For the first 5 min of P1, the SAHC was not rotated. This was done to ensure good signal quality. For the next 10 min, the SAHC was slowly rotated at 5 RPM. During +1Gz phases (P2, P4, P6 and P8), the gravitational vector along the body was as follows: head +0.3Gz, mediastinum +0.5Gz, and feet +1Gz with an approximate RPM of 16. During +2Gz phases (P3 and P7), the gravitational vector along the body was +0.6Gz for the head, +0.9Gz for the mediastinum and +2.0Gz for the feet with an approximate RPM of 26.4. During P5, the SAHC was rotated at 5 RPM.

The purpose of implementing a graded +Gz protocol was to reduce the occurrence of OI due to rapid inductions of +2Gz. Therefore, the first +1Gz phase (mild hyper-gravity) was used to reduce any sudden occurrence of an OI event and to prime the baroreflex system for a moderate gravitational stress stimulus. The +2Gz phases induced moderate +Gz and the desired cardiovascular reactions. A reduction back to +1Gz allowed for mild +Gz exposure, with the addition of absorbing any carry-over effects of the +2Gz phase. P5 allowed for a return to baseline conditions before starting the second round. The two rounds of +Gz were set up in order to ascertain whether these reactions could be replicated, adding to the principle of interval training.

In case of signs of orthostatic intolerance (OI) during the GIT, the +Gz level was reduced or the protocol was terminated completely. These included perfuse sweating, confusion, dizziness, nausea, a narrowing of the visual field or the presence of hemodynamic decompensation criteria, such as new-onset ECG abnormalities, an abrupt drop in MAP of >20 mmHg or a critical narrowing of the pulse pressure. Additionally, each subject could press a "panic button" if they wished to abort the test. Direct verbal and visual contact was maintained between the flight physician and the test subject via camera and microphone.

# Cardiovascular Recording

Mean arterial blood pressure (MAP, mmHg) and pulse pressure waves were monitored continuously and non-invasively via finger plethysmography (Portapres <sup>R</sup> , Finapres Medical Systems BV monitoring, Amsterdam, the Netherlands). The collected blood pressure waves were analyzed with Beat-scope <sup>R</sup> software. Systolic and diastolic blood pressure (SBP, mmHg and DBP, mmHg respectively) as well as stroke volume (SV, ml) were extracted from the continuous blood pressure waveform. Heart rate (HR, bpm) was recorded via a 3-lead ECG. Cardiac output (CO, L/min) and systemic vascular resistance (SVR, mmHg·min/l) were calculated via the Wesseling formula (Veerman et al., 1995). To control for significant baseline anthropometric gender differences, CO, SV, and SVR were converted to cardiac index (CI L/min/m<sup>2</sup> ), stroke volume index (SVI ml/m<sup>2</sup> ) and systemic vascular resistance index (SVRI, mmHg·min/l/m<sup>2</sup> ).

# Data and Statistical Analysis

All recorded data was synced according to phase and grouped according to gender. The last 60 s from each phase were averaged and used for the statistical analysis. This time section was chosen primarily to provide continuity with a previous study based on the first part of the same experiment, (Habazettl et al., 2016). In that study, microvascular and cardiovascular responses from the last 60 s of each phase were analyzed in order to highlight the maximal response. Furthermore, cardiovascular parameters in most subjects showed oscillation patterns during the first 2–3 min of each phase, thereby exhibiting high variability thus not allowing for an accurate assessment of cardiovascular reactions. Secondly, the last 60 s were used with the goal of reflecting maximal cardiovascular response as well as steady state without the interference of acute changes in +Gz transitions. Diaz Artiles et al. (2016) recorded the above-mentioned cardiovascular trends during an SAHC study, where it was shown that the participants reached maximal response or steady state during the last 60 s of SAHC exposure and the variability during this period was at a minimum.

Statistical analysis of cardiovascular reactions during the GIT were done via repeated measures ANOVA using phase comparisons from baseline as a within and gender as a between

subject factor. To test for equality of variance between men and women the Levene's test was employed. Assumption checks were made using a Mauchly's test of Sphericity followed by Greenhouse-Geiser corrections. Effect size per cardiovascular parameter was calculated via partial Eta squared test. This was done for the whole group as well as independently for women and men. Changes from baseline were further analyzed using t-tests with Holm-Sidak post-hoc correction. To assess for differences in baseline, a student's t-test was used. All data is presented as mean and ± standard error of the mean (SEM). The level of statistical significance was defined at alpha = 0.05. Cardiovascular parameter synchronizing and averaging was performed using "R" statistical environment version 3.2.5 (R Core Team, 2013). Inferential statistics were performed using SPSS Statistics 23 (IBM, New York, USA) and JASP Version 0.8.2 for Mac OS (JASP Team, 2018). Graphics were created using Data Graph 4.2 software for Mac OS (Visual Data Tools, Inc., 2013).

# RESULTS

From the 28 subjects, 23 (11 women and 12 men) completed the GIT without any interruption. All five non-finishers (NF) were women and will be referred to as womenNF. One woman exhibited an OI event in both +2Gz phases, however she endured all +1Gz phases. Three other women experienced an OI event during P3, two of which requested to terminate their GIT exposure at that point. One went on to complete P4 before terminating her exposure. Another woman experienced an OI event during P3, however she completed all subsequent phases, and P7 at a reduced +Gz level (1.5+Gz instead of +2Gz).

# Baseline

Anthropometric data are represented in **Table 1**. Height, weight and BSA were all significantly greater amongst men than women who finished the GIT (p < 0.05). The womenNF were significantly taller than finisher women (p < 0.05). Baseline cardiovascular values were obtained from the last 60 s of P1 and are represented in **Table 2**. The womenNF exhibited significantly greater MAP, DBP, and SVRI (p < 0.05), and non-significantly greater SBP (p = 0.054) than women who finished the GIT. There were no significant differences found between men and women who finished the GIT regarding baseline cardiovascular parameters. However, men had slightly higher DBP and SVRI at baseline, whereas women exhibited a higher HR.

# Cardiovascular Reactions

All cardiovascular parameters passed the test of normality (p > 0.05). The respective effect sizes, F-statistics and pvalues for each parameter are summarized in **Table 3**. The cardiovascular responses for women and men during GIT are displayed as line plots in **Figures 2**, **3**. Across the seven observed cardiovascular parameters, there was a significant main effect of +Gz (p < 0.05) across all phases for women and men, with the exception of SBP (**Table 3**). No significant differences were found amongst cardiovascular parameters between women and men.

Masatli et al. Cardiovascular Reactions During SAHC Training


\*Denotes a significantly greater value in men compared to women (p < 0.05). #Denotes a significantly greater value in womenNF compared to the finisher women (p < 0.05).

TABLE 2 | Mean and SEM baseline cardiovascular data for women and men.


\*Denotes a significantly greater value in womenNF compared to the finisher women (p < 0.05). The p-value for SBP between womenNF and finisher women was nearly significant (p = 0.053).

# Central Blood Pressure Reactions

**Figure 2** details the central blood pressure reactions (MAP, DBP, and SBP) recorded during GIT in women and men. MAP in women was increased over baseline in P3 and P4 (p < 0.05), with MAP in P3 being higher than in P2 and P4 (p < 0.05). Throughout P6-P8, MAP was non-significantly increased over baseline in women. In men, MAP expressed a significant increase only during P6 and P8 (p < 0.01), while remaining nonsignificantly above baseline in all other phases. DBP in women was elevated throughout the first round (p < 0.01), and was higher during +2Gz phases than +1Gz phases (p < 0.001). During the second round, DBP increased over baseline only during P7 (p < 0.01). DBP in this phase was also higher than P6 and P8 (p < 0.01). DBP in men was significantly elevated over P1 during P3, P4, and throughout P6-P8 (p < 0.05), while DBP was higher during P3 than +1Gz phases (p < 0.05). SBP showed no significant changes from baseline in both genders. Men tended to have overall higher MAP, DBP, and SBP than women, however there were no significant gender differences found. Women did show a lesser degree of variability compared with men for MAP and DBP, particularly during +2Gz phases.

# Vascular Resistance and Cardiac Reactions

**Figure 3** details SVRI, HR, CI, and SVI reactions apparent during GIT in both genders. SVRI in women was significantly higher than baseline throughout all +Gz phases (p < 0.05), with SVRI tending to be higher in the +2Gz phases. However, there was no statistical difference between +1Gz and +2Gz. SVRI amongst men, although elevated during +Gz, was not significantly different from baseline. SVRI was the parameter that showed the highest variability in men, and this variability was markedly greater in men over women.

Women showed a significantly higher HR over baseline throughout all +Gz phases (p < 0.05), with significantly higher HR levels during +2Gz phases than +1Gz phases (p < 0.001). Men expressed significant HR elevations over baseline during P2, P3, and P7 (p < 0.05, p < 0.001, and p < 0.001). HR was also significantly higher during the +2Gz phases than +1Gz phases (p < 0.001). Women displayed non-significantly higher HR values over men throughout all phases of GIT (p = 0.094). CI showed no significant decreases from baseline amongst women, while in men, CI demonstrated significant decreases from P1 during P2, P4, and throughout P6-P8 (p < 0.05). SVI in both genders was consistently lower than baseline during all +Gz phases than baseline (p < 0.001), and lower in +2Gz than +1Gz (p < 0.001). Only in men, SVI was significantly increased over baseline during P5 (p < 0.05). Finally, no significant differences were found when comparing the two rounds of GIT with each other. Despite the gender specific differences, there was an overall similarity in cardiovascular reactions for men and women. Both HR and SVRI increased with higher +Gz levels, while SVI and CI displayed an inverse relationship with the +Gz level.

# Finisher Women vs. WomenNF

While a majority of women participating in this study (11/16) did not experience any orthostatic event during GIT, five experienced an orthostatic event and did not complete the full study protocol (womenNF). During the first minute of the initial +2Gz phase (P3), womenNF exhibited a significant decrease in MAP, SBP and SVI compared to baseline (p < 0.05), with no subsequent increases in HR and SVRI. Only two women went on to complete the protocol albeit at a reduced +Gz level (+1/1.5/1 Gz). A statistical comparison was performed between womenNF and finisher women during P2 and P3 (n = 5 vs. n = 11). Compared with finisher women, womenNF exhibited higher MAP, DBP, SBP, and SVRI during P2 (p < 0.05). During P3, the only difference between these groups was SVI, which was significantly lower in womenNF compared with finisher women (p < 0.05).

# DISCUSSION

This study set out to determine gender specific cardiovascular reactions during intermittent +Gz interval protocol as a potential countermeasure against micro-g associated cardiovascular deconditioning. Upon conclusion of this study, it could be determined that women showed consistent significant increases in HR, and SVRI throughout the GIT, with significant DBP increases occurring during phases 2, 3, 4, and 7. On the other hand, men exhibited significant increases in DBP and HR only during the +2Gz phases with increases in DBP occurring in all but one +1Gz phase. No significant changes in SVRI for men were recorded. The lack of significant SVRI findings could be attributed to a high inter-group variability amongst the men involved in this study. SVRI in women did not show this extent of variation and responded in a uniform fashion. Amongst men,


TABLE 3 | Partial Eta-squared effect sizes, F-statistics, and p-values with Greenhouse-Geiser sphericity corrections per cardiovascular parameter for women, men, and all subjects for main effect of phase.

Partial Eta squared, F-statistics, and p-values comparing women and men with Type III sum of squares are reported in the last row for each parameter.

CI was significantly decreased throughout the GIT, while in women no significant changes were observed. As both genders had similar SVI reductions, this gender difference in CI is most likely due to the HR increases recorded amongst women over men. Finally, when comparing both rounds of the GIT, no significant differences were found. The cardiovascular reactions occurring in the first round were mirrored in the second round for both genders. This was the desired outcome of the interval protocol, and it can be assumed that a third round would have induced similar responses.

# Cardiovascular Reactions During GIT: Female Specific Reactions

To date, only few of studies dealt specifically with the topic of gender related effects of AG exposure. Two studies concluded that a single exposure to a graded AG protocol (from +0.6Gz to submaximal +Gz) significantly improved orthostatic tolerance amongst normovolemic and hypovolemic women (Evans et al., 2015; Goswami et al., 2015a). A third study found that 3 weeks of daily 35-min intermittent +Gz exposure coupled with exercise improved orthostatic tolerance amongst ambulatory women (Stenger et al., 2007). Other AG studies have focused on the pre-frontal cortical activity, which was decreased in women undergoing AG exposure with respect to men (Smith et al., 2013; Schneider et al., 2014). In the present study we aimed to investigate gender related differences in cardiovascular reaction during exposure to graded +Gz levels, while testing for the occurrence of a training effect during the second run. We found that women undergoing GIT exhibited significant increases in DBP, HR, and SVRI, which can be attributed to greater baroreflex sensitivity with respect to men during gravitational challenges (Hogarth et al., 2007). Another contributing factor for these findings could be plasma epinephrine, renin, and vasopressin levels, which have been recorded to be higher in women than in men are during gravitational challenges (Geelen et al., 2002). Although not measured in this study, increased plasma concentrations of the aforementioned neuroendocrine components could have led to the augmented SVRI, as well as HR activity. The low interindividual variability displayed by women as a group would

x-axis (P1-P9).

be advantageous when implementing future +Gz protocols, possibly negating the need to individualize a +Gz training profiles for women. A further reason for these reactions in women may be due to anthropometric factors, as a more compact body size in women may allow for increased sensitivity to gravitational challenges.

Although the majority of women (11/16) that participated in this study endured the GIT, five women did not follow suit and experienced an OI event. A predisposing factor for OI risk seems to be height and baseline central and systemic pressure, with taller women having higher MAP, DBP, SBP, and SVRI. This supports evidence that women have a higher propensity for OI during gravitational stress and the reasons for this have been discussed in prior research (Harm et al., 2001; Fu et al., 2004; Fong et al., 2007; Nordine et al., 2015) and indicate that anthropometric as well as physiological factors could play a role in this trend. It would appear that +2Gz exceeds the capacity for women with a certain height to maintain MAP, and a reduced +Gz profile would be beneficial. In order to ensure a 100% orthostatic tolerance rate amongst women, the +Gz limit for each person should be assessed prior to +Gz exposure, similar to the +Gz profile utilized by Goswami et al. (2015a). Alternatively, a modified +Gz profile could be used, as it appeared that 30% of women in this study could not tolerate full +2Gz for more than 60 s. These women may have benefitted from a starting +1Gz/+1.5Gz/+1Gz x2 profile.

# Cardiovascular Reactions During GIT: Male Specific Reactions

During GIT, men did not exhibit the same significant changes from baseline as women did. HR and DBP were significantly increased over baseline during +2Gz, but not SVRI. In men, a minimum of +2Gz seems to be needed in order to induce significant HR increases, a response which has also been recorded by Goswami et al., during SAHC training (Goswami et al., 2015b), Ueda et al. (2015) and Polese et al. (1992). A possible explanation as to why men require a stronger gravitational stimulus in order to trigger increases in HR may be due to higher resilience to venous pooling and central volume loss due to a greater amount of available plasma volume. Therefore, during +1Gz phases, while the decrease in functional blood volume is enough to trigger the beta sympathetic response in women, men may require a higher degree of SVI loss until HR increases become apparent. Surprisingly, SVRI in men showed no significant changes from baseline. However, as previously mentioned, this is more likely due to high inter-individual variability observed amongst men rather than low absolute values recorded. Possible explanations for these findings have been proposed by some

study groups, which include polymorphic genetic differences in vascular resistance regulation (Seasholtz et al., 2006), or elevated respiratory effort during +Gz, which induces a higher amount of venous return to the central circuit (McKenzie, 1993). Although these factors may offer reasons for varying intensities of SVRI regulation, they do not offer insight to the difference in variability observed between genders, as these studies included exclusively men. In order to train both active vasomotor and cardiac mechanisms in men, a higher +Gz-level gradient may be required. For future GIT protocols in men, an individualized +Gz limit should be determined prior to training, as to maximize the effects on the cardiovascular system.

# Cardiovascular Activity During GIT: Gender Differences

Although no significant gender differences in cardiovascular reactions were recorded, women tended to respond with heightened HR over men during the GIT. Other working groups have recorded similar HR responses in women compared to men during gravitational stress. These studies however used lower body negative pressure (LBNP) (Frey and Hoffler, 1988; Franke et al., 2000), and thus may not be directly comparable to SAHC studies. The gravitational vector applied by SAHC distributes a relatively equal transmural vascular pressure along the length of the body, whereas LBNP produces a strong transmural pressure directly below the LBNP seal (Dosel et al., 1998; Watenpaugh et al., 2004; Robertson, 2008).

The heightened HR response in women can be attributed to gender differences in the autonomic circuitry, myocardial structure, as well as hormonal status. Compared with men, women show increased parasympathetic withdrawal to the cardiac circuit during orthostatic stress thereby leading to an increase in HR (Evans et al., 2001; Huxley, 2007). In addition, the hormone estradiol may augment epinephrine sensitivity during orthostatic stress and thus be a contributing factor (Wenner et al., 2013; Gordon and Girdler, 2014). Further reasons for an increased HR in women could be differences in trans-mitral filling velocity and faster myocardial velocities (Nio et al., 2015). These factors would equate to a higher cardiac filling rate and faster myocardial contraction in women than in men, thus offering an account for differences in HR.

Regarding SVRI, no significant difference was found between genders, although men exhibited higher overall SVRI than women. This was reflected in higher MAP, DBP, and SBP recorded amongst men. Some groups have also observed significantly increased vasoconstrictor response in men over women during orthostatic challenges (Frey and Hoffler, 1988; Jarvis et al., 2010; Hachiya et al., 2012). However, more evidence suggests that peripheral vasoconstrictor activity does not differ between genders during orthostatic stress (Convertino, 1998; Franke et al., 2003; Kelly et al., 2004; Fu et al., 2005; Carter et al., 2015; Russomano et al., 2015; Patel et al., 2016). Certain authors have shown that during beta and muscarinic blockade, there is a predominant vascular regulation in men compared to dominant parasympathetic influence on heart rate regulation in women (Evans et al., 2001; Huxley, 2007). Those studies used different modalities to provoke orthostatic stress such as LBNP, head-up tilt (HUT), as well as pharmacological means. Previous studies that have recorded increases in vasomotor activity in men over women (Hachiya et al., 2012) have attributed these differences to an overall higher magnitude of sympathetic nerve activity to the periphery in men (Hart et al., 2009; Yang et al., 2012). It should be noted that SVRI is the total accumulation of vasoconstrictor activity in the circuit, which includes the extremities as well as the core (thorax, abdomen, and pelvis). Although not recorded in this study, studies have shown differences regarding vasomotor responses in different anatomical regions amongst men and women (Dart et al., 2002; Jarvis et al., 2010; Hachiya et al., 2012). Clearly, further AG studies are needed in order to critically examine gender differences regarding regional vascular activity during +Gz gradients via the use of Doppler sonography, laser Doppler, as well as near infrared spectroscopy.

# LIMITATIONS

While this study directly compared gender specific cardiovascular reactions during a newly devised graded intermittent AG training protocol, the findings should be interpreted with caution. Firstly, the small sample size used for this study, although of good size for human AG studies, makes drawing larger conclusions for the average population difficult. Also, as there is a lack of data concerning gender specific differences during SAHC exposure, more studies of this nature should be performed in order to ascertain if these gender specific reactions can be replicated. Moreover, this study did not set out to induce maximal +Gz limits in the human test subjects nor was the +Gz gradient tailored for each individual test subject. Had maximal +Gz profiles been deployed, or individual +Gz profiles been used, the results may have been different. Lastly, serum catecholamine, testosterone, estrogen, progesterone, and hemoglobin levels were not measured in this study. Thus, their role in the gender specific cardiovascular findings of this study can only be speculated upon. Additionally, no restrictions were placed on women concerning menstrual cycle phase, nor oral-contraceptive use upon partaking in this study. Certain authors have suggested that these factors may have an effect on autonomic functioning in women during gravitational stress (Wenner et al., 2013). In order to strengthen the findings of this study, more studies involving gender responses during +Gz exposure via SAHC, particularly with a similar +Gz profile, are warranted.

# CONCLUSIONS

This study showed that gender specific cardiovascular reaction patterns are indeed apparent during an intermittent AG session, while no significant differences between women and men were found. The gender specific cardiovascular differences in response to +Gz emphasize the importance of implementing AG countermeasure strategies specifically for women and men. Since micro-g induced cardiovascular deconditioning leads to vascular dysregulation upon reintroduction of standard gravity (Waters et al., 2002), a potential countermeasure protocol should ensure adequate vasomotor stimulation in both genders. This GIT, which lasted for 45 min, consistently stimulated vasomotor activity amongst women with SVRI increasing in all phases and DBP during the +2Gz phases. Men may require higher +Gz gradients in order to exhibit similar reactions. Also, only 70% of women finished the entire +Gz exposure. In order to ensure a 100% tolerability rate as well as ensuring the desired cardiovascular reactions needed to overcome orthostatic instability upon re-introduction of gravity modified or individualized +Gz profiles should be deployed, as was demonstrated by previous AG studies (Goswami et al., 2015a). Also, further testing in conjunction with bed-rest, immersion, and isolation studies are needed to see if this GIT training can mitigate cardiovascular deconditioning prior to manned space flight implementation.

# AUTHOR CONTRIBUTIONS

ZM and MN wrote the manuscript and analyzed the data with the help of KB, BH, SM, and AW. KB, AW, BH, SM, and MN performed statistical analysis. AS and OO implemented the data collection. HH, H-CG and OO conceived and planned the study. MM, HH, H-CG, OO, AW, SM, and KB provided expertise, reviewed, and approved the final manuscript.

# FUNDING

This study was supported by the German ministry for economic affairs and energy (Grant #50WB1123) and the German Aerospace Agency (DLR).

# ACKNOWLEDGMENTS

The authors wish to express their sincere gratitude to all subjects that participated in this study, as well as the DLR for their outstanding professionalism and for allowing access to their state of the art facilities. Furthermore, the authors want to thank the German ministry for economic affairs and energy for their generous financial support.

# REFERENCES


for gender differences. Am. J. Physiol. Heart Circ. Physiol. 286, H449–H457. doi: 10.1152/ajpheart.00735.2002


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Masatli, Nordine, Maggioni, Mendt, Hilmer, Brauns, Werner, Schwarz, Habazettl, Gunga and Opatz. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Limb Skin Temperature as a Tool to Predict Orthostatic Instability

Oliver Opatz<sup>1</sup> \*, Michael Nordine<sup>1</sup> , Helmut Habazettl<sup>1</sup> , Bergita Ganse<sup>2</sup> , Jan Petricek<sup>3</sup> , Petr Dosel<sup>3</sup> , Alexander Stahn1,4, Mathias Steinach<sup>1</sup> , Hanns-Christian Gunga<sup>1</sup> and Martina A. Maggioni1,5

<sup>1</sup> Charité – Universitätsmedizin Berlin, Institute of Physiology, Center for Space Medicine and Extreme Environments Berlin, Berlin, Germany, <sup>2</sup> German Aerospace Center (DLR- Deutsches Zentrum für Luft- und Raumfahrt), Institute of Aerospace Medicine (Institut für Luft- und Raumfahrtmedizin), Cologne, Germany, <sup>3</sup> Institute of Aviation Medicine, Military University Hospital Prague, Prague, Czechia, <sup>4</sup> Division of Sleep and Chronobiology, Department of Psychiatry, Perelman School of Medicine at the University of Pennsylvania, Philadelphia, PA, United States, <sup>5</sup> Department of Biomedical Sciences for Health, Università degli Studi di Milano, Milan, Italy

Orthostatic instability is one of the main consequences of weightlessness or gravity challenge and plays as well a crucial role in public health, being one of the most frequent disease of aging. Therefore, the assessment of effective countermeasures, or even the possibility to predict, and thus prevent orthostatic instability is of great importance. Heat stress affects orthostatic stability and may lead to impaired consciousness and decrease in cerebral perfusion, specifically during the exposure to G-forces. Conversely, peripheral cooling can prevent orthostatic intolerance – even in normothermic healthy subjects. Indicators of peripheral vasodilation, as elevated skin surface temperatures, may mirror blood decentralization and an increased risk of orthostatic instability. Therefore, the aim of this study was to quantify orthostatic instability risk, by assessing in 20 fighter jet pilot candidates' cutaneous limb temperatures, with respect to the occurrence of G-force-induced almost loss of consciousness (ALOC), before and during exposure to a push-pull maneuver, i.e., head-down tilt, combined with lower body negative pressure. Peripheral skin temperatures from the upper and lower (both proximal and distal) extremities and core body temperature via heat-flux approach (i.e., the Double Sensor), were continuously measured before and during the maneuver. The 55% of subjects that suffered an ALOC during the procedure had higher upper arm and thigh temperatures at baseline compared to the 45% that remained stable. No difference in baseline core body temperature and distal limbs (both upper and lower) skin temperatures were found between the two groups. Therefore, peripheral skin temperature data could be considered a predicting factor for ALOC, prior to rapid onset acceleration. Moreover, these findings could also find applications in patient care settings such as in intensive care units.

Keywords: lower body negative pressure, skin temperature, blood pooling, acceleration, orthostatic hypotension

# INTRODUCTION

G-force (+Gz)-induced loss of consciousness (GLOC) occurs during exposure to strong forces of acceleration. The resulting stress on the cardiovascular system is caused by decentralization of blood to the lower limbs and the splanchnic vessels of the pelvis. Consequently, the cardiac preload is reduced and perfusion of the central nervous system decreases (Self et al., 1996). Similar to rapid

#### Edited by:

Geoffrey A. Head, Baker Heart and Diabetes Institute, Australia

#### Reviewed by:

Andreas Roessler, Medizinische Universität Graz, Austria Joyce McClendon Evans, University of Kentucky, United States

> \*Correspondence: Oliver Opatz oliver.opatz@charite.de

#### Specialty section:

This article was submitted to Integrative Physiology, a section of the journal Frontiers in Physiology

Received: 23 February 2018 Accepted: 16 August 2018 Published: 05 September 2018

#### Citation:

Opatz O, Nordine M, Habazettl H, Ganse B, Petricek J, Dosel P, Stahn A, Steinach M, Gunga H-C and Maggioni MA (2018) Limb Skin Temperature as a Tool to Predict Orthostatic Instability. Front. Physiol. 9:1241. doi: 10.3389/fphys.2018.01241

**294**

changes in posture, that cause an orthostatic reaction, these forces can lead to loss of consciousness and postural tone. However, some individuals exhibit a high degree of cardiovascular resilience to +Gz, and therefore have greater orthostatic tolerance than others. The detailed mechanism and the contributors to orthostatic tolerance have yet to be elucidated, however, Yang et al. (2012) demonstrated that inter-individual differences in sympathetic firing pattern could be involved in orthostatic reactions during lower body negative pressure (LBNP). Several prodromal states have been described to be pathognomonic for orthostatic intolerance including dizziness, hearing loss, nausea, discomfort, sweating, palpitations, and "grayout" and "blackout" events (Benni et al., 2003). Regarding the exposure to high accelerations during space-flight takeoff/landing or rapid in-flight maneuvers, orthostatic intolerance could lead to severe consequences, including a critical missioncompromising event. Therefore, a physiological parameter that would yield predictive information prior to the occurrence of an orthostatic event, would be invaluable, not only in aerospace medicine (Tripp et al., 2009), but also in clinical settings (Krakow et al., 2000). Current means to test for orthostatic intolerance – and GLOC-susceptibility are the tilt table, LBNP human centrifuge, and parabolic flight. Normally, cardiovascular parameters such as heart rate and non-invasive blood pressure are analyzed during these training methods (Durand et al., 2004). However, these values can show high inter-individual variation and there can be significant fluctuations in signal quality during high accelerations (Hanousek et al., 1997), which would provide unreliable and invalid physiological data. Further evidence by Simonson et al. (2003) showed that baseline heart rate and blood pressure demonstrated no significant differences amongst subjects with high versus low +Gz tolerance during LBNP testing. Thus, baseline hemodynamics seem to be unsuitable for predicting GLOC. However, Crandall et al. (2010) established that peripheral blood volume distribution caused by heat stress can reduce cerebral blood velocity. This distribution can thus be counted as an important factor in individual susceptibility to orthostatic events (Wilson et al., 2006). Earlier Crandall's group found that external cooling may also improve orthostatic stability in normothermic subjects (Durand et al., 2004). In this light, we propose here a novel approach to test for GLOC-susceptibility using baseline proximal and distal skin temperature of the limbs monitored prior to a gravitational challenge. We hypothesize that subjects displaying a higher limb skin temperature (Tskin) prior to push-pull maneuver, i.e., head-down tilt (HDT) followed by LBNP, would be more prone to GLOC.

# MATERIALS AND METHODS

# Subjects

Twenty male subjects, with a mean age of 28.2 ± 9.5 years, were tested by means of a standardized HDT/LBNP protocol and gave their written and informed consent before testing. As all participants were military pilots, the study was undertaken in compliance with the guidelines of the local Medical Military Advisory Board, which allowed the study and the use of the collected data. Mean anthropometric data for the subjects were the following: height 181.6 ± 6.7 cm, weight 80.2 kg ± 8.6, and BMI 24.3 ± 1.8 kg/m<sup>2</sup> . Each test subject was accustomed to daily high-workload endurance training. None had any cardiovascular, metabolic, or musculoskeletal pathologies. Exclusion factors were any history of syncope, arrhythmia, or other similar medical abnormalities. Each subject underwent a pre-test screening conducted by the flight physician. This included a physical exam and a resting 12-lead electrocardiogram. No reasons to exclude any subjects were found.

# Experimental Procedure

To test for orthostatic fitness, the combined HDT/LBNP test is used as a routine part of the Czech Air Force Centre of Flight Training (CLV) program. HDT/LBNP testing has been standardized for use at the Institute of Aviation Medicine (Dosel et al., 1998, 2007). The rapid change from the HDT to the head-up tilt with LBNP is known as the "push-pull maneuver" (Banks et al., 1994). LBNP followed a standardized phase profile which consists of four phases, as demonstrated in **Figure 1**. During baseline, subjects remained in an upright sitting position (Sitting); then, subjects were tilted head-down (HDT) for 2 min, then tilted back to head-up sitting position (Sitting) and simultaneously exposed to a negative pressure of −70 mmHg (onset at 2 s) for 2 min (Sitting/LBNP). Finally, the subjects were returned to baseline conditions without LBNP (Sitting). Upon entering the test facility, each test subject undressed down to their underwear and was fitted for instrumentation. Air temperature was kept constant at 27.0◦C ± 0.6 with a humidity of 40– 50%. No subject reported being cold or too warm during the experiment, nor exhibit any shivering or sweating prior to testing. Temperatures inside the LBNP chamber at baseline ranged with a median of 26.3◦C between 24.2 and 27.2◦C. Test subjects were placed in the LBNP device upright, seated position. A pressure seal was fitted around the participant at anterior superior iliaclevel to ensure an air-tight seal. To monitor Tskin, a total of 5 thermometer probes were placed on each subject at various regions: (i) the anteromedial aspect of the left arm bisecting

FIGURE 1 | Scheme of the head-down tilt and lower body negative pressure (HDT/LBNP) testing. The depicted phases are assigned to posture of the subject and pressure in the LBNP. Acceleration equivalents are related to cardiovascular reaction at comparable accelerations during flight or long-arm centrifuge runs. Duration relates to the time the subjects were actually exposed to the referred posture and pressure at a steady level, excluding transition phases.

the biceps muscle, (ii) the dorsal side of the left hand 3 cm distal to the carpal region, (iii) the left anterior medial thigh, halfway between the anterior superior iliac crest and the patella region, (iv) 3 cm distal to the tarsal region of the leg, and (v) a thermistor attached to the dorsal aspect of the foot. Peripheral temperatures were measured via the Heally system (SpaceBit Gmbh, Wiesenick, Germany) using thermistors soldered to a silver-plated 32 American wire gauge copper wire (7/40) with polyurethane insulation. Core body temperature was measured using a double heat flux sensor (Gunga et al., 2008, 2009; Mendt et al., 2017), which was affixed to the medial forehead of each test subject. The influence of fluctuating air temperatures on the skin temperature probes was mitigated by ensuring a constant ambient temperature and environment. No side effects such as skin reactions to the applied electrodes were observed. Cardiovascular parameters such as heart rate (HR, bpm) and blood pressure (BP systolic/diastolic, mmHg) were continuously recorded non-invasively via Portapres monitoring (BMI-TNO, Amsterdam, Netherlands). Once all monitoring equipment was affixed to each subject and an uplink was established, an "all clear" signal was given by the flight physician and the phase profile exposure began. If any subject experienced syncope or cardiovascular instability during the study (such as a HR of <50 bpm, mean arterial pressure of <60 mmHg, or systolic BP of <90 mmHg – the so-called "stop criteria"), the flight physician could immediately terminate the LBNP exposure. In this case, the LBNP would be switched off and normal atmospheric pressure to the lower extremities would be reestablished. These cardiovascular "stop criteria" are based on former investigations using LBNP and tilt tables (Hanousek et al., 1997). The experiment would also be interrupted immediately in case the subject described visual disturbances, such as blurred vision, tunnel vision, grayout, or blackout. In addition, the appearances of any autonomic symptoms such as sweating, nausea, or paresthesia would also halt the LBNP exposure.

# Data Analysis

Cardiovascular data were recorded continuously, with a resolution of 250 Hz for electrocardiography and a sample rate of 1 Hz as for BP data. Temperature data were also collected continuously at 1 Hz sample rate. For all variables, the values from the last 10 s of each 2 min phase (**Figure 1**) were averaged and taken into account for data analysis, in order to observe the maximum effect of each different posture/G level.

# Statistical Analysis

Data are reported as mean ± standard deviation (m ± SD), if not otherwise stated. Data were analyzed for normality via a Shapiro–Wilks test using SPSS (IBM Corporation, Armonk, NY, United States). Statistical comparisons were made using the more conservative Mann–Whitney U-test due to the non-normality of all data groups. Significance was defined as a p < 0.05. Data are reported using Tukey boxplots. The box represents the interquartile range (IQR) including the median, and the whiskers represent 1.5 × IQR above the third or below the first quartile. To determine the optimal cut-off temperatures and cardiovascular values to predict GLOC, receiver operating characteristics (ROCs) analyses were performed (Hanley and McNeil, 1982). To define cutoff values, we used the Youden's I to estimate the probability of an informed decision (Youden, 1950). The statistical power for this study was set at 80% to determine a difference in Tskin of ±0.7◦C for a minimum of 20 subjects.

# RESULTS

Of the 20 subjects, 11 (55%) exhibited an almost loss of consciousness (ALOC), leading to test termination by the flight physician. In all cases the reason for this was a strong decrease in systolic BP. Therefore, subjects were classified into two groups, those who suffered an ALOC and those who did not (NALOC), to allow a comparison. Overall the intervention induced similar changes in Tskin for all subjects (variation from 0.2 to 0.5◦C). Core body temperature did not exhibit fluctuations nor differences between ALOC/NALOC subjects. At baseline, mean core body temperature of all subjects was 36.9 ± 0.4◦C. Statistically significant baseline differences between ALOC and NALOC groups were found regarding Tskin only in the upper arm and thigh. Baseline Tskin at the upper arm in the NALOC group was 32.03 ± 1.08◦C as compared to 33.23 ± 0.66◦C in the ALOC group (p = 0.04). Baseline Tskin at the thigh was 31.34 ± 0.81◦C in the NALOC group as compared to 32.24 ± 0.77◦C in the ALOC group (p = 0.006). These values are displayed as boxplots in **Figure 2**. At baseline, diastolic BP was slightly higher in the NALOC group (p = 0.056). Statistical comparisons at baseline of core body temperature (p = 0.45), HR (p = 0.86), systolic BP (p = 0.32), and distal limb Tskin temperatures revealed also no significant differences between groups (as for foot p = 0.10 and as for hand p = 0.13).

During HDT, Tskin showed significant differences of upper arm and thigh between ALOC and NALOC group, i.e., NALOC exhibited significantly lower temperatures than ALOC subjects (p = 0.002 and p = 0.016, respectively) (**Figure 2**). Moreover, ALOC subjects exhibited a significantly lower diastolic BP than NALOC subjects (p = 0.004) (**Figure 3**).

During Sitting/LBNP, a significant increase of upper arm and thigh Tskin was found in the ALOC group with respect to NALOC (p = 0.002 and p = 0.01, respectively, see **Figure 2**). Furthermore, during this phase, a significant decrease in systolic BP was retrieved as for ALOC with respect to NALOC subjects (p = 0.007), whereas diastolic BP was similar between groups (p = 0.60); HR showed no difference between the groups (p = 0.54 in HDT and p = 0.18 in Sitting/LBNP, see **Figure 3**).

The ROC was used to show the diagnostic ability of the method and the discrimination thresholds. It revealed significant associations between cutaneous arm and thigh temperatures with ALOC. Optimal cut-offs to predict ALOC were 32.81◦C for arm temperature and 31.78◦C for thigh temperature (**Figure 4B**).

# Limitations

So far, results of this study could only be applied for a limited specific population groups (young healthy men, already

FIGURE 2 | Boxplots of absolute limb temperature data on the left (A) and standard deviations as Z-values on the right (B), in ALOC (i.e., almost loss of consciousness) versus NALOC (i.e., absence of almost loss of consciousness) subjects at baseline in the last 30 s while sitting, during head down tilt (HDT) and during sitting + LBNP (i.e., low body negative pressure). Significant differences (p < 0.05) found using the Mann–Whitney U-test are denoted by <sup>∗</sup> .

right (B) in the last 30 s while sitting, during head down tilt (HDT) and during sitting + LBNP (i.e., low body negative pressure). Significant differences (p < 0.05) found using the Mann–Whitney U-test are denoted by <sup>∗</sup> . ALOC, almost loss of consciousness; NALOC, absence of almost loss of consciousness (see also text for abbreviations).

trained as jet pilots), therefore further studies are needed to cover more individual parameters such as gender, age and a spectrum of anthropometric differences. In this context it is needed to study a greater number of subjects grouped by the described parameters to exactly define the efficacy of the method.

# DISCUSSION

The results of this study showed a link between individuals' proximal limb Tskin and orthostatic stability during a simulation of aerospace braking maneuvers in a temperature-controlled environment (i.e., push-pull maneuver).

curve; HR, heart rate (p = 0.86); Diastolic, diastolic blood pressure; Systolic, systolic blood pressure; Tcore, core body temperature; Tarm, upper arm temperature;

Other research has focused on the effects of external temperature regarding orthostatic stability during LBNP. It has been shown that orthostatic tolerance in men is decreased during heat stress (Kirsch et al., 1986; Rubinstein and Sessler, 1990; Robertson, 2008; Schlader et al., 2015). Moreover, wholebody heating during LBNP leads to a decrease in central venous pressure as well as pulmonary capillary wedge pressure (Wilson et al., 2007). Furthermore, the link between increases in skin blood flow during hyperthermia and orthostatic tolerance has been extensively studied by the group of Craig Crandall (Shibasaki et al., 2006; Wilson et al., 2007; Crandall et al., 2010). On the contrary, cooling of the periphery, and thus the simultaneous decrease in skin blood flow, does appear to have a positive impact on orthostatic tolerance in a normothermic person. Durand, indeed found that active skin cooling of the normothermic subject might improve orthostatic stability (Durand et al., 2004). In his study, environmental temperature was kept constant in order to minimize the effect of ambient temperature on peripheral vasomotor activity, demonstrating that, despite thermo-neutral ambient temperature conditions, Tskin does influence orthostatic stability. This is in concordance with our original reasoning that Tskin would correlate with cutaneous vasomotor activity and that higher temperatures would indicate increased perfusion to the periphery, thus depriving the central circuit of volume. However, this would need to be verified by direct measurement of vasomotor activity via near-infrared spectroscopy (NIRS), to concretely solidify the link between the two factors. During acceleration conditions (i.e., LBNP), a decrease in blood pressure activates the baroreceptor reflex which induces strong sympathetic activation (Convertino

Tthigh, thigh temperature (see also text for abbreviations).

et al., 2004). This sympathetic counter-response results in adrenergic α-receptor-mediated vasoconstriction as well as β1 receptor mediated HR increase. Maintenance of BP during orthostatic stress depends on the careful shifting of blood flow from non-essential organ systems to essential systems. The increased vasomotor activity plays a key role in maintaining mean arterial pressure via increases in arteriolar tone, which in turn increases systemic vascular resistance (Hachiya et al., 2010), while at the same time inducing a decrease in cutaneous perfusion. As previously reported by Rubinstein and Sessler (1990), it was shown that increased vasoconstriction in the fingertip had a significant negative correlation with skin surface temperature. In contrast, a study by Cotie et al. (2011) showed that cutaneous temperature changes did not reflect changes in femoral artery blood flow. The authors, however, conceded that bulk blood flow measurements may not be sensitive enough to reflect changes in cutaneous microvascular perfusion.

In our study, we exploited Rubinstein's findings and measured skin surface temperature at the limbs to estimate changes in skin perfusion via skin surface temperatures. As stated, our key findings concerning baseline Tskin during the HDT/LBNP procedure showed that proximal Tskin within the NALOC group were not significantly different from baseline, thus exhibiting no change in cutaneous perfusion across different phases of exposure. This trend was also seen in the ALOC subjects. It occurs when the central nervous system fails to receive adequate perfusion, resulting in a syncopal reaction (Wilson et al., 2006). The monitoring of brain perfusion during simulated hyper-gravity can be performed very effectively using NIRS, as published by Ryoo et al. (2004). While reduction of brain

perfusion is the most important resulting factor, peripheral blood volume distribution might be the most important predicting factor of subject's G-tolerance. Thus, we can only observe the effect, but not the cause, which is known to be venous blood pooling in the splanchnic compartment and lower limbs (Blaber et al., 2013) and insufficient cutaneous vasoconstriction to maintain blood pressure (Crandall et al., 2010). According to these findings the predictive peripheral temperature values during LBNP were determined using ROC curves. It could be demonstrated that heart rate and systolic blood pressure underperformed as predictive values. We hypothesized that lower baseline vessel tone, as reflected by higher Tskin, may also indicate lower ability to increase cutaneous vascular resistance during orthostatic challenges. Our results were able to confirm this finding – at least, amongst a small subject pool such as this one. Thus, peripheral Tskin data could be employed as a predicting factor for ALOC/NALOC prior to rapid onset acceleration. Significant differences were detected in baseline upper arm and thigh temperatures between the ALOC and NALOC groups (see **Figure 4**). Both temperatures predicted ALOC with a sensitivity of >0.8 and, in case of the upper arm temperature, with a specificity of >85%. The reason why only proximal and not distal limb temperatures were associated with ALOC may be due to the larger proximal density of blood vessels, different innervation of the proximal and distal limbs, and different receptor density. Such differences were also reflected by the considerably lower baseline temperatures at the distant limb site (i.e., an anatomical bordering effect). Other scenarios involve the variability of autonomic innervation of the proximal and distal limbs as well as different receptor density (Schiller, 2003). Due to the larger bulk of vasculature in the proximal limbs versus distal limbs, this could lead to a discrepancy with regards to autonomic vasculature innervation. Such differences were reflected by the considerably lower baseline temperatures at the distant limb site. Thomas (2011) reported that increases in vasomotor tone of the capacitance vessels would decrease tissue blood volume while, at the same time, increasing cardiac preload.

However, further studies on the concrete mapping of the distribution of autonomic fibers in the limbs are needed to confirm this. At baseline, there were no significant differences among subjects with respect to cardiovascular parameters. There was a trend in the ALOC subjects to exhibit lower diastolic blood pressure compared to NALOC subjects. Diastolic pressure is an indicator of vascular resistance (Lacolley et al., 1992) and test subjects tending to ALOC exhibited lower diastolic pressure, which would equate to lower vascular resistance. Thus, this is compatible with the higher Tskin in subjects prone to ALOC.

According to van Genderen et al. (2013) it could also be shown that the Pulse-Perfusion-Index (PPI) could be used to detect hypovolemia before it becomes clinically apparent. In contrast to

# REFERENCES

Banks, R. D., Grissett, J. D., Turnipseed, G. T., Saunders, P. L., and Rupert, A. H. (1994). The "push-pull effect". Aviat. Space Environ. Med. 65, 699–704.

our study van Genderen et al. (2013) used a stepwise descending LBNP pressure scenario where they found a decrease in PPI already at a pressure of −20 mmHg. However, the reduction of PPI prior to LBNP exposition as a predictor of syncopal events could only be shown at a single subject collapsed in the LBNP.

The measuring of peripheral temperatures could therefore be a more sensitive early indicator of vasomotor activity than the measurement of cardiovascular values alone. Once established as a sensitive and reliable tool, this method, coupled with classical cardiovascular monitoring, can be employed in real world military high G-force (or +Gz) maneuvers, Earth to orbit manned space missions, as well as civilian air transport. This technique is non-invasive, sensitive, and easy to employ. When combined with classical cardiovascular monitoring, it adds a new individual method to monitor volume distribution during orthostatic stress.

# ETHICS STATEMENT

All participants were military pilots, therefore the study was undertaken in compliance with the guidelines of the Czech Aviation Medicine Military Advisory Borad (Prague), which approved the protocol, allowed the study and the use of collected data. All subjects gave written informed consent in accordance with the Declaration of Helsinki.

# AUTHOR CONTRIBUTIONS

OO conceived the study and wrote the manuscript with MM. MN contributed to drafting the manuscript and to statistics. BG, JP, and PD performed data collection and data analysis. AS, H-CG, and MM contributed to the study design, and with HH and MS provided expertise and feedback. OO formatted, and with assistance of MM, revised manuscript.

# FUNDING

The study was supported by grant 50WB0720, 50WB0724, and 50WB1030 from the German Aerospace Center (DLR).

# ACKNOWLEDGMENTS

We would like to acknowledge the outstanding support and cooperation of all subjects and thank them for dedicating their time and effort toward the successful completion of the study. Furthermore, we wish to express our deep thanks to the team of the Czech Aeromedical Institute – especially Dr. Hanousek.

Benni, P. B., Li, J. K., Chen, B., Cammarota, J., and Amory, D. W. (2003). NIRS monitoring of pilots subjected to +Gz acceleration and G-induced loss of consciousness (G-LOC). Adv. Exp. Med. Biol. 530, 371–379. doi: 10.1007/978- 1-4615-0075-9\_34


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Opatz, Nordine, Habazettl, Ganse, Petricek, Dosel, Stahn, Steinach, Gunga and Maggioni. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Recent Progress in Space Physiology and Aging

Felice Strollo<sup>1</sup> \*, Sandro Gentile<sup>2</sup> , Giovanna Strollo<sup>3</sup> , Andrea Mambro<sup>4</sup> and Joan Vernikos<sup>5</sup>

<sup>1</sup> Diabetes Unit, San Raffaele Institute, Rome, Italy, <sup>2</sup> Campania University "Luigi Vanvitelli" and Nefrocenter Research Network, Naples, Italy, <sup>3</sup> Endocrinology Unit, FBF San Pietro Hospital, Rome, Italy, <sup>4</sup> Anesthesiology and Resuscitation Unit, "Misercordia" Hospital, Grosseto, Italy, <sup>5</sup> Thirdage LLC, Culpeper, VA, United States

Astronauts coming back from long-term space missions present with different health problems potentially affecting mission performance, involving all functional systems and organs and closely resembling those found in the elderly. This review points out the most recent advances in the literature in areas of expertise in which specific research groups were particularly creative, and as they relate to aging and to possible benefits on Earth for disabled people. The update of new findings and approaches in space research refers especially to neuro-immuno-endocrine-metabolic interactions, optic nerve edema, motion sickness and muscle-tendon-bone interplay and aims at providing the curious - and even possibly naïve young researchers – with a source of inspiration and of creative ideas for translational research.

#### Edited by:

Jack J.W.A. van Loon, VU University Amsterdam, Netherlands

#### Reviewed by:

Victor Demaria-Pesce, Institut National de la Santé et de la Recherche Médicale (INSERM), France Elisio Costa, Universidade do Porto, Portugal

> \*Correspondence: Felice Strollo felix.strollo@gmail.com

#### Specialty section:

This article was submitted to Integrative Physiology, a section of the journal Frontiers in Physiology

Received: 27 July 2018 Accepted: 16 October 2018 Published: 12 November 2018

#### Citation:

Strollo F, Gentile S, Strollo G, Mambro A and Vernikos J (2018) Recent Progress in Space Physiology and Aging. Front. Physiol. 9:1551. doi: 10.3389/fphys.2018.01551 Keywords: microgravity, aging, space, physiology, translational research

# INTRODUCTION

Astronauts coming back from long-term space missions present with different health problems potentially affecting mission performance and closely resembling those found in the elderly. The latter mostly involve the immune system, bones, muscles, eyes balance and coordination and the cardiovascular system (Strollo, 1999; Vernikos, 2004; Vernikos and Schneider, 2010). The identification of appropriate measures of astronaut performance and of individual genetic or acquired stress resilience features has thus become a research priority along with psychological investigations designed to enhance cognitive performance, sleep, and team-building<sup>1</sup> .

To succeed in such an ambitious task, it is absolutely necessary to start from where we find ourselves today considering that NASA's 60th anniversary is on October 1, 2018.

We are fully aware that this is not an exhaustive review and should be viewed as a partial synthesis of research findings in this field designed to stimulate the interest of young specialists in the direction we mentioned above. It also is designed to build on earlier reviews rather than repeat them. We will not deal with cardiovascular physiology (Norsk, 2005), which is by far the most studied system in space (Scano and Strollo, 1996), or with pulmonary physiology, an extremely important area where a few groups have done a lot of work (Prisk, 2014), as the reader can easily refer to recent dedicated reviews (Antonutto and di Prampero, 2003). We chose instead to point out the most recent advances in the

<sup>1</sup>A Midterm Assessment of Implementation of the Decadal Survey on Life and Physical Sciences Research at NASA; http: //nap.edu/24966. Last accessed July 24, 2018.

literature in areas of expertise in which specific research groups were proven to be creative, as well as with respect to aging and to possible benefits on Earth for disabled people.

# IMMUNE CHANGES AND ENVIRONMENTAL STRESS

fphys-09-01551 November 8, 2018 Time: 19:17 # 2

Since the very beginning of the space era immune defects were identified in response to real and simulated microgravity (Cogoli, 1993). Many research groups involved in this field confirmed those defects, such as increased virulence, antibiotic resistance, enhanced susceptibility to infections during space missions (Pierson et al., 2005; Stowe et al., 2011). They then explored in greater detail the possible underlying molecular and systemic mechanisms (Sonnenfeld and Miller, 1993). These data were used to identify appropriate countermeasures that might be most beneficial during long-duration missions per se (Sonnenfeld et al., 2003) as well as to develop specific preventative exercise protocols that could be performed during the mission (Walsh and Whitham, 2006). These could also be exploited for elderly people on the ground, most of whom undergo some kind of immune deficiency over time as well. Recently, after describing space-related immune system alterations and administering a questionnaire to many experts in the field, the need for longer duration missions devoted to immune research was suggested (Frippiat et al., 2016) along with new space-specific technologies to detect immune changes inflight in minute amounts of blood, reviewed in situ, with operational solutions onboard the ISS, with several potential individualized immune countermeasures for exploration missions in the context of precision medicine (Crucian et al., 2018). Animal research studied adult male mice that were exposed to chronic unpredictable mild psychosocial and environmental stressors (CUMS model) for 3 weeks. This period is long enough to simulate in mice a long human flight. This exposure induced an increase of serum IgA, a reduction of normalized splenic mass and reduced production of proinflammatory cytokines, as has been previously reported during or after space missions. However, they did not modify either major splenic lymphocyte sub-populations or their proliferative responses, suggesting that the observed inflight changes could be due to other factors such as gravity variations (Gaignier et al., 2018).

# MUSCLE-TENDON UNIT AND EXERCISE AS A COUNTERMEASURE

Sarcopenia is a typical aging-like effect of microgravity on the muscle-tendon unit structure and function that is still awaiting effective physical, nutritional, or pharmacological countermeasure. These could also be of benefit on Earth for the rehabilitation of injured elderly people. Since the early space flight era, the research in the field has been extensive. Many papers have dealt with the problem of muscle changes (Hargens et al., 2013; Stein, 2013; Ohira et al., 2015). From a nutrition point of view, protein /aminoacid supplementation has been studied as a possible solution that could be exploited also on Earth in inactive elderly people but, in the absence of exercise, no positive effect was obtained in microgravity when increasing either protein or leucine intake (Biolo et al., 2004; Backx et al., 2018). Branched chain aminoacids, however, were found to be somewhat effective, when tested during hind limb suspension in growing rats (Jang et al., 2015) and during bed rest in women (Dorfman et al., 2007).

After many years of poor results obtained with the exercise programs routinely carried out onboard the international space station, a possible solution may be provided by the horizontal sled jump method (Kramer et al., 2017, 2018) tested in Cologne during an ESA Bed Rest study. A similar device might be adapted in the future in terms of time/intensity of muscle involvement, to prevent sarcopenia in temporarily bed-ridden, non-frail elderly people (Vernikos, 2017). This "SPRINT" protocol of repeated short duration sessions of different kinds of high intensity aerobic and resistance exercise was deemed to represent both a promising inflight multi-system countermeasure and a potentially viable strategy for pre-habilitation before elective medical procedures on Earth (Ploutz-Snyder et al., 2018). However, compared with resistance training, whole-body vibration (WBV) is safer and more convenient while avoiding the risk of injury in older adults. A recent study using 12 Hz – 3 mm WBV (10 repetitions, each lasting 60 s, followed by a 30 s resting period per session, 3 times a week for 3 months) showed significantly positive effects on muscle mass, physical fitness (standing on one foot, flexibility, sit-to-stand, and grip strength) in sarcopenic institutionalized elderly people (Chang et al., 2018).

As far as basic mechanisms are concerned, unloading/disuse sarcopenia is still poorly understood and should be further investigated. Effective signal manipulation methods are needed that are applicable to muscle wasting prevention in the elderly. In mice. mechanical unloading reduced the expression of irisin, a muscle-derived insulin-sensitizing and bone-enhancing cytokine, presumably through pathways involving osteocalcin, a bone morphogenic protein (BMP), and PI3K suggesting that irisin might be involved in muscle/bone relationships regulated by mechanical stress (Kawao et al., 2018). Tendon physiology still awaits greater understanding. Appropriate training should be taken into account in preventing unexplained injuries especially in elderly people trying to get fit after decades of sedentary life (Magnusson and Kjaer, 2018).

# INSULIN RESISTANCE AND ENDOCRINE CHANGES

Insulin sensitivity changes that typically accompany aging on Earth and that gradually may progress to diabetes (a condition of decompensated insulin resistance) have been repeatedly reported as a consequence of space flight (Tobin et al., 2002; Hughson et al., 2016) and bed rest (Bergouignan et al., 2011) and assumed to be partly due to the cephalad fluid shift but

especially dependent on unloading-related sarcopenia. Based on cross-sectional intervention studies (Lillioja et al., 1987; Prior et al., 2015), reduced muscle capillarization was thought to be strongly involved in metabolic disturbance (Stuart et al., 1988). However, a paper published more recently, clearly showed that after only 4 days of bed rest both insulin sensitivity and skeletal muscle fiber cross-sectional area decreased, while capillary density increased, in healthy young subjects (Montero et al., 2018). This suggests that muscle capillaries do not primarily influence insulin sensitivity through impaired nutrient delivery in the absence of aging or pathological conditions (Montero et al., 2016). The most relevant changes responsible for insulin resistance might involve the expression and/or activity of insulin receptor β-subunit, Akt, GLUT4, AMPK and many other intracellular signals known to be involved in the insulin effect (Lassiter et al., 2018). Eventually, liver dysfunction may further contribute to insulin-resistance during long duration missions (Jonscher et al., 2016; Gentile et al., 2016). Since the beginning of manned flights other possible endocrine defects involved in aging have also been investigated in space but uncertainties remain, probably due to the fact that all of them are extremely sensitive to the stress experienced by astronauts both in flight and upon re-entry. Therefore, many of the observed changes can reflect other factors rather than the effects of microgravity per se. This especially applies to fluid-electrolyte regulating hormones and the glucoactive adrenal hormones, cortisol/corticosterone, which continue to be reported as increased in flight by most authors (Macho et al., 2001) despite some conflicting results (Grindeland et al., 1990; Strollo, 2000). As for thyroid hormones, production was reported to be impaired in rats and monkeys kept at a low temperature for 2 days after a 12-day Cosmos 1887 flight and rat thyroid hyperactivity was reported up to 14 days after re-entry (Plakhuta-Plakutina et al., 1990), This led to the hypothesis of a typical microgravity-induced hypothyroid state with post-flight rebound, further supported by the increased TSH levels found in humans (Leach et al., 1977). In fact, this was contradicted when iodine was no longer added to drinking water for hygienic purposes as after the Spacelab 2 flight. Blunted thyroid C-cell calcitonin secretion (Plakhuta-Plakutina et al., 1988) and enhanced PTH secretion by the parathyroid glands (Plakhuta-Plakutina, 1979) were also reported in earlier flights. Testosterone was measured occasionally in male astronauts (Smith et al., 2012) and more often in bedrest conditions in healthy volunteers (Wade et al., 2005). Carefully designed dedicated studies are still needed to resolve conflicting results (Strollo et al., 1998b), that often show no changes in the human in bed-rest but point to possible negative effects not only in inflight tests on astronauts but also in experiments carried out in flight and on ground using cellular or animal models (Amann et al., 1992; Uva et al., 2005). Much more would be learned from future inflight endocrine signal measurements, which will require conditions and equipment specifically designed for space, such as miniaturization, use of dry chemistry methods or minimization of liquid handling procedures and quality control assurance of laboratory equipment analysers.

# OSTEOPOROSIS

A well-known negative consequence of the unloading in spaceflight is osteoporosis which also affects most of the geriatric population. Their study has been addressed by leading groups in the field since the very beginning of the space era. We encourage young readers to study thoroughly the original work (Morey-Holton and Globus, 1998; Alexandre and Vico, 2011; Cappellesso et al., 2015). With respect particularly to postflight rehabilitation needs, a very interesting finding is worth mentioning: loading during the early post-fracture stages, while matrix deposition and remodeling are prevalent, may enhance repair through the formation of additional cartilage and bone (Liu et al., 2018). In fact, not only bone but also cartilage problems have been reported in microgravity. These include smaller chondrogenic pellets, less proteoglycan synthesis and reduced dynamic stiffness of three-dimensional engineered cartilage constructs (Freed et al., 1997). These observations warrant further investigation because synovial joints, which contribute to articular cartilage, subchondral bone, meniscal fibro-cartilage, tendon and ligaments bathed in synovial fluid and enclosed in a fibrous capsule, may undergo damage, especially upon reloading on return to Earth Joint instability and compensatory connective tissue changes resulting in joint degradation would follow (Fitzgerald, 2017).

Space osteoporosis is characterized by a lower bone formation rate and enhanced bone resorption, which calls for careful and timely preventative strategies, including Vitamin D<sup>3</sup> and K<sup>2</sup> supplementation and antiresorptive drug administration (Iwamoto et al., 2005). Bisphosphonates are the most commonly used anti-bone-resorptive drugs on Earth. Although they have also been used to prevent space-related osteoporosis, the results are conflicting (Cavanagh et al., 2005). The reason for this depends on the fact that bone formation is coupled to bone resorption and therefore about one month after antiresorptive effects begin, a "frozen" bone remodeling condition occurs (Bauer et al., 2018). This seems to be particularly the case during unloading. However, as significant increases in bone microarchitecture can be achieved with low-dose anti-resorptive therapy without reducing bone formation, such an antiresorptive dosing approach could be used to preserve bone quality by limiting it to the period of disuse. This could provide astronauts with faster recovery and less adverse effects (Lloyd et al., 2008). Meanwhile, the current combination of bisphosphonates with exercise (Leblanc et al., 2013) seems to be the accepted compromise.

# MELATONIN AND SLEEP-UNRELATED FUNCTIONS

A progressive decrease in night-time melatonin secretion is observed during aging accompanied by and often causing severe sleep disturbances (Bubenik and Konturek, 2011; Duffy et al., 2015). A similar reduction in melatonin has been reported in microgravity (Holley et al., 1991). Night-time and daytime and sleep are disrupted in space flight by a variety of

factors, including circadian rhythm disturbances due to the added effects of continuous exposure to office-like dim light, sleepdisrupting noise levels in the spacecraft during rest and 90' day/night, light /dark cycles through the spaceship windows (Dijk et al., 2001). Furthermore, cosmonaut Atkov reported the discomforting absence of the sensation of putting the weight of his head down on a pillow in microgravity (Vernikos, personal communication). Others report although they seem to have slept enough hours they do not feel rested when they wake up. All of these may be responsible for disturbing normal variations in sleep-wake markers and therefore sleep and circadian rhythms need to be explored systematically both on the ground and in space, in humans and non-humans, for the sake of healthy sleep and consequent work efficiency and for better quality of life.

Great strides have been made in the last decade in the field of sleep, sleep deprivation and the body's timing mechanisms as to its relationship to the environment, that have not been applied to space.

Related to this a recent paper was published though apparently dealing with something else: the bone resorption marker aminoterminal cross-linked collagen I telopeptide (NTx) was found to be increased in 20 premenopausal women volunteering for continuous wake and dim light conditions as compared to normal environmental conditions, thus confirming both the existence of an endogenous circadian rhythm in NTx with a night-time peak and the synchronizing influence of environmental factors on it (St Hilaire et al., 2017). Lower bone mineral density (Kim et al., 2013) and even increased risk of hip and wrist fractures (Feskanich et al., 2009) have been reported in shift workers who are exposed to chronic circadian, sleep and melatonin disruption. Similar effects, though maybe only additive to the already known role of muscle and bone unloading, may be expected for others with chronic circadian misalignment, such as astronauts. These should be taken into account when dealing with integrative physiology, which in fact often opens unexpected connecting paths among seemingly unrelated systems.

# CNS CHANGES

Cognition and behavior may be impaired in microgravity due to changes in cerebrovascular circulation including the increased blood pressure caused by the redistribution of body fluid (Strollo et al., 1998a; Lakin et al., 2007). The anatomical configuration of the brain and cerebrospinal fluid (CSF) spaces were found to change as a consequence of long term space flight: magnetic resonance imaging (MRI) studies showed narrowing of the central sulcus, upward shift of the brain, and narrowing of CSF spaces at the vertex in most astronauts studied (Roberts et al., 2017; Van Ombergen et al., 2017b). The duration and clinical significance of such findings warrants further systematic study of previously observed anatomical brain changes during bed rest and evidence from functional MRI (fMRI) signs of performance related cortical reorganization (Roberts et al., 2010; Roberts et al., 2015). This may be relevant to vestibular patients and elderly people with multisensory deficit syndromes or in merely inactive individuals (Van Ombergen et al., 2017a).

Most pertinent functional studies concentrated on spatial orientation, object recognition, motion perception and high-level cognitive functions including learning, memory, reasoning and calculation (Kornilova, 1997; Leone, 1998; Koga, 2000; Shehab and Schlegel, 2000; Kelly et al., 2005). To investigate this further a 7-day head-down tilt (−6 ◦ bedrest) study involving 20 subjects concluded that the first three days in space should be considered as potentially critical for cognitive performance as based on mental rotation tests (Wang et al., 2017).

Postural stability is yet another aspect of CNS performance that is affected both by spaceflight as well as aging (Demertzi et al., 2016). Increased cortico-spinal drive from leg motor cortex to lower limb motor neurons was found following postural perturbations in the elderly along with impaired perceptual processing of sensory afferent signals, which form the basis of prolonged muscle response delays during perturbed balance (Ozdemir et al., 2018). A close association between neural and muscular factors for morphological and functional adaptation to space flight was found (Ohira, 2000). This led to the proposed theory that the whole postural control system is tied together by links between vestibular, visual and somatosensory information. On Earth, these develop and are kept spontaneously active through experience of inertial and gravitational reaction forces, while in microgravity they should be actively established for postural and perceptual stability (Mergner and Rosemeier, 1998). Not to mention lunar or planetary microgravity environments where gravity levels may be at or below sensory thresholds. According to this concept, biomechanics and multi-body dynamics should be taken into account, as well as, the feedback and possibly feed forward loops used for postural control, whose complexity may be impossible to resolve in the absence of current developments in computer science and robotics. To resolve such a sophisticated puzzle as postural control may still be far from being achieved. Precise mechanisms implemented by the brain on a neural or molecular level will probably not be achieved any time soon even with the most powerful calculation tools available.

# OPTIC NERVE EDEMA

Closely linked to changes occurring in the CNS, a threatening and still poorly managed side-effect of long duration space flight is optic nerve head (ONH) edema causing reduced visual acuity with both health and mission consequences. The overall eye defects range from changes in refractive error and varying degrees of disk edema to globe flattening, choroidal folds, and cottonwool spots (Mader et al., 2011; Taibbi et al., 2013) and are referred to as space flight-associated neuro-ocular syndrome (SANS) correlated with intra-orbital and intracranial magnetic resonance imaging both in-flight, as well as, in terrestrial ultrasonographic and ocular optical coherence tomography findings (Lee et al., 2017). How this happens is unknown and yet to be determined. It seems to be at least partly explained by increased intracranial pressure accompanying the cephalad fluid shift, venous outflow obstruction, blood-brain barrier breakdown, and disruption in CSF flow with local effects on ocular structures (Wiener, 2012),

individual differences in metabolism, and the vasodilator effects of carbon dioxide (Michael, 2018). A suitable treatment of ONH depends on a better understanding of the underlying mechanisms with research moving more in this direction. Using pre- and post-flight optical coherence tomographic scans of the ONH region, global and quadrant total retinal thickness, retinal nerve fiber layer (RNFL) thickness and choroidal thickness have been calculated: circumpapillary RNFL thickness was found to be increased by a median of 2.9 µm without any change in choroidal thickness. This imaging technique is now expected to allow the assessment of longitudinal changes and the development and testing of countermeasures in astronauts, as well as potentially in patients suffering from disk edema on Earth (Patel et al., 2018).

# VESTIBULAR DEFECTS AND MOTION SICKNESS

Another typical problem experienced by the elderly, as well as by astronauts and cosmonauts, especially at the beginning of their space flight, is motion sickness. Selection of candidates to minimize the risk has always been the preferred solution, nevertheless training procedures have also been investigated throughout the space program. Allocating less attention to the central field during visual motion stimulation has been recently found to be associated with more stable vection and higher resistance to motion sickness (Wei et al., 2018). Virtual reality application designers may use this finding to strengthen and stabilize vection, while reducing the risk of visually induced motion sickness.

On the other hand, impaired vestibular function in the elderly can result in worsened gaze stabilization with an increased risk of falling. Ocular-counterroll (OCR), a gaze-stabilizing mechanism involving ocular torsion in a compensatory direction to lateral roll-tilt of the head, is predominantly mediated by the otoliths at low frequencies of motion and has been shown to be a sensitive marker of such a defect. In fact, a reduction in OCR gain with aging correlates with increased postural sway, suggesting OCR loss may also be a useful predictor of fall risk (Serrador et al., 2009). A recent paper showed that low levels of subcutaneous electrical stochastic noise (SN), known to enhance weaker signals in nonlinear systems through the phenomenon of resonance, improves otolith-ocular function in aging people with impaired otolith responses without inducing hypersensitivity or other adverse effects in those with normal function. Electrical SN may then represent an effective and well-tolerated non-invasive alternative or add-on to existing rehabilitation strategies. Moreover, its noise-based mechanism is expected to require no adaptation by neural systems, thus theoretically making long-term treatment more feasible (Serrador et al., 2018). Another possible solution may be electro-acupuncture (EA), a method of applying bilaterally, an electric current by means of a special device to needles fitted to specific points, i.e., PC 6 (Pericardium 6), ST 36 (Stomach 36), and LI 4 (Large Intestine 4) at least half an hour before the exposure to disrupting vestibular stimuli (Fydanaki et al., 2017).

# SPACE ANEMIA

Anemia in space is also a typical aging-like consequence of microgravity requiring greater understanding not only for the sake of astronauts but also to improve the quality of life of millions of elderly people all around the world (Rizzo et al., 2012). In the early days if spaceflight it was believed to be a consequence of exposure to the high environmental oxygen concentrations required to compensate for low cabin pressure, and after the normalization of intra-vehicular air pressure/composition, it was hypothesized to reflect either "acute plethora" of red blood cells resulting from fluid shifts or bone calcium reabsorption dependent increased oxygen displacement from circulating hemoglobin (Hughes-Fulford, 1993; De Santo et al., 2005). Now that these mechanisms have been ruled out, anemia is believed to depend on compensatory blunted erythropoietin-release in response to decreased intravascular plasma volume with consequent hemo-concentration and/or enhanced haemolysis (Tavassoli, 1982; Hargens et al., 2013; Kunz et al., 2017). Actually, the results from a specifically designed head down bed rest study do not seem to support such a mechanism since circulating markers of haemolysis did not change throughout the test despite hemoconcentration (Trudel et al., 2017). This may be of high clinical significance since erythrocyte loss from prolonged bed rest partly explains the unpredictable anemia in bed-ridden patients or in people with decreased mobility (Guralnik et al., 2004), an extremely prevalent defect in the elderly population. An often-ignored finding that warrants further investigation is that erythrocytes recover four weeks after the bed rest-induced, decreased production rather than increased haemolysis (Ryan et al., 2016).

# PHARMACO-KINETICS and –DYNAMICS

Aging often alters pharmacokinetics and pharmacodynamics of life-saving drugs. Similarly, microgravity induces many physiological adaptations and therefore pharmacology in space is a work in progress because it requires continuous monitoring of the physiological changes and pharmacological behavior incrementally with each successive mission duration.

Besides possible conformational receptor-drug interaction changes and metabolic clearance rate at single organ levels, numerous factors unrelated to microgravity, may affect drug effects and thus interfere with normal physiology (Vernikos, 2018, in press):



Nasal route drug administration for local and systemic delivery of many drugs is an attractive strategy for clinicians as the nasal cavity is highly vascularized and provides a large surface area for drug absorption. It is especially good when targeting the brain because neural pathways such as the olfactory and trigeminal nerves may be directly accessed. However, polar molecules cannot cross the nasal mucosa, and therefore effective solutions must be devised to enable them to cross the blood-brain barrier. Actually in vitro studies showed that hypergravity may be an effective new strategy for that. Such a finding might be of great interest in space pharmacology (Vernikos, in press) where enhanced nasal drug delivery in microgravity may be desired (Kim et al., 2018).

# 3D-SCAFFOLDS OF STEM CELL DERIVED MULTICELLULAR SPHEROIDS

Both drug research and surgery nowadays largely rely on synthetic tissues: more and more organs fail as age increases and transplantations suffer from shortage of donors all over the world while bioethics asks more and more for cruelty-free drug research. That's why for the last ten years or so artificial organs have become the object of extensive research, which seems to pave the way to productive large-scale solutions. For a long time, the development of three-dimensional (3D) structures, socalled multicellular spheroids (MCSs), was of great interest to those wishing to investigate various biological processes such as growth, proliferation, differentiation, and drug effects beyond the limits imposed by isolated cells and even single-layer cell biology. The most important finding was that MCSs could be easily produced within hours when growing cells in the Rotating Wall Vessel (RWV) or the Random Positioning Machine (RPM). These devices which allow investigators to perform more and more fruitful experiments mimic the effects of microgravity on Earth providing a useful alternative to actual microgravity facilities like those provided by the KUBIK hardware initially and the BIOLAB onboard the International Space Station (ISS) thereafter, as well as, by Russian retrievable "Bion" or "Foton" satellites, followed by smaller ones based on the CubeSat standard or combining several cubes together (e.g., the United States "PharmaSat-1," the "GeneSat-1," or the "SporeSat"). Original stem cells and those derived from human umbilical cord blood progenitors are more and more widely used to engineer 3D constructs under low-gravity conditions. They represent ideal tools for Regenerative Medicine since, in contrast to mature cells, they are able to differentiate into several different cell types, thus allowing the production of a large variety of patient autologous tissues, can be cultured for a very long time while keeping costs low, by exploiting simulated microgravity without actually going into space (Grimm et al., 2018).

# EVA AND DECOMPRESSION SICKNESS (DCS) RISK

Decompression syndrome and altitude sickness were thought to affect only the young but nowadays more and more middle-aged people progressively becoming old and specific sports-naïve elderly people decide to spend their money in a different way than before to test their performance limits and feel younger than they are by getting involved in underwater or mountain activities. Although this might seem different from what happens in flight, space can be viewed as a suitable, yet involuntary, test bed for possible lifethreatening events faced by elderly people on Earth. Before the very first EVA (extravehicular activity) was carried out NASA realized that DCS due to variable degrees of venous and arterial nitrogen embolism either limited to tendons (causing "bends") or generalized (lung and brain lesions) was a risk that had to be mitigated. The current NASA spacesuit referred to as the EMU (extravehicular mobility unit) operates at 4.3 psia or 222 mm Hg above the vacuum of space whereas the Russian Orlan space-suit operates at 5.8 psia. Increasing the space-suit pressure and/or reducing cabin inert pressure are the two ways to decrease the pressure gradient between environments to help minimize risk of DCS as complete elimination of DCS is practically impossible. However, despite these concerns, there have been no recorded cases of DCS in astronauts in pressurized space-suits during EVA, in contrast to altitude chamber technicians, who are much more prone to symptoms or signs of DCS. This might be explained by potential operational and gravitational benefits of the spaceflight upward fluid shift and pre-breathing exercise that induce faster denitrogenation (Sherman and Sladky, 2018).

# CONCLUSION

Readers looking for a systematically organized, exhaustive review of space physiology may be disappointed. Numerous such reviews have been published by these and other authors. However, due to the relentless development of science, such reviews run the risk of soon being outdated whatever the scientific relevance of the subject and competence of the authors.

As stated in the introduction, our intention was to offer the reader an update of new findings and translational approaches in space research, as advanced techniques and theories become available and the cross-talk between Space, Earth and Aging physiology continues to enrich scientific and social relevance. We hope to have provided the curious – and even possibly naïve young researchers – with a source of inspiration and of creative ideas.

# AUTHOR CONTRIBUTIONS

fphys-09-01551 November 8, 2018 Time: 19:17 # 7

FS and JV wrote the paper based on their longstanding space research experience. SG provided the group with his experienced support with respect to issues related to metabolism and internal

# REFERENCES


medicine. GS provided the group with her experienced support with respect to issues related to geriatrics and endocrinology. AM realized a thorough search of the literature and supported the group in the fields of sleep, decompression sickness and EVA.



microgravity conditions – a conceptual model. Brain Res. Rev. 28, 118–135.doi: 10.1016/S0165-0173(98)00032-0



Tavassoli, M. (1982). Anemia of spaceflight. Blood 60, 1059–1067.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Strollo, Gentile, Strollo, Mambro and Vernikos. This is an openaccess article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# High-Intensity Exercise Mitigates Cardiovascular Deconditioning During Long-Duration Bed Rest

#### Edited by:

Andrew Blaber, Simon Fraser University, Canada

#### Reviewed by:

Piero Ruggeri, Università di Genova, Italy Joyce McClendon Evans, University of Kentucky, United States

\*Correspondence:

Martina A. Maggioni martina.maggioni@charite.de orcid.org/0000-0002-6319-8566

†These authors have contributed equally to this work

‡Paolo Castiglioni, orcid.org/0000-0002-8775-2605 Stefan Mendt, orcid.org/0000-0001-8227-9655 Anika Werner, orcid.org/0000-0002-9822-0348

#### Specialty section:

This article was submitted to Integrative Physiology, a section of the journal Frontiers in Physiology

Received: 31 March 2018 Accepted: 16 October 2018 Published: 19 November 2018

#### Citation:

Maggioni MA, Castiglioni P, Merati G, Brauns K, Gunga H-C, Mendt S, Opatz OS, Rundfeldt LC, Steinach M, Werner A and Stahn AC (2018) High-Intensity Exercise Mitigates Cardiovascular Deconditioning During Long-Duration Bed Rest. Front. Physiol. 9:1553. doi: 10.3389/fphys.2018.01553 Martina A. Maggioni 1,2 \* † , Paolo Castiglioni 3†‡, Giampiero Merati 2,3, Katharina Brauns <sup>1</sup> , Hanns-Christian Gunga<sup>1</sup> , Stefan Mendt 1‡, Oliver S. Opatz <sup>1</sup> , Lea C. Rundfeldt <sup>1</sup> , Mathias Steinach<sup>1</sup> , Anika Werner 1,4‡ and Alexander C. Stahn1,5

<sup>1</sup> Charité—Universitätsmedizin Berlin, Institute of Physiology, Center for Space Medicine and Extreme Environments Berlin, Berlin, Germany, <sup>2</sup> Department of Biomedical Sciences for Health, Università degli Studi di Milano, Milan, Italy, <sup>3</sup> IRCCS Fondazione Don Carlo Gnocchi, Milan, Italy, <sup>4</sup> Université de Normandie, INSERM U 1075 COMETE, Caen, France, <sup>5</sup> Department of Psychiatry, Perelman School of Medicine, University of Pennsylvania, Philadelphia, PA, United States

Head-down-tilt bed rest (HDT) mimics the changes in hemodynamics and autonomic cardiovascular control induced by weightlessness. However, the time course and reciprocal interplay of these adaptations, and the effective exercise protocol as a countermeasure need further clarification. The overarching aim of this work (as part of a European Space Agency sponsored long-term bed rest study) was therefore to evaluate the time course of cardiovascular hemodynamics and autonomic control during prolonged HDT and to assess whether high-intensity, short-duration exercise could mitigate these effects. A total of n = 23 healthy, young, male participants were randomly allocated to two groups: training (TRAIN, n = 12) and non-training (CTRL, n = 11) before undergoing a 60-day HDT. The TRAIN group underwent a resistance training protocol using reactive jumps (5–6 times per week), whereas the CTRL group did not perform countermeasures. Finger blood pressure (BP), heart rate (HR), and stroke volume were collected beat-by-beat for 10 min in both sitting and supine positions 7 days before HDT (BDC−7) and 10 days after HDT (R+10), as well as on the 2nd (HDT2), 28th (HDT28), and 56th (HDT56) day of HDT. We investigated (1) the isolated effects of long-term HDT by comparing all the supine positions (including BDC−7 and R+10 at 0 degrees), and (2) the reactivity of the autonomic response before and after long-term HDT using a specific postural stimulus (i.e., supine vs. sitting). Two-factorial linear mixed models were used to assess the time course of HDT and the effect of the countermeasure. Starting from HDT28 onwards, HR increased (p < 0.02) and parasympathetic tone decreased exclusively in the CTRL group (p < 0.0001). Moreover, after 60-day HDT, CTRL participants showed significant impairments in increasing cardiac sympathovagal balance and controlling BP levels during postural shift (supine to sitting), whereas TRAIN participants did not. Results show that a 10-day recovery did not compensate for the cardiovascular and autonomic deconditioning following 60-day HDT. This has to be considered when designing rehabilitation programs—not only for astronauts but also in general public healthcare. High-intensity, short-duration exercise training effectively minimized these impairments and should therefore deserve consideration as a cardiovascular deconditioning countermeasure for spaceflight.

Keywords: heart rate variability, hemodynamics, countermeasure, cardiovascular deconditioning, posture, longterm bed rest

# INTRODUCTION

Upcoming deep space missions such as Martian expeditions will require exposure to up to 1,000 days in microgravity (Horneck, 2006). Space agencies are thus investigating adverse health effects of long-term missions and their possible countermeasures in order to reduce detrimental consequences for astronaut health (Aubert et al., 2016; Bergouignan et al., 2016; Frippiat et al., 2016; White et al., 2016; Lang et al., 2017). Space analogs simulating prolonged gravity changes therefore play a crucial role (Ploutz-Snyder, 2016). Bed rest with −6 degrees headdown tilt (HDT) is one of the best conditions to mimic the effect of long-term weightlessness on the human body—even within the limitations of this model. HDT shifts fluids from the lower domain to the upper region of the body, similar to the fluid centralization observed in spaceflight (Pavy-Le Traon et al., 2007; Hargens and Vico, 2016; Watenpaugh, 2016). Bed rest models also allow for the investigation of some of the effects of immobilization, secondary to hospitalization, and physical inactivity. Indeed, elderly patients spend over 80% of their hospital admission confined to their bed (Vernikos and Schneider, 2010; Baczynska et al., 2016), and physical inactivity is one of the leading causes of death in Western countries (Blair, 2009). Therefore, investigating the physiological consequences of physical inactivity and designing effective countermeasures is essential for planning future long-term space missions as well as for public health and rehabilitation purposes. Weightlessness negatively affects several physiological functions. For example, it can cause deconditioning of the cardiovascular system, which may be characterized by higher resting heart rate with altered autonomic control associated with orthostatic intolerance (Blomqvist et al., 1994; Sigaudo et al., 1998; Fortrat et al., 2001; Custaud et al., 2002). Despite research spanning at least four decades on weightlessness-associated cardiovascular alterations (Pavy-Le Traon et al., 2007; Hargens and Vico, 2016), the exact time courses of changes in hemodynamic regulation and autonomic cardiovascular control induced by long-term spaceflight are not fully understood (Liu et al., 2015; Aubert et al., 2016). Moreover, several countermeasures for cardiovascular deconditioning have already been tested (e.g., volume loading, lower-body negative pressure, hypergravity; Wang et al., 2011; Stenger et al., 2012; Jeong et al., 2014; Li et al., 2017), but exercise is the most investigated countermeasure (Blaber et al., 2009; Petersen et al., 2016; Ploutz-Snyder, 2016). However, despite the consensus on physical activity as a countermeasure, the type and intensity of the exercise are undergoing further investigation. Common exercise countermeasures include aerobic (Pagani et al., 2009; Cavanagh et al., 2017; Demontis et al., 2017) and resistive exercise (Holt et al., 2016; Demontis et al., 2017), as well as in combination with whole-body vibration (Belavý et al., 2010). Recent findings show high-intensity interval training (HIIT) to be salutary in cardiovascularly compromised persons (Ramos et al., 2015; Fleg, 2016; Hussain et al., 2016), improving aerobic capacity, endothelial, and left-ventricular function, vasomotor function, and blood pressure (Hussain et al., 2016). So far, however, HIIT has rarely been implemented to counteract cardiovascular deconditioning and orthostatic intolerance in microgravity settings (Hughson et al., 1994; Greenleaf, 1997; Hastings et al., 2012; Hargens et al., 2013). This study therefore aimed to evaluate whether short-duration HIIT is an effective countermeasure against cardiovascular deconditioning and orthostatic intolerance induced by 60 days of head-down-tilt bed rest. To achieve this aim, we compared subjects doing HIIT with a control group and investigated 1) the time course of hemodynamic changes and adaptations of the cardiovascular autonomic control during 60-day HDT, and 2) the cardiovascular response to a postural test performed before and after the bed rest confinement. As for the HIIT exercise, we administered specific lower body resistance training that provides neuromuscular force solicitation: the reactive jump protocol in a sledge jump system (Kramer et al., 2010, 2017b).

# METHODS

This research was performed as part of the European Space Agency (ESA) sponsored study "**R**eactive jumps in a **S**ledge jump system as countermeasure during **L**ong-term bed rest— RSL Study" at the DLR :envihab (German Aerospace Agency (DLR), Cologne, Germany), between 2015 and 2016. Details related to the core project design, recruitment, randomization of volunteers, and training protocol are reported elsewhere (Kramer et al., 2017a,b). The study was conducted in accordance with the Declaration of Helsinki for Medical Research Involving Human Subjects (revision October 2013) and was approved by the ethics committee of the Northern Rhine Medical Association in Düsseldorf, Germany (see Kramer et al., 2017a). After the purpose, procedures, and known risks of the tests had been explained to the participants, each participant gave written informed consent. In brief, the study consisted of 15 days of baseline data collection (BDC−15 through BDC−1), 60 days of HDT bed rest (HDT1 through HDT60), and 15 days of recovery (R+0 through R+14). During the 60 days of −6 degrees HDT, the reactive jump training was administered as a countermeasure in one randomly selected subsample (TRAIN: training), whereas the other subsample (CTRL: control) did not perform any physical training (see section subjects below). The training protocol was performed using a sledge jump system (Novotec Medical GmbH, Pforzheim, Germany) composed of a lightweight sledge sliding on rails. Cylinders pull the sledge toward the plates with force exerted on the subject by adjusting the pressure settings within the cylinders. The participant was fixed supine to the sledge with shoulder straps and with feet on force plates. Participants would then perform countermovement jumps while receiving feedback on jump height and peak force from a monitor. Participants in the TRAIN group underwent the training protocol starting from HDT1. Each training session consisted of repetitive jumps and different series of countermovement jumps with an average load equal to or above 80% of the participant's body weight. Sessions lasted about 20 min, including preparation. Training took place in the afternoon between 2 and 6 pm, 5–6 times per week, for a total of 48 sessions during the 60-day bed rest. A comprehensive description of the sledge system, the training method, and timeline is reported elsewhere (Kramer et al., 2017a,b).

# Subjects

Data were collected from 23 young, healthy, male participants (baseline: age 29 ± 6 [m ± SD] years, weight 77 ± 7 kg, height 181 ± 6 cm), who were not involved in competitive or professional sport activities at the time of the study (see Kramer et al., 2017a,b for details on inclusion and exclusion criteria). Participants were randomly allocated to a training group (TRAIN, n = 12, age 30 ± 7 years, weight 78 ± 7 kg, height 181 ± 7 cm) or to a control group (CTRL, n = 11, age 28 ± 6 years, weight 76 ± 8 kg, height 181 ± 5 cm), and were matched based on anthropometry (Kramer et al., 2017a). One subject of the TRAIN group and one of the CTRL group were re-ambulated after 49 and 50 days of HDT (instead of 60 days), respectively, for medical reasons (Kramer et al., 2017a). Notably, this did not affect their completion of the recovery phase and all the scheduled measurements were collected. Accordingly, these subjects were included in the data analysis.

# Data Collection

To evaluate short-, mid-, and long-term exposure to HDT, autonomic cardiovascular and hemodynamic data were collected on the 2nd (HDT2), on the 28th (HDT28), and on the 56th (HDT56) day of HDT. To evaluate the response to a postural stimulus, we also collected the identical data 7 days before the start of HDT (BDC−7) and 10 days after the end of HDT (R+10, R+0 being the first day of recovery) in both sitting and supine positions (**Figure 1**). In each session, a time series of physiological data were recorded for 10 min. During the lay-to-sit challenge (in BDC−7 and R+10) data were first recorded for 10 min in the supine position and then for an additional 10 min immediately after the change of posture to the sitting position. Recordings included blood pressure at the finger artery (BP) measured via continuous finger plethysmography and one-lead electrocardiogram (ECG). Both were sampled at 200 Hz. Beat-by-beat stroke volume (SV), cardiac output (CO), and total peripheral resistance (TPR) were obtained using impedance cardiography (TensoScreen <sup>R</sup> , Medis Germany). All measurement sessions were conducted between 9 and 12 am (before lunch), at least 18 h after the previous training session. Participants were required to avoid caffeine consumption during the 4 h leading up to the measurements, and were instructed not to move, talk, or fall asleep during the recordings.

# Data Analysis

An expert operator visually inspected the ECG and BP signals, identifying and manually removing possible artifacts and premature beats. Beat-by-beat time series of normal-to-normal R-R intervals were derived from the ECG tracing for heart rate variability (HRV) analysis. Beat-by-beat values of systolic BP (SBP) and diastolic BP (DBP) were obtained from the BP signal. Beat-to-beat series of R-R intervals and DBP values were interpolated linearly at 10 Hz and resampled at 5 Hz. The Welch periodogram was estimated by splitting the resampled series in 90% overlapping Hann windows of 240 s in duration, computing the FFT spectrum in each window, and by averaging the spectra over all the windows. The final periodogram was smoothed with a broadband procedure that averages adjacent spectral lines with a moving average filter whose order increases with the frequency from 3 to 11 (Di Rienzo et al., 1996). Following the guidelines on HRV analysis (Task Force of the European Society of Cardiology the North American Society of Pacing

Electrophysiology, 1996), the power spectrum of R-R intervals was integrated over a very-low-frequency (VLF, 0.005–0.04 Hz), a low-frequency (LF, 0.04–0.15 Hz), and a high-frequency band (HF, 0.15–0.40 Hz). The ratio between LF and HF powers was calculated as an index of cardiac sympathovagal balance (i.e., LF/HF). The LF power was also derived for the DBP spectrum as an index of sympathetic modulations of the vascular tone (Castiglioni et al., 2007). The short-term fractal index DFA1 was estimated by applying detrended fluctuation analysis to the beat-by-beat R-R interval series (Peng et al., 1995) and by considering block sizes not larger than 16 beats. DFA1 reflects changes in the cardiac autonomic tone, which increases when the sympathovagal balance increases or the vagal tone decreases (Tulppo et al., 2001; Castiglioni et al., 2011). The mean breathing rate was evaluated as the central frequency of the power spectrum of ECG-derived respiration (EDR) signal. The EDR signal reflects the respiratory movements of the thorax as modulations of the amplitude of the QRS complex (Schmidt et al., 2017).

# Statistics

Descriptive statistics have been reported as means and standard deviations (m ± SD) unless stated otherwise. To evaluate

SE of percent changes from baseline values (with BDC−7 supine =100%) at each time point. Colored asterisks indicate significance compared to BDC−7 within the single group. Black asterisks show significant difference from baseline in the whole sample of participants (when only the factor Time - and not the factor Group or their interaction - is significant). The pound sign denotes significant differences between groups at each time point. (A) HR, Heart Rate; (B) SV, Stroke Volume; (C) CO, Cardiac Output; (D) TPR, Total Peripheral Resistance; (E) SBP, Systolic Blood Pressure; (F) DBP, Diastolic Blood Pressure. \*\*\*p <0.001, \*\*p < 0.01, \*p < 0.05, #p < 0.05.

the time course of cardiovascular changes during and after bed rest compared to baseline, we expressed data of HDT2, HDT28, HDT56, and R+10 as percentage changes from BDC−7 by dividing the values recorded in each HDT time point by the corresponding value measured in BDC−7 supine (which therefore corresponds to the 100% reference). A log transformation was applied to frequency-domain indices to attain normal distribution (Castiglioni et al., 1999). Because of the properties of the logarithm, the normalized variables were expressed as the difference between the log-transformed value in each time point and the log-transformed value in the baseline, which therefore corresponds to the reference zero level. Two-factorial linear mixed models were then used to assess the time course of cardiovascular changes in supine position within and between subjects. Subjects were entered as random factors and bed rest Time (HDT2, HDT28, HDT56, and R+10) and intervention Group (CTRL and TRAIN) were included as fixed factors. Significant effects of Time and Group or their interaction were followed up using contrasts (with BDC−7 as a reference for Time). When only the factor Time was significant, contrasts were performed irrespective of the intervention (i.e., CTRL and TRAIN groups were pooled).

As for the postural test (i.e., the shift from supine to sitting) performed before and after HDT, variables were expressed as delta scores. The delta score corresponds to the difference (1) between the value measured before bed rest (at BDC−7) and the respective value measured after bed rest (at R+10) per group (CTRL and TRAIN) and position (supine and sitting). Therefore, delta scores are changes from baseline values (i.e., BDC−7) and express the effects of bed rest according to each group and each posture. Mixed models were used to assess within-participant and between-participant differences in delta scores. Body position and intervention group were included as fixed factors and subjects as random factors. When the factor Posture, or Group, or their interaction was significant, contrasts were performed to follow up on single comparisons and corrected for multiple comparisons by a sequentially rejective correction procedure. To test whether bed rest had a significant effect in a given group and posture, contrasts were used to determine whether each delta score differed significantly from zero (non-zero contrasts). Covariance matrices were determined by restricted maximum likelihood (REML) estimation. P-values were obtained using Satterthwaite's approximation for denominator degrees of freedom. Normality and homogeneity were checked via visual inspections of plots of residuals against fitted values. The level of significance was set at α = 0.05 (two-sided) for all testing. All comparisons were corrected for multiple comparisons using a sequentially rejective correction procedure (Hochberg, 1988). To maximize sensitivity for detecting true differences while maintaining control over family-wise Type I errors, we followed the recommendation of choosing smaller, more focused families rather than broad ones (Westfall et al., 2011). All statistical analyses and graphical illustrations were carried out using the software package R (R Core Team, 2016) Mixed models were run using the packages lme4 and lmerTest (Bates et al., 2014; Alexandra Kuznetsova et al., 2016). Adjusted means were calculated using TABLE 1 | HDT time course: significance of Time and Group factors and their interaction based on linear mixed models analysis (see text for abbreviations).


the lsmeans package (Lenth, 2016) and htmlTable (Max Gordon, 2017). Figures were created using ggplot2, ggpubr, and cowplot (Wickham, 2016; Alboukadel Kassambara, 2017; Claus O. Wilke, 2017).

# RESULTS

# Time Course of Hemodynamic Variables During HDT

**Figure 2** shows percent changes in hemodynamic values from baseline (BDC−7 supine) measured during HDT and recovery. **Table 1** reports the results of the linear mixed models analysis. The factor Time and its interaction with Group were significant for HR. Accordingly, HR was higher than at baseline from HDT28 up to the recovery phase (R+10) in CTRL participants, while not exhibiting changes in the TRAIN group. In particular, the difference between groups in HR changes was statistically significant near the end of bed rest (HDT56) and during recovery (R+10). Time was also a significant factor for SV, which decreased during HDT, recovering only partially in R+10. The factor Group was marginally significant for CO, with values lower than baseline for only the TRAIN group. The factor Time and its interaction with Group were highly significant for SBP. **Figure 2** shows a marked fall in SBP during recovery in the CTRL group only. DBP showed a marginal significance for the interaction between Group and Time. No factors were significant for TPR (**Figure 2**) and EDR; the respiratory rate did not change significantly during and after HDT.

# Time Course of Autonomic Indices During HDT

**Figure 3** shows percent changes in autonomic indices from baseline. **Table 1** shows the results of the linear mixed models. Since percent changes of spectral powers were log-transformed

before the statistical test, they are reported as differences vs. zero, i.e., the baseline reference (see Methods). The factor Time and the interaction between Time and Group were highly significant for the HF and LF powers of RRI (**Table 1**). Both these powers had values lower than baseline at HDT28 and HDT56 for the CTRL group only. Time was a significant factor and the interaction between Time and Group was close to the 5% significance threshold. This was also the case for DFA1, whose profile mirrored the profile of the HF power, with a significant increase at HDT28 and HDT56 for the CTRL group only. In this case, however, the increase was consistently statistically significant during recovery as well. As for the LF/HF powers ratio, Time was the only factor close to statistical significance (**Table 1**). No factors were significant for the LF power of DBP.

# Hemodynamic Response to the Postural Test

**Table 2** reports descriptive statistics of hemodynamic data in supine and sitting positions at BDC−7 and R+10. Not only before but also after bed rest, the shift from supine to sitting posture appears associated with an increase in HR, DBP, and TPR, and with a decrease in SV and CO in both groups. However, the bed rest had a different effect on the amplitude of the changes in the two groups, as reported in **Figure 4** (i.e., delta scores of

LF of DBP: Low-Frequency power of Diastolic Blood Pressure; (F): LF/HF, Low-Frequency to High-Frequency powers ratio \*\*\*p < 0.001, \*\*p < 0.01, \*p < 0.05.


TABLE 2 | Postural test: mean (SD) in the supine (Sup) and sitting (Sit) position before (BDC−7) and after (R+10) bed rest (see text for abbreviations).

hemodynamic variables) and **Table 3** (i.e., factors significance based on linear mixed models). The HR delta score of the CTRL group in supine position was positive and significantly higher than in sitting position (**Figure 4A**), suggesting that bed rest increased HR more in the supine than the sitting position in the CTRL group. The factor Group and its interaction with Posture were significant (**Table 3**), and bed rest did not increase HR in the TRAIN group (delta scores were negative). A significantly different delta score between groups in the supine position was also found. Bed rest decreased SV in both groups (**Figure 4B**), independent of posture (**Table 3**). CO (the product of HR and SV) reflected the combination of HR and SV delta scores. Both factors and their interaction were significant (**Table 3**) and delta scores differed between the groups in the supine position and between positions in the CTRL group (**Figure 4C**).

The factors Posture, Group, and their interaction were also significant for TPR (**Table 3**). The negative TPR delta score of CTRL participants in the supine position indicates that bed rest decreased supine TPR only in the CTRL group (**Figure 4D**). Posture and its interaction with Group were also significant factors for SBP (**Table 3**). Bed rest decreased SBP more in supine than in sitting position and markedly more in the CTRL group (**Figure 4E**). Effects of bed rest were less pronounced on DBP (**Figure 4F**) than on SBP.

# Autonomic Response to the Postural Test

The shift from supine to sitting position was associated with an increase in the LF/HF powers ratio and DFA1, and a decrease in the HF power of RRI and the LF power of DBP—both before and after bed rest (**Table 2**). However, the prolonged bed rest period influenced these changes differently in the two groups (**Figure 5**). In fact, after bed rest, the HF power of RRI in supine position only increased in the TRAIN group (**Figure 5A**), and the DFA1 in sitting position only decreased in the CTRL group (with differences in delta scores between positions found only in the CTRL group, **Figure 5D**). Furthermore, bed rest decreased the sitting LF/HF index in CTRL participants only and influenced the supine LF power of DBP differently between groups (i.e., the delta score was positive for CTRL participants and negative for TRAIN participants, **Figures 5B** and **5C**).

FIGURE 4 | (A–F) Delta scores of hemodynamic variables, i.e., the difference (1) between the value measured before bed rest (at BDC−7) and the corresponding value measured after bed rest (at R+10, see text) for the supine (Sup) and the sitting (Sit) position: adjusted means ± SE. For each panel: control group (CTRL) red bars on the left-hand side, and training group (TRAIN) blue bars on the right-hand side. Asterisks above whiskers indicate delta scores significantly different from zero. Horizontal lines with asterisks indicate significant differences between supine and sitting delta scores within the same group. Pound signs indicate significant differences between groups at the same position. (A) HR, Heart Rate; (B) SV, Stroke Volume; (C) CO, Cardiac Output; (D) TPR, Total Peripheral Resistance; (E) SBP, Systolic Blood Pressure; (F) DBP, Diastolic Blood Pressure. \*\*\*p < 0.001, \*\*p < 0.01, \*p < 0.05, ##p < 0.01, #p < 0.05.

# DISCUSSION

Our study investigated the alterations induced by 60 days of HDT in hemodynamics and autonomic modulations of heart rate. To the best of our knowledge, this study involved the highest number of participants to investigate these specific effects during long-term HDT bed rest so far. Our main findings show that: (1) HR progressively increased some days after the start of the HDT; (2) changes in SV and HR vagal modulations appeared almost synchronously; and (3) alterations in these variables and SBP were detectable several days after the end of bed rest, indicating persistent cardiovascular deconditioning. Furthermore, this high-intensity/shortduration exercise alleviated the cardiovascular deconditioning, counteracting the autonomic alterations and improving recovery. Although the adopted exercise protocol mainly involved the lower part of the body, it likely influenced central mechanisms of cardiac modulation and appears to be a promising countermeasure for long-term spaceflight missions. Different exercise-based countermeasures have been tested so far to reduce cardiovascular deconditioning. A previous study showed that low-magnitude whole body vibration with



resistive exercise prevented the increase of the autonomic index of cardiac sympathovagal balance after an HDT period lasting 60 days (as was the case in our study) without, however, improving orthostatic tolerance (Coupé et al., 2011). Other studies were different from the present in their design and duration. Nonetheless, these studies highlight the crucial role of exercise as a countermeasure against cardiovascular deconditioning. The intermittent exposure to hypergravity coupled with ergometric exercise limited the decrease in parasympathetic activity after 14 days of HDT (Iwasaki et al., 2005). Daily rowing ergometry and biweekly strength exercise training after 5 weeks of HDT prevented orthostatic intolerance only when combined with volume loading (Hastings et al., 2012). Finally, supine cycling could counteract orthostatic intolerance after 18-day bed rest only in combination with plasma volume restoration (Shibata et al., 2010). However, the vast range of different study designs and countermeasures complicate a direct comparison with the sledge jump training protocol.

# Time Course of Hemodynamic Variables and Autonomic Indices During HDT

We described the time course of long-term HDT adaptations by comparing values during HDT with baseline measures in the supine position (0 degrees). Compared to sitting, the recumbent position is characterized by a fluid shift to the upper body, which increases stroke volume and stimulates baroreceptors and volume receptors, inducing cardiac vagal activation. By choosing this reference we could describe the isolated effect of the prolonged −6 degrees bed rest and this makes our results comparable with the cardiovascular changes observed during mid- and long-term spaceflight (Baevsky et al., 2007; Di Rienzo et al., 2008; Demontis et al., 2017). In the CTRL group, HR significantly increased from baseline starting from the 28th day of HDT and remained higher than baseline on the 56th day, in the recovery phase, and 10 days after the end of bed rest (R+10 supine; **Figure 2A**). A similar pattern was observed in another 60-day HDT study (Liu et al., 2015). Interestingly, this pattern was missing in the TRAIN group, where HR remained stable during and after HDT. The HR increase is a crucial feature of cardiovascular deconditioning, and the lack of this feature in the TRAIN group is a clear marker of the efficacy of the proposed training protocol as a cardiovascular countermeasure. A question arises about the type of training we adopted as short duration HIIT is not usually considered to act as a vagal enhancer. However, some data confirm that also this type of training may induce a parasympathetic adaptation of HR when performed in the supine position (Kiviniemi et al., 2014). An alternative hypothesis for the unchanged HR in TRAIN group after 28 days of bed rest might be constant increments in left atrial volume that could have induced a bradycardic response by stretching the sinus node. However, this unlikely occurred in our study because we found a significant reduction in SV in both groups (**Figure 2A**). SV depends on contractility, on arterial blood pressure, and on atrial pressure. Since contractility did not change (as sympathetic indices of HRV remained stable in the TRAIN group) and arterial blood pressure did not change from HTD28 onwards, we may hypothesize that atrial pressure, although not directly measured, was chronically reduced (perhaps via reduced blood volume and hence preload) in both groups of subjects. A previous echocardiographic study during bed rest confirmed that the left ventricular end-diastolic volume (a surrogate of cardiac preload) progressively decreases throughout 60 days of bed rest (Westby et al., 2016). Hence, it appears that the main reason for the unchanged HR in the trained subjects could be a training-induced enhancement of vagal tone. SV progressively decreased in both groups during HDT, reaching a minimum at the end of bed rest (i.e., about 90% of the baseline value). The SV reduction is in line with the literature on HDT, which reports a decrease in plasma and blood volume by 10 to 30% within the first 24 to 72 h of confinement (Convertino, 2007). A decrease in SV suggests a dehydration condition, which was probably due to different reasons. One factor was the increased renal sodium excretion and thus reduced water retention (Convertino, 2007). Another factor could be related to tissue compression in lying position that dehydrated areas of weight bearing because of greater interstitial flow into the microcirculation (Hargens and Vico, 2016). The reduced daily physical activity might also have been a cause of dehydration (Convertino, 2007).

The training not only failed in counteracting the decrease in SV, but it might have even accelerated it. On the 28th day of bed rest, SV was 92% of the baseline value in both groups, but on the 2nd day it was equal to 100% in CTRL participants and decreased to 91% in TRAIN participants. Moreover, on the 2nd day of bed rest SBP decreased significantly only in the TRAIN group, and DBP changes tended to be lower in the TRAIN when compared to the CTRL group (**Figures 2E** and **2F**). This contrast suggests that exercise training might quickly influence the hemodynamic balance, inducing post-exercise hypotension likely due to an early blood volume reduction. We speculate that training might accelerate the loss of plasma

Horizontal lines with asterisks indicate significant differences between supine and sitting delta scores within the same group. Pound signs indicate significant differences between groups at the same position. RRI, R-R intervals; (A) HF, High-Frequency power; (B) LF/HF, Low-Frequency to High-Frequency powers ratio; (C) LF, Low-Frequency power; (D) DFA1, short-term fractal index by detrended fluctuations analysis. \*\*\*p < 0.001, \*p < 0.05, #p < 0.05.

volume, as demonstrated by recent works showing that lower limb exercise (cycling) produced a different adaptation of the autonomic sympathetic tone in supine vs. upright position (Ried-Larsen et al., 2017). Supine position activated the stretch receptors of heart, veins, and pulmonary circulation, increasing central blood volume, and blunting the metaboreflex activation. Such a response reduced heart contractility, limiting SV, and reducing blood pressure via a non-renal mechanism. At the same time, this reaction could be responsible for a reduction in plasma renin activity and therefore for an early increase in water loss (Ried-Larsen et al., 2017). Thus, a potential additive effect on the cardiovascular system of acute blood volume changes and sympathetic response to exercise might occur in the first days of training and HDT. We can suggest that, in order to prevent such additive effect, the exercisebased countermeasures should start at full load a few days after exposure to microgravity or its analog. The only other time point at which we observed a significant difference in blood pressure between groups was 10 days after the end of HDT (R+10), when CTRL participants showed a decrease in supine SBP compared to baseline of 17 mmHg (see **Table 2**), a marked phenomenon of hypotension absent in the TRAIN group. As will be discussed later, this suggested that prolonged bed rest had long-term effects on blood pressure control mechanisms, potentially leading to orthostatic hypotension at the restoration of normal gravity conditions. The implemented training protocol showed positive effects on these cardiovascular modifications. With respect to the cardiac autonomic modulations of HR, previous studies documented a decrease in HRV total power and vagal indices in early and chronic HDT (Fortrat et al., 1998; Sigaudo et al., 1998; Pavy-Le Traon et al., 2007), as well as contrasting findings on the sympathetic cardiovascular control (Hughson et al., 1994; Sigaudo et al., 1998; Fortrat et al., 2001; Ferretti et al., 2009). In our study, the HF power decreased significantly in the CTRL group on the 28th and 56th day of bed rest (**Figure 3**). The breathing rate was stable before, during, and after HDT, always falling within the HF band (**Table 2**). In our experimental set-up, the HF power thus correctly represented the respiratory component of the parasympathetic modulations of HR. Therefore, our data indicated that HDT induced a substantial reduction in vagal modulations of HR in the respiratory band. The LF/HF powers ratio is an index of cardiac sympathovagal balance: in the CTRL group, it tended to increase on the 28th and 56th day (+9 and +13% after log transformation, **Figure 3F**). Additionally, DFA1 quantifies changes in the sympathovagal balance, but unlike the LF/HF powers ratio, it considers fractal components of the HR dynamics not related to the amplitude of the oscillations. Concurrently with the HF power reductions (**Figure 3A**), DFA1 also tended to increase in the CTRL group on the 28th and 56th day of bed rest and remained higher than baseline even 10 days (i.e., R+10) after the end of bed rest (**Figure 3D**). The LF power reflects both the vagal and sympathetic cardiac modulations; in CTRL participants it decreased in a similar fashion as the HF power, suggesting the predominance of the vagal withdrawal compared to possible sympathetic activation (**Figure 3B**). The VLF power reflects the cardiac modulations of different humoral and thermoregulatory mechanisms superimposed on the autonomic cardiac control; the time course of the VLF power during bed rest was remarkably similar in the two groups, suggesting that the effects of training on the changes in HR variability during bed rest mainly regards the fast-vagal modulations of HR (**Figure 3C**). Unlike DFA1 or the LF/HF powers ratio, the power of DBP oscillations in the LF band is a measure of sympathetic vascular control not influenced by parasympathetic modulations. This index did not show any significant effect of bed rest (**Figure 3E**). Therefore, the analysis of changes occurring during bed rest in the cardiovascular autonomic indices suggested a reduction of vagal heart rate modulations and an increase in the sympathovagal balance without evidence of an altered sympathetic tone in the CTRL group. This effect was detectable up to 10 days after the end of bed rest. The trends were different in the TRAIN group without apparent alterations during the HDT period with regard to any autonomic index. Therefore, our results suggested that during HDT the reactive jumps training reduced the cardiac autonomic deconditioning.

# Hemodynamic and Autonomic Response to the Postural Test

The postural test allowed for the evaluation of the effects of the 60-day HDT on the cardiovascular system as it operates around different working points. In the supine position, the upper part of the body contains a larger volume of fluids than in sitting position. The fluid shift from sitting to supine posture is expected to increase the volume of the large vessels and the filling of the heart chambers, stimulating volume receptors which induce an autonomic response. The descriptive statistics of hemodynamic variables and autonomic indices in **Table 2** showed higher SV and vagal index, and lower HR, sympatho/vagal balance indices, and TPR in supine compared to a sitting position. Our results indicate that in the recovery phase 10 days after bed rest, the cardiovascular deconditioning affects some variables more in one position than in the other. For instance, this is the case with HR (**Figure 4A**) because the HR delta score of the CTRL group significantly differed from zero in only the supine position. Such a finding points out that after HDT without countermeasures, the cardiovascular deconditioning affects HR more in supine than in sitting position. HR delta scores of TRAIN participants were indeed closer to zero in both positions and lower than those of the CTRL group in the supine position (**Figure 4A**). This finding confirms the efficacy of the administered training protocol as a countermeasure for cardiovascular deconditioning. In contrast to HR, SV showed the same significant reduction in the recovery phase, between −5 and −7 mL, independent of the posture in both groups, suggesting that the training protocol did not affect the loss of body fluids caused by HDT. Interestingly, the postural test was able to detect a significant reduction in TPR in the CTRL group (**Figure 4D**), but in the supine position only. The difference between the two postures was also highly significant in the CTRL group. It is possible that the fall in supine TPR induced by the 60-day HDT was responsible for the significant reduction in supine SBP observed at R+10 only in the CTRL group (**Figure 2E**)—a hypotensive effect also evidenced by the negative delta score of supine SBP (**Figure 4E**). The significant differences between supine and sitting delta scores of TPR, SBP, and DBP in CTRL participants further highlighted that longterm effects of head-down-tilt bed rest depended on posture. In this regard, the crucial point is that the sledge jump training protocol showed positive effects. No significant delta scores were observed in the TRAIN group for TPR and BP in the supine position, indicating the ability of the administered training protocol to accelerate recovery after HDT bed rest. Interestingly, in the TRAIN group the significantly positive delta score of TPR in sitting position (**Figure 4D**) indicates that exercise training might even have improved the capability to increase the total peripheral resistance in sitting position. An improved endothelial function induced by training (Ashor et al., 2015) could mediate such an effect. The autonomic indices (**Figure 5**) also shows the effects of HDT related to the posture. When sitting in the recovery phase, the CTRL group had lower sympathovagal activation and vagal withdrawal (**Table 2**). This is demonstrated statistically by significant negative delta scores of the LF/HF powers ratio and DFA1 and by a significant positive delta score of the HF power in sitting position (**Figure 5**). This phenomenon, suggests an impaired autonomic response to a postural shift after prolonged bed rest, a possible marker of orthostatic intolerance. By contrast, the TRAIN group had unchanged (i.e., null) delta scores for the indices of sympathovagal balance in both supine and sitting positions, whereas the index of vagal modulations of HR (HF power, **Figure 5A**) increased after bed rest in a similar manner for both postures. These findings therefore suggest that the proposed training protocol allowed for a faster recovery of the physiological autonomic responses to posture changes.

# CONCLUSION

Considering that a supine position on Earth mimics acute cardiovascular effects of weightlessness, which induces a robust vagal activation immediately after the fluid shift to the upper body, we described the time course of long-term changes by comparing HDT to supine baseline recordings. Our data revealed different dynamics of cardiovascular adaptations. The acute autonomic changes induced by the supine position persisted throughout short-term HDT exposure and were strongly attenuated during mid-term and long-term HDT exposure, whereas SV adaptations showed an enduring trend. The administered training protocol appeared to mitigate some of the autonomic cardiovascular adaptations occurring during mid- or long-term HDT exposure and to accelerate their recovery after bed rest. The hypotensive phenomenon observed in the TRAIN group only during short-term HDT exposure, however, suggests that administering this exercise with initially light but progressively heavier loads during the first days of bed rest is an effective countermeasure.

# AUTHOR CONTRIBUTIONS

MM, PC and AS wrote the manuscript and processed the data. AS designed and directed the project. KB helped supervise the project and performed data collections with support from AW and SM. PC and MM preprocessed data for statistical analyses. AS performed statistical analyses and prepared the figures and tables. PC, MM, GM, and AS drafted the manuscript. H-CG, LR, MS, and OO provided critical feedback and contributed to the interpretation of the results. All authors discussed the results and contributed to the final manuscript.

# REFERENCES


# FUNDING

This investigation was supported by ESA (European Space Agency) and by the German Aerospace Center (Deutsches Zentrum für Luft- und Raumfahrt; DLR) through grant 50WB1525.

# ACKNOWLEDGMENTS

We thank Edwin Mulder, Alexandra Noppe, Melanie von der Wiesche, Wolgang Sies, and the entire DLR team for their operational, technical and logistic support. We also thank the European Space Agency (ESA) for providing the opportunity to participate in this study. We thank Dorothée Grevers for editing and proofreading the manuscript. In addition, we acknowledge support from the German Research Foundation (DFG) and the Open Access Publication Fund of Charité - Universitätsmedizin Berlin.

low-level paraplegia. Am. J. Physiol. Regul. Integr. Comp. Physiol. 292, R1502– R1509. doi: 10.1152/ajpregu.00273.2006


series on neurophysiology research priorities. NPJ Microgravity 2:16023. doi: 10.1038/npjmgrav.2016.23

Wickham, H. (2016), ggplot2: Elegant Graphics for Data Analysis (Use R!), 2nd Edn. New York, NY: Springer Verlag.

**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Maggioni, Castiglioni, Merati, Brauns, Gunga, Mendt, Opatz, Rundfeldt, Steinach, Werner and Stahn. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Critical Role of Somatosensation in Postural Control Following Spaceflight: Vestibularly Deficient Astronauts Are Not Able to Maintain Upright Stance During Compromised Somatosensation

#### Edited by:

Recep A. Ozdemir<sup>1</sup>†

William H. Paloski<sup>4</sup>

, Rahul Goel<sup>2</sup>†

\*

Jack J. W. A. van Loon, VU University Amsterdam, Netherlands

#### Reviewed by:

Joyce McClendon Evans, University of Kentucky, United States Irene Di Giulio, King's College School, United Kingdom Giovanni Bertolini, University of Zurich, Switzerland

#### \*Correspondence:

William H. Paloski William.h.paloski@nasa.gov †These authors have contributed equally to this work

#### Specialty section:

This article was submitted to Integrative Physiology, a section of the journal Frontiers in Physiology

Received: 19 June 2018 Accepted: 08 November 2018 Published: 27 November 2018

#### Citation:

Ozdemir RA, Goel R, Reschke MF, Wood SJ and Paloski WH (2018) Critical Role of Somatosensation in Postural Control Following Spaceflight: Vestibularly Deficient Astronauts Are Not Able to Maintain Upright Stance During Compromised Somatosensation. Front. Physiol. 9:1680. doi: 10.3389/fphys.2018.01680 <sup>1</sup> Beth Israel Deaconess Medical Center, Harvard Medical School, Boston, MA, United States, <sup>2</sup> Department of Health and Human Performance, University of Houston, Houston, TX, United States, <sup>3</sup> Neurosciences Laboratory, Johnson Space Center, National Aeronautics and Space Administration, Houston, TX, United States, <sup>4</sup> Human Research Program, Johnson Space Center, National Aeronautics and Space Administration, Houston, TX, United States

, Millard F. Reschke<sup>3</sup>

, Scott J. Wood<sup>3</sup> and

The free-fall of orbital spaceflight effectively removes the gravitational vector used as a primary spatial orientation reference on Earth. Sustained absence of this reference drives adaptive changes in the internal perception-action models of the central nervous system (CNS), most notably in the processing of the vestibular otolith inputs. Upon landing, the return of the gravitational signal triggers a re-adaptation that restores terrestrial performance; however, during this period, the individual suffers from a functional vestibular deficiency. Here we provide evidence of a transient increase of the weighting of somatosensory inputs in postural control while the CNS resolves these vestibular deficiencies. Postural control performance was measured before and after spaceflight in 11 Shuttle astronauts and 11 matched controls and nine elderly who did not experience spaceflight. A quiet-stance paradigm was used that eliminated vision, modulated the lower extremity somatosensory cues by subtly modulating the orientation of the support surface beneath feet of subjects in all groups. Additionally, in astronauts and matched controls, we challenged the vestibular system with dynamic head tilts. Postural stability on the landing day (R+0) was substantially decreased for trials with absent visual and altered somatosensory cues, especially those also requiring dynamic head tilts ( ± 5 ◦ @ 0.33 Hz) during which 20/22 trials ended prematurely with a fall. In contrast, none of the astronauts fell during eyes-closed, dynamic head tilt trials with unaltered somatosensory cues, and only 3/22 trials resulted in falls with eyes-closed and altered somatosensory cues, but static upright head orientation. Furthermore, postural control performance of astronauts was either statistically not different or worse than that of healthy elderly subjects during the most challenging vestibular conditions on R+0. Overall, our results demonstrate a transient reweighting of sensory cues associated with microgravity-induced vestibular deficiencies, with a significant increase in reliance

on somatosensory cues, which can provide an effective reference even without vision and with dynamic vestibular challenges. The translation of these results to aging population suggests that elderly individuals with visual and vestibular deficits may benefit from therapeutic interventions enhancing sensorimotor-integration to improve balance and reduce the risk of falling.

Keywords: sensory reweighting, postural control, spaceflight, somatosensory inputs, elderly

# INTRODUCTION

All neurophysiological systems including sensorimotor networks controlling reflexive and coordinated voluntary motor behaviors have evolved to function in Earth's gravity (Anken and Rahmann, 2002; Clement and Reschke, 2008). In the context of human upright stance and locomotion, gravitational sensory inputs from vestibular otolith organs serve as the primary sensory modality to establish a vertical spatial reference (Shumway-Cook and Woollacott, 2007; Macpherson and Horak, 2013; Pfeiffer et al., 2014) critical for controlling upright stance and terrestrial navigation. Adaptation of this essential spatial reference in response to sustained microgravity during spaceflight drives adaptive changes in the central nervous system (CNS) and leads to modification of internal models governing the input-integration-output characteristics of relevant sensorimotor repertoire (Clement and Reschke, 2008). The neural reorganizations that happen during spaceflight help to mitigate space motion sickness and optimize motor performance in the microgravity environment. However, these neural reorganizations are maladapted to gravitational constraints and, thus, significantly disrupt coordinated motor behaviors immediately upon returning to Earth (Paloski et al., 1992).

Upright stance control depends on the continuous integration of vestibular, visual, and somatosensory afference, and any ambiguous or disrupted inputs from one of these sensory modalities may cause destabilization of standing balance (Jacobs and Horak, 2007). Previous studies have consistently reported an increase in body sway and impaired upright stance control in astronauts following prolonged exposure to microgravity (Paloski et al., 1992, 1993; Wood et al., 2015). Misinterpretation of otolith signals (the otolith tilttranslation reinterpretation –OTTR-hypothesis), for example, has been proposed as possible mechanism of microgravityinduced maladaptive vestibular reorganization that degrade postural control and spatial orientation in astronauts while they are re-adapting to the return of gravitational inputs during early post-flight period (Young et al., 1984; Parker et al., 1985). One compensatory strategy the CNS is capable of employing during this maladapted early post-flight period can be the dynamic update of relevant internal models through sensory reweighting (Peterka, 2002). Sensory reweighting is an adaptive filtering process that regulates the relative contribution of each sensory modality to the internal model by down-weighting ambiguous afferences (e.g., vestibular) while up-weighting reliable sensory modalities to maximize overall gain and reduce signal-tonoise ratio (Polastri et al., 2012). For example, if the surface conditions are firm and stable, somatosensory inputs from the feet mechanoreceptors and ankle proprioceptors are more reliable than when standing on a soft and compliant surface.

Numerous previous studies have reported functional contribution (i.e., maintaining upright stance) of somatosensory inputs to postural control in healthy young adults with normal (Vuillerme et al., 2008) or disrupted vestibular function (Dietz et al., 1992; Mittelstaedt, 1992; Clapp and Wing, 1999; Kavounoudias et al., 2001; Bringoux et al., 2003; Modig et al., 2012), as well as in the elderly (Pasma et al., 2015) and clinical populations (Bronstein, 1999; Vaugoyeau et al., 2008). Research has shown increased reliance on somatosensory information under alcohol intoxication in healthy young adults (Modig et al., 2012), higher somatosensory weights in older adults with visual impairments (Pasma et al., 2015) and patients with unilateral vestibular loss (Peterka et al., 2011), and compensatory effects of using electro-tactile biofeedback during altered vestibular inputs in healthy young adults (Vuillerme et al., 2008). These studies have greatly improved our knowledge of the relative use of somatosensory information for maintaining postural control in various sensory contexts. However, the functional role of somatosensory inputs, signaling orientation of the body relative to surface-vertical, for stabilizing postural control during the early post-flight recovery period in vestibularly deficient astronauts is not well understood (Lowrey et al., 2014; Strzalkowski et al., 2015). Additionally, microgravity-induced musculoskeletal deconditioning and the transient vestibular deficiency can provide a unique model to better understand underlying mechanisms of impaired postural control in the elderly individuals with increased fall risks, and develop preventive rehabilitation protocols utilizing principles of sensory reweighing.

Thus, the primary goal of this study was to investigate the role of somatosensory inputs on postural control performance during disturbed/impaired vestibular conditions in vestibularly deficient astronaut subjects immediately after spaceflight. We administered dynamic head tilts during postural control tasks to further distort the accuracy of vestibular inputs following exposure to microgravity. Previous studies have shown that

**Abbreviations:** AP, anterior-posterior; CDP, computerized dynamic posturography; CNS, central nervous system; COM, center of mass; COP, center of pressure; EQ, equilibrium score; FAM, familiarization; HD, head dynamic; HE, head erect; iTTB, integrated time to boundary; JSC, Johnson Space Center; KSC, Kennedy Space Center; L-x, x days before launch of spaceflight; OTTR, otolith tilt-translation reinterpretation; PAR-Q, physical activity readiness questionnaire; R+x, x days after return from spaceflight; SOT, sensory organization tests; TTB, time to boundary.

standard sensory organization tests (SOTs) may not be sensitive enough to detect subtle upright stance control dysfunctions in patients with vestibular disorders (Paloski et al., 2006; Mishra et al., 2009; Honaker et al., 2016), who may compensate upright stance control performance by task vigilance during SOTs. Therefore, modified SOTs with dynamic head movements have been suggested for better fall risk diagnosis during functional postural control performance assessments, both in elderly (Pang et al., 2011) and clinical populations (Mishra et al., 2009). Furthermore, when vision is absent, introducing an additional experimental challenge to the vestibular system by dynamic head movements would also allow us to better understand the compensatory role of sensory reweighting as a function of the availability of reliable somatosensory inputs. Therefore, we hypothesized that: availability of reliable somatosensory cues, in the absence of vision, will mitigate destabilizing effects of both impaired (due to microgravity) and distorted (due to head tilts) vestibular function during upright stance control. As a secondary purpose, we compared postural control performance from pre-flight and return day sessions' of astronauts with healthy elderly individuals to better understand aging-related sensorimotor aspects of increased body sway.

# MATERIALS AND METHODS

# Subjects

Postural control performance during disturbed vestibular function and compromised somatosensory inputs was systematically monitored in 11 astronauts (7 males, 4 females; age range 38–49 years) before and after short-duration (11–13 days) Shuttle flights, and 11 matched controls who followed the same timeline as astronauts but did not fly into space. Postural control performance of astronauts before and immediately after returning from spaceflight was also compared with nine healthy elderly subjects (3 males, 6 females; age range 73–86 years) to infer sensory mechanisms underlying postural control impairments in the elderly population. Each astronaut subject was a first-time flier. Each control subject was matched with an astronaut subject in terms of age (±4 years), sex, height (±5 cm), weight (±5 kg), and postural control performance [same quartile of Composite Equilibrium Score (EQ)]. Time spacing between pre- and post- "flight" sessions for controls was the same as that of matched astronauts. Postural control performance was assessed using computerized dynamic posturography (CDP). All subjects were participating in the CDP testing for the first time; hence any learning effects should have been similar for all groups. All astronauts passed NASA spaceflight physical examination prior to their missions, and control subjects had passed an Air Force Class III physical examination within 12 months of beginning the study. None of the astronaut or control subjects reported any history of balance or vestibular abnormalities. Elderly subjects were selected among those with no known neurological, cardiovascular, vestibular or musculoskeletal disorders, and no history of falls for at least 6 months before the start of the study. Overall health status of elderly subjects was screened by using the Physical Activity Readiness Questionnaire PAR-Q (Canadian Society for Exercise Physiology, 2002). Experimental protocols and voluntary participation procedures were explained to all subjects before they gave their written consent. All subjects were consented before inclusion. The selection criteria and experimental procedures for astronaut and control subjects were approved by the NASA Johnson Space Center (JSC) Committee for Protection of Human Subjects. The study protocol for elderly subjects was approved by the Institutional Review Board of the University of Houston.

To establish pre-flight postural stability baseline data, each astronaut subject participated in four pre-flight testing sessions at JSC, occurring 141 (± 35), 133 (± 35), 50 (± 7), and 14 (± 1) days before launch (mean ± standard error of mean (SEM)), designated as familiarization (FAM), L-60, L-30, and L-10 sessions, respectively (**Figure 1**). The first pre-flight session, which occurred at least two days before the second pre-flight session, was considered FAM training for the postural tests using the standard SOTs, and data from this session were excluded from the analyses. The other three pre-flight sessions were considered to be independent estimates of a putatively stable individual. While nominally scheduled for 60, 30, and 10 days before the flight, launch schedules often shifted after one or more sessions had been completed. Since we did not have any reason to believe that the usual performance of astronauts would have been affected by launch delays, we accepted the actual timing while still classifying them based on expected timing. The first post-flight session (R+0) was performed at the Kennedy Space Center (KSC), Florida within 2–5 h after return from spaceflight using an experimental setup identical to that at JSC. All subsequent post-flight sessions were performed at JSC at 2 (R+2), 3 (R+3) and 7 or 8 (R+7/8) days after return from spaceflight. The second (L-60) and third (L-30) pre-flight sessions, and the third (R+3) post-flight session included the experimental postural tasks before and after exposure to shortradius centrifugation. However, only pre-centrifugation data are presented here. Astronaut and control subjects were instructed to avoid exposure to other unusual motion environments, strenuous physical activities or other experiments that might disrupt their recovery of balance function. Elderly subjects participated in a single session which included both familiarization and testing trials.

# Experimental Procedures and Data Collection

Postural control performance was evaluated using a modified CDP system (Neurocom Balance Manager, Natus Medical Incorporated, Pleasanton, CA, United States). The CDP system enabled SOT procedures, documented elsewhere (Paloski et al., 1992). During the FAM session, three trials of each of the six SOT conditions were carried out to familiarize astronaut and control subjects with the CDP system and to assess their baseline postural control performance which was also one of the metrics used to match astronaut and control subjects. During each experimental session (L-60, L-30, L-10, R+0, R+2, R+3, R+7/8), twelve 20 s trials were conducted with eyes closed using a combination of modulation of the sway of the support surface

session. All trials were conducted with eyes closed.

and head tilt conditions. The support surface was either fixed (SOT-2) or sway-referenced in the sagittal plane (SOT-5) in direct proportion to the estimated instantaneous center-of-mass (COM) sway angle (i.e., a gain of 1 was used). Subjects were instructed to either maintain the position of their head static and upright (head erect, HE) or perform continuous ± 20◦ dynamic (head dynamic, HD) pitch tilt oscillations at 0.33 Hz paced by an audible tone transmitted through lightweight headphones. Two trials were conducted for each of the HE and HD head conditions during both SOT-2 and SOT-5 support-surface conditions. Schematic representations of the experimental setup and postural task conditions used for astronaut and control subjects are shown in **Figure 1**. The trial order was counterbalanced across astronaut and control subjects and held constant across sessions. Throughout each trial, the astronaut or control subject was instructed to maintain a stable upright posture with arms folded across the chest. White noise supplied through headphones masked external auditory orientation cues. Infrared markers placed on the headset frame were used to quantify head position using an OptoTrak System (Model 3020, Northern Digital Inc., ON, Canada). Before beginning each dynamic head tilt trial, the test operator used real-time head position display information to guide the subject in achieving the desired head tilts, providing corrective instruction. Amplitude and phase of head pitch position during dynamic head tilts were obtained from both sinusoidal curve fits and detection of the maximum and minimum position in each cycle. Phase shift was made relative to the sinusoidally varying audio tone. Every dynamic head tilt trial began only after the operator has ascertained that the head tilts are approximately ± 20◦ . Once the dynamic head tilt trial began, the audible tone continued, but no feedback was provided for the remainder of the trial.

For specific details of the experimental procedures used with elderly subjects, please see Ozdemir et al. (2018). In brief, elderly subjects performed postural tasks with no head tilts under three different sensory conditions: (1) stable surface with eyes-open (SOT-1), (2) stable surface with eyes-closed (SOT-2), and (3) sway-referenced surface with eyes-closed (SOT-5). Two 90 s long trials were performed with 30 s testing duration for each of the three conditions continuously, without having a break among sensory conditions within each trial. Only the first 20 s of the data from the SOT-2 and SOT-5 conditions of the first trial were

FIGURE 2 | Representative time to boundary (TTB) time series for an astronaut and an elderly subject. Effects of spaceflight on postural control performance can be observed from head erect (HE) trials obtained from a representative astronaut subject before spaceflight (L-10) and immediately after landing (R+0) for both SOT-2 (A) and SOT-5 (B) conditions. Similarities in postural control performance between spaceflight and aging can be observed from head dynamic (HD) trials obtained on return day (R+0) in a representative astronaut subject and head erect (HE) trials obtained in a representative elderly subject for both SOT-2 (C) and SOT-5 (D) conditions. Black lines with shaded regions depict the TTB trace in anterior-posterior direction over each trial truncated at 10 s. Shaded areas indicate the iTTB (absolute) for each trial. The iTTB (%) is computed by dividing the iTTB (absolute) by the total area shown in the plot. The shaded region in (D) for an astronaut subject also show the time instance when TTB value hit zero, indicating a fall.

analyzed and presented here to compare with the performance of the astronaut subjects in their first session (L-60) and on return day (R+0).

# Data Reduction and Analyses

For the experimental sessions, the subject's center of pressure (COP) was computed directly from force transducers in the support platform sampled at 100 Hz. The extremes of the feet defined the base of support. The COM was estimated based on a 2nd order low-pass Butterworth filter (0.85 Hz cutoff) applied to the COP (Ozdemir et al., 2013). Time-to-boundary (TTB) was calculated by dividing the instantaneous anterior-posterior (AP) COM distance to the boundary of the base of support by the instantaneous COM velocity (Forth et al., 2007). The TTB at each instance suggest the time it would take to reach the boundary of the base of support if you were to continue to move in the same direction and at the same speed. A higher value of TTB implies more stability. This measure has the advantage of combining both the spatial and temporal aspects of sway by also evaluating the influence of velocity (Haddad et al., 2006), and is sensitive to changes in stability limits with aging and support surface compliance (Haibach et al., 2007). The primary postural control performance measure was the integrated area of TTB (iTTB) below an arbitrary 10 s threshold that represents an estimate of relative stability over the entire trial (**Figure 2**). The iTTB is expressed as a fraction of the total area beneath the threshold (i.e., 10 s × trial duration) and is not affected by "falls" (subject raising a foot or arm to maintain balance), which are discrete events

that cannot be considered part of the continuous EQ distribution. A lower value of iTTB represent higher stability.

The primary goal of this study was to investigate the role of somatosensory inputs on postural control performance in vestibularly deficient astronaut subjects immediately after spaceflight. SOT-5 trials for which the peak support surface tilt was less than 1◦ were discarded both in astronaut and in control subjects, as we do not expect any distortion of somatosensory cues in those trials. We chose to use 1◦ to discard trials as Peterka (2002) observed reweighting only when support surface perturbations were greater or equal to 2◦ . Such trials were identified by plotting peak support surface sway as a function of peak AP COM sway for SOT-5 trials across all head conditions for data from astronaut and control subjects. All data analyses were performed using customized MATLAB (MathWorks, Natick, MA, United States) scripts and functions.

Owing to sensory reweighting, we hypothesized that the availability of reliable somatosensory inputs, a stable platform in SOT-2, would reduce the destabilizing effects of HD tilts on postural control performance. To test this hypothesis, we calculated a series of sensory ratios using control and astronaut data from the R+0 day session. First of all, to follow similar interpretation of sensory ratios as used by ratios obtained using EQ (Neurocom, 2009), we subtracted iTTB (%) from 100 for these analyses, such that higher (100 – iTTB) % represents good balance. Somatosensory index (SI) was then calculated as a ratio of (100 – iTTB) % of HD by HE conditions and represents the subject's ability to use input from the somatosensory system to maintain balance. Somatosensory reweighing index (SRwI) was calculated as a ratio of SI of SOT-2 by SOT-5. The SRwI measure helped us to test our primary hypothesis as it represents how well the availability of somatosensory inputs can compensate for destabilizing effects of HD. Vestibular index (VI) was calculated as a ratio of (100 – iTTB) % of SOT-5 by SOT-2 conditions and represents the subject's ability to use input from the vestibular system to maintain balance. We also calculated somatosensory change index (SCI) as a ratio of SI of astronaut by control representing how much SI changed due to spaceflight, and vestibular change index (VCI) as a ratio of VI of astronaut by control representing how much VI changed due to spaceflight. SCI was calculated only for SOT-2 and not SOT-5 as during SOT-2, subjects are supposed to rely more on somatosensory inputs. Similarly, VCI was calculated only for HD and not HE as HD has been shown to be a more sensitive test to assess vestibular changes (Jain et al., 2010).

# Statistical Analyses

Preliminary analyses revealed that there was no statistical difference between iTTB values of first and second trial and thus data from only the first trial of each postural testing condition (SOT-2 and SOT-5; HE and HD as applicable) were used in statistical analyses for all groups (Control, Astronaut, Elderly). For all the "within-subject" comparisons (learning effect in control and astronaut subjects, and effects of spaceflight in astronauts), we used pairwise comparisons. If the underlying data were normally distributed [assessed using Shapiro–Wilk test (p > 0.05)], paired t-tests were carried out, else Wilcoxon Signed-rank tests were carried out. Independent sample t-tests (if data were normally distributed), or Mann–Whitney U-tests were performed for all the "betweensubject" comparisons (control vs. astronaut pre-flight, control vs. astronaut return day, elderly vs. astronaut pre-flight, elderly vs. astronaut return day). Statistical significance was accepted at p < 0.05 (SPSS version 21, SPSS Inc., Chicago, IL, United States).

# RESULTS

**Figure 2** illustrates the time series data during different postural control performance conditions from a representative astronaut subject and an elderly subject.

# Learning Effects in Control Subjects

For the SOT-2 HE condition (**Figure 3**. blue triangles), there was a significant improvement (i.e., lower iTTB) at R+0 in comparison to L-60 (Z = 2.134, p = 0.033). However, no change was observed from R+0 to R+7/8 sessions (p > 0.05), indicating that postural control was fine-tuned from L-60 to R+0 sessions and was stabilized after R+0. For the SOT-2 HD condition, postural control performance was stable across sessions. During SOT-5, significant learning effect was observed between L-60 and L-10 in the HE condition [t(8) = 3.161, p = 0.013], and then the performance remained stable for the following sessions. For the HD condition during SOT-5, there was a decreasing trend. However it was not significant (p > 0.05), and thus we can conclude that the postural control performance was stable across sessions.

# Postural Control Performance Before and After Spaceflight

Results of the independent sample t-tests showed no significant (p > 0.05) difference in postural control performance between control and astronaut subjects during any of the pre-flight sessions (L-60, L-30, and L-10) for either head condition (HE, HD) or support surface condition (SOT-2, SOT-5). Nor were any learning effects observed for astronaut subjects during pre-flight sessions for either head condition or support surface condition (p > 0.05).

To understand the effects of spaceflight on postural control, a series of paired comparisons were carried out between postflight and pre-flight sessions' in astronaut subjects. Since no learning effect was observed in astronaut subjects pre-flight, data of L-60 was used to compare with the data of the post-flight sessions of astronauts. For SOT-2, postural control performance was significantly reduced, i.e., higher iTTB, in the R+0 session when compared to the pre-flight session [R+0 vs. L-60 – HE: t(10) = 3.020, p = 0.013; HD: t(9) = 3.763, p = 0.004] for both head conditions (**Figure 3**, red triangles). For the HE condition in SOT-2, postural control performance became similar to the pre-flight level at the R+2 (R+2 vs. L-60: p > 0.05) session and remained stable during the following two post-flight sessions (p > 0.05 for R+2 vs. R+3, and R+3 vs. R+7/8). For the

HD head condition in SOT-2, postural control performance returned to the pre-flight level only at the R+7/8 [R+2 vs. L-60: t(9) = 3.540, p = 0.006; R+3 vs. L-60: t(10) = 3.370, p = 0.007; R+7/8 vs. L-60: p > 0.05] session indicating a slower recovery in postural control performance during HD trials.

For SOT-5 trials, postural control performance significantly deteriorated in R+0 session when compared to the pre-flight session [R+0 vs. L-60 – HE: Z = 2.073, p = 0.038; HD: t(9) = 6.539, p < 0.001] for both head conditions (**Figure 3**, red circles). Pairwise comparisons showed that in the HE condition, postural control performance returned to the preflight level at the R+2 session (R+2 vs. L-60: p > 0.05) and remained stable during the following sessions (p > 0.05 for R+2 vs. R+3, and R+3 vs. R+7/8). In the HD head condition, however, postural control performance was still impaired at the R+2 session compared to the pre-flight sessions (R+2 vs. L-60: Z = 2.803, p = 0.005) and returned to the preflight level only at the R+3 session and remained stable after that (p > 0.05 for R+3 vs. L-60, and R+3 vs. R+7/8), indicating longer recovery time when vestibular system was challenged.

# The Role of Somatosensory Inputs

For both astronaut and control subjects, we used data from the R+0 session to examine whether the availability of reliable somatosensory information could compensate for dynamic head tilt related performance decrements in balance control especially when the vestibular system is in a maladapted state due to microgravity effects on vestibular functioning.

First of all, we compared SI between the SOT-2 and SOT-5 conditions in control subjects (**Figure 4B**) and found that the SI for SOT-2 was significantly higher than the SI for SOT-5 (Z = 2.756, p = 0.006). This confirms, as expected, the importance of a stable, veridical, Earth-fixed reference for somatosensory inputs in HD compared to HE. Next, we compared the effects of spaceflight on the SI for SOT-2 (**Figure 4C**) and found that the SI was significantly lower in astronauts on the return day than in controls [t(19) = 2.404, p = 0.027]. While this might suggest a reduction in reliance on somatosensory cues, it seems more

TABLE 1 | Number of falls across the head (HE and HD) and support-surface (SOT-2 and SOT-5) postural test conditions in 11 astronaut subjects during R+0, R+2, R+3, and R+7/8 sessions and in 11 control subjects during the R+0 session.


likely that there could be some inaccuracies in somatosensory processing associated with spaceflight or that the alterations in the vestibular system associated with spaceflight were too profound to be fully compensated for by the somatosensory system. Furthermore, we compared the effects of spaceflight on SRwI (**Figure 4D**) and found SRwI to be nearly an order of magnitude higher in astronaut subjects on the return day than in control subjects (Z = 2.746, p = 0.005). This suggests a much higher reliance on somatosensory cues after spaceflight, even when they are inaccurate, confirming our primary hypothesis.

We then compared VI in control subjects between the two head conditions (**Figure 4E**) and found that the VI of HE was significantly higher than that for HD (Z = 2.756, p = 0.006). This confirms that, as expected, reliance on vestibular input decreases with HD. As expected, VI during HD for astronaut subjects on the return day was significantly lower than that in controls (**Figure 4F**, Z = 3.380, p < 0.001), clearly demonstrating a decreased reliance on vestibular inputs early after spaceflight.

Finally, we assessed relative decrements in performance after spaceflight associated with the vestibular and somatosensory systems by comparing SCI and VCI on the return day (**Figure 4G**). We found that the VCI was significantly lower than the SCI (Z = 2.803, p = 0.005), suggesting that the relative decrement in reliance on vestibular inputs was far greater than that for somatosensory inputs, resulting in a relative increase in reliance on somatosensory inputs.

Another functional performance metric is the number of fall (loss-of-balance) incidences observed under each test condition (**Table 1**). None of the subjects lost balance on any trial of SOT-2. The only two fall incidences observed in control subjects occurred during SOT-5 trials with HD (**Table 1**; bottom row). Conversely, on return day, all 11 astronaut subjects fell on at least one of two HD trials during SOT-5, and three astronaut subjects fell on one of the two HE trials during SOT-5. By R+2, recovery was well underway, as the incidence of falls on SOT-5 trials with HD decreased to 5/22, and beyond that, recovery was essentially complete, with only one fall observed in each of the final two test sessions.

# Astronaut vs. Elderly Comparisons

To gain better insights regarding postural control impairments in the elderly subjects, we compared postural control performance of elderly subjects during SOT-2 and SOT-5 trials (only HE) with the astronaut subjects pre-flight (L-60) and immediately after return (R+0), in both head

conditions (**Figure 5**). Pre-flight comparisons showed that postural sway was significantly higher in elderly subjects when compared to the HE [SOT-2: t(18) = -3.437, p = 0.007, SOT-5: t(17) = -5.810, p < 0.001], and the HD [SOT-2: t(18) = -2.347, p = 0.031, SOT-5: t(18) = -2.279, p = 0.035] conditions in astronaut subjects. The R+0 performance comparisons for SOT-2 trials showed no significant difference in performance between the HD condition in astronaut subjects and the HE condition in elderly subjects (p > 0.05), while in the HE condition in astronauts, postural sway was still significantly lower than that of elderly subjects in the HE condition [t(18) = -2.370, p = 0.029]. For SOT-5 trials on R+0, however, astronaut performance in the HD condition was significantly worse [t(17) = 5.190, p < 0.001] than that of elderly subjects in the HE condition, and no significant differences were found in HE trials between astronauts and elderly subjects (p > 0.05). Overall comparisons show that astronauts on R+0 (i.e., with a maladapted vestibular system) perform better than elderly subjects only when somatosensory cues are reliable. Our results also show that astronauts on the return day perform comparable to the elderly when vestibular inputs are disrupted through HD or when somatosensory cues are compromised, and perform worse than the elderly when vestibular inputs are disrupted through HD in compromised somatosensory condition.

# DISCUSSION

The current study was designed to examine somatosensory contributions to upright stance control performance during distorted and maladapted vestibular functioning following shortterm spaceflight in astronauts. Consistent with previous studies (Paloski et al., 1992; Jain et al., 2010), astronauts' postural control performance was significantly degraded on the return day (R+0), in all postural control tasks and head conditions. Although somatosensory contributions to postural control may also have been degraded in astronauts, short-duration spaceflight primarily impaired vestibular functioning such that vestibularly deficient astronauts were able to maintain their upright stance when somatosensory cues were relatively reliable on a stable surface but inevitably fell when a sway-referenced surface further challenged the reliability of somatosensory cues in the absence of vision. Considering the incidence of falls during sway-referenced postural tasks, our analyses demonstrated the critical role of reliable somatosensory cues on functional upright postural control when the vestibular system is maladapted during the early post-flight period. Finally, comparable postural control performance between elderly and vestibularly deficient astronauts during challenging vestibular conditions on the return day supports current aging literature and suggests that therapeutic strategies enhancing sensorimotor integration can improve postural control performance in older adults.

# Spaceflight Disrupts Sensorimotor Control of Balance

Spaceflight causes distinct sensorimotor reorganizations due to the sustained absence of gravitational sensory inputs and lack of mechanical loading to the musculoskeletal system. Although adapted for a long-duration stay in space, these sensorimotor reorganizations are maladaptive to function in Earth's gravity and, thus, pose serious postural control and locomotor challenges for astronauts by significantly increasing their risk of falling while they are restoring terrestrial performance during the early post-flight period. One critical aspect of monitoring postflight sensorimotor re-adaptation is to employ valid testing paradigms to detect subtle changes in functional sensorimotor performance and determine the time course of sufficient recovery for astronauts such that they can safely return to their daily life activities. During the early history of human spaceflight, astronauts were allowed to return to their daily routines and duties two days after landing based on the results of standard clinical examinations (McCluskey et al., 2001; Clark, 2002). More recent reports (Jain et al., 2010; Wood et al., 2015)

different between panels.

employing both standard and modified SOTs, and quantifying standing postural control performance with EQ scores showed substantially higher fall rates and longer recovery period during dynamic head movements with unstable support than standard SOTs with head erect and fixed-support. Our results, using iTTB as a postural control performance metric, further extend these previous reports that fall incidences and significantly higher body sway can still be observed in some astronauts when compared to their pre-flight levels, during dynamic head movements, on the last follow-up (R+7/8) examination. For example, during standard SOTs (HE) pre-flight postural control performance was found to be restored on R+2 both for SOT-2 and SOT-5. However, SOT trials with HD clearly showed that the pre-flight performance level was not achieved until R+3 for SOT-5, and R+7/8 for SOT-2, suggesting that fall risks are still present for certain astronauts up to a week following return from spaceflight during compromised visual and somatosensory conditions (e.g., walking in dim light or in dark, and over a compliant surface like sand). Due to our small sample size, however, prolonged recovery for HD during SOT-2 should be interpreted with caution since a closer examination of the data show higher variability in post-flight sessions, although no outliers were detected.

# Somatosensory Functioning Is Less Affected by Spaceflight and Critical During Maladapted Vestibular Functioning in Astronauts

Our analyses on the sensory ratios showed that the destabilizing effects of dynamic head movements on upright stance control change, as a function of reliable somatosensory inputs with respect to the surface-vertical, considerably in healthy controls but critically in vestibularly deficient astronauts. Specifically, when control subjects performed dynamic head movements blindfolded on a fixed support surface (SOT-2), providing reliable somatosensory inputs regarding body orientation, SI index was very high (0.95 ± 0.05) indicating that the performance difference between HE and HD is negligible in SOT-2 (**Figure 4B** red bar). However, when the dynamic head movements were performed on a sway-referenced support surface (SOT-5), compromising the reliability of somatosensory inputs, SI index was significantly lower (0.76 ± 0.06), suggesting a notably destabilizing effect of HD on upright stance performance even in healthy controls for SOT-5 trials (**Figure 4B** blue bar). Additionally, we also recorded two fall incidences during SOT-5 HD condition, and no fall during SOT-5 HE condition (**Table 1**), suggesting that the availability of reliable somatosensory cues may compensate for disrupted vestibular inputs in healthy controls.

However, this somatosensory driven compensation may become critical in vestibularly deficient astronauts immediately after spaceflight. Although we observed decreased SI and VI indices (**Figures 4C,F**) in astronauts, suggesting both impaired somatosensory and vestibular functioning, the degree of impairment was substantially higher in vestibular functioning following spaceflight. By comparing the ratio of changes in SI and VI between astronauts and healthy controls (**Figure 4G**), we showed a relatively high SCI but substantially decreased VCI, meaning astronauts were almost as good as healthy controls to utilize reliable somatosensory but were unable to use vestibular cues for compensating the destabilizing effects of HD. This suggests that somatosensory inputs are still relatively reliable sensory feedback source for vestibularly deficient astronauts, and thus they rely more on the less affected sensory system (somatosensory cues) to monitor their standing balance in the absence of vision while the CNS resolve transient vestibular deficiencies immediately upon return. We further supported these findings by showing an increased reliance into somatosensory weights in vestibularly deficient astronauts immediately after spaceflight (**Figure 4D**). In fact, availability of relatively reliable somatosensory cues was crucial for astronauts such that 20 out of 22 trials (% 90.9) resulted in falls when the validity of somatosensory cues for referencing gravitational vertical is further challenged during SOT-5. On the other hand, no single fall was observed when somatosensory cues could be used to infer gravitational vertical during SOT-2 trials. Thus the primary finding of this study is the critically functional role of somatosensory inputs from foot sole cutaneous receptors and ankle joint proprioceptors for maintaining upright stance in vestibularly deficient astronauts following spaceflight. Considering all the analyses we performed along with fall incidences, the difference between falling and standing for an astronaut during maladapted vestibular functioning seems to heavily depend on the reliability of somatosensory cues monitoring body sway with respect to the gravitationalvertical.

Promising hypotheses have been proposed such that the perceptual mechanism of vestibular (mal)adaptation following prolonged exposure to microgravity is explained mainly by reinterpretation of otolith inputs (Wood et al., 2015). Since otolith graviceptors only respond to translations in space, but not tilts, prolonged exposure to microgravity results in neglecting afferent signals from head tilts during spaceflight, and thus any head tilt is perceived as translation immediately upon returning to the Earth (Young et al., 1984; Parker et al., 1985). With an impaired vestibular function on the return day, performing dynamic head pitch movements on a fixed support surface poses further challenge to the postural control system by causing a unique ambiguity across sensory channels such that while vestibular inputs would be transmitting translation signals, somatosensory inputs from foot soles and ankle proprioceptors would be transmitting COP displacements and rotational torques of body sway, respectively. Owing to its plasticity, the CNS can resolve such sensory conflicts through sensory reweighting. Various form of this dynamic and compensatory sensory readaptation process has been increasingly investigated over the last two decades in postural control research (Peterka, 2002; Jeka et al., 2008; Vuillerme et al., 2008; Peterka et al., 2011; Modig et al., 2012; Polastri et al., 2012; Asslander and Peterka, 2014). In his seminal study, for example, Peterka (2002) monitored body sway characteristics on a sway-referenced platform with a progressive increase in surface sway angle. His results showed that blindfolded healthy subjects initially rely on the vertical surface reference to control body orientation for up to two

degrees of surface sway angle but switches to a gravitational vertical reference by moving in the opposite direction to surface sway at larger sway amplitudes to maintain upright stance, indicating an accuracy based dynamic up-weighting and downweighting of vestibular and somatosensory cues, respectively.

Although majority of studies have focused on sensory reweighting mechanisms of vestibular and visual inputs, recent studies consistently reported importance of somatosensory reweighting by evidencing increased reliance to mechanoreceptor inputs from the foot sole, ankle proprioceptor, and/or tactile cues to compensate for the destabilizing effects of different forms of disrupted or impaired vestibular and visual functioning on upright stance control (Vuillerme et al., 2008; Modig et al., 2012; Lowrey et al., 2014; Strzalkowski et al., 2015). Our results further extend these findings that reliable somatosensory cues from a firm and stable surface are important in healthy adults to stabilize upright stance during disrupted vestibular function (i.e., HD condition), but crucial in vestibularly deficient astronauts to prevent them from falling during the early period of post-flight recovery.

Although our data suggest a transient increase of the weighting of somatosensory inputs to upright stance control, caution should be taken to understand falls during dynamic head tilts in vestibularly deficient astronauts. It is obvious that, in addition to the absence of reliable somatosensory cues, many other sensory and musculoskeletal factors might have also contributed to the inevitable falls observed in astronauts during dynamic head movements on sway-referenced surface conditions. For example, as previously reported in patients with the bilateral vestibular loss (Peterka, 2002), vestibularly deficient astronauts might have failed to employ functional sensory reweighting during sway-referenced support surface condition (SOT-5) which optimally requires down-weighting of somatosensory and up-weighting of vestibular inputs. It is likely that astronauts' increased reliance on somatosensory cues was maladapted during SOT-5 trials such that they were unable to switch from a surface reference to a gravity reference. Another important aspect to consider in falling astronauts is microgravityinduced musculoskeletal deconditioning: that can cause a decline in muscle stiffness and loss of force and power due to prolonged unloading; which may all contribute to falls (Wood et al., 2015). Considering all these mechanisms, however, we cannot rule out increased reliance on somatosensory cues when vestibular inputs are disrupted in astronauts during upright stance control task on the return day since we have observed the same strategy in control subjects with intact vestibular and musculoskeletal functioning. Our comparisons for the effects of dynamic head movements within the same sensory condition (**Figure 4**) strongly supports increased weighting of somatosensory inputs such that dynamic head movements do not further destabilize postural control performance on a fixed surface, but only on the sway-reference surface condition.

# Sensorimotor Impairments in the Elderly

In many ways, microgravity-induced physiological adaptations including overall deconditioning in musculoskeletal and cardiovascular systems, and a general decline in sensorimotor functions resemble the physiology of aging (Vernikos and Schneider, 2010). In this respect, spaceflight can be considered a unique model (Young, 1993) to probe underlying mechanisms of aging-related postural control impairments as older adults are long known to have increased body sway and thus highly prone to falls (Horak, 2006). Comparing postural control performance between elderly subjects and astronauts, our data suggest that the increased body sway in elderly subjects, in the absence of vision, can be attributed to vestibular deficiencies mainly, but also to somatosensory deficits. Post-flight comparisons indicate vestibular deficiencies in elderly subjects such that astronauts perform either comparable to or worse than elderly individuals on a sway referenced platform with HE and HD trials, respectively. No differences were also found between astronauts and elderly when astronauts perform HD trials on a stable support surface. Altogether post-flight comparisons suggest that astronauts perform a lot more comparable to elderly during the most challenging vestibular conditions (**Figure 5** right panels). However, when somatosensory cues are reliable during the fixed support surface condition, better postural control performance in astronauts either during pre-flight comparisons with HD trials or post-flight comparisons with HE trials suggest somatosensory deficits or compromised reweighting of somatosensory cues in elderly subjects. Alternatively, a general decrease in musculoskeletal function with age should also be considered as an important factor for increased postural sway in the elderly during challenging sensory conditions. Overall, our findings are in agreement with many previous reports arguing that the vestibular system degenerates the most with age (Horak et al., 1989; Sturnieks et al., 2008; Barin and Dodson, 2011; Faraldo-Garcia et al., 2012; Liston et al., 2014b), and sensory re-weighting is slower and/or compromised during challenging upright stance conditions in healthy elderly adults when compared to young adults (Pavlou et al., 2004; Jeka et al., 2010).

Many implications can be derived from this study. A firm, stable surface for standing and walking can compensate for challenges associated with reduced vision (darkness, smoke-filled cabin) and dynamic head movement requirements immediately after landing, which can be very critical for safe egress. The translation of these results to aging population suggests that elderly individuals with visual and vestibular deficits may benefit from therapeutic interventions (Liston et al., 2014a) enhancing sensorimotor integration to improve postural control and reduce the risk of falling.

# ETHICS STATEMENT

This study was carried out in accordance with the recommendations of "The Protection of Human Research Subjects, NASA Flight IRB" with written informed consent from all subjects. All subjects gave written informed consent in accordance with the Declaration of Helsinki. The study protocol for astronaut and control participants was approved by NASA Flight IRB, and for elderly participants was approved by the IRB of the University of Houston.

# AUTHOR CONTRIBUTIONS

fphys-09-01680 November 23, 2018 Time: 15:50 # 12

RO helped with data analyses, interpretation of results, and writing of the manuscript. RG also helped with data analyses, interpretation of results, and writing of the manuscript. MR and SW helped with the study design and data collection. WP designed the study, and helped with data collection, interpretation of results, and revision of the manuscript.

# REFERENCES


# ACKNOWLEDGMENTS

We wish to thank members of the Neurosciences Laboratory at NASA Johnson Space Center for data collection and analysis support, and the mission integration coordinators for implementation support. Finally, we thank all subjects for their willing participation and insightful feedback.



**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Ozdemir, Goel, Reschke, Wood and Paloski. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# The Role of Enhanced Cognition to Counteract Detrimental Effects of Prolonged Bed Rest: Current Evidence and Perspectives

#### Uros Marusic1,2 \*, Voyko Kavcic<sup>3</sup> , Rado Pisot<sup>1</sup> and Nandu Goswami2,4 \*

1 Institute for Kinesiology Research, Science and Research Centre Koper, Koper, Slovenia, <sup>2</sup> Department of Health Sciences, Alma Mater Europaea – European Center Maribor, Maribor, Slovenia, <sup>3</sup> Institute of Gerontology, Wayne State University, Detroit, MI, United States, <sup>4</sup> Head of Research Unit: "Gravitational Physiology, Aging and Medicine", Otto Loewi Research Center of Vascular Biology, Immunity and Inflammation, Medical University of Graz, Graz, Austria

#### Edited by:

Ovidiu Constantin Baltatu, Anhembi Morumbi University, Brazil

#### Reviewed by:

Alessandro Tonacci, Istituto di Fisiologia Clinica (IFC), Italy Omar Šerý, Masaryk University, Czechia

#### \*Correspondence:

Uros Marusic umarusic@outlook.com Nandu Goswami nandu.goswami@medunigraz.at

#### Specialty section:

This article was submitted to Integrative Physiology, a section of the journal Frontiers in Physiology

Received: 21 May 2018 Accepted: 11 December 2018 Published: 23 January 2019

#### Citation:

Marusic U, Kavcic V, Pisot R and Goswami N (2019) The Role of Enhanced Cognition to Counteract Detrimental Effects of Prolonged Bed Rest: Current Evidence and Perspectives. Front. Physiol. 9:1864. doi: 10.3389/fphys.2018.01864 Prolonged periods of physical inactivity or bed rest can lead to a significant decline of functional and cognitive functions. Different kinds of countermeasures (e.g., centrifugation, nutritional, and aerobic interventions) have been developed to attempt to mitigate negative effects related to bed rest confinement. The aim of this report is to provide an overview of the current evidence related to the effectiveness of computerized cognitive training (CCT) intervention during a period of complete physical inactivity in older adults. CCT, using a virtual maze navigation task, appears to be effective and has long-lasting benefits (up to 1.5 years after the study). Moreover, enhanced cognition (executive control) reduces decline in the ability to perform complex motor-cognitive dual-tasks after prolonged period of bed rest. It has been demonstrated that CCT administration in older adults also prevents bed rest stress-related physiological changes [these groups showed minimal changes in vascular function and an unchanged level of brain-derived neurotrophic factor (BDNF)] while control subjects showed decreased peripheral vascularization and increased plasma level of the neurotrophin BDNF during a 14-day bed rest. In addition, the effects of CCT are evident also from the brain electrocortical findings: CCT group revealed a decreased power in lower delta and theta bands while significant increases in the same EEG spectral bands power were found in control subjects. If we consider an increase of power in delta band as a marker of cortical aging, then the lack of shift of EEG power to lower band indicates a preventive role of CCT on the cortical level during physiological deconditioning induced by 2-week bed rest immobilization. However, replication on a larger sample is required to confirm the observed findings. Applications derived from these findings could be appropriate for implementation of hospital treatment for bed ridden patients as well as for fall prevention programs.

Keywords: non-pharmacological countermeasures, aging, cognitive training, geriatrics, falls prevention

# INTRODUCTION

fphys-09-01864 January 21, 2019 Time: 17:53 # 2

Hospitalization and prolonged bed rest represent major risk factors for older persons, often resulting in irreversible deterioration in functional status and a significant decline in the quality of life (Blain et al., 2016; Bousquet et al., 2017; Goswami, 2017; Goswami et al., 2017). The proportion of older persons (defined as ≥65 years) is increasing worldwide, and is projected to double by the year 2030 with an expansion in life span of another 10 years by the year 2050 (United Nations, 2013). In addition, the number of super-olds (≥80 years) is expected to triple, from 22 to 61 million from year 2008 to 2060, respectively (European Commission, 2009). It has been shown that the greater the age, the stronger the association of bed confinement with functional loss and failure to recover during hospitalization (Covinsky et al., 2003) or in the nursing home setting (Fortinsky et al., 1999). As a consequence of this trend in population aging, higher demands for public health and aging services together with concomitant increases in health care costs are foreseen (Blain et al., 2016; Bousquet et al., 2017; Goswami, 2017).

The bed rest model was first introduced in the 1960s to simulate acute adaptations to the microgravity environment involved in space flights (Adams et al., 2003; Goswami et al., 2015a). Indeed, bed rest confinement, especially during prolonged hospitalization, could be modeled on the so-called bed rest protocol where healthy participants spend a number of days in a horizontal or (more extreme) head-down tilt bed rest condition. The negative adaptations of the cardiovascular system were observed to be similar in spaceflight and bed rest confined persons (Goswami, 2017). The negative effects of bed rest confinement, however, occur ten times faster than those that arise due to the normal aging process (Vernikos and Schneider, 2009). Previous research has mainly focused on the effects of prolonged bed rest on cardiovascular and musculoskeletal physiology (Pavy-Le Traon et al., 2007; Pisot et al., 2016; Gao et al., 2018). With improved technological development and increasing availability of brain imaging techniques, changes and adaptations of the central nervous system can be more effectively studied and the application of these techniques has become of greater interest in bed rest research (Koppelmans et al., 2017; Van Ombergen et al., 2017a,b; Gandarillas and Goswami, 2018). Typically, past bed rest studies included young and healthy participants and were designed to simulate the microgravitational environment and its accompanying effects, relevant for space flight missions (Blaber et al., 2013; Goswami et al., 2013; Cvirn et al., 2015; O'shea et al., 2015; Waha et al., 2015). Both horizontal and headdown tilt bed rest protocols have been implemented. Important differences in the two protocols such as in the adaptation in tissue fluid redistribution and hydrostatic pressures, however, need to be taken into account (Hargens and Vico, 2016).

The aim of this review is to summarize the current scientific evidence regarding bed rest-related impacts on mainly cognitive outcomes of healthy adults, and to highlight the importance of non-physical countermeasures such as cognitive training. The primary application of such scientific knowledge is therefore to provide the basis for the development of possible countermeasures both for the negative consequences of bed rest confinement and space flight microgravity, in particular in regard to negative bed rest effects in older persons during hospitalization. Such countermeasures are potentially extendable to addressing important consequences of the aging process in general. Emphasis is placed in this review on cognitive countermeasures to prevent not only cognitive but also sensorimotor adaptations that occur during acute and chronic situations involving hospitalization, prolonged physical inactivity, and the general aging process.

# SEARCH STRATEGY AND STUDY SELECTION

Scientific literature in English language was acquired through searches conducted on PubMed/MEDLINE (NLM), Embase, and Web of Science databases until October 1, 2018. A search for manuscripts on "cognitive training" and "bed rest" (with specific deviations of keyword combinations, such as "hospitalization," "cognitive intervention," "brain training," "mental training," etc.) was conducted and yielded 807 results. Furthermore, reference sections of included manuscripts were inspected to identify additional manuscripts of interest. Our results showed that there were no other studies besides "Bed Rest Study – PANGeA, Valdoltra 2012 (Slovenia, EU)," which investigated effectiveness of cognitive interventions during bed rest. Therefore, the final search consisted of four manuscripts (Goswami et al., 2015b; Marusic et al., 2015, 2018a; Passaro et al., 2017) that represent the same bed rest campaign, while each manuscript reported results based on different outcome measures.

# IMPACT OF REDUCED PHYSICAL ACTIVITY AND BED REST ON FUNCTIONAL AND COGNITIVE OUTCOMES

The positive role of physical activity in maintaining effective mobility is well-established by research and promoted by the popular media (Hui and Rubenstein, 2006; Muller et al., 2016). Reduction of the appearance of illness and chronic disease, improvements in gait and balance, as well as reduction in the risk of falls are among the most important advantages of engagement in physical activity by elderly people (Kovacs et al., 2013). A recent review highlighted the fact that up to 82% of total brain gray matter volume can be modified by engaging in physical activity (Batouli and Saba, 2017). Also, several types of cognitive-motor interventions have been previously applied in an older adult population showing positive improvements in functional and cognitive performance (van Iersel et al., 2007; Fraser et al., 2017), which is recently supported by brain structural adaptations also in subjects with mild cognitive impairment (Maffei et al., 2017). However, seniors often have limited access and fewer opportunities to engage in physical exercise programs with a 50% dropout rate in such activities in the first 3–6 months (Allen and Morey, 2010). Regardless of the issues of limited access and/or lack of motivation for engaging in

physical exercises, older adults are often forced to limit or even completely eliminate physical activity due to injuries or surgeries, which results in a specific syndrome referred to as the "disuse syndrome" (Bortz, 1984). Among the main characteristics of this syndrome are premature aging, skeletomuscular fragility, obesity, cardiovascular vulnerability and depression (Bortz, 1984).

In general, frail seniors are often limited by a disability- or a disease-related burden preventing their physical activity, and for them engaging in real-world situations involving complex locomotion where higher cognitive-resource demands are required can be a challenge. Reduced levels of physical activity have also been shown to be associated with increased risk of cognitive impairment and various types of dementia (Laurin et al., 2001). Interestingly, the results of bed rest studies do not always point toward the same conclusions. Over the years, various stressors that could lead to cognitive impairment after periods of hospitalization have been proposed. The increased levels of stress created by hospitalization itself with accompanying alternations of stress-response hormones and neuro-chemicals, as well as delirium, medications and polypharmacy, and depression are important factors in this regard (for a detailed review, see Mathews et al., 2014).

The majority of bed rest trials in which cognitive performance was assessed were carried out in a head-down tilt position and yielded contradictory results. For example, impaired cognition was found in a study of Lipnicki et al. (2009). These authors reported alterations of cognitive processes associated with decision making after 50 days of head-down tilt bed rest. After a similar length (45 days) of head-down tilt bed rest, Liu et al. (2012) observed worsening of executive functions. On the other hand, a 16-day head-down bed rest study did not report any changes in executive functions (Ishizaki et al., 2009). Moreover, in other studies, neither 17 days, nor 60 or 90 days of head-down tilt bed rest affected general cognitive functioning (Shehab et al., 1998; Seaton et al., 2009).

There is a scarcity of literature available reflecting cognitive outcomes after horizontal bed rest. After 14-day horizontal bed rest with young adults, (Dolenc and Petri, 2013) observed a minor improvement in mental visualization and no change in other assessed cognitive functions. In the same study, older individuals showed significant impairments in delayed recall (Dolenc and Petri, 2013), which was, however, not the case for those subjects who had cognitively stimulating environment during bed rest (Marusic et al., 2018a).

As extensively summarized in the "bed rest and cognitive functioning review" by Lipnicki and Gunga (2009), results from experimental studies with healthy young and older individuals are also not conclusive and do not unequivocally point in the same direction. More specifically, only eight of 17 bed rest studies included in that review reported significant detrimental effects on cognitive performance. Six studies reported unchanged cognitive functioning after bed rest, whereas three studies surprisingly showed improvements in cognitive performance. In the latter case, task exposure and practice effects could mask the underlying detrimental effect of bed rest on cognitive functioning (Lipnicki and Gunga, 2009), suggesting that eliminating practice effects in neuropsychological tests is important for the better evaluation of bed rest on cognitive functioning.

The majority of bed rest trials assessing cognitive performance before and after bed rest fail to address the underlying adaptation of the brain and subsequent correlational analysis between behavioral and neural outcome measures. Differences also exist among bed rest designs which vary in terms of amount of days that the participants were bedridden, type of the bed rest protocol used (e.g., horizontal or head-down tilt), as well as the motivation behind the studies. Some of these studies aimed at replicating spaceflight conditions, lack of sensory-motor stimulation and immobilization, and/or post bed rest recovery (Lipnicki and Gunga, 2009; Marusic et al., 2014b). Thus, an open question remain how CCT could be used as a general approach for improving cognitive performance in bed rest confined older subjects.

# COGNITIVE TRAINING AS A POSSIBLE COUNTERMEASURE DURING PROLONGED BED REST

Cognitive training aimed at optimizing cognitive functioning and/or slowing brain aging has been extensively used, especially with healthy older adults. It generally involves guided practice on tasks representing different domains of cognition in order to increase or maintain particular cognitive functions such as memory or attention. Cognitive training programs are commonly run as a time-limited, daily sessions for a specified period of intervention (e.g., 1 hour per day for 5 days a week for a total of 20 sessions). The training tasks are often designed to present an increasing challenge to cognitive abilities and thereby induce learning. A variety of tasks and approaches have been used for cognitive training (for a detailed review, see Tardif and Simard, 2011), with most of the reviewed studies reporting significant improvements in cognitive functions associated directly to those trained (e.g., Ball et al., 2002; Willis et al., 2006; Klusmann et al., 2010; Bahar-Fuchs et al., 2017), while most studies demonstrated only a limited transfer to other cognitive functions and/or activities of daily living (Ball et al., 2002; Unverzagt et al., 2007).

In the past decades, several reviews have shown beneficial effects of cognitive interventions in healthy older adults (Papp et al., 2009; Martin et al., 2011; Tardif and Simard, 2011; Mowszowski et al., 2016; Mewborn et al., 2017; Webb et al., 2018). Each concluded that cognitive training can effectively improve various aspects of objective cognitive functioning, such as memory performance, executive functioning, processing speed, attention, fluid intelligence, and subjective cognitive performance. In a recent meta-analysis, authors Marusic et al. (2018b) reported the generalization of cognition-based interventions to a distal untrained domain, such as gait performance. The influence of cognition on mobility control in older adults has been shown previously (Heuninckx et al., 2005). This knowledge has opened

new perspectives for cognitive training interventions for older population in general or those older individuals who are reluctant or not able to follow a physical activity intervention.

Recently, there is an increased use of CCT, which allows structured practice on standardized, and cognitively challenging tasks. CCT has several advantages over traditional drill and practice methods, including visually appealing interfaces, efficient and scalable delivery, the ability to measure performance and response time changes in multiple methods, and the ability to constantly adapt training content and difficulty to individual performance. The advantage of performing CCT in a supine/horizontal position opens new perspectives for implementing such a protocol in hospital/rehabilitation institutions. For a summary of non-physical approaches, readers are referred to Marusic and Grosprêtre (2018). The next section summarizes the impact of CCT during prolonged bed rest as a novel tool for mitigating negative effects of hospitalized older patients in acute phase after injury/surgery and/or in the subsequent rehabilitation process.

# OUTCOMES OF THE CCT EFFECTS DURING BED REST

In this section we review the current evidence related to potential cognitive countermeasures in relation to extreme environments (e.g., experimental bed rest or hospitalization), which has not received much attention. A recent pilot study (Marusic et al., 2018a) showed the effectiveness of CCT intervention during bed rest in older adults. Sixteen healthy older male individuals (mean age of 60 years) were randomly assigned to an intervention and an active control group. Results revealed that CCT using virtual spatial navigation was effective and exerted long-lasting effects (up to 1.5 years after the study) as evaluated by improved performance on the virtual navigation task which was specifically targeted (Marusic et al., 2018a). In the same study, there were significant transfer effects of CCT, specifically on executive functions, attention, and processing speed (Marusic et al., 2018a). It was also observed that there was a detectable decline in the ability to perform complex motorcognitive dual-tasks in the control group of older adults, but that CCT reduced the negative impact of bed rest on these integrated tasks, indicating better outcomes for the cognitively active intervention group (Marusic et al., 2015). Consequently, participants who followed the CCT protocol started their 28-day rehabilitation period (immediately after 14-day bed rest) from a higher functional and cognitive level.

Additionally, these data also revealed that CCT-related effects were also observed on the peripheral vascular function, perfusion assessments (Goswami et al., 2015b) as well as in the level of plasma brain-derived neurotrophic factor (BDNF) (Passaro et al., 2017). Older adults who underwent CCT did not show bed rest stress-related physiological changes (e.g., minimal changes in vascular function and increased level of BDNF) which indicate a preventive role of CCT during physiological deconditioning induced by 2-week bed rest immobilization. The mechanism of augmented BDNF levels after bed rest was attributed to a protective overshooting of the brain to counteract the bed rest-related negative effects (Soavi et al., 2016).

In addition to the above-mentioned findings, the electroencephalographic (EEG) recordings were used to evaluate the effects of 14-days bed rest on the brain neuroelectric activity. EEG results obtained with baseline eye-closed recordings showed that older adults who underwent CCT showed decreased power in lower delta and theta bands while control subjects showed significantly increased power in the same EEG spectral bands (Marusic et al., 2014a,b; Marušic, ˇ 2015). A so-called global "slowing" of the baseline, intrinsic EEG [e.g., increases in power in the slower delta range (2– 4 Hz)], occurs with aging (Rossini et al., 2007). Vecchio et al. (2013) observed "slowing" of EEG in healthy older individuals progressing to mild cognitive impairment and probable Alzheimer's disease. We concluded that increased spectral power in baseline EEG in control subjects but not in intervention subjects supported the notion that the CCT prevented negative effects of 14 days bed rest on brain baseline neuroelectric activity (i.e., increased power in EEG lower spectral bands) indicative of brain aging. Moreover, when analyzing event-related potentials (early perceptual processing of a stimuli), additional neuronal recruitment for the same amount of processing was observed only in the control group while participants in CCT group did not show the same trend (Marušic, 2015 ˇ ). In the same study, greater working memory enhancements (reduced P200 latency component) were observed in the CCT group, as compared to the controls.

In addition to CCT, other non-physical/cognitionbased interventions, such as action observation and motor imagery might also be incorporated into experimental bed rest research. Motor imagery represents the mental simulation of an action without any corresponding motor output (Decety, 1996), while action observation (observing someone else's movement) is known to activate the brain mirror neurons (Nedelko et al., 2010). The combination of both techniques induces even greater activity in motor areas of the brain as compared to either intervention alone (Taube et al., 2015). To date, no study has tested these nonphysical techniques during a bed rest, which might open new perspectives for mitigating bed rest-related adaptation of the central nervous system (Van Ombergen et al., 2017b; Marusic and Grosprêtre, 2018).

# CONCLUSION AND FUTURE DIRECTIONS

Overall, CCT interventions, developed from an underlying brain-based model (Marusic et al., 2018a), show that cognitive engagement during bed rest can trigger changes not only at

the behavioral level, but also at the peripheral physiological (peripheral perfusion and blood BDNF level) and neuroelectric level. Thus, CCT intervention might represent a new promising approach for mitigating possible bed rest-associated physiological, functional, and cognitive declines, especially when motor execution is constrained or limited (e.g., during acute hospitalization) and may particularly be of special value for addressing the impact of bed confinement in older persons. Finally, it may also represent a promising research avenue as well as an option for a practical implementation in hospital settings and fall prevention programs.

# REFERENCES


Bortz, W. M. II (1984). The disuse syndrome. West. J. Med. 141:691.


# AUTHOR CONTRIBUTIONS

UM drafted and wrote the article. VK performed to article drafting and final corrections. RP did final corrections. NG contributed to the article idea, drafting, and final corrections.

# ACKNOWLEDGMENTS

The authors acknowledge the financial support from the Slovenian Research Agency (research core funding no. P5-0381).



studies (1986-2006). Eur. J. Appl. Physiol. 101, 143–194. doi: 10.1007/s00421- 007-0474-z


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Marusic, Kavcic, Pisot and Goswami. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# The Promise of Stochastic Resonance in Falls Prevention

#### Olivier White1,2, Jan Babicˇ 3 , Carlos Trenado<sup>4</sup> , Leif Johannsen<sup>2</sup> and Nandu Goswami<sup>5</sup> \*

1 INSERM UMR1093-CAPS, Université Bourgogne Franche-Comté, UFR des Sciences du Sport, Dijon, France, <sup>2</sup> Acquired Brain Injury Rehabilitation, Faculty of Medicine and Health Sciences, School of Health Sciences, University of East Anglia, Norwich Research Park, Norwich, United Kingdom, <sup>3</sup> Laboratory for Neuromechanics and Biorobotics, Jožef Stefan Institute, Ljubljana, Slovenia, <sup>4</sup> Leibniz Research Centre for Working Environment and Human Factors TU Dortmund (ifADO), Institute of Clinical Neuroscience and Medical Psychology, University Hospital Düsseldorf, Düsseldorf, Germany, <sup>5</sup> Otto Loewi Research Center for Vascular Biology, Immunology and Inflammation, Medical University of Graz, Graz, Austria

Multisensory integration is essential for maintenance of motor and cognitive abilities, thereby ensuring normal function and personal autonomy. Balance control is challenged during senescence or in motor disorders, leading to potential falls. Increased uncertainty in sensory signals is caused by a number of factors including noise, defined as a random and persistent disturbance that reduces the clarity of information. Counter-intuitively, noise can be beneficial in some conditions. Stochastic resonance is a mechanism whereby a particular level of noise actually enhances the response of non-linear systems to weak sensory signals. Here we review the effects of stochastic resonance on sensory modalities and systems directly involved in balance control. We highlight its potential for improving sensorimotor performance as well as cognitive and autonomic functions. These promising results demonstrate that stochastic resonance represents a flexible and non-invasive technique that can be applied to different modalities simultaneously. Finally we point out its benefits for a variety of scenarios including in ambulant elderly, skilled movements, sports and to patients with sensorimotor or autonomic dysfunctions.

### Edited by:

Ovidiu Constantin Baltatu, Anhembi Morumbi University, Brazil

### Reviewed by:

Aiguo Song, Southeast University, China Alessandro Giuliani, Istituto Superiore di Sanità (ISS), Italy Lianchun Yu, Lanzhou University, China

> \*Correspondence: Nandu Goswami

nandu.goswami@medunigraz.at

### Specialty section:

This article was submitted to Integrative Physiology, a section of the journal Frontiers in Physiology

Received: 27 September 2018 Accepted: 11 December 2018 Published: 28 January 2019

### Citation:

White O, Babic J, Trenado C, ˇ Johannsen L and Goswami N (2019) The Promise of Stochastic Resonance in Falls Prevention. Front. Physiol. 9:1865. doi: 10.3389/fphys.2018.01865 Keywords: stochastic resonance, balance disorder, orthostatic intolerance, aging, falls

# THE CHALLENGE OF HEALTHY AGING

Life expectancy is increasing globally and functional contributions to society post-75 years are highly attainable. Over one third of adults over the age of 65 fall each year (Sattin et al., 1990). Two million patients present annually to emergency units, worldwide, complaining of dizziness upon standing up or after an accidental fall (Goswami, 2017; Goswami et al., 2017). Falls are ranked within top three causes in terms of years lived with disability in most regions of the world (WHO). Falls are not only associated with injury and morbidity, but also to reductions in physical, psychological, and social capacities (Myers et al., 1996; Blain et al., 2016; Bousquet et al., 2017). The direct cost of falling exceeds \$10 billion a year in the United States, with almost 9,500 deaths per year attributed to falling alone (Myers et al., 1996). The total costs of falls in the population of the United Kingdom has been estimated from more than two billion GBP to up to 4.4 billion GBP per year by the NHS (Public Health England with the National Falls Prevention Coordination Group Member Organisations, 2017) and is predicted to rise further. Critically, the consequences of a fall after 75 years old are much worse than between 65 and 75 years old and include fracture, frequent hospitalizations, increased morbidity and mortality. Preventing the occurrence of such events is therefore of paramount importance.

Epidemiological studies have shown that 30–70% of falls occur while walking on level ground. However, simple it appears, bipedal ambulation is demanding in many ways and gait relies on

**343**

complex sensorimotor integration. The act of balancing on our lower extremities is not learned until about 10–12 months of age. A healthy sensory system is necessary for successful postural control and locomotion. Indeed, the central nervous system must maintain accurate estimates of the position of the body in space and of the limbs in relation to each other (proprioception). Efficient postural reflexes must be triggered when external perturbations are detected. Furthermore, appropriate cardiovascular responses that keep blood pressure stable after the transition from lying/sitting to standing are required. The act of balancing needs to become automatic so that other tasks will not jeopardize it by interfering. During senescence, impairments in one or more of these systems, including cognitive functions, lead to increased risk of falls. Falls in elderly and impaired individuals are a complicated phenomenon comprising multi-factorial intrinsic and extrinsic risks (Shumway-Cook et al., 1997). Intrinsic factors, or those related to the individual, include a decreased performance in the balance control system, with loss of mobility being a strong indicator for increased fall risk. In order to maintain stability, adequate levels of vision, vestibular function, musculoskeletal function, and proprioception are all required. Extrinsic factors, or those pertaining to environmental hazards, contribute significantly to fall incidents and include obstacles to trip over, poor lighting, slippery surfaces, or inappropriate furniture. Hence, an in-depth understanding of the multitude of neuromuscular, cognitive, sensory, sociological and environmental factors that contribute to balance control are necessary for early diagnosis and treatment of elderly who present significant risks of falls.

Since the process of balance control relies on many physiological factors and sensory signals, maintaining the quality of these signals at their optimal level is fundamental. Here, we review how a highly promising phenomenon called stochastic resonance could play an important role toward addressing balance control, and consequently, toward falls prevention.

# STOCHASTIC RESONANCE: WHEN NOISE BECOMES AN ALLIED TO THE BRAIN

To ensure optimal control of a system, adequate and accurate information inputs are required. For instance, navigating a computer mouse to a button on the screen entails the estimation of target position but also the hand/mouse position. These are available only through (biological) sensors that also embed uncertainties and are sometimes available as very weak delayed signals. Stochastic resonance can help enhancing detection and processing of a weak signal blurred by the many sources of uncertainties and perturbations.

# The Concept of Stochastic Resonance

In linear systems, optimal performance is obtained in the absence of noise. However, restricting models of natural systems to noise-free and linear systems is not realistic. First, noise is ubiquitous and can never be completely eliminated. Second, systems with thresholds characterize a large class of non-linear systems, e.g., from neurons to behaviors (such as the perception of sensory stimuli). This is where the concept of stochastic resonance becomes interesting: it describes any phenomenon where the presence of noise in a non-linear system improves the quality of an output signal compared to when there is no noise (McDonnell and Ward, 2011). Intuitively, noise is usually thought of as detrimental and associated with words such as nuisance or undesirable, and the concept of it being useful is apparently contradictory. The rationale behind stochastic resonance is concisely illustrated in the work of Gammaitoni et al. (1998). Let us consider a marble moving in a symmetric double well potential V(x) (**Figure 1A**). Without any external actuation force, the mass remains stuck in one of the wells. A potential barrier (1V) prevents the marble from swapping positions between wells. When a small periodic driving force is submitted to the system, it generates oscillations of the deepness of each well. Still, the potential barrier – although decreased compared to its resting state – is too high too allow transitions. An appropriate small dose of noise (stochastic) injected into the system will constructively (resonate) combine with the driving force and statistically allow periodic transitions between the wells. Hence, adding a small noise to a weak input signal can make the resultant signal surpass the threshold of a given neuron or neural circuit and hence provide useful information about the input weak signal for the central nervous system. On the other hand adding a too large unrelated noise signal will make the output of the threshold detection non-linear system useless in terms of providing information about the weak input signal. A trade-off must then be sought between these two extremes (**Figure 1B**). The graphical representation of such trade-off, namely a U-like curve represents the signature of the stochastic resonance phenomenon (**Figure 1C**).

By adopting this concept of stochastic resonance, it becomes possible to detect a weak signal using noise. In that case, the signal exceeds the detection threshold stochastically, generating an information that is transmitted to the output. This concept has been successfully demonstrated and applied to optimize logic operations in genetic regulatory networks blurred by various forms of noise (Wang and Song, 2016; Zhang and Song, 2018). The presence of a control noise in terms of a forcing signal enhanced the reliability of the logical function implemented in those networks. Furthermore, stochastic resonance is also effective to detect subthreshold signals at the single neuron level and can even be optimized in neuronal networks (Chen et al., 2008). Interestingly, this phenomenon does not only enhance the detection of subthreshold – but also superthreshold – signals in feedforward neuronal networks (Stocks, 2000; Yu et al., 2016). In biological systems, a threshold is reached when biological sensors (cutaneous mechanoreceptors in hands or feet, proprioceptors in muscles and joints, hair cells in the inner ear, etc) receive a signal. However, the amplitude of that signal must be large enough to elicit a response to the central nervous system that will eventually, once processed by the brain, trigger an action (e.g., a postural adjustment to prevent a fall). Unfortunately, these thresholds increase with age (Wells et al., 2003) hampering their detection. Stochastic resonance therefore makes the signal detectable again,

as if this technique actively adapted the sensitivity of the suboptimal sensor. In that way, the system can maintain the same responsiveness to hazardous situations. Therefore, shortening reaction latencies through decreased information processing time puts a system in better conditions to circumvent unwanted effects, such as falls (Toledo et al., 2017).

Stochastic resonance can be applied to a range of physiological systems. This technique has been shown to improve detection of low tactile stimuli in the hand mechanoreceptors (Collins et al., 2003; Moss, 2004; Stein et al., 2005; Trenado et al., 2014c) and various motor functions (Richardson et al., 1998; Kitajo et al., 2003; Aihara et al., 2008; Mulavara et al., 2011; Trenado et al., 2014a). Furthermore, noisy (stochastic) stimulation of the vestibular system has the potential to improve motor functions (Pan et al., 2008; Samoudi et al., 2014; Lee et al., 2015), postural stability (Pavlik et al., 1999; Pal et al., 2009; Mulavara et al., 2011; Samoudi et al., 2014; Inukai et al., 2018), may prevent orthostatic intolerance (symptoms when standing upright) and cardiovascular responses (Soma et al., 2003; Yamamoto et al., 2005; Tanaka et al., 2012), and possibly also auments cognitive functions (Yamamoto et al., 2005; Pan et al., 2008; Wilkinson et al., 2008; Kim et al., 2013). Since stochastic resonance emerges in any thresholdactivated system, its effects may be relatively independent of any underlying pathology affecting perceptual uncertainty in sensory systems. All the above mentioned physiological systems are relevant for maintaining a stable body balance. While the effect of stochastic resonance may be small in absolute terms, utilizing it in situations where margins are important can lead to large benefits. The principle of system enhancement by stochastic resonance is well documented, and several small studies indicated improvements in more than one domain of balance control during exposure to sensory noise via different sensory modalities (multisensory stochastic resonance). However, this promising technique has thus far only been tested in a limited number of patients and healthy controls and for durations not exceeding 24 h. In the following sections, we review in detail these effects and identify emerging applications.

# The Vestibular System

The technique of applying stochastic vestibular stimulation instead of using a more traditional square-wave or sum of sinewave signals is relatively new. Several studies have focused on performance improvement with stochastic resonance applied on the vestibular system such as body responses in posture, balance, and gait (Fitzpatrick et al., 1996; Pavlik et al., 1999; Scinicariello et al., 2002). In addition, stochastic vestibular stimulation at imperceptible levels improves stability during balance tasks in normal, healthy subjects (Mulavara et al., 2011, 2012). Similarly, these stimulations also improve ocular stabilization reflexes in response to whole-body tilt and postural balance performance on

an unstable compliant surface in Parkinsonian patients (Pal et al., 2009; Samoudi et al., 2014).

The vestibular system is connected to spinal, cerebellar and cerebral motor control structures and can be selectively activated with external electrodes. A constant current stimulator developed to ensure imperceptibility of electrical stimulation of the vestibular system by subjects has been engineered (Mulavara et al., 2011). A series of evaluations on the efficacy of stochastic vestibular stimulation during standing and walking on unstable surfaces and on perception of tilt sensation in otolith-canal (intra-vestibular) conflict scenarios has also been done (Wuehr et al., 2016a,b). These researchers have evaluated the frequency characteristics of the electrical stimulus to optimize balance performance of subjects standing on a compliant surface with their eyes closed. Low imperceptible amplitude bipolar binaural electrical stimulation of the vestibular organs was applied using a constant-current stimulator through electrodes placed on the mastoid processes in the range of 30–330 µA root mean square (RMS) in 30 µA RMS steps. Using measures of head and trunk stability and a multivariate optimization criterion, and by employing a white noise-based stochastic stimulation signal, these researchers have shown that the stochastic vestibular stimulation in the range of 30–120 µA RMS improved balance performance in normal healthy subjects as well as in patients with Parkinson's disease. More recently, it was also shown that noisy galvanic vestibular stimulation can enhance roll vestibular motion perception (Keywan et al., 2018). Additionally, cross-planar improvements in balance performances in the range of 5–26% in normal healthy control subjects have been seen. These results indicate that stochastic vestibular resonance may be sufficient to provide a comprehensive countermeasure approach for improving postural stability.

# Haptics and Somatosensory Perception Are Relevant for Balance Control

For the control of body sway, human postural control system can flexibly utilize different sensory channels, such as nonplantar skin receptors. Uncertainty about the state of the body or a limb, such as its position in space can lead to erroneous motor planning and, consequently, injuries or falls. Should additional tactile and proprioceptive information become available, however, more accurate state estimates based on multisensory integration and prediction may decrease the likelihood of unsuccessful behaviors (Wolpert and Ghahramani, 2000; Todorov, 2004).

Mechanically non-supportive fingertip contact (<1 N) with an earth-fixed reference reduces body sway in quiet standing with eyes closed and may be even more efficient than vision alone (Holden et al., 1994; Jeka and Lackner, 1994; Lackner et al., 1999). A review by Baldan et al. (2014) on the effect of light touch on postural sway in individuals with balance problems due to aging, brain lesions or other motor or sensory deficits concluded that light touch leads to very reliable body sway reductions irrespective of the underlying balance impairment. Most studies demonstrated the benefits of haptic feedback exclusively in a quiet standing posture but improvements have also been reported in more dynamic postural activities, such as balance compensation following a mechanical perturbation (Dickstein and Laufer, 2004; Johannsen et al., 2007, 2017; Forero and Misiaszek, 2013), overground and treadmill walking (Dickstein and Laufer, 2004; Fung and Perez, 2011; Forero and Misiaszek, 2013; Kodesh et al., 2015) and staircase negotiation (Reeves et al., 2008; Reid et al., 2011).

Reductions in body sway during upright standing observed with light fingertip contact to an external stationary reference can be further reduced by the application of vibratory noise to the touch contact (displacement <0.1 mm) (Magalhães and Kohn, 2011). A typical experimental setting that can be used to investigate these aspects is discussed in detail in Magalhães and Kohn (2011). Briefly the participant's posture is measured by means of a force plate. The area covered by the projection of the trajectory of the center of mass serves as performance index. Measurements can be performed in different sensory conditions involving the provision of visual feedback or not, and the availability of an external reference for fingertip light touch. Values obtained in experimental conditions are compared to a defined baseline performance. In the stochastic resonance condition, vibratory noise stimuli is applied to the fingertip. Power of low frequency components of body sway (<0.5 Hz) were particularly responsive to the vibratory noise stimulation (Magalhães and Kohn, 2011). This effect of vibratory noise stimulation on the frequency components of body sway, however, seems to depend on the relative vibratory amplitude. Kimura et al. demonstrated that vibratory stimulation with an amplitude of the 50% of each participant's vibrotactile threshold caused reduction in the power of high-frequency components (>1 Hz) of body sway in contrast to greater stimulation amplitudes (Kimura et al., 2012).

Aging is accompanied by a constant decline of the sense of touch: between the age of 20 to 80 years, acuity thresholds increase by approximately 1% per annum with the fingertips tending to be one of the most vulnerable body parts (Wells et al., 2003). This loss of sensory acuity impacts on various aspects of function in the elderly, including the manual handling of objects and control of postural stability. These impairments reduce quality of life and independence because elderly lose dexterity and might have falls with serious physical injury. However, older adults retain the ability to use fingertip contact for augmentation of body sway feedback despite reductions in their tactile sensitivity (Tremblay et al., 2004). In fact, the efficacy of tactile feedback for sway reduction seems to be even greater in older compared to younger adult participants perhaps due to more severe aging-related loss of somatosensory information in the distal parts of the lower extremities (Baccini et al., 2007). Differential effects noted between young and elderly indicate that elderly people gain more in motor control performance than do young people with the application of noise to the feet (Priplata et al., 2003). The benefits of tactile information for the control of body balance in quiet stance as well as dynamic activities is robust and has been demonstrated in a number of neurological disorders.

# Body Balance

fphys-09-01865 January 24, 2019 Time: 19:21 # 5

Increased postural instability and falls risk due to impaired somatosensation is associated with reduced plantar sensation in older adults, patients with diabetic neuropathy and stroke patients. In the past one and a half decades, however, the effect of stochastic resonance by noise enhanced stimulation on balance control has been demonstrated as a very robust phenomenon. For example, a meta-analysis by Woo et al. concluded that enhanced noise stimulation of the lower limbs results in moderate to high effects sizes on parameters of postural regulation and stability (Woo et al., 2017). Changes in postural performance with noise enhanced stimulation were evaluated with respect to diverse postural activities, such as standing and walking, the body location of the applied stimulation, such as the soles of the feet and more proximal regions of the lower extremities and the fingertips. In the majority, the stimulation modality was vibrotactile but electrical and auditory stimulation have also been used. Finally, target populations ranged from healthy young adults to patients with neurological impairments and reduced somatosensation. A systematic review by Bagherzadeh Cham et al. (2016) on the benefits of subthreshold vibratory noise stimulation of the soles of the feet in older adults and diabetic individuals concluded that vibratory stimulation improves balance and gait performance in these populations.

Mechanoreceptors in the soles of the feet, providing information about pressure distribution and shear forces, pose an important sensory modality for the control of posture during quiet standing as well as during walking. This is evidenced by systematic alterations of the center of pressure trajectories during quiet standing during suprathreshold vibrotactile stimulation of the plantar foot zones as well as plantar electrical stimulation in neurologically healthy individuals (Kavounoudias et al., 1998, 1999, 2001; Roll et al., 2002). These involuntary drifts away from the equilibrium state during suprathreshold vibrotactile stimulation may be a cause why suprathreshold vibrotactile noisy stimulation has a destabilizing effect on postural control in young and older adults (Simeonov et al., 2011).

In contrast to suprathreshold levels of vibratory stimulation, Priplata et al. demonstrated that subthreshold vibratory noise stimulation improves body sway stability in patients with peripheral and central somatosensory deficits such as following diabetic neuropathy and stroke (Priplata et al., 2006). This was also demonstrated for a more dynamic postural task such as Timed-Up-and-Go (Podsiadlo and Richardson, 1991), in which subthreshold (70–85% of the sensory threshold) plantar vibrotactile stimulation increased complexity of sway dynamics and mobility (Zhou et al., 2016). Not only performance in the Timed-Up-and-Go improves during plantar subthreshold vibrotactile stimulation but temporal gait variability is also reduced in older adults (Lipsitz et al., 2015). The individual falls risk seems not to play a role with respect to the benefit on gait variability (Galica et al., 2009). Stephen and coworkers showed, however, that the effect of noise-enhanced vibrotactile stimulation on gait variability in older adults is dependent on participants' initial variability levels (Stephen et al., 2012). Individuals with relatively low initial gait variability showed increases, while for those who expressed high gait variability initially, variability was strongly reduced. This shows that not every individual will gain an advantage of subthreshold vibrotactle noise stimulation. The frequency and the amplitude of plantar vibrotactile stimulation is relevant, as these affect longrange correlations in stride length and stride interval during walking in healthy young adults (Chien et al., 2017).

Kelty-Stephen and Dixon (2013) reanalyzed the original data by Priplata et al. (2002) and found that the effects of subthreshold noisy vibrotactile stimulation on body sway dynamics are modulated by the degree of autocorrelations present in body sway, which could explain differences in interindividual responsiveness. It is therefore crucial to adjust the stimulation parameters to an individual's specific requirements. For example, Wells et al. applied an adaptive psychophysical procedure to assess participants' vibrotactile thresholds of the foot soles in younger and older adults and demonstrated that subthreshold vibratory noise stimulation increased their plantar sensitivity (Well et al., 2005). Further, they suggested that their procedure allows a reliable a priori determination of participants' optimal stimulation noise level to augment plantar vibrotactile sensitivity.

Subthreshold vibrotactile noise stimulation augments postural performance under more difficult task conditions too. Noise-enhanced vibrotactile stimulation benefits gait variability in terms of stride width in a fatigued state, for example during inclined treadmill walking with 30% added body weight (Miranda et al., 2016). Dettmer et al. (2015) tested young and older adults in an intersensory conflict situation by sway-referencing the visual input, thereby removing any visual sway-related information. In this context, older adults have been shown to show dramatically increased postural instability due to overreliance on visual information for body sway control (Simoneau et al., 1999). Vibrotactile noisy stimulation was found to improve balance performance particularly in the older adults (Dettmer et al., 2015). This demonstrates the stimulation's potential to attenuate age-related visual overreliance and susceptibility to intersensory conflict. However, in a follow-up study which increased the challenge of balance control during standing by the addition of a secondary working memory task, Dettmer et al. (2016) did not find an enhanced effect of vibration despite moderate correlations between the vibrotactile plantar sensitivity and postural parameters.

A limitation of the capacity of subthreshold vibrotactile noise stimulation to reduce body sway seems to exist, however, when the postural context becomes too complex. Keshner et al. (2014) challenged control of body balance with two kinds of disruptive stimuli alone and in combination in healthy young adults: a continuous optokinetic visual field rotation around the pitch axis, which imposed intersensory motion conflict, and a mental calculation task imposing additional attentional demands. While plantar subthreshold noisy vibrotactile stimulation reduced body sway during either mental calculation or visual stimulation, the increased task complexity of the visual-conflict-dual-task combination diminished the influence of subthreshold vibrotactile stimulation on the dynamics of body sway. Also, it is important to consider,

that the sensitivity of the soles of the feet to vibrotactile stimulation can be affected by the general postural context (Mildren et al., 2016).

The alternative to subthreshold vibrotactile noisy stimulation of mechanoreceptors is electrical stimulation muscle proprioception. For example, subthreshold electrical noisy stimulation of the lower leg muscles (tibialis anterior, triceps surae) in seated and standing participants expressed a connection between reduced variability in isometric plantar flexion force and reduced body sway (Magalhães and Kohn, 2012, 2014). Cutaneous receptors respond to electrical noise stimulation as, when applied at the knee, it reduces sway during single-leg standing in older adults (Gravelle et al., 2002). Finally, in terms of balance control, auditory white noise stimulation reduces the variability in both the lower and higher frequency components of postural sway in young and older adults as well (Ross and Balasubramaniam, 2015; Ross et al., 2016).

# Hand Function and Tactile Sensitivity

As mentioned above, subthreshold vibrotactile noise stimulation to the fingertips augments balance performance. It also improves performance in manual fine motor control. For example, Sueda et al. (2013) demonstrated that subthreshold vibratory noise applied to the grip of a surgical device improves the tactile sensitivity mediated by the device. Similar effects were reported by Sawada et al. (2015). Furthermore, Beceren et al. (2013) demonstrated that adjusted subthreshold vibratory noise to the fingertips reduces the sensitivity threshold. Interestingly, normal vibration of the fingertip differed from the tangential vibration in their study. While normal vibration seemed to stimulate mainly fast-adapting Type I fibers, the tangential vibration also resulted in triggering slow-adapting Type II receptors (Beceren et al., 2013).

The augmenting effects of subthreshold vibratory noise stimulation may not be restricted to the actual body location of vibratory stimulation but can radiate to other locations as well. In stroke patients, for example, Enders and colleagues reported augmented fingertip sensitivity during stimulation at the wrist and dorsal skin surface of the paretic hand (Enders et al., 2013). Remarkably, fine motor dexterity of the paretic hand also seems to benefit from this remotely induced increased fingertip sensitivity (Seo et al., 2014). Suprathreshold vibration leads to reduced fingertip sensitivity, while the remote benefits of subthreshold stimulation seemingly do not depend on the exact location on the vibrated hand (Lakshminarayanan et al., 2015). The later finding led to the conclusion that any remote effects stem from central, possibly supraspinal origins. In a further study to follow-up the neurological basis of the remote effects of vibratory noise stimulation, Seo et al. reported increased somatosensory evoked potentials in the primary sensorimotor and premotor cortices during fingertip stimulation with simultaneous subthreshold wrist vibration of 60% of the vibrotactile threshold (Seo et al., 2015). Interestingly, it has also been suggested that improvement of sensorimotor performance during a visuomotor task via vibrotactile stimulation of the index finger is consistent with an increase in cortical motor spectral power and corticomuscular coherence (Trenado et al., 2014b).

Stochastic resonance can also influence motor learning. From the perspective of motor control, an error signal is exploited by the central nervous system to update its motor policy and reduce the discrepancy between observed and predicted errors (Diedrichsen et al., 2005). For instance, when playing darts, visual errors help us derive the best strategic change to eventually aim at the target. Improved signal quality can refine the internal state estimate and the accuracy of the error signal, therefore contributing to increased learning rates. As such, stochastic resonance can potentially facilitate movement learning processes that is based on integration of sensory inputs and sensorymotor coupling. Mendez-Balbuena et al. (2012) showed that an individually determined optimal level of mechanical noise significantly improves sensorimotor performance in a static force compensation task involving the index finger (Mendez-Balbuena et al., 2012). Effects of stochastic resonance were also studied in the context of sensorimotor performance during exercise induced muscle damage. Gleeson (2017) studied how application of mechanical vibrations to the biceps femoris muscle influences sensorimotor performance of target force replication (Gleeson, 2017). The study showed that subthreshold mechanical vibrations compensate the negative effects of muscle damage and suggested that more efficient means of delivering stochastic resonance (e.g., functional electrical stimulation) would be needed to improve the effectiveness to enhancing sensorimotor performance under adverse conditions of exercise stress.

# Visual Information

Vision is highly relevant for balance control as well (Figueiro et al., 2011). Stochastic resonance has been shown to enhance visual perception in humans by addition of pixelnoise to static scenes (Simonotto et al., 1997; Moss, 2004). Likewise, visual noise increased visual contrast detection sensitivity around threshold level and thereby improved pattern recognition and perception of ambiguous 3-D figures (Leopold et al., 2002; Sasaki et al., 2006). Also, it was demonstrated that adding background white pixel-noise to a random dot motion stimulus improved ability of participants to discriminate among motion's direction (Treviño et al., 2016).

Remarkably, applying weak perturbations directly on the cortex also showed positive effects. Online low intensity transcranial magnetic stimulation facilitates detection of weak motion signals. In contrast, high intensities lead to impairment in detection. Thus, it was suggested that transcranial magnetic stimulation acts by adding noise to neuronal processing (Schwarzkopf et al., 2011). Recently, administration of transcranial random noise stimulation (100–640 Hz zeromean Gaussian white noise with intensities ranging from 0 to 1.5 mA) to the occipital region of human participants resulted in detection accuracy of visual stimuli that followed an inverted U-shape function (**Figure 2**). The authors interpreted this effect as a stochastic resonance phenomenon (van der Groen and Wenderoth, 2016).

FIGURE 2 | Participants were tested on a visual perception task in three different conditions. (A) Participants fixated on the center of the screen and a visual stimulus was randomly presented either in the first (shown here) or second interval. After the second interval, participants had to indicate which interval contained the stimulus. In the main experiment, the stimulus contrast was fixed to yield either 60% detection accuracy (subthreshold group) or 80% detection accuracy (suprathreshold group). (B) Representation of the three different experiments. In the tRNS–noise and tRNS–control experiments, no noise was presented on the screen. (C) Representative data of individual participants. The participants in the visual–noise (left) and tRNS–noise (middle) experiments show a peak in their detection performance when noise was added to a subthreshold stimulus (orange line), but not to a suprathreshold stimulus (blue line) (van der Groen and Wenderoth, 2016).

# CURRENT CHALLENGES, RECOMMENDATIONS AND FUTURE PERSPECTIVES

There is promise that combined stochastic resonance approaches may improve motor functions in older persons and in patients. There are, however, a number of challenges and broader open questions that prevent real applications from emerging.

Although targeting one sensory modality is already sufficient to demonstrate improvements on a certain performance outcome, the effect of combining more than one sensory modality has yet to be evaluated comprehensively within the context of falls and balance control. The potential interferences between modalities should be carefully addressed in appropriate experimental designs. Moreover, every individual may respond differently to the same stimulation. Theoretical work is needed to identify the relevant parameters that would allow for a

personalized, online and optimal intervention. In the future, devices exploiting stochastic resonance should adopt smart approaches that will lead to restoration of impaired functions and that will be used intermittently.

While stochastic resonance is well documented theoretically, we lack clear understanding of its mechanisms and possible connections to other similar observed effects in biological (and non-biological) systems. The broadest possible definition of stochastic resonance is that it occurs when randomness has a positive role in a signal-processing context (McDonnell and Abbott, 2009). Initially, stochastic resonance was considered to be restricted to the case of periodic input signals. In fact, it now is widely used as an all-encompassing term, whether the input signal is a periodic sine wave, aperiodic or even chaotic. Depending on the input signal, the output performance takes the form of SNR (periodic) or mutual information (aperiodic) in function of noise level. Stochastic resonance reduces to vibrational resonance when the input signal is a high-frequency periodic force (Landa and McClintock, 2000; Deng et al., 2010) and to chaotic resonance when a system responds to a weak signal through the effects of chaotic activities (Nobukawa et al., 2017).

Over the course of evolution, the brain learned to integrate the effects of gravity – that cannot be removed – to optimize motor actions (Crevecoeur et al., 2009; White, 2015; Rousseau et al., 2016). Similarly, noise is ubiquitous at all levels, from cellular to perceptual and motor, and it is virtually impossible to remove it completely (Faisal et al., 2008). Motor strategies need to be set in ways that make use of the inherent noise to obtain an optimal response.

The concept of noise is closely related to the one of variability. Interestingly, the theoretical model of optimal movement variability (Stergiou and Decker, 2011) is based on the complementary concepts of complexity and predictability. The optimal state of a biological system may be characterized by chaotic temporal variations in the steady state output that correspond to maximal complexity (Lipsitz and Goldberger, 1992). Any deviation from healthy state, that may be induced by senescence or disease, causes a loss in complexity. Similarly to stochastic resonance, this effect has been observed in a very broad panel of contexts. In addition, a lack of practice results in high disorder (randomness or no predictability) and excessive practice leads to high order (periodic signal or maximal predictability). A system has the propensity to adapt best to external disturbances at an intermediate state of predictability. The human body behaves as a non-linear dynamical system and exhibit a complex structure associated to an infinite repertoire of behaviors at different time scales. A decrease of complexity of a system results from either a reduction in the number of structural components or an alteration in the coupling function between these components. For instance, the upper limb can become rigid with senescence, hence suppressing degrees of freedom and, consequently, reducing its complexity. A holistic approach to study these mechanisms requires to associate specific measurements to these two concepts. The Hurst exponent is well suited to reflect predictability and the Minkowski fractal dimension provides good measurability of the "apparent rugosity" of fractals and reflects complexity. These two indexes are correlated and converge to the same value if measurement time is set to infinity. However, they yield complementary information in more realistic contexts such as in gait analysis (Dierick et al., 2017).

The measure of performance should not be restricted to signal detection or some SNR. Instead, it is more sensible to asses variations in functions with changing noise level. Stochastic resonance would then predict that, for some non-zero noise, the function will work optimally. Similarly, optimal movement variability would predict that for some complexity/predictability, the function will be optimal as well. Theoretical questions remain open as to whether these two optimum are the same and how stochastic resonance and optimal movement variability relate to each other, possibly in the context of system stability (Boulanger et al., 2018).

Successful stochastic resonance studies mentioned in previous sections underline the potential to develop field prototypes. McDonnell and Abbott point out that ". . . if it can be established that SR plays an important role in the encoding and processing of information in the brain, and that it somehow provides part of the brain's superior performance to computers and artificial intelligence in some areas, then using this knowledge in engineering systems may revolutionize the way we design computers, sensors, and communications systems." (McDonnell and Abbott, 2009). A compromise between technological factors (e.g., footprint, power consumption), device versatility and end-user acceptability must be sought. A few projects dedicated to improving balance specifically or use vibrations with stochastic resonance in an attempt to improve some function have been conducted. For instance, two emblematic projects have demonstrated the use of mechanical vibrations to improve balance (Balancing Act) and writing (ARC Pen). Subjects with retinal disorder and impaired vision also showed enhancement of letter recognition at the same level as the one observed in normal sighted individuals when tested with visual prosthesis using stochastic resonance (Itzcovich et al., 2017). Nevertheless, going from a prototype to a clinically validated product is a long-lasting story that demands involvement of patients and increasing awareness about the safety and efficiency of the developed technology. Furthermore, in a world of connected smart technologies, stimulation delivery schedules should be dynamically optimized in function of well-defined behavioral factors and provide regular meaningful feedback reports to the patient. This last point is central as it ensures that the patient remains convinced of the usefulness of the technology.

# AUTHOR CONTRIBUTIONS

All authors contributed to writing of this manuscript and conceptualized the review. OW conceived and designed the manuscript, conducted the literature search, interpreted the data for the work, and drafted the manuscript. JB and CT revised the work critically for important intellectual content. LJ and NG contributed to designing of the manuscript, interpreted the data and helped in editing the written manuscript.

# REFERENCES

fphys-09-01865 January 24, 2019 Time: 19:21 # 9


# FUNDING

Funds for open access publication were received from the Medical University of Graz.



stochastic resonance. J. Neurosci. 32, 12612–12618. doi: 10.1523/JNEUROSCI. 0680-12.2012



neurodegenerative disorders. Ann. Neurol. 58, 175–181. doi: 10.1002/ana. 20574


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 White, Babiˇc, Trenado, Johannsen and Goswami. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Galanin and Adrenomedullin Plasma Responses During Artificial Gravity on a Human Short-Arm Centrifuge

Julia Winter<sup>1</sup> , Charles Laing1,2, Bernd Johannes<sup>1</sup> , Edwin Mulder<sup>1</sup> , Bianca Brix<sup>3</sup> , Andreas Roessler<sup>3</sup> , Johannes Reichmuth<sup>3</sup> , Joern Rittweger1,4 \* and Nandu Goswami<sup>3</sup> \*

<sup>1</sup> Department of Aerospace Physiology, Institute for Aerospace Medicine, German Aerospace Center e.V. (DLR), Cologne, Germany, <sup>2</sup> Centre for Human and Aerospace Physiological Sciences, King's College London, London, United Kingdom, <sup>3</sup> Gravitational Physiology and Medical Research Unit, Physiology Division, Otto Loewi Center for Research in Vascular Biology, Immunity, and Inflammation, Medical University of Graz, Graz, Austria, <sup>4</sup> Department of Pediatrics and Adolescent Medicine, University of Cologne, Cologne, Germany

# Galanin and adrenomedullin plasma responses to head-up tilt and lower body negative pressure have been studied previously. However, to what extent short-arm human centrifugation (SAHC) affects these responses is not known. In this study, we assessed how the application of variable gradients of accelerations (1Gz) via shifting of the rotation axis during centrifugation affects selected hormonal responses. Specifically, we tested the hypothesis, that cardiovascular modulating hormones such as galanin and adrenomedullin will be higher in non-finishers (participants in whom at least one of the pre-defined criteria for presyncope was fulfilled) when compared to finishers (participants who completed the entire protocol in both sessions) during SAHC exposure. Twenty healthy subjects (10 women and 10 men) were exposed to two g-levels [1 G<sup>z</sup> and 2.4 G<sup>z</sup> at the feet (Gz\_Feet)] in two positions (axis of rotation placed above the head and axis of rotation placed at the heart level). Elevated baseline levels of galanin appeared to predict orthostatic tolerance (p = 0.054) and seemed to support good orthostatic tolerance during 1 Gz\_Feet SAHC (p = 0.034). In finishers, 2.4 Gz\_Feet SAHC was associated with increased galanin levels after centrifugation (p = 0.007). For adrenomedullin, the hypothesized increases were observed after centrifugation at 1 Gz\_Feet (p = 0.031), but not at 2.4 Gz\_Feet, suggesting that other central mechanisms than local distribution of adrenomedullin predominate when coping with central hypovolemia induced by SAHC (p > 0.14). In conclusion, baseline galanin levels could potentially be used to predict development of presyncope in subjects. Furthermore, galanin levels increase during elevated levels of central hypovolemia and galanin responses appear to be important for coping with such challenges. Adrenomedullin release depends on degree of central hypovolemia induced fluid shifts and a subject's ability to cope with such challenges. Our results suggest that the gradient of acceleration (1Gz) is an innovative approach to quantify the grade of central hypovolemia and to assess neurohormonal responses in those that can tolerate (finishers) or not tolerate (non-finishers) artificial gravity (AG). As AG is being considered as a preventing tool for spaceflight induced deconditioning in future missions, understanding effects of AG on hormonal responses in subjects who develop presyncope is important.

Keywords: syncope, orthostatic intolerance, gravity gradient, post-flight orthostatic intolerance, human spaceflight

### Edited by:

Ovidiu Constantin Baltatu, Anhembi Morumbi University, Brazil

## Reviewed by:

Joyce McClendon Evans, University of Kentucky, United States Satoshi Iwase, Aichi Medical University, Japan

#### \*Correspondence:

Joern Rittweger joern.rittweger@dlr.de Nandu Goswami nandu.goswami@medunigraz.at

#### Specialty section:

This article was submitted to Integrative Physiology, a section of the journal Frontiers in Physiology

Received: 16 July 2018 Accepted: 22 December 2018 Published: 01 February 2019

#### Citation:

Winter J, Laing C, Johannes B, Mulder E, Brix B, Roessler A, Reichmuth J, Rittweger J and Goswami N (2019) Galanin and Adrenomedullin Plasma Responses During Artificial Gravity on a Human Short-Arm Centrifuge. Front. Physiol. 9:1956. doi: 10.3389/fphys.2018.01956

**355**

# INTRODUCTION

fphys-09-01956 January 30, 2019 Time: 17:47 # 2

Central hypovolemia leads to changes in hemodynamic variables as well as in vasoactive endocrine hormones. These include the classical volume regulating hormones such as renin, angiotensin, aldosterone (RAAS), atrial natriuretic peptide (ANP) and vasopressin. Common stimuli for exploring the effect of central hypovolemia are HUT, LBNP (Goswami et al., 2008, 2019) or HUT in combination with LBNP (Bondanelli et al., 2003; Hinghofer-Szalkay et al., 2011; Goswami et al., 2012; O'shea et al., 2015). Specifically, a rapid elevation of plasma adrenaline and noradrenaline and, after a 10- to 20-min delay, renin-angiotensin system activation, leading to elevated plasma renin activity, angiotensin II and aldosterone is seen (degli Uberti et al., 1996; Convertino et al., 1998; Rossler et al., 1999; Hinghofer-Szalkay et al., 2006, 2011). Arginine vasopressin (AVP), synthesized in the hypothalamus and stored in the posterior pituitary gland, is released in response to hypotension and hypovolemia with some delay (Hirsch et al., 1993; Bichet, 2016). Its main functions include control of water body homeostasis via increases in water reabsorption in the kidney thus increasing the blood pressure (Chopra et al., 2011; Bichet, 2016). Unsurprisingly, AVP plays a key role in the development of orthostatic intolerance: Levels of plasma AVP increase several fold when the endpoint of cardiovascular stability (presyncope) is reached (Hinghofer-Szalkay et al., 2011). For details of the responses RAAS and vasopressin during artificial gravity (AG) application the reader is referred to Yang et al. (2011).

Since the last 20 years, research has now extended to include other hormones that are altered during orthostatic loading and/or central hypovolemia. These include galanin and adrenomedullin (degli Uberti et al., 1996; Bondanelli et al., 2003; Hinghofer-Szalkay et al., 2006, 2011; O'shea et al., 2015). Galanin is a peptide hormone released by the neurohypophysis. Evidence suggests that plasma galanin is increased during orthostatic challenge thus highlighting it's potential as a marker of presyncope (Hinghofer-Szalkay et al., 2006, 2011; O'shea et al., 2015). Galanin responses have been proposed as reflectors of sympathetic drive (Hinghofer-Szalkay et al., 2006) and appear closely related to heart rate increases (degli Uberti et al., 1996; Bondanelli et al., 2003). Adrenomedullin was first detected in the adrenal gland in human pheochromocytoma but it also seems to originate in endothelial cells (Kitamura et al., 1993; Nishikimi et al., 2001). Acting via paracrine pathways, adrenomedullin exerts rapid, strong and long-lasting vasodilatory effect by increasing NO-production in endothelial cells (Kitamura et al., 1993; Eto, 2001; Ueda et al., 2005). Other studies have reported that changes in plasma adrenomedullin during orthostatic challenge commensurate with the duration and intensity of the central hypovolemia (Rossler et al., 1999; Hinghofer-Szalkay et al., 2011; O'shea et al., 2015). Due to its vasodilatory effects, adrenomedullin is believed to play an important role in the development of presyncope: adrenomedullin increases have been associated with decreases in orthostatic tolerance times (Gajek et al., 2004).

# Short-Arm Human Centrifugation (SAHC): Effects and Benefits

Short-arm human centrifugation (SAHC) offers a novel approach to explore the effect of volume shift upon hormonal control of the cardiovascular system. The gradient acceleration field in SAHC provides a hydrostatic pressure profile that increases quadratically with the distance from the center of rotation, which contrasts with the linear profile in constant 1g-fields. Furthermore, in our study the placement of axis of rotation results in two different pressures of fluid shift to lower extremities and two different grades of central hypovolemia. SAHC is believed to cause central hypovolemia by passive blood shifting toward the lower limbs leading to reduced venous return, cardiac output, and inadequate cerebral perfusion with oxygenated blood (Arthur and Kaye, 2000), which could lead to the development of a presyncopal situation. Presyncope is characterized by symptoms like dizziness, light-headedness, sweating, and results ultimately in transient loss of consciousness (Goswami et al., 2015; Shen et al., 2017).

Short-arm human centrifugation induced centrifugal acceleration, which is imparted in the head-to-toe body axis (z-axis), has been proposed as an effective countermeasure against detrimental effects of microgravity (Clément and Bukley, 2007; Clement et al., 2016). Evidence suggests, that exposure to centrifugation, could improve orthostatic tolerance (Goswami et al., 2015). As post-flight orthostatic intolerance occurs commonly when astronauts return from space (Blaber et al., 2011; Goswami et al., 2015; Verma et al., 2016), artificial gravity (AG) has been proposed as a countermeasure against post-spaceflight orthostatic intolerance.

Since RAAS and vasopressin responses during AG application have been previously reported (see Yang et al., 2011), in this study the endocrinological response of galanin and adrenomedullin during varying grades of central hypovolemia induced by SAHC were studied. Specifically, short-term hormonal concentrations to SAHC by application of two different centers of rotation and two different g-levels were investigated in a crossover design in healthy women and men. We tested the hypothesis, that cardiovascular modulating hormones such as galanin and adrenomedullin were higher in finishers (participants who completed the entire protocol in both sessions) when compared to non-finishers (participants in whom at least one of the pre-defined criteria for presyncope was fulfilled) during SHAC exposure.

# MATERIALS AND METHODS

# Design and Setting

The study was conducted in autumn 2014 at the Institute for Aerospace Medicine of the German Center for Aerospace Medicine (DLR) in Cologne, Germany. The protocol (Lfd Nr. 2014123) was approved by the ethics committee of the Aerztekammer Nordrhein (Northern Rhine Medical Council), Duesseldorf, Germany and conformed to the Declaration of

**Abbreviations:** HUT, head-up tilting; LBNP, lower body negative pressure; SAHC, short-arm human centrifugation.

Helsinki. Subjects gave their written informed consent before inclusion into the study.

# Subjects

Twenty healthy normotensive female and male subjects participated in the study. Subjects were excluded from study participation if they had practiced drug abuse or, were smokers. Further exclusion criteria were arterial hypertension, diagnosed diabetes mellitus, any muscle or joint disease, herniated disk, chronic back pain, history of epileptic seizures or heart disease or possession of a cardiac pacemaker, or who were pregnant or had histories of orthostatic intolerance.

# Experimental Protocol

fphys-09-01956 January 30, 2019 Time: 17:47 # 3

The protocol consisted of two visits with five passive spins of centrifugation of 10 min duration during each visit. An 8-week wash-out-period was incorporated between the two visits to avoid training effects. Subjects were centrifuged lying on their back, and with their head orientated toward the axis of rotation.

After being positioned on the centrifuge, subjects were secured by a harness and instrumented for monitoring of continuous physiological variables (including ECG, pulse oximetry, and beat-to-beat finger blood pressure control). Ten minutes before centrifugation commenced, data recording started for baseline assessment (supine resting period on the centrifuge nacelle), and contiguous centrifugation spins were separated by a 25-min break (including ramp-ups and ramp-downs and position changes). After the second centrifuge spin, there was a 25-min break, during which toilet usage was allowed, and standardized muesli bar and clear water (5 ml per kilogram weight) were provided. After another 10 min of baseline assessment, the subsequent spins took place. At the end, a 20-min recovery phase was included (**Figure 1A**).

A randomized schedule was used for the applied gradients (1 G<sup>z</sup> or 2.4 G<sup>z</sup> at feet (Gz\_Feet), center of rotation placed above the head or at the heart level). A g-level of 2.4 Gz\_Feet was the greatest possible for the shortest subject (at 38 rpm). Anthropometric parameters of each subject were used to determine the two positions of axis of rotation. Measured specifically was the distance between apex of the heart and vertex of the head, as this distance was used to determine the length of radius that was used during centrifugation in each subject. When the axis of rotation was located above the head position, the distance between axis of rotation and vertex of the head was equal to the distance between the vertex of the head and apex of the heart. The other rotational axis used was located at the apex of the heart (**Figure 1B**).

Presyncope was defined as a simultaneous blood pressure and heart rate drop for at least 5 seconds with impaired vision, lightheadedness or feeling of faintness experienced by the subject [following clinical criteria of the "Guidelines for the diagnosis and management of syncope" of the European Society of Cardiology (version of 2017), Shen et al., 2017]. In case of development of presyncopal signs and symptoms, the centrifugation protocol was immediately terminated by the medical doctor. An occurrence of presyncope during any spin of centrifugation defined a subject as belonging to the group of non-finishers, independent of the visit or spin of centrifugation during which the presyncope occurred. Nevertheless, the subject could continue with the centrifuge protocol. In case of a further presyncope in another spin during the same visit, the procedure of centrifugation was stopped, because no more than two presyncopal events per subject and visit were allowed in this study (Laing et al., 2016).

# Effect of Centrifugation

Centrifugal acceleration (Gz) is a function of angular velocity (ω) and radius (r): 1G<sup>z</sup> = ω <sup>2</sup> ∗ r. Thus, by manipulation of either of the two variables, at least theoretically, any g-level can be achieved. The gradient of acceleration can be conceptualized as the 1Gz-difference between representative points along the z-axis of the body, e.g., the head and the foot (**Figure 1B**). Moreover, the gradient 1G<sup>z</sup> depends on both ω and r, thus implying that varying 1G<sup>z</sup> will lead to different fluid pressure distributions in the body. In other words, to be meaningful the G<sup>z</sup> challenges during centrifugation need.

# Blood Sample Protocol

Venous blood was taken from an 18-Gauge cannula placed in the left cubital region. Blood was taken twice during baseline, immediately after first centrifugation and immediately before second centrifugation (i.e., 19 min after stopping the centrifuge). Furthermore, blood was taken three times (immediately, 9 and 19 min after stopping) following the last spin of centrifugation. No blood was sampled in between single spins (**Figure 1A**). In case of an aborted centrifugation due to presyncope development, additional three blood samples were taken: immediately after the abort and at 9 and 19 min after aborting the centrifugation. Venous blood samples were taken in 9 ml EDTA monovettes and immediately transferred into prechilled (4◦C) aprotinin-EDTA tubes. The samples were cooled on crushed water ice until further processing. After spinning the aprotinin-EDTA blood samples in a 4◦C temperaturecontrolled blood centrifuge (Heraeus Multifuge 1 S-R centrifuge) for 10 min at 3000 rpm, blood plasma was distributed in 0.5 ml portions in Eppendorf tubes and stored in a −80◦C fridge until further processing at the Gravitational Physiology and Medicine Research unit, Medical University of Graz, Austria.

# Hormonal Analyses

After study completion, plasma galanin was measured using a commercially available radioimmunoassay (RIA) kit (Peninsula Laboratories International Inc., Belmont, CA, United States), after trifluoroacetic acid extraction. The eluate was stored at −80◦C until the day assay was carried out. This RIA has an advantage that it does not cross-react with similar peptides, including insulin, secretin, substance P, or vasoactive intestinal peptide (VIP). The minimum detectable substance is 1.5 pmol l−<sup>1</sup> (O'shea et al., 2015). The Shionogi Inc. (Osaka, Japan) immunoradiometric assay (IRMA) was used to measure plasma adrenomedullin concentrations of mature-type adrenomedullin (m-AM). Adrenomedullin is present in the plasma as biologically active, mature-type (amidated at the

carboxy terminus), and a biologically intermediate type, which prior enzymatic amination is glycine-extended (O'shea et al., 2015). Using the one-step, two side IRMA process, this essay is able to detect m-AM quickly and reliably without cross-reacting with the intermediate type or similar peptides. The minimum detectable concentration of the substance is 0.5 pmol l−<sup>1</sup> (O'shea et al., 2015).

# Data Processing and Statistical Analyses

Data and statistical analyses were performed with SPSS Statistics 21. Linear mixed effect (LME) models fitted by Restricted Maximum Likelihood estimation (REML) with gender, medical doctor, Gz\_Feet-load and position as fixed effects and subject ID as random effect were constructed in order to assess position, and Gz-load and gender effects. The dependent variables were plasma

concentration of adrenomedullin or galanin. Data are given as means and standard deviation (SD). The level of statistical significance was set to α = 0.05.

# RESULTS

All subjects completed the study. Of the twenty subjects, nine subjects could complete the entire protocol in both sessions (finishers), whilst centrifugation had to be interrupted in eleven subjects, because of at least one of the pre-defined criteria for presyncope was fulfilled (non-finishers, **Table 1**). Nonfinishers included five female and six male subjects; in five subjects, two presyncopal episodes occurred per visit (two times in female and three times in male subjects). Thirteen presyncopal events occurred after centrifugation with axis of rotation above the head whereas no presyncopal events appeared with axis of rotation positioned at the heart level. The medical doctor's influence, which determined the subjects as non-finishers or finishers, was not revealed as significant (p > 0.05).

## Galanin Response to Centrifugation

Finishers tended to have slightly greater galanin plasma concentrations than non-finishers at baseline (p= 0.054,**Figure 2**), but absolute galanin level changes were comparable between finishers and non-finishers following 1 Gz\_Feet centrifugation. In both groups, galaninlevelswerelower after1Gz\_Feet centrifugation with rotation axis placed at the heart compared with rotational axis placed above the head (p = 0.034, **Figure 3A**), suggesting an effect of gradient 1Gz. Indeed, galanin concentrations after rotation above the head with 1 Gz\_Feet were almost comparable with baseline values in both groups.

Following exposure to centrifugation with 2.4 Gz\_Feet, galanin plasma concentration in finishers was significantly influenced by the applied gradient 1G<sup>z</sup> (p = 0.007, **Figure 3B**). As with 1 Gz\_Feet centrifugation, highest galanin levels were measured during 2.4 Gz\_Feet centrifugation with axis of rotation positioned above the head in comparison with axis of rotation placed at the heart level.

# Adrenomedullin Response to Centrifugation

Baseline adrenomedullin plasma concentration were not different between finishers and non-finishers (finishers 5.5 ± 3.0 pg/ml, non-finishers 5.7 ± 2.7 pg/ml, p = 0.8). After 1 Gz\_Feet centrifugation, a significant interaction between development of presyncopal symptoms and position of centrifugation was found

(p = 0.031, **Figure 4**). Finishers had greatest adrenomedullin levels after centrifugation with axis of rotation located at the heart level, and non-finishers had greatest levels after centrifugation with axis of rotation placed above the head.

After exposure to centrifugation with 2.4 Gz\_Feet, neither the applied position nor the applied Gz-level led to any differences in adrenomedullin levels (p ≥ 0.14).

## Effect of Gender on Hormonal Response

Galanin and adrenomedullin data showed no gender effect at baseline (p > 0.079 in all cases), neither after 1 Gz\_Feet centrifugation with axis of rotation placed above the head nor at the heart level (p > 0.81 in all cases) nor after centrifugation with 2.4 Gz\_Feet (p > 0.362 in all cases).

# DISCUSSION

Short-arm human centrifugation with various locations of rotational axis represents a new method for studying the impact upon fluid shift to lower extremities and the accompanying endocrinological responses. As renin, aldosterone

TABLE 1 | Subject's anthropometry data, mean ± SD.


The table shows the measured anthropometry data of finisher and non-finisher subjects, separated by gender. BMI, body mass index; SD, standard deviation.

(p = 0.007) in only the finishers. AOR, axis of rotation.

and vasopressin responses during AG application have been previously studied (Yang et al., 2011), in this study we assessed additional hormones that are altered during orthostatic loading and central hypovolemia.

# Galanin Changes During Centrifugation and at Presyncope

baseline. AOR, axis of rotation.

Elevated baseline levels of galanin appeared to predict orthostatic tolerance (p = 0.054) and seemed to support good orthostatic tolerance during 1 Gz\_Feet SAHC (p = 0.034). Firstly, finishers showed higher galanin baseline levels as compared to non-finishers. In former studies, galanin was suggested as a stabilizer of cardiovascular responses during acute orthostatic challenge in HUT (Bondanelli et al., 2003). Although the pre-determined level of statistical significance of α = 0.05 was not reached in this current study, our results for galanin baseline concentration expand earlier findings. Our data suggest that galanin levels at baseline/during resting state and during SAHC application can predict the development of presyncopal signs and symptoms in an individual. Therefore, galanin should be considered, in combination with other clinical measurements (e.g., tilt table testing, Schellong test) (Stewart et al., 2017) as a marker for predicting presyncope during central hypovolemia induced by SAHC.

Secondly, we observed an influence by SAHC on galanin plasma concentrations. Existing literature is inconclusive regarding the role of galanin in orthostatic intolerance: researchers have reported increased galanin levels at the point of presyncope and a high galanin level association with orthostatic intolerance (Hinghofer-Szalkay et al., 2011; O'shea et al., 2015) while others have reported that in patients with HUT-induced syncope galanin does not change either preceding – or during – loss of consciousness (Bondanelli et al., 2003). The data reported from Bondanelli et al. (2003) suggest that endogenous galanin plays a key role in the adaptive responses to acute orthostatic challenge and prevents syncope in susceptible persons. In agreement with previously published studies, during central

hypovolemia created by 1 Gz\_Feet SAHC, increased plasma levels of galanin were seen.

Thirdly, galanin levels in finishers after centrifugation at 2.4 Gz\_Feet were significantly increased in comparison to baseline and 1 Gz\_Feet centrifugation values. Independent of the used position, SAHC with such high g-levels represents a high stress level for the cardiovascular system. Our galanin results after 2.4 Gz\_Feet may reflect a high level of sympathetic activity, which is in accordance with galanin's role in preventing presyncope, which was reported in previous studies (Bondanelli et al., 2003). In comparison to finishers, in non-finishers galanin levels were not increased during SAHC; this in turn, could have contributed to their development of presyncope.

Our results appear to disclaim our hypothesis related to plasma galanin levels before SAHC application, after 1 Gz\_Feet and after 2.4 Gz\_Feet SAHC. In contrast to former studies, our results are indicative of galanin's stabilizing effect for coping with graded central hypovolemia when compared to non-finishers during SAHC exposure.

# Adrenomedullin Changes During Centrifugation and at Presyncope

No significant differences in adrenomedullin were seen at baseline between finishers and non-finishers. It has been reported that adrenomedullin acts as an effective paracrine vasodilator showing a specific dose-dependent releasing during orthostatic stimulus (Kitamura et al., 1993; Rossler et al., 1999; Eto, 2001; Bondanelli et al., 2003). Our results are in accordance with previous studies, which reported no difference before application of the orthostatic stimulus between finishers and non-finishers (Kitamura et al., 1993; Gajek et al., 2004).

Furthermore, we observed a significant interaction in adrenomedullin concentration changes between finishers and non-finishers after 1 Gz\_Feet centrifugation dependent on the applied position. Finishing the protocol without developing syncope was related to stable adrenomedullin plasma levels after centrifugation with axis of rotation placed above the head in comparison with increased adrenomedullin levels in nonfinishers at the same position. Our results are in accordance with recent studies, which reported that orthostatic intolerance (assessed using HUT + graded LBNP until presyncope) following 21-day of bedrest was associated with increased adrenomedullin levels (O'shea et al., 2015). Inversely, in nonfinishers, adrenomedullin plasma levels were comparable to baseline levels and increased in orthostatic tolerant subjects after centrifugation with axis of rotation placed at heart level. Our results suggest that adrenomedullin plasma concentrations after SAHC are dependent on the degree of central hypovolemia and ability to cope with it. Further, adrenomedullin plasma changes after SAHC may not be always comparable to those obtained using other orthostatic loading stressors (Gajek et al., 2004; O'shea et al., 2015). Moreover, our findings of adrenomedullin concentrations after exposure to centrifugation at 2.4 Gz\_Feet suggest that it might have no relevance to coping responses during such exposures. Based on previous evidence, we speculate that factors other than local distribution of adrenomedullin may be involved in coping with central hypovolemia caused by 2.4 Gz\_Feet SAHC (Nishikimi et al., 2001; Gajek et al., 2004; Hinghofer-Szalkay et al., 2011).

In summary, following 1g Gz\_Feet SAHC with axis of rotation placed above the head (that is, which induces a high degree of central hypovolemia) adrenomedullin levels increased in non-finishers. However, following SAHC with axis of rotation placed at the heart level (that is, which induces a lower central hypovolemia), adrenomedullin plasma concentrations in nonfinishers were reduced in comparison to finishers. Following 2.4g Gz\_Feet SAHC, adrenomedullin plasma levels showed no significant different between non-finishers and finishers. The adrenomedullin data obtained in our study only partially support our proposed hypothesis.

# Gender Differences in Hormonal Responses

Interestingly, neither the number of presyncopal episodes nor the hormonal concentrations of the subjects during our centrifuge protocol showed any statistically significant gender difference. This contrasts with a former study, which suggested gender differences in hormonal concentrations during presyncope in galanin (Hinghofer-Szalkay et al., 2006). The differences in the results might be explained by the stress of SAHC itself, which represents a different orthostatic stimulus in comparison to HUT and LBNP used in the other study. Due to limited data available in gender specific responses to central hypovolemia (Evans et al., 2018), future studies should focus on gender differences (Clement et al., 2016).

# Limitations

As the time required to complete a single centrifuge run was long, the subjects were not spun at the same time of the day. Therefore, effects of circadian rhythms on the both of these hormones could not be accounted for. However, studies that have previously examined adrenomedullin responses have not reported any effects of circadian rhythms on adrenomedullin levels (Nishikimi et al., 2001; Nishikimi, 2005; Clement et al., 2016). Similarly, we could not find any study that examined the effects of circadian rhythms on galanin.

Furthermore, the total sample size (n = 20) is probably too small to detect subtle gender differences. However, this study is the first to report involvement of at least one the two under-studied hormones, namely galanin and adrenomedullin, during graded central hypovolemia induced by G<sup>z</sup> loading. The hormonal data obtained in this pilot study can, therefore, serve as a basis for sample size calculations in future epidemiological studies.

# CONCLUSION AND FURTHER DIRECTIONS

We assessed whether galanin and adrenomedullin plasma concentration changes during central hypovolemia observed in other studies are transferable to SAHC, especially with varying gravitational loading and varying positions of axis of rotation of the centrifuge. We observed that galanin plasma levels are not only important during orthostatic challenge but could

also predict development of presyncope in subjects undergoing centrifugation. Therefore, we recommend in future studies the measurement of baseline galanin levels – in combination with other clinical tools (e.g., questionnaire, tilt-test or Schellongtest) – to confirm galanin's ability to predict development of presyncope in subjects. Furthermore, as falls occur during changes in posture, especially in older persons (Goswami et al., 2018, 2019; Trozic et al., 2019), our data could have clinical relevance. Galanin and adrenomedullin could act as biomarkers for predicting which older persons can potentially develop orthostatic intolerance.

In addition, it appears that SAHC offers a novel approach to explore the effect of volume shift upon hormonal control of the cardiovascular system. It should also be noted that the gradient acceleration field in SAHC will engender a hydrostatic pressure profile that increases quadratically with the distance from the center of rotation, which contrasts with the linear profile in constant 1g-fields. Based on our novel results, the gradient of acceleration 1G<sup>z</sup> appears to be optimal for application of centrifugal acceleration.

# AUTHOR CONTRIBUTIONS

JW conducted the study, performed the statistical analysis and interpreted the data for the work, and drafted the work

# REFERENCES


and revised it critically for important intellectual content. CL and EM conceived and designed the study and conducted the study. BJ performed the statistical analysis. JoRe, AR, and BB drafted the work and revised it critically for important intellectual content. JR and NG conceived and designed the study, interpreted the data and helped in editing the written manuscript.

# FUNDING

This study was supported by DLR, Cologne and the Institute of Physiology, Medical University of Graz, Austria.

# ACKNOWLEDGMENTS

We would like to thank Mr. Andreas Jantscher, Institute of Physiology, Medical University of Graz, Austria for carrying out the hormonal measurements. We would also like to thank Mr. Ulrich Limper, Department of Aerospace Medicine, Institute for Aerospace Medicine, German Aerospace Center (DLR), Cologne, Germany for his persistent support of this manuscript.

hypotensive and vasodilating peptides. Peptides 22, 1693–1711. doi: 10.1016/ S0196-9781(01)00513-7



**Conflict of Interest Statement:** Parts of this manuscript were presented at the "1st Human Physiology Workshop" in Cologne, Germany, December 10, 2016 and at the "16. Tag der Forschung der Universität Duisburg-Essen" in Essen, Germany, November 17, 2017 by Julia Stroetges and during the "Aerospace Conference 2016" in Montana, United States, March 5–12, 2016 by CL (Laing et al., 2016).

The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

The reviewer JE declared a past co-authorship with one of the authors NG to the handling Editor.

Copyright © 2019 Winter, Laing, Johannes, Mulder, Brix, Roessler, Reichmuth, Rittweger and Goswami. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Stress Related Shift Toward Inflammaging in Cosmonauts After Long-Duration Space Flight

Judith-Irina Buchheim<sup>1</sup> , Sandra Matzel<sup>1</sup> , Marina Rykova<sup>2</sup> , Galina Vassilieva<sup>2</sup> , Sergey Ponomarev<sup>2</sup> , Igor Nichiporuk<sup>2</sup> , Marion Hörl<sup>1</sup> , Dominique Moser<sup>1</sup> , Katharina Biere<sup>1</sup> , Matthias Feuerecker<sup>1</sup> , Gustav Schelling<sup>1</sup> , Detlef Thieme<sup>3</sup> , Ines Kaufmann1,4, Manfred Thiel<sup>5</sup> and Alexander Choukèr<sup>1</sup> \*

<sup>1</sup> Laboratory of Translational Research "Stress and Immunity", Department of Anesthesiology, Hospital of the University of Munich, LMU, Munich, Germany, <sup>2</sup> Institute of Biomedical Problems, Russian Academy of Sciences, Moscow, Russia, 3 Institute of Doping Analysis and Sports Biochemistry, Dresden, Germany, <sup>4</sup> Department of Anesthesiology, Hospital Munich-Neuperlach, Munich, Germany, <sup>5</sup> Department of Anesthesiology and Surgical Intensive Care Medicine, University Medical Center Mannheim, Medical Faculty Mannheim, Heidelberg University, Mannheim, Germany

Edited by:

Andreas Roessler, Medical University of Graz, Austria

# Reviewed by:

Marli Cardoso Martins-Pinge, State University of Londrina, Brazil Shisan Bao, University of Sydney, Australia

> \*Correspondence: Alexander Choukèr alexander.chouker@ med.uni-muenchen.de

#### Specialty section:

This article was submitted to Integrative Physiology, a section of the journal Frontiers in Physiology

Received: 31 March 2018 Accepted: 24 January 2019 Published: 19 February 2019

#### Citation:

Buchheim J-I, Matzel S, Rykova M, Vassilieva G, Ponomarev S, Nichiporuk I, Hörl M, Moser D, Biere K, Feuerecker M, Schelling G, Thieme D, Kaufmann I, Thiel M and Choukèr A (2019) Stress Related Shift Toward Inflammaging in Cosmonauts After Long-Duration Space Flight. Front. Physiol. 10:85. doi: 10.3389/fphys.2019.00085 Space flight exerts a specific conglomerate of stressors on humans that can modulate the immune system. The mechanism remains to be elucidated and the consequences for cosmonauts in the long term are unclear. Most of the current research stems from short-term spaceflights as well as pre- and post-flight analyses due to operational limitations. Immune function of 12 cosmonauts participating in a long-duration (>140 days) spaceflight mission was monitored pre-, post-, and on two time-points in-flight. While the classical markers for stress such as cortisol in saliva where not significantly altered, blood concentrations of the endocannabinoid system (ECS) were found to be highly increased in-flight indicating a biological stress response. Moreover, subjects showed a significant rise in white blood cell counts. Neutrophils, monocytes and B cells increased by 50% whereas NK cells dropped by nearly 60% shortly after landing. Analysis of blood smears showed that lymphocyte percentages, though unchanged pre- and post-flight were elevated in-flight. Functional tests on the ground revealed stable cellular glutathione levels, unaltered baseline and stimulated ROS release in neutrophils but an increased shedding of L-selectin post-flight. In vitro stimulation of whole blood samples with fungal antigen showed a highly amplified TNF and IL-1β response. Furthermore, a significant reduction in CD4+CD25+CD27low regulatory T cells was observed post-flight but returned to normal levels after one month. Concomitantly, high in-flight levels of regulatory cytokines TGF-β, IL-10 and IL-1ra dropped rapidly after return to Earth. Finally, we observed a shift in the CD8<sup>+</sup> T cell repertoire toward CD8<sup>+</sup> memory cells that lasted even one month after return to Earth.

Conclusion: Long-duration spaceflight triggered a sustained stress dependent release of endocannabinoids combined with an aberrant immune activation mimicking features of people at risk for inflammation related diseases. These effects persisted in part 30 days after return to Earth. The currently available repertoire of in-flight testing

**364**

as well as the post-flight observation periods need to be expanded to tackle the underlying mechanism for and consequences of these immune changes in order to develop corresponding mitigation strategies based on a personalized approach for future interplanetary space explorations.

Keywords: space flight, anandamide, stress, regulatory T cells, CD8<sup>+</sup> memory T cells, inflammaging

# INTRODUCTION

Long-duration missions to the Moon and Mars are the next ultimate goals for human space exploration. One of the major concerns is the effect of extreme environmental conditions in space, such as microgravity, radiation, confinement, circadian rhythm disruption, physiological as well as psychological stress, on the human immune system. Increased susceptibility to infection in cosmonauts was already reported during the Apollo era, when a surprisingly high incidence of infectious diseases or inflammation related symptoms were reported on board or after spaceflight (Mermel, 2013). Many studies have investigated this issue and data from space analog research has helped greatly to understand single aspects of the multitude of stressors encountered in space (reviewed in Pagel and Chouker (2016)). Today, there is a growing body of evidence showing that the immune system undergoes a variety of changes during and after space travel such as shifts in leukocyte distribution (Crucian et al., 2008, 2013, 2015), monocyte and granulocyte function (Kaur et al., 2004, 2005), changes of cytokine release in plasma and in response to functional stimulation (Crucian et al., 2000, 2014). Most of the available data on the human immune system, however, is based on pre- and post-flight analyses due to procedural limitations or is based on short-term spaceflight missions (Gueguinou et al., 2009; Crucian et al., 2011). There is only limited knowledge of immune alterations that occur during a long-duration mission and the impact of these immune changes after re-adaptation to the conditions on Earth (Crucian et al., 2015). Chronic exposure to stress can exert profound alterations on the immune system. Classical stress hormones like cortisol driven by the hypothalamic-pituitary-adrenal (HPA) axis or the catecholamines norepinephrine and epinephrine have been shown to interfere with immune responses (Sorrells and Sapolsky, 2007). High cortisol levels, again, trigger other stress response systems such as the endocannabinoid system (ECS). Once activated, it helps the human body to adapt to stressful conditions (Hill et al., 2010). Short-term exposure to altered gravity, such as during a parabolic flight, induces the increased release of compounds of the ECS such as anandamide/Narachidonylethanolamine (AEA) and 2-arachidonoylglycerol (2- AG) in blood (Chouker et al., 2010). Recently, the capability of the ECS to alter immune function is studied intensively and it was shown that endocannabinoids (EC) can exert a strong suppression of innate and adaptive immune responses (Buchheim et al., 2018). The effects of long-term spaceflight on EC blood levels and their participation in the adaptation process aboard are unknown. Interestingly, a constant exposure to a stressful environment together with a constant exposure to antigens and an ongoing inflammatory response can result in a pathogenic phenotype showing features of premature immune aging, with increased cytokine responses and the accumulation of memory and effector T cells (Fagiolo et al., 1992; Franceschi et al., 2000; De Martinis et al., 2005). This so called "inflammaging" is a low-grade, pro-inflammatory state that renders its host at risk for latent viral infections, allergic or autoimmune disease as seen in elderly patients (Ravaglia et al., 2003; Pawelec and Gouttefangeas, 2006; Bauer and Fuente Mde, 2016; Sanada et al., 2018). There are indications that space travel might promote such an agerelated inflammatory phenotype prematurely as crew members spend a considerably long time in such an environment and viral shedding and skin exanthema occur frequently in flight (Mehta et al., 2014; Crucian et al., 2016).

As part of the European Space Agency (ESA) and the Russian space agency joint project IMMUNO, a systematic immune function analysis from both innate and adaptive immune responses in 12 cosmonauts was carried out before and after their long-duration space mission (increment duration > 140 days) to the International Space Station (ISS) to gain more insight into immune alterations during the stressful and unique conditions of space. The results from this long-duration spaceflight study increases our current understanding of the functional significance of the immune system changes which occur after long-duration space travels and provide insights in preparation of future exploration class missions.

# MATERIALS AND METHODS

# Study Design and Ethical Approval

This study was carried out in accordance with the recommendations from the ethical standards of the appropriate institutional and national committees and with the World Medical Association Helsinki Declaration of 1975 (revised in 2008). The protocol was evaluated by the Russian ethical board (Biomedicine Ethics Committee) at the Institute of Biomedical Problems (IBMP), Moscow, and was awarded the protocol number MBI-29 under the Title "IMMUNO." The protocol was also approved by the ESA medical board and the Russian Space Agency (Roscosmos). All subjects gave written informed consent in accordance with the Declaration of Helsinki. Russian space crew members employed by the Russian Space Agency (ROSCOSMOS) are referred to as "cosmonauts" whereas space travelers employed by NASA or ESA are referred to as "astronauts." In this manuscript, we will use the term cosmonauts throughout. A total of 12 male cosmonauts (median age 46 years (41/51)) participating in long-term spaceflight missions (median mission duration 162 days (142/181)) were recruited for the IMMUNO study. Sample collection occurred at a maximum of 6 time-points. Sample collection on ground was performed at the Gagarin Cosmonaut Training Centre (GCTC) near Moscow or the DLR training facility in Cologne, Germany. Baseline data collection (BDC) was scheduled 25 days before flight (L-25) and post-flight sampling occurred on day 1 (R+1), day 7 (R+7) and day 30 (R+30) after landing. Two inflight time-points were scheduled for flight day 90 (F+90) and flight day 150 (F+150).

# Stress Questionnaire

fphys-10-00085 February 15, 2019 Time: 17:48 # 3

The current stress test (CST) is a validated, standardized, selfestimating, one-page paper test that can be completed within 1 min (Chouker et al., 2001). The test is composed of six pairs of contradictory feelings of increasing intensities between which the subject must decide. From the sum of values obtained, the final CST score is calculated, which may range from 1 (no stress) to 6 (maximal stress).

# Cortisol in Saliva

Saliva samples were collected in the morning (7:00–8:00 AM) and in the evening (7:00–8:00 PM) before teeth brushing or food ingestion. Saliva was collected by chewing on a sterile cotton swab for 30–45 s (Salivettes <sup>R</sup> , Sarstedt, Germany) and samples were stored frozen on ground and inflight until further analysis. Free cortisol levels in saliva were quantified with an automated immunoassay system based on the principle of electro-chemiluminescence (Elecsys 2010, Roche, Mannheim, Germany) at the Institute of Clinical Chemistry, Hospital of the University of Munich, Germany.

# Blood Samples

Blood was drawn into blood tubes containing either EDTA or heparin as anticoagulants. The blood was aliquoted into different portions. Three different blood smears per time-point were performed according to standard procedures on ground and inflight. Fresh whole blood samples were transported to Germany on the same day for blood phenotype and immune function analysis while the EDTA anti-coagulated blood sample was centrifuged and immediately frozen on-site at –80◦C until further analysis. Aboard the ISS, blood samples were stored at – 80◦C until download to Earth. Transfer of samples from the landing site to the lab were performed on dry ice.

# Endocannabinoid Concentrations in Blood

Blood samples were quantified for the EC N-arachidonylethanolamine/anandamide (AEA) and 2 arachidonoylglycerol (2-AG) using high-performance liquid chromatography tandem mass spectrometry technique as published elsewhere (Vogeser and Spohrer, 2006; Chouker et al., 2010).

# Differential Blood Count

Absolute white blood cell count and the percentage of each white blood cell type were measured in whole blood samples (Institute of Clinical Chemistry, Hospital of the University of Munich, Germany).

# Surface Adhesion Marker

Polymorphonuclear leukocytes (PMNs) were extracted from whole blood samples through 40 min of separation in Histopaque <sup>R</sup> (Cat No. 10771, Sigma-Aldrich, Germany) and stained for adhesion molecules β2-integrin (CD18) and L-selectin (CD62L) (all antibodies from BD Heidelberg, Germany) as previously described (Thiel et al., 2001). Data were expressed as MFI.

# Determination of TNF/fMLP-Induced Production of Radical Oxygen Species (ROS)

Isolated PMNs (∼1 × 10<sup>5</sup> ) were incubated with dihydrorhodamine 123 (DHR123) (1 µM) (MoBiTec GmbH, Goettingen, Germany) in 0.5 mL HBSS buffer to detect the production of ROS as previously described (Kaufmann et al., 2012). After priming of the cells with TNF (10 ng/mL) for 5 min, PMNs were stimulated with fMLP (10−<sup>7</sup> M) for another 5 min. As a positive control, cells were stimulated with PMA. Activation of cells was terminated by putting the tubes in ice water. Production of ROS was determined by flow cytometry (BD, Heidelberg, Germany). DHR fluorescence (530 nm) was measured for gated PMN on a FACScan 9235 (Becton Dickinson Immocytometry Systems, Germany) and analyzed through Cell Quest Pro software (BD Biosciences, United States). Data were collected for 5000 events from each sample and are expressed as mean fluorescence intensity (MFI).

# Monocyte Subset Analysis

Monocyte subset analysis was performed using FACS Calibur flow cytometer (Becton Dickinson, United States) with monoclonal antibodies: anti-CD14-FITC, anti-CD14-PE, anti-CD11b-FITC, anti-CD18-PE, anti-CD54-FITC, anti-CD36- FITC, anti-CD16-PE, anti-CD206-PE (IO test, Beckman Coulter, France). 50 µl EDTA whole blood was stained with 5 µl of the relevant antibodies for 20 min at room temperature. In order to reduce Fc-receptor mediated binding of test antibodies, 2.5 µl Human Seroblock reagent (BIO-RAD, United States) were added for 10 min at room temperature. Red blood cells were lysed with 250 µl OptiLyse C Lysing Solution (Beckman Coulter, France) before analysis.

# Immunophenotyping

Blood cells were stained for flow cytometry with 50 µl EDTA whole blood and 5 µl of the relevant antibodies each for 15 min. Red blood cells were lysed with 450 µl BD Lysing Solution (BD 349202, Heidelberg, Germany) before being analyzed by BD Calibur flow cytometer. Live cells were gated based on forward and side scatter profiles. A comprehensive 4-color flow cytometry antibody matrix (BD MultitestTM, Heidelberg, Germany) was formulated for peripheral blood immunophenotype analysis. This panel assessed the major leukocyte and lymphocyte subsets, T cell subsets, memory/naïve and central memory T cell subsets, and constitutively activated T cell percentages. For regulatory T (Treg) cells, the following antibodies were used for staining: anti-human CD3 APC, CD4 PerCP, CD25 FITC, CD127 PE

(All from BD Biosciences, Heidelberg, Germany). Analysis was performed using BD Multiset and Cellquest Software. For detection of lymphocyte sub-populations, heparinized blood (50 µl) was incubated with antibodies against the cell-surface markers for 10 min at room temperature in the dark to differentiate CD3<sup>+</sup> (T-lymphocytes), CD3+CD4<sup>+</sup> (T-helper cells), CD3+/CD8<sup>+</sup> (T-suppressor cells/cytotoxic T-cells), CD3−CD19<sup>+</sup> (B-lymphocytes), CD16+CD57<sup>+</sup> (mature natural killer lymphocytes), CD3+CD8+CD45RA<sup>+</sup> and CD3+CD8+CD45RO<sup>+</sup> naïve/memory CD8+ T-cells, CD3+CD4+CD45RA<sup>+</sup> and CD3+CD4+CD45RO<sup>+</sup> naïve/memory CD4<sup>+</sup> T-cells. All antibodies (Multitest 3/8/45/4; Multitest 3/16/+56/45/19; anti-CD3-FITC; anti-CD8-FITC; anti-CD45RA-FITC; anti-CD57-FITC; anti-CD8-PE; anti-CD25-PE; anti-CD28-PE; anti-CD45RO-PE; anti-CD45-Percp; anti-CD3-APC; anti-CD4-APC; anti-CD8-APC) were obtained from BD Life Sciences, Germany. BD FACS lysing solution (BD Lyse; BD Life Sciences, Germany) was added to lyse RBC and incubated for 3 min at room temperature in the dark prior to flow cytometry analysis using a BD FACSCanto System (BD Biosciences; Germany).

# Intracellular Glutathione

Peripheral blood was collected in EDTA- or sodium heparincontaining tubes. PBMCs were isolated by Histopaque <sup>R</sup> density gradient centrifugation (Sigma-Aldrich, Germany) following the manufacturer's instructions and were washed once (centrifuged at 235 × g for 15 min) in PBS (Sigma-Aldrich, Germany). Pellets were re-suspended in RPMI-1640 medium, labeled with cell-specific, fluorescence-marked antibodies (CD4<sup>+</sup> or CD8<sup>+</sup> lymphocyte subsets were identified by binding of PerCP-conjugated monoclonal antibodies, all BD Heidelberg, Germany) and incubated with the dye Mercury-Orange (Sigma-Aldrich, Germany) for 5 min at room temperature. Intracellular GSH concentrations were determined by flow cytometry using BD FACSCalibur Flow Cytometer.

# Cytokine Release Assay

Heparinized whole blood (500 µl) was supplemented with equal amounts of RPMI Medium and different concentrations of each stimulating antigen (Fungal antigen mixture, Aspergillus fumigatus and pokeweed mitogen (PWM)) and incubated for 48 h at 37◦C with 5% CO2. The PWMstimulated plasma sample was used as a positive control for whole blood cytokine release. After each time point of incubation, the supernatants were collected and stored at –20◦C until further analysis. The supernatants were then analyzed simultaneously for 27 cytokines (Bio-Plex Pro Human Cytokine Assay; Bio-Rad Laboratories, Germany), including IL-1β, IL-1ra, IL-2, IL-6, IFN-γ, TNF in accordance with the manufacturer's instructions. The samples were analyzed using a Bio-Plex 200 instrument with Bio-Plex BioManager analysis software. The concentrations of cytokines were measured by comparing the bead color of the bound antibodies and the MFI from each set of beads against a verified standard curve.

# Statistics

Statistical evaluation and plotting of the data set was done using SPSS 24 (IBM, United States) and SigmaPlot 12 (Systat Software Inc., United States). Data was tested for normal distribution using the Kolmogorov-Smirnov Test. Box-Cox transformation of the data was performed only when residuals of the tested variable were found to be non-normally distributed. The data was then analyzed using a linear mixed effects (LME) model, where the time-points measured were regarded as fixed and subjects as random effects. A value of P < 0.05 was considered as statistically significant.

# RESULTS

# Analysis of Stress Through Self-Evaluation Questionnaire and Biochemical Markers

To investigate whether exposure to the special environment aboard the ISS is perceived as stressful by cosmonauts, we used subjective and objective test methods. The subjective stress level was evaluated in the morning and evening using the CST questionnaire. The mean calculated preflight scores were 2.01 (±0.66) in the morning and 2.39 (±0.63) in the evening (**Figure 1A**). Over the time-course of the mission, the values moderately increased but did not significantly differ between daytime or mission time-points, indicating a personal perception of low stress during the mission. Similar results were obtained when we measured cortisol in saliva samples. Circadian differences with slightly higher levels in the morning and lower levels in the evening were observed, however, no significant changes were detected over time. The evening values showed a slight increase of cortisol levels on F+90 but without statistical significance (**Figure 1B**). The blood concentrations of the endocannabinoids N-arachidonylethanolamine/anandamide (AEA) and 2-arachidonoylglycerol (2-AG) were measured as another objective stress parameter. AEA was found to be significantly increased at the second inflight time point (F+150) compared to ground (**Figure 1C**).

# Alterations of Peripheral Blood Leukocyte Phenotypes After Landing

Next, we wanted to evaluate the effect of long-term spaceflight on leukocyte populations. The total number of leukocytes (WBC) was significantly increased immediately after landing (R+1). Compared with baseline, neutrophil and monocyte numbers increased by 50% at this time-point. Absolute numbers of lymphocytes remained constant pre- and post-flight (**Figure 2A**). Due to procedural limitations, we were able to study leukocyte changes inflight only by analyzing blood smears generated percentages. The relative counts obtained showed an increase of neutrophils and monocytes post-flight, confirming the results obtained based on absolute numbers. When we analyzed the monocyte population further on ground, we found a significant decrease of CD14+CD16+on R+1. On R+7 this decrease was still observed but did not reach statistical significance

due to high inter-individual variations (data not shown). Interestingly, the relative lymphocyte population inflight was significantly increased and returned to normal numbers postflight (**Figure 2B**). When studying the lymphocyte subgroups after return in whole blood samples, an increase in B-cells was seen, whereas T cells showed no alteration. In particular, natural killer (NK) cells dropped by 60% post-flight (**Figure 2C**).

# Activation of Polymorphonuclear Leukocytes (PMN) After Spaceflight

Since neutrophil counts rose in relative and absolute terms on R+1, we turned to look at their state of activation and function. The capability to produce reactive oxygen species (ROS) was tested, which remained stable after spaceflight in response to a strong (PMA) or physiologic stimulus fMLP/TNF (**Figure 2D**). Similar results were obtained when we analyzed the amount of cellular antioxidant glutathione which was unaltered in all cell types over all time-points (data not shown.) Furthermore, the expression of adhesion molecules β2-integrin (CD18) and L-selectin (CD62L) on the surface of PMN was characterized. The expression of β2-integrin was already elevated preflight and remained so until 30 days post-flight without significant changes (**Figure 2E**). On the other hand, a notable decrease in cellsurface-bound L-selectin molecules was detected upon return (R+1), indicating an increased shedding of L-selectin and thus strong activation of PMNs (**Figure 2F**).

# Amplified TNF Response to an in vitro Exposure to Fungal Antigen

To analyze the overall cytokine immune response to an infection, we challenged whole blood samples with distinct antigens for 48 h. The production of pro-inflammatory cytokines, TNF, IL-2, IL-6, IFN-γ as well as IL-1β and IL-1ra, were monitored in culture supernatants in the presence of fungal antigen mixture or Aspergillus antigen (**Figure 3**). At R+1, the secretion of TNF was highly and significantly increased compared to all other time-points in response to fungal antigens, TNF production increased by 10-fold (**Figure 3A**). At time point R+7, we could still observe elevated levels of TNF, though it was not statistically significant. Similar results were obtained for Aspergillus antigen (**Figure 3B**). IL-1β release was also markedly increased after stimulation with fungal antigens. Comparing these results with the measured cytokine plasma concentrations, IL-1β was significantly increased in flight but returned to baseline concentrations at R+1 and could therefore not account for the higher amounts detected after stimulation. For IL-2, IL-6 and IFN-γ responses, no significant difference from baseline was detected. In non-challenged control samples, no alteration of TNF, IL-2 and IFN-γ levels were measurable throughout the study interval (data not shown).

# Regulatory T-Cell Subsets and Cytokine Profiles Display a Reduction of the Anti-inflammatory Capacity

Since we observed higher leukocyte numbers and a cytokine response overshoot following space flight, we looked further into the T cell population. Among the T cell subsets characterized, no significant difference was detected for CD4<sup>+</sup> memory cells and CD4<sup>+</sup> naïve cells (data not shown). Interestingly, antiinflammatory regulating T cells (Tregs) (CD4+CD25+CD127low) were significantly reduced by nearly 30% after spaceflight on R+7 (**Figure 4A**). Anti-inflammatory cytokine IL-1ra (**Figure 4B**) as well as regulatory cytokines IL-10 (**Figure 4C**) and TGFβ (**Figure 4D**) were found to be reduced after spaceflight but highly increased inflight, suggesting a pro-inflammatory immune status with a concomitant reduction in the antiinflammatory capacity.

# Repertoire Shift of CD8<sup>+</sup> T Cells

When looking at the CD8<sup>+</sup> T cell compartment, we found that CD8<sup>+</sup> naïve cells remained largely unaltered by spaceflight (**Figure 5A**), while CD8<sup>+</sup> effector T cells were found to be transiently diminished directly after return (**Figure 5B**) but reverted to normal values on day 7. Additionally, we observed an increase in the CD8<sup>+</sup> memory cell repertoire. This shift persisted until day 30 after return, the last time-point of the study, indicating a prolonged change in the CD8<sup>+</sup> T cell compartment. Values at R+7 did not reach statistical significance due to interindividual differences (**Figure 5C**).

# DISCUSSION

This study aimed to assess immune function in ISS crew members before and after long-duration spaceflight by evaluating stress levels and through performing a variety of analyses on both the innate and adaptive immune systems. Our study revealed a general activation of stress response systems as evidenced by high levels of AEA inflight. We detected a proinflammatory state characterized by an aberrant peripheral blood leukocyte distribution, elevated neutrophil activity and highly amplified TNF and IL-1β response against fungal antigen stimulation. Moreover, we observed a reduction in anti-inflammatory capacities since regulatory T cells, as well as regulatory cytokines IL-1ra, IL-10 and TGF-β, were found to be reduced post-flight. Finally, a shift in the CD8<sup>+</sup> cell repertoire with elevated CD8<sup>+</sup> memory T-cell counts that persisted 30 days post-flight argues for a prolonged alteration in adaptive immunity.

# Stress Response Systems in Chronic Stress

Stress can be regarded as the reaction of an organism to changes in the surrounding environment (Selye, 1956). An acute stress response, the so called "fight or flight" reaction, leads to the redistribution of blood flow to distinct organs in order to ensure survival in case of a life threatening event (Cannon, 1929). The hypothalamus directs the fast response (in seconds) via the sympathetic nerve system, where the medulla of the adrenal glands release catecholamines (epinephrine and norepinephrine), and concurrently a slower response via the HPA axis, resulting in a release of cortisol. In this study, we asked cosmonauts about their personal experiences of stress using the CST questionnaire

FIGURE 2 | Leukocyte phenotype after long-duration space flight. (A) Absolute count of leukocytes (WBC) and subsets pre and post-flight, data expressed as mean ± SEM, n = 11 <sup>∗</sup> = P < 0.05 vs. L-25 control (B) Relative counts (%) of neutrophils lymphocytes and monocytes evidenced by blood smear analysis, data expressed as mean ± SEM, n = 8 on F+90 and n = 10 on F+150 <sup>∗</sup> = P < 0.05 vs. L-25 control (C) Relative counts (% lymphocytes) of lymphocyte subsets pre and post-flight, Data expressed as mean ± SEM, n = 12 <sup>∗</sup> = P < 0.05 vs. L-25 control (D) Expression of adhesion molecule β2 integrin (CD18) on the surface of PMNs. Data is expressed as median ± 95th and 5th percentile (E) Expression of adhesion molecule L-selectin (CD62L) on the surface of PMNs, data is expressed as median ± 95th and 5th percentile, n = 12, <sup>∗</sup> = P < 0.05. (F) Production of H2O<sup>2</sup> measured after stimulation with PMA (positive control), with TNF-α/fMLP (physiologic stimulus) or without stimulus (Control), data expressed as mean ± SEM, n = 12.

expressed as median ± 95th and 5th percentile, n = 11,<sup>∗</sup> = P < 0.05 vs. L-25 control (C) Quantification of IL-1β after 48 h of stimulation with fungal antigen, data is expressed as median ± 95th and 5th percentile, n = 5 <sup>∗</sup> = P < 0.05 (D) Quantification of IL-1β in Plasma, data is expressed as median ± 95th and 5th percentile, n = 9 on F+90, n = 11 on F+150, <sup>∗</sup> = P < 0.05.

and they did not report acute strain. Cortisol levels were moderately affected and are in line with the CST results. The endocannabinoid AEA, however, was highly increased inflight, demonstrating a biological stress response to the environment aboard the ISS. The ECS is an important stress response system and has multiple roles in a myriad of physiological processes of stress (Chouker et al., 2010; Dlugos et al., 2012; Hauer et al., 2013; Neumeister et al., 2015), metabolism (Campolongo et al., 2009), sleep and activity patterns (Richard et al., 2009; Feuerecker et al., 2012). Most importantly, endocannabinoids are potent immune modulators (Sardinha et al., 2014; Buchheim et al., 2018). They act via the endocannabinoid receptors (CB) 1 and 2, which are not only expressed in the central and peripheral nervous systems (CB1), but also on the surface of many leukocytes (CB2). Our data suggest that AEA is a good measure for chronic stress and does not necessarily translate into psychological burden as seen in the low self-evaluation questionnaire (CST) scores. After an event causing acute stress such as landing on Earth (R+1), EC levels remained low. It appears that EC release is stimulated by conditions in space and not by the acute stress exerted during landing, which is well in line with the knowledge that acute and chronic stresses exert very different effects on the immune system (Moynihan, 2003). Correspondingly, our data show similarities to previously reported results, where the ECS was responsible for the modulation of constant baseline ROS release (without stimulus) in neutrophils and did not alter ROS production after stimulation with fMLP and TNF (Buchheim et al., 2018).

# Leukocyte Recruitment Triggered by an Acute Stressor

F+90, n = 10 on F+150; all panels: data is expressed as median ± 95th and 5th percentile, <sup>∗</sup> = P < 0.05 vs. L-25 control.

The immune system reacts to acute stress by releasing leukocytes (Meehan et al., 1993; Stowe et al., 2003; Dhabhar et al., 2012). Neutrophils belong to the first line of innate immune defense and are the first to arrive at sites of infection. Functional defects of neutrophils are characterized by poor wound healing and recurring infections. In this study, we observed neutrophil activation following spaceflight (R+1) characterized by a decrease in L-selectin CD62L expression on the cell surface of neutrophils. Neutrophils are tasked with the production of cytotoxic granules containing ROS and their release during oxidative burst (Bian et al., 2012; Liu et al., 2012; Kolaczkowska and Kubes, 2013). ROS release remained stable and antioxidative levels of glutathione were preserved under physiologic stimuli. It would be of great interest to examine ROS release under in-flight conditions together with assessing AEA levels, since the ECS plays a decisive role in controlling oxidative stress and thus in maintaining cellular homeostasis (Lipina and Hundal, 2016). Due to procedural limitations, this is currently not possible in space. One might speculate that the high levels of AEA could be responsible for the stable ROS production observed over the course of this study. Previous investigations following short-duration spaceflight also reported changes of neutrophil function demonstrated by enhanced chemotactic activity after landing, increased neutrophil adhesion to endothelial cells and significantly changed L-selectin expression (Stowe et al., 1999). Interestingly, L-selectin expression on neutrophils was significantly increased after short-duration spaceflight (Stowe et al., 1999), in contrast to our findings following long-duration spaceflight. This difference may result from the cumulative effects of long-duration mission related factors (i.e., microgravity, radiation, or re-adaptation to earth environment) on neutrophil activation. The observed changes in the peripheral blood

leukocyte distribution post-flight (R+1) are largely consistent with previous reports on immune changes following short-term flights (Crucian et al., 2008) and a pilot study after long-term spaceflight (Crucian et al., 2015). Directly following landing, we observed elevated leukocyte counts including both neutrophils and monocytes, as well as B-cells. This phenomenon may at least in part be triggered by the landing process, which induces dramatic acute physical strain on the human body owing to the combination of microgravity, hyper-gravity and fierce vibrations.

# Hypersensitivity Toward Fungi and Proinflammatory Phenotype

The percentual increase in lymphocytes in blood smears inflight indicate that the adaptive immune system was also activated in long missions and not only just a component of the innate immunity as previously reported (Crucian et al., 2015). Since fungal colonization of the ISS is a known burden (Novikova et al., 2006), it is particularly interesting to understand the interactions of stress, fungal antigen load aboard and the resulting immune response in the long term. When we challenged the immune system with fungal antigens upon return from long-term spaceflight, an overshoot of immune response with pro-inflammatory cytokines TNF and IL-1β which continued to be elevated 7 days after return to Earth. Moreover, IL-1β levels were significantly elevated in flight but dropped rapidly post-flight, indicating that the measured overshoot response was not biased simply due to higher plasma concentrations of the cytokine. Previously, we have shown that under ECS activation the fungal antigen response is abrogated (Buchheim et al., 2018). Here, we observe that plasma cytokine concentrations do not necessarily correlate with the cytokine response after antigen stimulation. Additionally, we observed a significant reduction in NK cells at R+1. NK cells are critical in the defense against fungal infections and conversely, fungi can exert immunosuppressive effects on NK cells (Schmidt et al., 2017). It was reported that NK

cell cytotoxicity was reduced after short-term shuttle missions without a drop in absolute numbers (Mehta et al., 2001).

In accordance with the boosted response to fungal antigens, we observed a reduction of Treg cells upon return to Earth, suggesting a compromised function in suppressing T cell proliferation and pro-inflammatory cytokine secretion. Many investigations have demonstrated that Treg cells are capable of down-regulating antigen-specific T-cell responses (Dejaco et al., 2006). The changes in anti-inflammatory plasma cytokine levels here mirror a Treg suppressed state. We also observed a significant increase in the regulatory cytokines IL-1ra and IL-10 inflight which dropped rapidly after return. The changes in concentrations of TGF-β were even more striking. This reflects the important role TGF-β plays in the function of Tregs and is thought to be critical for the maintenance of tolerance in the immune system (Ouyang et al., 2010; Kehrl et al., 2014). Concomitantly, we observed a shift of cells toward CD8<sup>+</sup> memory T cells, which persisted until the end of the study, whereas the CD4<sup>+</sup> T cell naïve and memory populations remained largely unaltered. It has been shown that CD8<sup>+</sup> T cells are more unstable and increase in the immune aging phenotype (Goronzy et al., 2007). This does not necessarily refer to chronological age but rather, is an expression of the imbalance of pro-and anti-inflammatory mechanisms (Ventura et al., 2017). Such alterations, should they persist, could lead to diseases associated with immune imbalance including chronic inflammation, autoimmune or other inflammation related diseases (Franceschi et al., 2007).

# Immune Response Shift in Chronic Stress Can Trigger Inflammaging

The effects of spaceflight on the immune system have two aspects: acute versus chronic stress and inflight versus post-flight. A strong acute stress reaction to a potentially life-threatening event such as the return to Earth serves as an immune activator that mobilizes immune cells and activates neutrophils regardless of the overall immune status. Chronic exposure to stress leads to an activation of similar mechanisms over a longer period of time. In the acute stress response, T cells are redistributed to areas of actions such as the blood stream or the skin, because skin damage is a likely event during a fight or flight situation. If, however, this alarm signal is maintained over a longer period of time, fewer T cells will respond to the recruitment leading to a reduced number of T cells in the skin, which has been shown in delayed type hypersensitivity (DTH) skin tests during spaceflight (Dhabhar and McEwen, 1997). Others have also noted a reduction in T cell proliferation and activity in the skin (Chang et al., 2012; Sonnenfeld, 2012). Similarly, delayed mucosal wound healing and a decrease in the key cytokine IL-1β can be observed during chronic stress (Marucha et al., 1998). A typical pattern in chronic stress is a shift to a T helper cell type 2 response through the decrease in proinflammatory cytokines (TNF, IL-1β, IL-6, INF-γ) and higher levels of anti-inflammatory cytokines (IL-10, TGF-β) (Marshall et al., 1998; Kang and Fox, 2001). These shifts were also observed in our study inflight, showing that cosmonauts were exposed to stress although they might not have perceived it as such. High levels of AEA inflight indicate a form of adaptation to this environment and the body's efforts in maintaining cellular homeostasis. If we had been able to conduct a stimulated antigen response also inflight, a reduction of cytokines after antigen stimulation would be expected (Buchheim et al., 2018). The technical limitations aboard the ISS currently prevent us from carrying out these tests in space. Prolonged exposure to this very specific environment does not only lead to immunological adaptation but triggers shifts in cellular compartments of the immune system. The data on Tregs and CD8<sup>+</sup> cells together with the functional tests suggest a special conglomerate of stressors inflight, triggering a hyperinflammatory and aging immune phenotype (inflammaging) that persists even after 1 month after the return to Earth. It will be valuable to prolong the post-flight observation period for long-term effects such as shifts in the T cell repertoire. This could identify crew personnel at risk for developing such diseases and encourage appropriate preventive measures as well as further monitoring.

In summary, the results of this study indicate long-duration spaceflight could trigger inflammaging, which may expose cosmonauts to risks for hypersensitivity diseases such as allergies or autoimmune diseases. Further on-board immune functional immune testing and a longer post-flight observation period are necessary to understand the underlying mechanism and the consequences for cosmonaut health in the long-term. It is then necessary to develop corresponding mitigation strategies based on a personalized approach for interplanetary space explorations.

# AUTHOR CONTRIBUTIONS

AC, MT, MF, IK, DT, and GS designed the study. SM, MR, GV, IN, and SP performed and supported the data sampling. SM, MH, KB, and DM performed the sample analysis. J-IB performed data analysis. J-IB and AC drafted the manuscript.

# FUNDING

This work was supported by the DLR on behalf of the Federal Ministry of Economics and Technology (BMWi 50WB0719, 50WB0919, and 50WB1319), the European Space Agency (ESA, ELIPS 3 and 4 program), the Russian Space Agency (Roscosmos) and the program of fundamental research (theme 65.1) of the Institute for Biomedical Problems (IBMP).

# ACKNOWLEDGMENTS

We explicitly thank all the helping hands, operators, scientists and administrators at Roscosmos, IBMP, TsNIIMash in Russia, at ESA, CNES and DLR as well as the NASA Kennedy Space Centre and the Johnson Space Centre, who made this project possible. Also, the support of Lufthansa German Airlines in transporting laboratory samples is acknowledged. We also thank Andrea Boltendahl, Camilla Ladinig and Iva Kumprey (former staff in

the laboratory of A. Choukèr, Ludwig Maximilian University) for their support. Our highest appreciation is expressed to the ISS crew members who participated and carried out these studies

# REFERENCES


with outstanding professionalism and enthusiasm. In honor and memory of the cosmonaut, vice director of the IBMP and IMMUNO-co-PI Prof. Boris Morukov who passed away in 2015.



**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Buchheim, Matzel, Rykova, Vassilieva, Ponomarev, Nichiporuk, Hörl, Moser, Biere, Feuerecker, Schelling, Thieme, Kaufmann, Thiel and Choukèr. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Cellular Responses of Human Postural Muscle to Dry Immersion

*Boris S. Shenkman1† \* and Inessa B. Kozlovskaya2†*

*1 Myology Laboratory, State Scientific Center of Russian Federation – Institute of Biomedical Problems, Moscow, Russia, 2 Department of Sensory-Motor Physiology and Countermeasures, State Scientific Center of Russian Federation – Institute of Biomedical Problems, Moscow, Russia*

Support withdrawal has been currently considered as one of the main factors involved in regulation of the human locomotor system. For last decades, several authors, including the authors of the present paper, have revealed afferent mechanisms of support perception and introduced the concept of the support afferentation system. The so-called "dry immersion" model which was developed in Russia allows for suspension of subjects in water providing the simulation of the mechanical support withdrawal. The present review is a summary of data allowing to appreciate the value of the "dry" immersion model for the purposes of studying cellular responses of human postural muscle to gravitational unloading. These studies corroborated our hypothesis that the removal of support afferentation inactivates the slow motor unit pool which leads to selective inactivation, and subsequent atony and atrophy, of muscle fibers expressing the slow isoform of myosin heavy chain (which constitutes the majority of soleus muscle fibers). Fibers that have lost a significant part of cytoskeletal molecules are incapable of effective actomyosin motor mobilization which leads to lower calcium sensitivity and lower range of maximal tension in permeabilized fibers. Support withdrawal also leads to lower efficiency of protective mechanisms (nitric oxide synthase) and decreased activity of AMP-activated protein kinase. Thus, "dry" immersion studies have already contributed considerably to the gravitational physiology of skeletal muscle.

Keywords: gravitational unloading, postural muscle, support afferentation, muscle atrophy, muscle stiffness, contractile properties, signaling pathways, plantar mechanical stimulation

# INTRODUCTION

One of the key spaceflight factors having a considerable effect on the regulation of the human locomotor system is support withdrawal. In recent years, several authors, including the authors of the present paper, have revealed afferent mechanisms controlling support perception and introduced the notion of the support afferentation system (Kozlovskaya et al., 2008). Experiments in spaceflight involving space crew members and animals as well as on-ground simulation studies with volunteers and animals show that support withdrawal has a significant effect on motor control mechanisms, in particular, changing the motor unit pools activation pattern by inactivating slow motor units, and leads to reflex tone decrease in postural and mixed postural/locomotor muscles (Kozlovskaya et al., 1988). Similar experiments show profound skeletal muscle atrophy and the associated change in the rate of protein

### *Edited by:*

*Nandu Goswami, Medical University of Graz, Austria*

#### *Reviewed by:*

*Giovanna Calabrese, Università degli Studi di Catania, Italy Sue Bodine, Roy J. and Lucille A. Carver College of Medicine, United States*

#### *\*Correspondence:*

*Boris S. Shenkman bshenkman@mail.ru † These authors have contributed equally to this work*

#### *Specialty section:*

*This article was submitted to Integrative Physiology, a section of the journal Frontiers in Physiology*

*Received: 04 April 2018 Accepted: 14 February 2019 Published: 11 March 2019*

#### *Citation:*

*Shenkman BS and Kozlovskaya IB (2019) Cellular Responses of Human Postural Muscle to Dry Immersion. Front. Physiol. 10:187. doi: 10.3389/fphys.2019.00187*

**377**

synthesis and degradation, alterations in the signaling mechanisms of protein turnover regulation, alterations in gene expression, including myosin heavy chain isoforms, and so forth (Baldwin et al., 2013). However, all such alterations were discovered in the course of studies conducted either in spaceflight or in on-ground simulations. Under actual microgravity, apart from support withdrawal, other biomechanical factors exist, including axial unloading and the increased ballistic component of movements. In another analog model (unilateral lower limb suspension), human contralateral limbs bear the whole weight, and tendon tension in the unloaded limb is different from tension observed in real microgravity (Lee et al., 2006). In head-down tilt bed rest studies, the support afferentation is not completely withdrawn. The virtually complete removal of support afferentation can only be attained through full submergence of the body in water, i.e., under the conditions of water immersion. The so-called "dry immersion" model which was developed in Russia (Shulzhenko and Vil-Vilyams, 1976; Navasiolava et al., 2011) allows for suspension of subjects in water for a particular period of time (3–56 days) while avoiding negative effects of prolonged water exposure on human skin. Under the conditions of dry immersion postural muscles exhibited decreased maximal voluntary and evoked maximal force and reflectory tone (Kozlovskaya et al., 1988; Grigor'ev et al., 2004), as well as alterations in motor unit recruitment and the parameters of spinal reflexes (Kirenskaia et al., 1986; Sugajima et al., 1996; Shigueva et al., 2015).

Evidence provided by studies on molecular and cellular mechanisms of postural muscle plasticity under gravitational unloading have been thoroughly reviewed in recent years (Kachaeva and Shenkman, 2012; Baldwin et al., 2013; Bodine, 2013; Ohira et al., 2015). However, these studies are focused on the properties of molecular mechanisms regulating the intensity of anabolic and proteolytic pathways in muscle fibers and alterations in contractile responses in muscle fibers of different types as well. It is taken for granted that all effects of microgravity are induced either by disuse of postural muscles or by the removal of resistive component of motor activities, which is often summarized as the "strain vs. activity" problem (Falempin and Mounier, 1998; Ohira et al., 2015). At the same time, these reviews do not cover integrative mechanisms leading to the inactivation of muscle fibers under microgravity. Dry immersion studies involving humans finally give us the possibility to understand the role of support afferentation withdrawal in the development of the three basic components of muscle plasticity under the conditions of real or simulated microgravity, namely hypogravity-induced atrophy (i.e., loss of muscle mass), hypogravity-induced atonia (i.e., sharp decline in stiffness of the whole muscle and single muscle fibers), and alterations in myosin phenotype (alterations in myosin gene expression leading to the domination of fast myosin heavy chain isoform expression).

Also, dry immersion studies allow for detection of the mechanisms that trigger atrophy and atony events at the earliest stages of gravitational unloading.

# ATROPHY DEVELOPMENT

The first evidence of muscle mass loss induced by real microgravity was obtained in animals. For instance, in rats, 12–14-day flights aboard biosatellites or in space vehicles (the Space Shuttle) led to 35–40% decrease in soleus muscle mass and 15–20% decrease in fast muscles mass (Baldwin et al., 2013). Also, cross-sectional area (CSA) of slow-twitch muscle fibers (in soleus muscle) decreased by over 40% after spaceflight (Baldwin et al., 2013). In rhesus macaques (*Macaca mulatta*), 12–14-day flights led to 30–40% decrease in soleus muscle fiber size and comparable decrease in vastus lateralis muscle fiber size according to our findings (Belozerova et al., 2003; Shenkman et al., 2003). The aim of these experiments, along with the measurement of muscle fiber atrophy in monkeys after spaceflight, was also to compare alterations in the structure and several other properties of muscle fibers in monkeys induced by real microgravity in spaceflight and on-ground model (a flight simulator capsule). Thus, the aforementioned studies provided for the first time the analysis of the effects of various spaceflight factors on atrophic changes in postural soleus muscle and nonpostural vastus lateralis muscle fibers. It was found that the profound atrophy of soleus muscle fibers occurred only in animals exposed to the conditions of real spaceflight, while atrophic changes in vastus lateralis muscle occurred in animals under the conditions of both real spaceflight and on-ground flight simulator capsule. Thus, the evidence showed that while the atrophy of soleus muscle was mainly induced by the removal of terrestrial gravity, the atrophy of vastus lateralis muscle was, at least partially, induced by its decreased contractile activity due to restricted mobility in the space vehicle.

The study of astronauts' skeletal muscle tissue before and after spaceflight was initially impeded by the fact that under real spaceflight conditions, astronauts are subjected not only to microgravity but also to physical exercises utilized as a countermeasure to adverse spaceflight effects. Moreover, the protocols of such exercises could not be standardized for a long time. The research team led by V. R. Edgerton was the first to analyze the vastus lateralis muscle tissue of three astronauts after a 5-day flight and of five astronauts after an 11-day flight aboard Space Shuttle crafts (Edgerton et al., 1995). After 5 days of flight, the range of alterations in the muscle fiber CSA was 11% for slow-twitch muscles and 24% for fasttwitch muscles. After 11 days of flight, the alteration range was 16–36%, with the fast-twitch muscle atrophy also more profound in this case.

We were able to analyze for the first time human vastus lateralis muscle fiber after 3-day and 7-day "dry" immersion (Shenkman et al., 1999). After 3 days of immersion, the evidence reliably showed a 7.5% decrease in CSA of slow-twitch muscle fibers, with the 8.9% (*p* < 0.05) decrease for the fast-twitch muscles. After 7 days of immersion, the decrease of the slowtwitch muscle fiber CSA was by 17.3%, with 15.3% for the fast-twitch muscles. Comparison of this evidence with that obtained by the Edgerton team shows that the CSA alteration ranges under immersion and in flight over a short period of time differ insignificantly and that the atrophic alterations in the slow and fast-twitch muscle fiber CSA (with even a slightly more profound changes in fast-twitch muscle, especially in flight) do not significantly differ both under real microgravity and immersion.

Later, we were able to obtain evidence of the decrease in human soleus muscle fiber CSA (Grigor'ev et al., 2004; Shenkman et al., 2004). After 7-day immersion, without exposure to other factors, we observed a decrease in the slow-twitch muscle fiber CSA by, on average, 24%, while no significant alterations in the fast-twitch muscle fiber CSA occurred. These alterations are difficult to compare with the similar data obtained by Widrick et al. (1999) and Trappe et al. (2009), who studied the Shuttle Vehicle and International Space Station crew members, due to the obviously different durations and conditions of exposure. Additionally, some of the latter studies involved astronauts performing 90-day and even longer flights with a significant amount of in-flight physical exercise.

It is desirable to compare the data that we obtained in immersion studies with the data on muscle fiber size recorded in head-down tilt (HDT) experiments of various durations. The atrophic alteration values found in our studies after 7 days of immersion were usually recorded on the 14th, 30th, or 60th day of the HDT (Hikida et al., 1989; Bamman et al., 1998; Clarke et al., 1998; Shenkman et al., 2000). Sadly, we could not find publications containing measurements of muscle fibers recorded during shorter-term HDT bed rest studies. Comparison of muscle fiber measurements made in immersion studies with data obtained in unilateral lower limb suspension (ULLS) studies shows the following. Alterations in the morphological characteristics of soleus and vastus lateralis muscle fibers after short-term exposure to ULLS are less profound compared to those after 7-day dry immersion (Hather et al., 1992; Brocca et al., 2015). After 14-day ULLS, no alterations in vastus lateralis muscle fiber size were found (Deschenes et al., 2002). Furthermore, the distribution of alterations in various muscles differs from such found under hypokinesia or immersion. In ULLS, a decrease in vastus lateralis muscle fibers CAS is more pronounced than that of soleus muscle (Hackney and Ploutz-Snyder, 2012). It can be suggested that in ULLS model, the functional unloading for soleus muscle is less pronounced than that for other muscles. This suggestion is confirmed by studies of functional activity of the portions of the Achilles tendon using functional MRI. Evidently, the strain in the middle portion of the Achilles tendon does not decrease under exposure to ULLS (Lee et al., 2006). Consequently, we can conclude that, for unknown reasons, the extent of soleus muscle unloading in ULLS is less pronounced than in spaceflight.

Thus, after 3–7 days of "dry" immersion rapid atrophic alterations occur that are similar to those found in astronauts after short-term spaceflights (at least in vastus lateralis muscle). This phenomenon can be described as rapid atrophy and suggesting that the removal of support afferentation triggers rapid and profound changes in postural and mixed postural/ locomotor muscles. Also worthy of note is the fact that, as opposed to the rarely described effects of short-term bed rest and unilateral lower limb suspension, soleus muscle atrophy after 7-day immersion is only significant in slow-twitch muscle fibers, which suggests a direct dependence of the state of these fibers on the intensity of afferent signaling.

# CONTRACTILE PROPERTIES OF SINGLE PERMEABILIZED FIBERS

Animal experiments (rats and monkeys) have repeatedly shown that exposure to real microgravity for 7–14 days leads to decreased contractile properties of single permeabilized fibers (Holy and Mounier, 1991; Stevens et al., 1993; Fitts et al., 2000). These experiments demonstrated a decrease in soleus muscle fibers maximal tension (sometimes decrease in specific tension), the decrease of calcium sensitivity (the right-hand shift of the Ca-tension curve), and the increase of the unloaded shortening velocity. Studies of muscle fiber physiology in astronauts began in the late 20th century. The shortest spaceflight analyzed by researchers lasted 17 days. Analysis of the contractile properties of permeabilized and calcium-stimulated fibers showed that maximal tension decreased by (on average) 20%, while the unloaded shortening velocity increased by 30% (Widrick et al., 1999). It is worth noting that, unlike experiments on animals exposed to real or simulated microgravity, the aforementioned study did not reveal alterations in calcium sensitivity in humans. The pronounced shift of the Ca-tension curve was only recorded for one crew member whose muscle fibers exhibited the largest decrease in maximal tension. According to the same researchers, a decrease in maximal tension in human soleus muscle fibers following 17-day bed rest did not exceed 13% (Widrick et al., 1997).

In 2002 and 2009, we studied the effect of 7-day "dry" immersion on the contractile properties of human soleus muscle fibers. In the first series of experiments, maximal tension of soleus muscle fibers decreased by 32% (Shenkman et al., 2004), and in the second series of experiments, it decreased by 26% (Ogneva et al., 2011a,b). No significant alterations in specific tension were noted. Both series of experiments exhibited consistent and significant right-hand shifts of the Ca-tension curve (according to рСа50) which indicates the sufficient decrease in calcium sensitivity of the myofibrillar apparatus (**Figure 1**). Note that in both series of experiments, alterations in the aforementioned parameters were more pronounced than those exhibited by astronauts although the duration of exposure was approximately 2.5 times shorter in the former case. The similar result had been reported in relation to isokinetic dynamometry studies conducted before and after a 7-day spaceflight and a 7-day "dry" immersion (Kozlovskaia et al., 1984). The decrease in maximal voluntary strength was found to be more pronounced under the conditions of immersion than in spaceflight. The authors explained these unexpected results by the fact that while the amount of contractile activity in human lower limb muscles is smaller in spaceflight than on the ground (although the authors of the latter paper do not provide exact data on the amount of contractile muscle

activity in spaceflight), it is nevertheless still larger in spaceflight than under the conditions of immersion hypokinesia. This suggestion has recently been corroborated by the data obtained from the study of the crew members' physical activity in spaceflight (Fraser et al., 2012). The larger decrease in the contractile properties of human single soleus muscle fibers under the conditions of immersion than in spaceflight may be accounted for by the similar suggestion.

# TRANSVERSAL STIFFNESS OF MYOFIBRILLAR APPARATUS

We used atomic force microscopy for the first study of the transversal stiffness of the various compartments of human soleus muscle fibers after 7 days of "dry" immersion (Ogneva et al., 2011a,b). The measurements were performed according to the original protocol (Ogneva et al., 2010) on single permeabilized fibers in the relaxed state and at the peak of isometric activity with *pCa* = 4.2. Some of the fibers were treated with Triton X-100 to lyse membrane structures and make the myofibrillar apparatus accessible to the cantilever. After the immersion, in relaxed fibers, the myofibrillar stiffness decreased twofold in the region of the Z-disc and threefold in the region of the M-disc. Between the Z-disc and the M-disc (the so-called half-sarcomere region), myofibrillar stiffness decreased more than twofold. Our experiment showed the approximately identical decrease of active fiber stiffness in all the regions of the myofibrillar apparatus that was examined: up to 0.4-fold of its initial value. Thus, the differences between the stiffness values in relaxed and active fibers were less significant after the immersion than before it. This evidence is in accordance with the measurements of the contractile properties of permeabilized fibers (see above) which show the decrease of both maximal isometric tension and calcium sensitivity in muscle fibers. It may be suggested that exposure to "dry" immersion leads to the malformation of cross bridges in calcium-activated fibers. To understand the causes of this phenomenon, it is necessary to analyze the effects of support withdrawal on myofibrillar cytoskeletal proteins.

# MYOFIBRILLAR CYTOSKELETAL PROTEINS

Titin is the chief scaffold protein connecting the basic contractile and regulatory elements of the sarcomere (thin and thick filaments, Z-discs and M-discs) and, according to recent research, acting as mechanosensor in muscle fibers (Lange et al., 2005). Decrease in titin content under unloading was first discovered in rats by Kasper and Xun (2000). Similar evidence was found at the Lille University (Toursel et al., 2002). Also in that year, we found decrease in titin content and increase in the content of its proteolytic fragment, T2, in rat soleus muscle after 14-day unloading (Shenkman et al., 2002). Since titin is considered to be a part of the series elastic component and decline in fiber stiffness is observed as early as after 3 days of hindlimb unloading (Ogneva, 2010), its breakdown or increase in elasticity could be expected already after 2 or 3 days of unloading. Apparently, however, this is not the case. Studies conducted in collaboration between our laboratory and Podlubnaya group in 2008 (Ponomareva et al., 2008) showed that the basic titin isoform, N2A, typical for skeletal muscles, remains intact after 3 days of unloading. Earlier, finding no changes in titin content after 3 days of unloading, Goto had discovered decrease, rather than increase, in elasticity in the elastic region of the titin molecule located between the Z-disc and the N2A domain (including the spring PEVK region) after such unloading (Goto et al., 2003). Recently, the explanation of this phenomenon has been suggested as a result of several studies (Nishikawa et al., 2012), which showed that increase in calcium ions content in fibers [occurring under gravitational unloading (Ingalls et al., 1999, 2001; Shenkman and Nemirovskaya, 2008)] leads to the strong binding of titin N2A domain to thin filaments. Even more recently, a 40% decrease in titin content and an increase in the T2 proteolytic fragment content were found to have occurred in mice gastrocnemius medialis muscle after a 30-day spaceflight on the Bion-M1 satellite (Ulanova et al., 2015). The same study provided the first evidence of a significant increase in titin phosphorylation level. Sharp decrease in titin content and increase in T2 content was first discovered after 7-day "dry" immersion in human soleus muscle (**Figure 2**; Shenkman et al., 2004). Experiments on single permeabilized fiber specimens show that partial titin breakdown is caused by the activity of calpains (calcium-dependent cysteine proteases), particularly μ-calpain (Murphy et al., 2006). Studies on passive single muscle stiffness in animals show that titin has a crucial role in determining the elastic properties of muscular tissue (Fukuda et al., 2008). Consequently, it may be suggested that decrease in the intrinsic muscle stiffness after "dry" immersion may be partially caused by decreased titin content.

Another effect of "dry" immersion on human soleus muscle is a decrease in the content of nebulin, an important sarcomere skeleton protein acting as scaffold for thin filaments (**Figure 2**; Shenkman et al., 2004). The function of nebulin is clearly demonstrated by nebulin knockout experiments in which muscle ultrastructure exhibits severe malformation and shorter length of actin filaments and even significant partial loss of thin filaments (Li et al., 2015). Similar malformations of thin filaments were registered by Riley and Fitts in their studies of human soleus muscle biopsy samples after 17-day spaceflight and 17-day bed rest (Riley et al., 1998). The authors note that alterations in thin filaments are accompanied, under such conditions, by a decrease in maximal tension and increase in unloaded shortening velocity in single permeabilized fibers. It is possible that structural alterations in thin filaments and the corresponding alterations in stiffness of myofibrillar apparatus are caused precisely by significant partial loss of nebulin molecules pool which we discovered in human soleus muscle after "dry" immersion and in gastrocnemius medialis muscle in mice after long-term spaceflight (Ulanova et al., 2015).

After 7-day "dry" immersion soleus muscle samples exhibited a 20% decrease in the content of desmin (Ogneva et al., 2011a,b), a protein interlinking myofibrils in the Z-disc region and other cellular compartments (mitochondria, myonuclei, etc.) (see review by Capetanaki et al., 2007). Earlier decrease in desmin level was shown in unweighting experiments on rats (Enns et al., 2007; Ogneva et al., 2010).

Thus, under the conditions of "dry" immersion, a phenomenon is observed that was discovered in unweighting experiments on rodents: decrease in cytoskeletal protein content in the sarcomere. It primarily concerns elastic giant proteins such as titin and nebulin. Significant partial loss of these proteins, which contribute to passive resistance in strained muscular tissue or in contracted muscle, necessarily affects muscle fiber stiffness (see above). Also, a number of cytoskeletal proteins is involved in the regulation of the actin-myosin interfilament spacing and, consequently, in the formation of cross-bridges. For this reason, contraction in normal muscle fibers, with

intact sarcomeric cytoskeleton, cannot be described as fully unloaded even in experiments involving withdrawal of resistive load. Although at present, it is difficult to measure intrinsic resistance in muscles caused by cytoskeletal proteins, we suppose it could have a certain effect on both the contractile properties and signaling regulatory mechanisms. Partial loss of intrinsic resistance observed in gravitational unloading (particularly in "dry" immersion) necessarily affects signaling processes.

# SUBSARCOLEMMAL CYTOSKELETON AND SARCOLEMMAL PERMEABILITY

It has been suggested that a number of mechanosensory events in any cell, including muscle fibers, is localized in the cortical compartment adjacent to the cell membrane (Durieux et al., 2007). Consequently, investigation of the state of cell cortex proteins and the state of the sarcolemma is necessary. Using an atomic-force microscope, we were able to register alterations in sarcolemmal (cortical) cytoskeleton stiffness in the Z-disc and M-disc regions and in the so-called half-sarcomere region (i.e., between the discs) in human soleus muscle after 7-day "dry" immersion. 61.7% stiffness decrease was exhibited in cortical cytoskeleton. While before the immersion calcium activation led to (approximately) 200% increase in cortical cytoskeleton stiffness (apparently, through force transduction from myofibrils to cortical cytoskeleton *via* cytoskeletal network), after 7-day immersion there was only 80–120% increase in active fiber stiffness (Ogneva et al., 2010). However, such a dramatic decrease in cortical cytoskeleton stiffness was not caused by changes in dystrophin content because after immersion no sarcolemmal dystrophin disruptions were significantly exhibited (Gasnikova et al., 2004). Note that a decrease in dystrophin content and the increased number of fibers exhibiting sarcolemmal dystrophin disruptions were observed in rat soleus muscle after 2-week, or longer-term, hindlimb suspension (Chopard et al., 2001; Gasnikova and Shenkman, 2005). It is possible that sarcolemmal dystrophin degradation occurs only under prolonged gravitational unloading. Current studies suggest that dystrophin, having certain stiffness (Gumerson and Michele, 2011), prevents macromolecules from passing through the membrane. However, muscle creatine phosphokinase is found in human blood samples taken from people regularly engaged in moderate physical activity. 7-day exposure to "dry" immersion is significantly known to lower creatine phosphokinase content in blood (Gasnikova et al., 2004). This evidence is in accordance with the evidence provided by a study of creatine phosphokinase content in the blood of astronauts after long-term spaceflights. That study also registered considerable decrease in the content of that enzyme in systemic blood during spaceflight (Markin et al., 1998). Decrease in circulating creatine phosphokinase levels during spaceflight or "dry" immersion may indicate a lesser disruption of sarcolemma and its cytoskeleton, possibly as a result of decrease in muscle fiber contractile activity. Decrease in this muscle contractile activity is also evidenced by electromyographic data (Miller et al., 2004). In another simulated rat hindlimb suspension study, Christine Kasper (Kasper, 1995) showed for the first time that exposure of animals to such conditions leads to lower macromolecular permeability of the sarcolemma; however, during the acute period of recovery after unloading, albumins dyed with Evans blue were seen to intensively penetrate soleus muscle fibers sarcolemma. In our experiments, during the recovery period, an enhanced albumin transport across the sarcolemma was accompanied by increased number of sarcolemmal dystrophin disruptions (Gasnikova and Shenkman, 2005).

Thus, apparently, support withdrawal and postural muscle inactivation under the conditions of immersion can prevent sarcolemmal disruption and lower the intensity of macromolecular transport across the sarcolemma.

# SIGNALING AND PROTECTIVE MECHANISMS

Signaling mechanisms controlling changes in muscle mass and muscle contractility have become the object of intense studies in recent years (for review see Schiaffino et al., 2013). For instance, these studies have elucidated changes in certain signaling pathways in postural soleus muscle. It has been shown, in particular, that subcritical depolarization of the sarcolemma occurs at the early stages of the muscle contractile activity decrease and is caused by decreased electrogenic activity of the α2 isoform of the Na, K-ATPase (Kravtsova et al., 2015). With membrane potential decreased to −40 mV, part of voltagegated L-type calcium channels may become activated and calcium ions are accumulated in the myoplasm (Ingalls et al., 1999, 2001). Simultaneously, nitric oxide production, heat shock protein content, and calpastatin expression are decreased during unloading (Enns et al., 2007; Lomonosova et al., 2011, 2012). Deficiency of these three signaling factors being inhibitors of calpain activities promotes the μ-calpain activation and cytoskeletal proteins degradation. Additionally, decreased level of nitric oxide leads to ubiquitin ligases gene expression activation (Lomonosova et al., 2011). The E3 ubiquitin ligases expression also intensifies in response to IRS-1 (insulin receptor substrate-1) ubiquitinylation and degradation with subsequent Akt dephosphorylation (Nakao et al., 2009). The Akt dephosphorylation leads to dephosphorylation of FOXO1 and FOXO3 and their translocation into myonuclei. These transcription factors bind DNA to activate the E3 ubiquitin ligases expression. Also early stages of gravitational unloading result in a significant decrease in the rate of protein synthesis, 28S ribosomal RNA expression, dephosphorylation of glycogen synthase kinase 3β (GSK3β) (Mirzoev et al., 2016) as well as increased eukaryotic elongation factor 2 (eEF2) phosphorylation (Krasniy et al., 2013). Recent studies have also showed a decrease in PGC1α gene expression which contributes to activation of ubiquitin ligases gene expression at the early stages of unloading (Cannavino et al., 2014).

As shown above, much has been understood about early events in skeletal muscle in laboratory rodents exposed to gravitational unloading (using hindlimb suspension model). However, there is still a lack of data on signaling mechanisms initiating muscle atrophy in humans during the early stages of unloading. The first data on such events were obtained employing unilateral lower limb suspension which, judging by data on the distribution of atrophic changes among muscles (Hackney and Ploutz-Snyder, 2012), does not completely simulate gravitational unloading under real microgravity. However, shortterm limb suspension causes increased ubiquitin ligases gene expression and decreased FAK phosphorylation level (Gustafsson et al., 2010; Flück et al., 2014). These data correspond well with the data obtained earlier in rat studies.

So far, there exist few studies on signaling mechanisms in human soleus muscle after dry immersion. After 7-day dry immersion, a dramatic decrease in neuronal nitric oxide synthase (nNOS) level was observed (Moukhina et al., 2004). More recently, a decrease in the level of this enzyme, although less pronounced, was observed after 3 days of immersion (Vilchinskaya et al., 2015). After 3 days of exposure, decrease in phosphorylation level of this enzyme was also observed (Vilchinskaya et al., 2015). Reduction of the level and degree of nNOS had been observed earlier in suspension studies involving rats and in bed rest studies involving humans (for review see Shenkman et al., 2015). In a 14-day rat suspension experiment we demonstrated a decrease in nitric oxide level in soleus muscle (Lomonosova et al., 2011). More recently we found a strong evidence of nitric oxide level decrease in rat soleus muscle already after 24-h suspension (Vilchinskaya et al., 2015, unpublished observation). It is possible that decrease in nitric oxide level is caused, at least in part, by decreased gene expression, or intensified degradation, of neuronal NO synthase. Nitric oxide is the endogenous inhibitor of calpain activity (Michetti et al., 1995). Consequently, decrease in nitric oxide level may lead to calpain activation (like in rat tail suspension studies—see above) and, accordingly, degradation of certain cytoskeletal proteins. Indeed, after 3 days of immersion, we found a significant decrease in desmin level in human soleus muscle (Vilchinskaya et al., 2015). Among kinases catalyzing NO synthase phosphorylation, we focused on AMP-activated protein kinase (AMPK) (Chen et al., 2000) whose activity is known to increase in response to AMP accumulation (for review see Mounier et al., 2015 which is usually caused by high energy expenditure) and decrease in response to high-energy phosphates phosphorylation [which can be caused by muscle inactivity (Ohira et al., 2015)]. We supposed that under the conditions of short-term "dry" immersion the degree of AMPK phosphorylation may decrease. Indeed, after 3 days of immersion phosphorylated AMPK level decreased significantly (Vilchinskaya et al., 2015). Thus, dry immersion studies provided us with the first evidence of AMPK phosphorylation decrease after short-term unloading. These data were confirmed in rat hindlimb suspension study (Mirzoev et al., 2016). Considering the importance of AMPK in the regulation of energy metabolism, gene expression and protein turnover, it may be suggested that this phenomenon may contribute to triggering a chain of events leading to the formation of the atrophic signaling pattern.

Neuronal NO synthase phosphorylation may also occur in response to the kinases of the IGF-1/Akt signaling pathway (Hinchee-Rodriguez et al., 2013). During the first week of hindlimb suspension rodents exhibited the ubiquitination and degradation of IRS-1 (insulin receptor substrate), which regulates the phosphorylation degree of the downstream protein kinases of the PI3K signaling pathway (Nakao et al., 2009). In our study, after 3 days of "dry" immersion no significant changes in IRS-1 were found (Vilchinskaya et al., 2015). We expect that future studies will show whether AMPK dephosphorylation at the early stages of unloading causes a decrease in nNOS activity and, consequently, a decrease in NO production and calpain activation. These and other aspects of atrophic and atonic signaling under short-term "dry" immersion are still need to be studied.

# THE ROLE OF SUPPORT AFFERENTATION IN THE DEVELOPMENT OF HYPOGRAVITY MUSCLE SYNDROME

The direct influence of support afferentation on human locomotor functions was first shown in the Soviet-Cuban joint experiment aboard the Soviet space vehicle. That experiment involved plantar mechanical stimulation (Hernandez-Korwo et al., 1983). In subsequent "dry immersion" studies, a modified plantar stimulation device was used that allowing for a long-term series of stimulation. These studies have shown, in particular, that plantar stimulation under the conditions of immersion allows for maintaining the normal level of electrical activity and reflectory transversal stiffness in soleus muscle (for review see Grigor'ev et al., 2004).

The following protocol has been used during our experiments: plantar stimulation, with pressure equaling 40 kP, was performed every day over the course of 6 h, for 20 min at the beginning of each hour, reproducing two natural modes of locomotion: slow walking (75 steps per min) and fast walking (120 steps per min), each for the duration of 10 min. After 7-day immersion with plantar mechanical stimulation no decrease in slow-twitch muscle fiber cross-sectional area and no discernible shifts in the proportion of fibers expressing slow and fast myosin heavy chain isoforms were observed in soleus muscle (Shenkman et al., 2004). Thus, atrophy development was prevented without usage of intense running or resistance exercise. Plantar stimulation allowed to prevent decrease in maximal isometric tension and calcium sensitivity in permeabilized muscle fibers (**Figure 1**; Litvinova et al., 2004; Shenkman et al., 2004; Ogneva et al., 2011a,b). These findings give evidence that muscle activity induced by support afferent stimulation could prevent malformation of cross-bridges.

In regard to studies on transversal stiffness of myofibrillar apparatus (using atomic force microscopy with permeabilized fibers pre-treated with Triton X-100), in relaxed fibers after support stimulation under the conditions of 7-day immersion decrease in stiffness was registered only in the Z-disc region (30%). In all the other sarcomere regions no significant alterations in transversal stiffness were registered as compared with pre-immersion values (Ogneva et al., 2011a,b). While the use of support stimulation did not completely prevent stiffness decrease in active fibers (with pCa = 4.2), its range across sarcomere regions was 15–25%. Thus, a decrease in active fiber stiffness was significantly less pronounced after immersion combined with support stimulation than after "pure" immersion (Ogneva et al., 2011a,b). Apparently, muscle activity allowed preserving the stiffness of myofibrillar apparatus by preventing malformations in cross-bridges and degradation of sarcomere cytoskeleton proteins. The latter suggestion is corroborated by data on titin and nebulin content in human soleus muscle after "dry" immersion involving support stimulation. In such immersion studies, members of the group exposed to support stimulation exhibited only a slight decrease in titin and nebulin content, while members of the group that was not exposed to support stimulation exhibited up to 40% decrease in titin and nebulin content (**Figure 2**; Litvinova et al., 2004; Shenkman et al., 2004). Decrease in desmin content was also not found in subjects exposed to support stimulation. As the degradation of the aforementioned cytoskeletal proteins is usually believed to be caused by μ-calpain, it is possible that muscle activity induced by afferent stimulation initiates the endogenous mechanism of calpain inhibition. Such a mechanism may be involved in maintaining the high level of nitric oxide synthase activity and NO being the endogenous inhibitor of calpain activity (see above). In our study, plantar mechanical stimulation not only prevented a decrease in nNOS content but also contributed to its increase as compared with the pre-immersion level (Moukhina et al., 2004). Further research will show the accuracy or inaccuracy of our hypotheses concerning the mechanism that allows support afferentation to ensure the consistent, albeit low, level of postural soleus muscle activity and to maintain the normal state of cytoskeleton and the systems of actomyosin motor mobilization.

We have also shown that some parameters that had been altered in response to gravitational unloading do not change in response to support stimulation. Thus, in an active muscle fiber, dramatic decrease in sarcolemma stiffness can be completely prevented by support stimulation, apparently, because this parameter depends on force transduction from myofibrils to cortical cytoskeleton. However, the transversal stiffness of relaxed fiber sarcolemma under the conditions of immersion remained decreased even when support stimulation had been performed during immersion (Ogneva et al., 2011a,b). Similarly, despite support stimulation, the content of α-actinin-1, a sarcolemmal cytoskeleton protein, remains decreased (Ogneva et al., 2011a,b). It is possible that the force that muscle fiber generates when exposed to afferent stimulation is insufficient for macromolecular permeability since creatine phosphokinase content in blood also remained decreased under such conditions (Gasnikova et al., 2004).

# CONCLUSION

Apparently, studies concerning cellular responses of human skeletal muscle to real microgravity (in spaceflight) remain scarce. This makes on-ground studies, especially those involving humans, particularly important. The present review is a summary of data that allows us to appreciate the value of the "dry" immersion model for the purposes of studying cellular responses of human skeletal muscle, particularly postural muscle, to gravitational unloading. The model proves to be especially useful to study the interaction between systemic mechanisms (primarily, ones of the central nervous system) and local (electrophysiological and biomechanical) mechanisms of hypogravity muscle syndrome development.

For instance, our studies showed the crucial role of support afferentation withdrawal in muscle alterations under hypogravity. These studies corroborated our hypothesis that the withdrawal of support afferentation inactivates the slow motor units pool (Kirenskaia et al., 1986) which inevitably leads to selective

FIGURE 3 | The hypothetical scheme of support withdrawal consequences in postural muscle. The scheme supposes that the exposure to weightlessness leads to the withdrawal of support afferentation. The support withdrawal induces the decline in slow motor unit activity and consequently the decline in mechanical activities of the slow-twitch muscle fibers. This decline means the reduced mechanical reflectory and then intrinsic muscle stiffness. The disuse of slow-twitch muscle fibers leads to reduced protein synthesis and increased protein breakdown, followed by the morphological signs of muscle atrophy. Simultaneously in disused fibers the alteration of myosin heavy chain isoforms expression pattern is followed by the slow-to-fast phenotypic transition. ST—slow-twitch.

inactivation, and subsequent atony and atrophy, of muscle fibers expressing the slow isoform of myosin heavy chain (which constitutes the majority of soleus muscle fibers) (**Figure 3**; Shenkman et al., 2004). Fibers that have lost a significant part of cytoskeletal molecules (Litvinova et al., 2004; Shenkman et al., 2004) are incapable of effective actomyosin motor mobilization (Litvinova et al., 2004; Ogneva et al., 2011a,b) which leads to a lower calcium sensitivity and lower range of maximal tension in calcium-induced contraction. Support withdrawal also leads to lower efficiency of protective mechanisms (NO synthase) and decreased activity of AMP-activated protein kinase (Vilchinskaya et al., 2015).

Dependence of the main anabolic and catabolic signaling pathways on the state of support afferentation system is still needs to be assessed, and the immersion model seems best suited for this purpose. Also, these studies (particularly involving plantar mechanical stimulation) provide convincing evidence that it is possible to maintain the key parameters of the internal environment of muscle fibers by means of low-intensity contractile activity without substantial external resistance. Data on the state of cytoskeletal proteins as well as contractile and stiffness properties of myofibrils show that the intrinsic resistance of muscle fibers found under the conditions of normal gravity is capable of maintaining the intracellular homeostasis even at low levels of contractile activity.

# REFERENCES


Thus, "dry" immersion studies, albeit rare as compared with studies using other models of microgravity, have already considerably contributed to the gravitational physiology of skeletal muscle. Future research, no doubt, will have even greater potential.

# AUTHOR CONTRIBUTIONS

Both authors listed have made a substantial, direct and intellectual contribution to the work, and approved it for publication.

# FUNDING

The work was supported by grants from the Russian Foundation for Basic Research (No. 17-29-01029, No. 16-29-08320) and the Program for Basic Research of the SSC RF–IBMP RAS (No. 65.3 and No. 63.1).

# ACKNOWLEDGEMENTS

The technical assistance of Ekaterina Dodonova and Timur Mirzoev is gratefully acknowledged. We apologize to colleagues whose studies were not cited owing to space limitations.


kinase expression in rat m. soleus under early stage of hindlimb unloading. *Dokl. Biochem. Biophys.* 453, 283–285. doi: 10.1134/S1607672913060021


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Copyright © 2019 Shenkman and Kozlovskaya. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.*

# Dry Immersion as a Ground-Based Model of Microgravity Physiological Effects

Elena Tomilovskaya<sup>1</sup> \*, Tatiana Shigueva<sup>1</sup> , Dimitry Sayenko<sup>2</sup> , Ilya Rukavishnikov<sup>1</sup> and Inessa Kozlovskaya<sup>1</sup>

<sup>1</sup> RF SSC – Institute of Biomedical Problems, Russian Academy of Sciences, Moscow, Russia, <sup>2</sup> Center for Neuroregeneration, Houston Methodist Research Institute, Houston, TX, United States

Dry immersion (DI) is one of the most widely used ground models of microgravity. DI accurately and rapidly reproduces most of physiological effects of short-term space flights. The model simulates such factors of space flight as lack of support, mechanical and axial unloading as well as physical inactivity. The current manuscript gathers the results of physiological studies performed from the time of the model's development. This review describes the changes induced by DI of different duration (from few hours to 56 days) in the neuromuscular, sensory-motor, cardiorespiratory, digestive and excretory, and immune systems, as well as in the metabolism and hemodynamics. DI reproduces practically the full spectrum of changes in the body systems during the exposure to microgravity. The numerous publications from Russian researchers, which until present were mostly inaccessible for scientists from other countries are summarized in this work. These data demonstrated and validated DI as a ground-based model for simulation of physiological effects of weightlessness. The magnitude and rate of physiological changes during DI makes this method advantageous as compared with other ground-based microgravity models. The actual and potential uses of the model are discussed in the context of fundamental studies and applications for Earth medicine.

Edited by:

Nandu Goswami, Medical University of Graz, Austria

#### Reviewed by:

Hanns-Christian Gunga, Charité – Universitätsmedizin Berlin, Germany David Andrew Green, European Astronaut Centre (EAC), Germany

> \*Correspondence: Elena Tomilovskaya finegold@yandex.ru

#### Specialty section:

This article was submitted to Integrative Physiology, a section of the journal Frontiers in Physiology

Received: 21 April 2018 Accepted: 04 March 2019 Published: 27 March 2019

#### Citation:

Tomilovskaya E, Shigueva T, Sayenko D, Rukavishnikov I and Kozlovskaya I (2019) Dry Immersion as a Ground-Based Model of Microgravity Physiological Effects. Front. Physiol. 10:284. doi: 10.3389/fphys.2019.00284 Keywords: dry immersion, motor control, gravity unloading, support withdrawal, supportlessness

# INTRODUCTION

Observations performed after space flights of different durations have shown that weightlessness causes body deconditioning, which is demonstrated by changes in most of the physiological systems (Gazenko et al., 1986; Kornilova, 1987; Kozlovskaya et al., 1988, 1990; Berger et al., 1992; Grigoriev and Egorov, 1992; Edgerton and Roy, 1996; Reschke et al., 1998; Clement et al., 2003). The necessity to study the mechanisms of hypogravitational effects on the human body led to the exploration of experimental models of weightlessness.

Immersion into a liquid analogous by density to human body tissues (Gazenko et al., 1972; Shulzhenko and Vill-Villiams, 1975) and antiorthostatic hypokinesia (head down tilt bedrest, HDBR) (Kakurin, 1968; Genin and Sorokin, 1969) have been proposed to be the most appropriate environment for the physiological studies of hypogravity effects.

Support withdrawal, local load elimination and the proximity of biomechanical conditions of motor activity organization to those in weightlessness from the beginning of space era

determined that immersion would be chosen as the best model for training of working operations in weightlessness.

In the early 1960s, scientists started to study the physiological effects of immersion to determine its capacity to simulate the effects of weightlessness (Beckman et al., 1961; Graveline et al., 1961). It has been shown that immersion reproduces microgravity-induced changes in the human body's motor (Ovsyannikov, 1972) and cardio-vascular (Shulzhenko and Vill-Villiams, 1976) systems and affects other physiological functions. The usage of immersion as a valuable model was limited only by discomfort and possible risks of long term skin contact with the liquid environment.

# HISTORY OF DI MODEL DEVELOPMENT

At the beginning of the 1970s two investigators from the Institute of Biomedical Problems (Russia), K. B. Shulzhenko and I. F. Vil-Villiams, developed a method of long term immersion, dry immersion (DI), with the use of special waterproof and highly elastic fabric (Shulzhenko, 1975; Shulzhenko and Vill-Villiams, 1975). A test subject wearing a shirt and trunks was put on waterproof fabric and immersed into a deep bath up to the neck level, in a supine position. The area of the fabric's surface considerably exceeded the area of the water surface. The folds of the waterproof fabric allowed the person's body to be enveloped from all sides freely (**Figure 1**).

The high elasticity properties of the fabric artificially created conditions similar to zero gravity via floatation. Thus, DI reproduced three effects of weightlessness: physical inactivity, support withdrawal and elimination of the vertical vascular gradient. Since its development, DI has been the main model in Russia for studying the effects of weightlessness lasting 5–7 days, similar to the duration of short-term flights on orbital space stations.

Most of the results described in this review come from studies where healthy male volunteers were placed, individually or in pairs, in a supine position in a bath with dimensions of

FIGURE 1 | Overall view of dry immersion facilities at IBMP. Image credit IBMP/Oleg Voloshin.

200 × 100 × 100 (300 × 300 × 200) cm. The bath was filled with water at a temperature that was kept constant at 33 ± 0.5◦C. The daily routine was specified in accordance with the schedule of studies, countermeasure procedures (if the experiment included them), including 8 h of sleep, 3–4 meals, a medical supervision program and experimental studies. The research participants were taken out of DI for 15–20 min each day for sanitary and hygienic procedures with the usage of a special lift rising from the bottom of the bath.

In 1974, E. B. Shulzhenko and I. F. Vil-Villiams performed the longest 56-day DI experiment with two volunteers, which convincingly proved the applicability and safety of the DI model for reproducing long-term microgravity effects. The results of this experiment showed that 56 days of DI decreased the redundant capabilities of the blood circulation system, providing deconditioning of the body as a whole. The changes of the functional state of the blood circulation system demonstrated signs of an adaptation processes with the predominance of parasympathetic tone and decline of resistance to g-loading in the "head-hip" direction and low body negative pressure (Shulzhenko and Vill-Villiams, 1976; Vil-Villiams and Shulzhenko, 1978).

In early years of the DI's use, the main direction of research was a description of its effects on different functions of the human body. Attention of the researchers was primarily focused on vital functions, including the activities of the cardiovascular and respiratory systems and metabolism. These studies revealed similarities in the depth, dynamics and direction of DIinduced physiological changes to those observed in space flights (Kamenskiy et al., 1976; Aleksandrova et al., 1980; Fomin, 1981; Ivanov et al., 1983; Kozinets et al., 1983; Orlov, 1985; Yarullin et al., 1987; Genin and Galichiy, 1995; Mikhailov et al., 1995).

It has been shown that the subjective experience of exposure to DI at its initial stage is perceived as a comfortable and pleasant state of relaxation (Shulzhenko and Vill-Villiams, 1976; Shulzhenko et al., 1983a; At'kov and Bednenko, 1987; Navasiolava et al., 2011a; Tomilovskaya, 2013a,b). However, heaviness in the head and nasal congestion in some cases were observed later (presumably due to a redistribution of the body's fluids in the cranial direction). During the first 2–3 days of DI, most of test subjects reported the development of abdominal pain (Rukavishnikov et al., 2014b) and back pain of differing intensity (Rukavishnikov et al., 2017a,b); on the 3rd or 4th day of DI, all of these signs disappeared. Sleep disorders (mostly due to back and abdominal pain), loss of appetite, meteorism and constipation are also observed in the initial stages of DI (Shulzhenko, 1975; Fomin, 1981; Shulzhenko et al., 1984; Tomilovskaya, 2013a,b). Under conditions of "double baths," when 2 subjects shared the same water pool, body movements of one subject were sometimes followed by the symptoms of space motion sickness for the other one (Barmin et al., 1983; Struhal et al., 2002; Kreidich et al., 2007), similar to those registered during the 1st days of space flight (Kornilova, 1997; Williams et al., 2009) and bedrest (Greenleaf, 1984; Baum and Essfeld, 1999).

In the 1970s the water immersion was used broadly as a fast and convenient model for assessing the efficacy of countermeasure means and methods. Those studies were

performed at the Russian Scientific Testing Institute of Aviation and Space Medicine by A. M. Genin, I. D. Pestov, V. I. Stepantsov, A. V. Eremin and other specialists of space mission medical support. Later, this research was entrusted to the newly created Institute of Biomedical Problems. However, the impossibility of long term stay in the water immersion imposed serious restrictions to the range of tested countermeasure means. The DI method opened the possibility for long-duration exposures and started a new line of studies.

# SIMILARITY OF DI- AND HDBR-INDUCED EFFECTS

In parallel to the DI experiments, HDBR research has been developing and the model has achieved worldwide popularity (Watenpaugh, 2016). This is not surprising because the organization of HDBR is relatively easier than DI, since the test subjects are available for the studies for almost 24 h a day at all the stages of experiment. Comparison of the data has shown similarity of the changes observed in these two models (Aleksandrova et al., 1980; Chaika et al., 1982; Grigoriev et al., 2004). However, the dynamics and the depth of these changes significantly differ (Kozlovskaya et al., 1984; Grigoriev et al., 2004). For instance, according to Shenkman et al. (1997), the decrease in size of the slow and fast types of muscle fibers reached 15–18% by the 7th day of DI; similar changes were observed only after 60 days of HDBR. Very recent studies of cardiovascular, postural and neuromuscular changes under conditions of 21 days of HDBR and 3 days of DI revealed the similarity of their effects, suggesting that DI has an influence that is seven times stronger than HDBR (Tomilovskaya et al., 2018).

HDBR and DI reproduce the lack of axial loads, redistribution of body fluids in a cranial direction (to a greater extent in HDBR), hypokinesia and immobilization. The only substantial difference between these models is the level of support deprivation: in DI, support loads are totally eliminated, while HDBR provides the redistribution of support loads from the soles of the feet to the larger areas of the back and sides of the torso. The data from an experiment with combined exposure to HDBR and DI (alternating 12 h of HDBR and 12 h of DI) revealed significantly greater changes in periods of DI and their smoothing during HDBR sessions, confirming the leading role of support afferentation in the tonic postural system and motor acts regulation (Kozlovskaya et al., 1984).

Therefore, at the end of the 1970s, the DI model became the core component of studies focused on the effects of microgravity on the sensory-motor system. The motor system is considered to be the most gravity-dependent system of the human body. One of the main gravity-depended functions of the motor system is the maintenance of vertical posture and the position of body segments in the gravitational field. These functions are provided in mammals including humans by the tonic (postural) muscle system. The absence or sharp decline of gravity, eliminating the necessity of the aforementioned activities in weightlessness, greatly alters the functional and structural properties of tonic muscles. Orbital space flights have confirmed this suggestion and revealed the main factors which contribute to these effects of microgravity (Kozlovskaya, 2007). These factors are: a decrease of mechanical loads, a decline of axial and support loads, changes in the biomechanics of movements and changes in sensory system activities. The DI model reproduces accurately and to a full extent axial and support unloading, and significantly changes the sensory system's activity. Exposure to DI eliminates the factor of support but at the same time does not directly affect the other sensory systems, which provides a definite advantage in comparison to other microgravity models and allows to study the role of support afferentation in the body systems' activities.

In recent years the model of DI has gained popularity: a number of investigators from different countries (DLR in Germany, Angers University and Caen University in France) have taken part in the Russian DI research at IBMP. The results of these studies confirmed the supposition of the effects of support withdrawal on the activities of almost all body systems.

# DI EFFECTS ON THE NEUROMUSCULAR SYSTEM

Exposure to DI results in dramatic decline in muscle forcevelocity properties, as well in muscle tone, and composition. Muscle effects, in particular a decrease in force-velocity properties, have been revealed even in very short term space flights (Kakurin et al., 1971; Bryanov et al., 1976; Kozlovskaya et al., 1984, 1988; Bachl et al., 1992); however, only under DI conditions their metrics and temporal dynamics could be studied in detail. Comparisons between the depth of changes after 7 days of DI and space flights of the same duration have revealed their close similarity (Grigorieva and Kozlovskaya, 1983; Kozlovskaya et al., 1984). DI experiments allowed not only to describe these phenomena but to get a better understanding about the nature of the decline in force-velocity properties of skeletal muscles (up to 30%), which cannot be explained by the muscle atrophy requiring much greater exposure times.

Subsequent studies of evoked and voluntary maximal contraction (MVC) strength of shin muscles confirmed this suggestion (Koryak, 1998, 1999, 2001). The results of the research have shown that MVC amplitude decreases similarly after 7 days of space flight, 7 days of DI and 4 months of HDBR. However, the amplitude of evoked responses declined insignificantly, indicating that the decrease of muscle force-velocity properties was caused by lower intensity of central efferent command (Koryak, 1998; Koryak, 2001).

One of the acute responses to transition to weightlessness is a flexor posture, which points to a decrease of the extensor muscle activity involved in the maintenance of vertical posture on Earth (Thornton, 1987; Reschke et al., 1998; Kozlovskaya, 2007). In fact, neurological observations performed immediately after short-term space flights in crew members of the first space expeditions revealed distinct extensor atonia (Bryanov et al., 1976). The results of studies of shin muscle transverse stiffness performed after 7-day space flights revealed a 15–20% decrease of muscle tone in m.gastrocnemius and m.soleus on the 2nd day after landing (Kozlovskaya et al., 1988).

This phenomenon was studied in detail in DI experiments on the ground, demonstrating the rapid decrease (up to 40–50%) of transverse stiffness in all three heads of the shin extensor the lateral and medial m.gastrocnemius and most prominent – in m.soleus during the first 6 h of exposure (Gevlich et al., 1983; Kozlovskaya et al., 1988; Miller et al., 2004, 2005, 2010). The decrease of transverse stiffness of single muscle fibers was also revealed in m.soleus after 7-day DI (Ogneva et al., 2011). Studies of transverse stiffness in the shin flexor (m.tibialis anterior) did not reveal significant changes under DI conditions (Gevlich et al., 1983; Kozlovskaya et al., 1988; Miller et al., 2010).

Based on the above-mentioned data, the reflex nature of the muscle tone decline and its connection to the changes in motor units' activity was suggested by Kirenskaya et al. (1986). This was confirmed in the studies of the recruitment order of motor units (MUs) in a task involving small effort plantar flexion (10–12% from maximal voluntary contraction) under conditions of 7 days of DI and 120 days of HDBR (Kirenskaya et al., 1986; Khristova et al., 1988; Kozlovskaya et al., 1988). The results of these studies allowed to conclude that support unloading causes the changes of recruitment order of shin extensor MUs through suppression of small (tonic) MUs recruitment and facilitation of recruitment of large (phasic) ones. Later, this suggestion was also confirmed in studies of DI of different duration (Shigueva et al., 2015b).

It is worth noting that revealed phenomena indicate that Henneman's law of the MUs' recruitment order is reversed under microgravity conditions. Classic Henneman's law (or size principle) states that under load, motor units are recruited from smallest to largest. In practice, this means that slow-twitch, lowforce, fatigue-resistant muscle fibers are activated before fasttwitch, high-force, less fatigue-resistant muscle fibers (Henneman et al., 1965). During DI, however, a motoneuron size is not the only factor defining the recruitment order.

In parallel to the neurophysiological research, a wide program of muscle morphological studies has been performed under DI conditions. Decreases in the size of both slow and fast types of muscle fibers reached 5–9% after 3 days and 15–18% after 7 days of DI (Shenkman et al., 1999; Shenkman, 2016; Shenkman et al., 2017). Meanwhile, the analogous changes after 60 days of HDBR reached 14.2 and 12.4% for slow and fast types of fibers, respectively (Shenkman et al., 2004a). Thus, the depth of size reduction of muscle fibers under conditions of short-term DI was commensurate to that under conditions of long term HDBR. This fact indicates the important role of support withdrawal in the genesis of hypogravitational muscle atrophy. It is important to note that the possibility of support-dependent control of atrophic processes has been discussed in a number of scientific papers (De-Doncker et al., 2000; Nemirovskaya and Shenkman, 2002; Ilyina-Kakueva and Kaplansky, 2005). However, this hypothesis was not validated in direct human experiments with stimulation of the support input. Oganov et al. (1980) first demonstrated the shift in the ratio of slow and fast myosin forms toward the fast ones under conditions of gravitational unloading, although these changes were not so evident in HDBR (Ilyina-Kakueva et al., 1979; Kuznetsov and Stepantsov, 1989, 1990; Berg, 1996).

More recent studies performed under strict conditions of maintenance of subjects' supine position reported significant atrophy (up to 10%) of type I muscle fibers and increased portion of hybrid, typeI/II muscle fibers after 3 days of DI (Demangel et al., 2017). Molecular studies have also revealed changes in muscle signaling responses at the early stages of gravitational unloading (Vilchinskaya et al., 2015, 2016).

In summary, the neuromuscular effects of DI include:


# DOES SUPPORT WITHDRAWAL TRIGGERS HYPOGRAVITATIONAL MOTOR SYNDROME DEVELOPMENT?

The DI studies created a new branch in motor physiology research which can be termed as gravitational motor physiology. Studies of the functions and segments of the motor system both at the cellular and systemic levels have shown that gravity is deeply influential in the motor system. Primarily it affects muscles, or the effector system. The data from cell together with neurophysiological studies allowed scientists to suggest that gravity is a factor which affects the tonic postural system development. The deep skin sensitivity represented by clusters of Fatter-Paccini bodies is the trigger of this system which provides information on the presence or absence of gravity (Otelin et al., 1976).

Since there are no any other direct afferent changes in DI, support withdrawal is what promotes the sharp development of both cellular and systematic changes. This statement was illustrated by the scheme suggested by I. B. Kozlovskaya and B. S. Shenkman (**Figure 2**).

To confirm the trigger role of support afferentation in the control of the structural-functional properties of the tonic muscle system, the influence of mechanical stimulation of the soles support zones in locomotor regimens (75 and 120 steps per minute, pressure of 40 kPa in each sole) was tested. Stimulation was provided by Compensator of Support Unloading (CSU) developed jointly by IBMP, "Zvezda" and "VIT." CSU provides alternating pressure of 0.5 kg/cm<sup>2</sup> (40 kPa) on the support zones of the soles (heel and metatarsal areas).

The results of these studies confirmed the hypothesis that support afferentation plays the trigger role in the system of postural tonic control. The aforementioned changes in leg muscle functions were not observed in participants in whom the mechanical stimulation of the soles support zones during DI exposure was applied (Kozlovskaya et al., 2007a,b).

A decrease of transverse stiffness in m.soleus in this group reached significance only by the 6th day of DI; at the same time decrease of EMG amplitude at rest and maximal isokinetic force of m.triceps surae as well as signs of hyperreflexia and changes in MU recruitment order were not registered at all (Grigoriev et al., 2004; Khusnutdinova et al., 2004; Miller et al., 2004,

2005, 2010; Netreba et al., 2004, 2005, 2006; Kozlovskaya et al., 2007a,b; Shigueva et al., 2015a; Zakirova et al., 2015). Analogous effects of mechanical stimulation of the soles support zones were demonstrated in the structure of the muscular apparatus: no decrease of myofibrils' sensitivity to free calcium ions, which is regularly observed under conditions of gravitational unloading; no changes in the cross sectional area of muscle fibers; and no decrease of the number of fibers containing slow isoforms of heavy myosin chains (Grigoriev et al., 2004; Moukhina et al., 2004; Shenkman et al., 2004a,b).

The natural consequence of the results described above was an offer to include the method of artificial support stimulation into the space flight countermeasure system for counteracting the early, acute motor system changes in weightlessness. Therefore, the device for mechanical stimulation of the soles support zones was modified in accordance with ISS requirements (the "CSU" device manufactured by IBMP and "Zvezda"), and at the end of 2017, it was delivered to the space station.

# DI EFFECTS ON THE SENSORY-MOTOR SYSTEM

The DI studies have discovered that sensory-motor control systems are deeply affected when unloaded. First, the accuracy of movement control decreases (Repin, 1981; Kreidich et al., 1982; Grigorieva and Kozlovskaya, 1985; Kirenskaya et al., 1985; Sosnina et al., 2016). It has also been shown that the cortical organization of voluntary movements is changed (Kirenskaya et al., 2006; Tomilovskaya et al., 2008), and that the systems of posture and locomotion control suffer greatly (Melnik et al., 2006; Shpakov et al., 2008; Sayenko et al., 2016; Amirova et al., 2017). Similarly, studies of resting state brain activity revealed the changes similar to those obtained in a real space flight, indicating that support withdrawal is the main contributor to brain activity alterations and decline in the function of corticospinal tract function in weightlessness (Kuznetsova et al., 2015; Lazarev et al., 2018). Interestingly, support stimulation under DI conditions also eliminated these changes, although to a lesser extent than peripheral ones (Kozlovskaya et al., 2007a,b).

Vestibular function and visual-manual tracking is also significantly changed under DI conditions despite the absence of any direct influence on the vestibular system (Repin, 1981; Kreidich et al., 1982; Barmin et al., 1983; Kornilova et al., 2004, 2008, 2011a,b, 2013, 2016; Zobova et al., 2008).

For instance, in the studies of the static torsional otolithcervical-ocular reflex (OCOR), dynamic vestibular-cervicalocular reactions (VCOR), vestibular reactivity (VR), and spontaneous eye movements after 5- and 7-day DI a significant decrease in OCOR (gOCOR was 0.1, compared to the background gOCOR value of 0.25) was detected alongside a simultaneous significant increase in the VCOR/VR parameters in 28% of subjects on next day after DI. The findings have demonstrated that the support withdraw and the deficit of proprioception affected rather the accuracy of visual tracking than manual tracking. The obtained results confirm the development of sensory deprivation (and afferent deficit) under the DI exposure (Kornilova et al., 2016).

At the same time, the sensitivity to vestibular signals is sharply increased: the thresholds of oculographic response to galvanic stimulation decreased to 0.68–0.73 mA on the day 2 after 7 day DI in comparison to 0,92–1,05 mA before DI (Repin, 1981). Similar findings were observed in DI experiments performed in primates (Eron et al., 2000; Badakva et al., 2007). It is worth noting that no signs of motion sickness were reported during DI (Tomilovskaya, 2013b).

The sensitivity to proprioceptive signals was also changed: exposure to DI was associated with the development of H-reflex system hypersensitivity, which was demonstrated by a decrease of the soleus H-reflex thresholds and an increase of relative H-reflex amplitudes (Kozlovskaya et al., 2007a; Saenko, 2007; Zakirova et al., 2015).

As mentioned above, motion sickness was not observed during DI (Tomilovskaya, 2013b). Authors suggested that this may be due to rather motionless state of the participants. To test this, researchers began to investigate another type of water immersion – "suit immersion" (SI) — developed by IBMP specialists at the end of the 1980s (Genin et al., 1988a,b). SI, which used a thin-walled wetsuit "Forel" ("Trout" in Russian) equipped with an inflatable neck pillow to create postural, vestibular and operational loads directly during immersion, differed from DI by exclusion of strict hypokinesia and maintenance of vertical posture (Genin et al., 1988a,b; Larina et al., 1999; Larina and Lakota, 2000).

Although there is a difference in the body position between DI and SI, that is supine vs. vertical position, respectively, hemodynamics and water-salt exchange parameters have been shown to be similar (Genin et al., 1988a,b; Larina et al., 1999). However, during the first hours of SI, the development of vestibulo-autonomic syndrome, characterized by perception of body and head movements/rotations, was revealed (Genin et al., 1988b). In addition, the wide spectrum of illusions up to shortterm loss of the awareness of body orientation were observed during SI. Following SI, a decrease of passive orthostatic probe tolerance, impaired postural control, an increase of EMG cost of standard motor tasks, and a significant decrease of physical and operator capacity were observed (Kreidich et al., 2007).

In summary, the effects of DI on the sensory-motor system include:


# DI EFFECTS ON THE CARDIORESPIRATORY SYSTEM

Studies of respiratory function have shown an increase of respiratory resistance, greater inspiratory and reduced expiratory reserve volumes, as well as changes in the thoracic-abdominal ratio in breathing motions and shifts in voluntary respiration regulation (Dyachenko et al., 2010; Mikhailovskaya et al., 2011; Popova et al., 2011). In addition, increased amplitude of breathing motions with reduced frequency has been noted, together with distinct breathing-related heart rate fluctuations (Semenov et al., 2011).

Furthermore, after 5 days of DI a significant increase in the markers of oxidative stress such as expired amines, chiefly butylamine, 2-cyanacetimide, some aldehydes, polyols, phenol, phenylacetylene, ketones, butyl acetate and a significant decrease in fatty acids has been observed (Mardanov et al., 2011). The mentioned changes indicate the alterations in metabolism and weakening of antioxidant system.

Studies of cardiovascular system changes in DI have revealed the subsequent involvement of electrical (increase of amplitude of QRS complex) and then – energy metabolic (decrease of heart rate and water-electrolyte balance alterations) processes changes in the myocardium; the most prominent changes were registered on the 5th day of DI (Ivanov et al., 2011). Autonomic regulation also appears to shift in the direction of the sympathetic component (Iwase et al., 2000; Eshmanova et al., 2008, 2009; Kabulova and Eshmanova, 2008).

Exposure to DI even of short duration (7 h) decreases orthostatic tolerance, and as the period of immersion is prolonged, the proportion of participants with orthostatic hypotension increases (Iwase et al., 2000; Vinogradova et al., 2002b; Eshmanova et al., 2009; Navasiolava et al., 2011a). All the studies revealed an increase in heart rate during vertical stance after DI of different duration (Navasiolava et al., 2011a,b). Recent studies performed after 3-days DI have shown decrease of systolic blood pressure, stroke volume and baroreflex sensitivity, pronounced tachycardia during low body negative pressure (LBNP) tilt test (De Abreu et al., 2017). The time of orthostatic tolerance in this study dropped from 27 to 9 min after DI; 9 of 12 subjects were unable to complete the test. Maximal LBNP value in the control studies reached −60 mmHg; 9 of 12 subjects have completed the test at the level of −30 mmHg. After DI exposure the LBNP level of −30 mmHg was reached by only 1 subject of 12 (De Abreu et al., 2017).

The relationship between changes in the water-electrolyte balance and cardiovascular responses revealed changes in electrophysiological propagation of myocardial excitation, and an increased variance of natural small oscillations of the electric potential of the heart (Larina et al., 2011). At the same time there was registered the increase of the brain natriuretic peptide BNP (as well as inactive NTproBNP secreted in an equimolar ratio), which is released by cardiomyocytes in response to stretching upon an increase in the ventricular pressure. The authors suggest that the increased level of NTproBNP during the recovery period following DI, reflects the degree of heart deconditioning during DI. The revealed intense tachycardia during orthostasis after DI, as well as the revealed changes in the electrophysiological characteristics of the myocardium and hypovolemia (one of the factors of development of the cardiovascular deconditioning under microgravity) during DI support the hypothesis of deconditioning induced increase in the heart load in the period of recovery. Within the first 24 h, the weight loss, rapid loss of fluid and sodium, and increase in the free cortisol content in urine were also observed. The changes in the water–electrolyte balance could cause metabolic–energy shifts that required activation of the respective regulatory mechanisms. A considerable (more than twofold) growth of the centralization index at the end of the experiment shows that inclusion of the central regulation mechanisms in the adaptation processes was an appropriate response aimed at compensation of the primary changes in the water–electrolyte balance and hemodynamics. The described changes appeared within the first 24 h of DI,

and were manifested in the following days of the experiment. The changes in autonomic regulation and the electrophysiology of the myocardium during the experiment gradually increased, and reached maximum at the end of the 7th day of DI. Thus, it has been shown that the structures regulating the metabolic processes and, potentially, higher autonomic centers, from molecular–cellular to systemic levels, may be involved in the cardiorespiratory response occurring during the first 24 h of DI. The revealed alterations in the myocardial excitation warrants further research (Larina et al., 2011).

In summary, the effects of DI on the cardiorespiratory system include:


# HEMODYNAMIC CHANGES IN DI

Exposure to DI has also been associated with endothelial dysfunction, including an increase in circulating endothelial microparticles (Larina et al., 2008; Navasiolava et al., 2010, 2011b). Ultrasound analysis of peripheral hemodynamics have shown that the linear velocity of blood flow along main arteries and low extremity veins decelerates in the course of DI (Moreva, 2008; Moreva et al., 2018). At the same time, increased intensity of tissue perfusion and a higher number of capillaries in the upper extremities during 5-day DI were registered (Suvorov et al., 2017). Other authors, however, reported an increase of middle cerebral vein velocity and jugular vein, portal vein and thyroid volume, but these findings were obtained only during the first 2 h of DI (Arbeille et al., 2017). Another recent study failed to reveal changes in blood flow in either cerebral artery, but internal carotid and vertebral arteries conductance decreased significantly on the 2nd and 3rd day of DI (Ogoh et al., 2017).

It was also shown that DI significantly affects blood supply in working muscles. The peak bloodstream in the calf muscles decreases 7–20% after DI exposure, while post-contraction hyperemia noticeably increases (Vinogradova et al., 2002b). In studies of the blood flow of the tibia performed under conditions of physical loading, it was suggested that exposure to DI strengthened the concurrent relationships between local working hyperemia and central vasoconstrictive effects directed toward the maintenance of circulatory homeostasis (Stoida et al., 1998). Physical loading after 7-day DI was also followed by changes in energy and metabolism as indicated by decrease of creatine phosphokinase activity, and the changes of levels of cortisol, triglycerides, insulin and inorganic phosphate in blood plasma (Buravkova et al., 2003). However, no changes in metabolic reflex regulation of hemodynamic parameters under conditions of local static muscle work were registered (Bravy et al., 2008).

In summary, the effects of DI on hemodynamics include:


# DI EFFECTS ON INTRACRANIAL PRESSURE (ICP)

Since DI was proposed to reproduce distribution of body fluids similar to that during space flight, some research has been dedicated to the assessment of intracranial pressure. However, the data of ultrasound studies were characterized by high variability, and did not lead to definite conclusions (Arbeille et al., 2017). Use of the otoacoustic emission method through registration of electrical response of outer hair cells of the cochlea — DPMC (Phase Shift of Microphonic Cochlear Potential) — showed that body position and DI do affect the DPMC (and thus ICP). Changes in ICP were similar during a supine-to-DI position change and supine-to-antiorthostatic position change. Following DI, a tendency to decreased phase shift of the otoacoustic response during transition of the body position from a vertical to antiorthostatic one, was revealed, which could be associated with the ICP increase (Avan et al., 2013; Rukavishnikov et al., 2013).

A study of optic nerve sheath diameter (ONSD) and cerebral autoregulation revealed a strong negative correlation between these two parameters. The authors suggested that a persistent elevation of ICP can be attributed to poor cerebral autoregulation recovery during DI (Kermorgant et al., 2017).

# DI EFFECTS ON RENAL AND DIGESTIVE SYSTEMS

Studies of the excretory function of the kidneys together with the classical increase of dieresis revealed an increase of protein and glucose excretion, although not exceeding physiological standards (Vorontsov et al., 2011, 2014). No shifts of glomerular filtration or tubular reabsorption were registered (Nesterovskaya et al., 2008). At the same time, the impedance measurement revealed a decrease of general body fluid and the volume of extracellular fluid as well as the volume of circulating plasma (Larina et al., 2008; Noskov et al., 2011).

Other studies report that exposure to DI causes rapid fluid and sodium loss, which increases the Na/K ratio in urine. The accompanying body weight loss is due to a decrease in the level of body hydration. An increase in the level of free cortisol in urine only on the 1st day of DI may be due to adaptation to the new conditions. The most pronounced changes were observed on the 1st day of DI. On the following days, a new level of water– electrolyte balance is established. The absence of changes in the renin or aldosterone level on days 3 and 7 was evidence that the major redistribution of fluids had been completed by that time

and regulation of the water–electrolyte balance had stabilized. The decrease in urine osmolality during DI characterizes an enhanced excretion of water but not electrolytes and shows that the regulatory mechanisms perceive water consumption during DI as excessive. In addition, this suggests a decrease in the concentration capacity of the kidneys (Larina et al., 2011).

Studies of the digestive system have revealed deceleration in the hepatic venous flow and signs of plethora in the abdominal venous system (Solovieva et al., 2016). Elevated levels of several substances were detected in the blood: pepsinogen, pancreatic amylase, bilirubin total (due to its unconjugated fraction), insulin, and C-peptide. The 13C-methacetin breath test has shown a slowdown in the rate of 13C-methacetin inactivation and a reduction in the hepatic metabolic capacity (Solovieva et al., 2016). The study of the liquid food evacuation during immersion has shown that its rate did not change substantially, therefore, reductions in the rates of methacetin inactivation and metabolism during immersion could be associated, mainly, with a slower assimilation of the preparation and a decrease in the venous blood flow of the liver. A reduction in the metabolic capacity of the liver in the conditions of immersion reflects a decrease in the inactivation of toxic metabolites, including unconjugated bilirubin, which may explain the increase of its content in blood.

Further studies have demonstrated the stability of liquid evacuation from the stomach and acceleration of the chymus transit along the small intestine hinder evacuation of the large intestine content, which is the primary cause for inhibition of gastrointestinal evacuatory activity in DI (Afonin and Goncharova, 2009; Afonin et al., 2013). The elevated gastrointestinal electrical activity may be related to the increased gastric secretion and elevated intestine tone in fasting test subjects, and displayed a close similarity to the changes induced by caffeine stimulation, long-term bed rest or space flight (Afonin and Sedova, 2012). At the same time, clear signs of dysbiotic changes such as significant reduce of fecal lactoflora and developed shifts in the microbial landscape (increase of the number of staphylococcus, yeast microflora and Gramnegative Bacillus) have been revealed (Ilyin et al., 2008; Solovieva et al., 2011).

In summary, the effects of DI on the renal and digestive systems include:


# DI EFFECTS ON BIOCHEMICAL PARAMETERS OF BLOOD AND URINE

Biochemical tests revealed changes in blood and urine protein content (Trifonova et al., 2010; Pastushkova et al., 2011; Vorontsov et al., 2014). Specifically, there is a decrease of muscle and cardiac constellation ferments activity, increase of lipid metabolism derivatives, increase of erythrocytes concentration, decrease in glucose tolerance and hypercholesterolemia, and increase of insulin and serum NT-proBNP levels, which is a proxy measure for brain natriuretic peptide (Ivanova et al., 2008; Markin et al., 2008; Navasiolava et al., 2011c; Coupe et al., 2013; Pastushkova et al., 2014; De Abreu et al., 2017). The new series of DI studies introduced the analysis of blood plasma proteome (Pastushkova et al., 2012). On DI days 2 and 3, growth of peak areas was observed in fragments of complement system proteins C3 and C4, high-molecular kininogen and fibrinogen. Significant increases of the peak area of apolipoprotein CI (reduced form with segregated threonine and proline) and C4 enzymes of the complement system, and fibrinogen on the 1st day after the experiment can be related to changes in motor activities of the subjects.

Analysis of the hemostasis system did not reveal any significant changes, although a tendency in decrease of antithrombin III activity (ATIII), protein C and plasminogen after exposure to 7-day DI was observed (Kuzichkin et al., 2010). Studies of antioxidant protection and lipid peroxidation rates did not reveal changes in their values during adaptation to DI. However, the return to normal activity after DI was associated with the development of a significant stress reaction, as evidenced by the strengthening of the lipoperoxidation, and a decrease in the functional activity of the antioxidant defense system (Zhuravleva et al., 2012).

In summary, the effects of DI on the biochemical composition of blood include: changes in the concentration of protein and lipid derivatives.

# DI EFFECTS ON SKELETAL SYSTEM

It is not surprising that there are not many studies dedicated to bone marker changes in DI, given that the skeletal system is quite inert and changes occurring throughout a relatively short exposure to DI, such as 7 days, are not expected. There were no effects of 7-day DI on bone formation or increase of bone resorption in the study of Kopp (2008). A recent study performed during 3-day DI, however, revealed significant decrement of bone formation markers, namely total procollagen type I N- and C-terminal propetides and osteoprotegerin (Linossier et al., 2017).

# DI EFFECTS ON IMMUNE SYSTEM

Studies of the DI effects on innate and acquired immunity revealed high variability of changes during DI (Berendeeva et al., 2011; Ponomarev et al., 2013). Observed negative shifts in congenital immunity at the end of 5-day DI and during recovery, suggest that there is an increased risk of infectious diseases, for instance caused by changes in normal microflora which normally does not cause pathological effects (Ponomarev et al., 2013). There were changes in the content

of immunoglobulin (sIgA, IgA, and IgM) in the gingival fluid, which can reflect inflammation process in the periodontal tissue. Presence in the oral cavity of periodontopathogenic type of bacteria, namely, Prevotella intermedia, Tannerella forsythia, Treponema denticola, Aggregatibacter actinomycetemcomitans, Porphyromonas gingivalis, was also noted (Nosovsky et al., 2017).

A study of the neuroendocrine regulation during DI has demonstrated a decrease of triiodothyronine and cortisol levels, as well as an increase of prolactin and thyroxine concentration (Nichiporuk et al., 2011; De Abreu et al., 2017). The rate of changes in immunity parameters among different subjects demonstrated the highest frequency in non-anxious and extravert individuals on day-5 in DI (Nichiporuk et al., 2011).

# EFFECTS OF DI ON THERMOREGULATION

Exposure to short and long-term space flights is followed by gradual increase of core body temperature by 1◦C over 2.5 months of flight, which was suggested to be associated with augmented concentrations of interleukin-1 receptor antagonist, a key anti-inflammatory protein (Stahn et al., 2017). Unfortunately, there no published studies on thermoregulation during DI, except for one report on the lack of significant changes in the body temperature on the day 3 of DI (De Abreu et al., 2017).

# DI RESULTS IN THE BACK PAIN PHENOMENON

A recent series of DI experiments explored the potential mechanisms of the back pain which is regularly observed at the beginning of space missions and under ground-based conditions. Almost all DI study participants report back pain during the first 2–3 days of DI (Rukavishnikov et al., 2014a,b, 2015, 2017a,b; Kozlovskaya et al., 2015; Treffel et al., 2017). Retrospective analysis of the data from nine of the most recent DI experiments conducted at IBMP (2006–2018) demonstrated that 70 out of 87 participants reported back pain with an intensity ranging from 4–5 to 9–10 points on a ten-mark subjective scale (see **Figure 3**) (Rukavishnikov et al., 2014b).

The findings also revealed a decrease of back extensor transverse stiffness (Rukavishnikov et al., 2017a,b, 2018). Data obtained during magnetic resonance imaging (MRI) also showed that vertebral column length significantly increased in the neck, thoracic and lumbar parts of the spine (Kozlovskaya et al., 2015; Rukavishnikov et al., 2015; Treffel et al., 2016a,b). At the same time, the height of intervertebral disks increased, the disks swelling was observed, the neck and thoracic kyphosis were flattened, as were the thoracic and lumbar lordosis. MRI analysis revealed a decreased cross sectional area of mm. quadratus lumborum, multifidus, and erector spinae at the level of L4–L5 (Rukavishnikov et al., 2017b). It has been proposed that each of these factors can contribute to the hypogravitational back pain, including intervertebral disk swelling, spine lengthening, influence on the yellow ligament

of the spine and back muscle atonia (Treffel et al., 2016b; Rukavishnikov et al., 2017b).

# CONTRIBUTION OF DI MODEL TO SPACE FLIGHT COUNTERMEASURE SYSTEM DEVELOPMENT

As indicated at the beginning of this review, DI has been developed to investigate in depth the mechanisms of the effects of microgravity on the human body. Logically, DI has become instrumental in the development of a countermeasure system to prevent negative physiological effects of space flights. The history and extent of the studies employing the DI model and leading to the tests, comparison, and optimization of various countermeasure approaches, worth to be discussed in a separate review. In this section, we focus on the major outcomes which originated the design of the current countermeasure programs for use on board the International Space Station.

In the late 1970s, studies of short arm centrifuge were performed (Shulzhenko and Vill-Villiams, 1976; Grigoriev and Shulzhenko, 1979; Vil-Villiams and Shulzhenko, 1980; Clement and Pavy-Le Traon, 2004). The experiments performed under conditions of 13- and 56-day DI have shown that application of short arm centrifuge (SAC) 0,5–1 Gz rotations in the course of DI was followed by maintenance of back-up capabilities of cardiovascular system and external respiration and increase of resistance to 5 min +3 Gz rotations in "head-pelvis" direction. Combination of SAC sessions with water-salt supplements and 2340 kg physical loads (hand and leg exercises with expanders) increased countermeasure effects in this case (Shulzhenko and Vill-Villiams, 1976; Kokova, 1983).

Later, the number of DI studies aimed to examine the effects of different countermeasure means were performed. For instance,

the studies of the efficacy of antigravity suit of bladderless type in maintenance of orthostatic tolerance have been carried out under conditions of 7-day DI (Shulzhenko et al., 1983b; Vil-Villiams et al., 1996). With the suit "Centaur" on, the 20 min tilt test at 70 degrees induced less changes in blood pressure as compared with that without the suit. The maximum heart rate significantly decreased, minimum stroke volume and pulse pressure increased. As such, using DI model the authors revealed that the anti-G suit helps to increase orthostatic tolerance and can be used following space flights.

Early DI studies demonstrated that the application of 25 mm Hg negative pressure to the lower body (LBNP) can counteract the decrease in the orthostatic tolerance (Genin and Pestov, 1974). The water-salt additives combined with LBNP and physical training, increased the capacity of vascular bed and help to retain body fluids, sodium, and chloride during DI (Grigoriev and Shulzhenko, 1979).

The positive effect of hip occlusion cuffs in preventing orthostatic intolerance after 18-h immersion strongly depended on the duration of break between occlusion sessions: the best effect has been demonstrated in the case of 1-min break (the pressure consisted 25 mm Hg, the duration of training session – 7 h) (Genin and Pestov, 1974). The mechanisms of the countermeasure effects of occlusion cuffs revealed in DI experiments, are based on the pooling of blood in vessels of the lower extremities due to the occlusion of superficial veins and hence – decrease of blood volume that is shifted in cranial direction (At'kov and Bednenko, 1987).

During 28-day DI study, the efficacy of isolated and combined effects of SAC and veloergometry have been examined. The study has shown that veloergometry training in the regimen of 600 kg/min (10 min training − 10 min break) for 60 min twice a day was followed by increase of cardiac input and decrease of total peripheral resistance. Combination of veloergometry training with SAC (0,8–1,9 Hz twice a day for 60 min per session) enforced the countermeasure effect which was assessed by cardiovascular system parameters (Vil-Villiams and Shulzhenko, 1980; Vil-Villiams, 1994).

Effects of low and high-frequency electrostimulation of leg muscles have been studied in the series of DI experiments (Koryak et al., 2008; Kozlovskaya, 2008; Eshmanova et al., 2009; Koryak, 2014, 2018; Solovieva et al., 2016). These studies demonstrated that daily use of low frequency stimulation can counteract the decrease of force-velocity properties in m. triceps surae, especially when the intensity of stimulation is rather high (more than 13 V). Comparative evaluation of the efficacy of low-frequency stimulation of the lower limb muscles versus high frequency stimulation during 7-days DI, showed superiority of the low-frequency stimulation in prevention of motor dysfunctions (Koryak et al., 2008; Kozlovskaya, 2008). During these experiments, positive effects of neuromuscular electrical stimulation were also associated with unloading of the right heart and revealed by HD ECG parameters and growth of the "myocardium" index (Eshmanova et al., 2009). The application of high frequency electromyostimulation prevented elevation in pepsinogen, pancreatic amylase, and bilirubin, predominantly within its unconjugated fraction, as well as an increase in insulin secretion; however, it did not affect the ultrasound patterns of the hemodynamic rearrangement in both the liver and the abdomen (Solovieva et al., 2016).

A wide program of DI studies was dedicated to the effects of mechanical stimulation of the plantar mechanoreceptors in locomotor regimens. These studies confirmed the hypothesis on the trigger role of support (weight bearing) afferentation in the hypogravitational motor syndrome development (Grigoriev et al., 2004; Kozlovskaya et al., 2007b; Layne and Forth, 2008). Application of the plantar mechanostimulation during six 20-min sessions per day (with the stimulation frequency of 75 and 120 steps/min and pressure 40 kPa) eliminated the negative effects of gravitational unloading on sensory-motor and neuromuscular systems, including the decline of extensor muscle stiffness and decrease of the maximal voluntary force; a significant decrease of the absolute force of the isometric contraction of single skinned muscle fibers; a prominent decline of the tonic muscle fibers transversal size; and the transformation of the myosin phenotype from slow to fast one (Popov et al., 2003; Grigoriev et al., 2004; Khusnutdinova et al., 2004; Litvinova et al., 2004; Miller et al., 2004; Moukhina et al., 2004; Shenkman et al., 2004a,b; Netreba et al., 2005; Kozlovskaya et al., 2007a,b). The results of studies of biomechanical characteristics of locomotions brought to the conclusion that the rates of mechanical foot stimulation applied in the experiment did not change energy expenditure in the muscles; however, they moderated the amplitude of angular knee joint movements following 7 days of DI (Melnik et al., 2006). Application of the support stimulation was also effective in preventing venous compliance increase and orthostatic intolerance (Vinogradova et al., 2002a).

Countermeasure effects of the soles support zones stimulation has been also demonstrated in the visual-vestibular function. The group of subjects without support stimulation exhibited marked omnidirectional deviations in the eye tracking parameters, whereas in the group of subjects with support stimulation, these parameters were similar to the baseline values. Support stimulation also stabilized the pursuit function of the eye making it less variant. However, there was no uniformity in the subjects' reaction to stimulation, which infers that the methods of improving the eye pursuit function should be personally "molded" (Kornilova et al., 2004).

Rather few data can be found in the literature concerning the studies of countermeasure effects of different pharmacological means with use of DI model. In the studies of pain sensitivity under conditions of DI, no reduction in the morning pain sensitivity (which is typical for normal conditions) has been revealed. Such analgesic as ketorolac had no effect on pain sensitivity, when determining the pain threshold by method of thermo algometry. The authors noticed that DI substantially altered the pharmacokinetics of ketorolac, increasing the rate of absorption of the drug and reduction of its relative bioavailability and retention time in the blood. These results indicate that pain therapy schemes may be administered differently during space flight as compared with 1G (Baranov et al., 2015).

In 7-day DI experiments the pharyngeal microflora in 22 healthy volunteers has been studied. For prophylactic pharyngeal dysbiosis, two probiotic drugs were used: oral -

"lactobacterin dry" and local -"lactobacterin immobilized on collagen." Administration of oral probiotic was accompanied with growth of pharyngeal opportunistic microflora preventing translocation of intestinal microflora. Local probiotic, on the contrary, decreased opportunistic microflora in the pharynx, but was associated with gastrointestinal dysbiosis. It was concluded that the combination of topic and oral probiotics provides the maximally effective prophylaxis of pharyngeal dysbiosis during DI (Kryukov et al., 2007; Ilyin et al., 2008). In another 7-DI study, the effects of sydnocarb [3- (beta-phenylisopropyl)-N-phenyl carbamoyl sydnonimine] were investigated. The subjects exercised on a bicycle ergometer before and after water immersion. During exercises, ECG, heart rate, minute respiration volume, oxygen consumption, carbon dioxide production, cardiac output, oxygen pulse were recorded. The subjects who received placebo showed a significant decrease of oxygen consumption at a maximum workload. Those who were given sydnocarb maintained normal oxygen consumption during bicycle ergometry. The drug increased the workload per kg body weight, maintained physical work capacity, and improved the cardiovascular function after DI exposure (Anashkin and Belyaev, 1982). The positive effect of the pharmaceutical – alleviation of myocardium strain – was a result of amlodipin (slow calcium channel inhibitor) assistance to the coronary blood flow; however, it had a negative side-effect on orthostatic stability (Eshmanova et al., 2009).

The DI model was actively used during the development of the "Penguin" axial loading suit (Barer et al., 1998). The studies have revealed increased EMG amplitude in the lower extremities muscles in subjects who wore Penguin suit while exercising on a bicycle ergometer. The same studies revealed increased heart rate and metabolic rate values in the subjects during performing the step loading veloergometric test with Penguin use. Other DI studies have shown that the "Penguin" suit was effective in preventing hypogravitational hyperreflexia, peripheral microcirculation changes, height increase, back pain development and back muscle atonia, and a decrease in force-velocity properties (Barer et al., 1998; Shigueva et al., 2015b; Rukavishnikov et al., 2015; Suvorov et al., 2017).

# CONTRIBUTION OF DI MODEL TO THE TERRESTRIAL MEDICINE

The findings on the physiological effects of DI on various systems of the human body, found their clinical application starting from the 1980s. First, the effects of DI in muscle relaxation, decreasing skeletal muscles tone and eliminating muscle spasticity, were noted (Kozlovskaya I. B. and Semenova K. A., unpublished observations).

The DI has been applied in edema therapy, especially intractable ones induced by cardiovascular etiologies as well as by burns, liver cirrhosis, and renal disease. In individuals with the mentioned syndromes, the effects of 1.5–3 times diuresis increase after 4–6 h post DI was maintained during 2 days (Yunusov et al., 1985; Ivanov and Bogomazov, 1988).

During the 1990s, several studies utilized DI in therapy of hypertensive crisis (Kirichenko et al., 1988; Evdokimova et al., 1989; Ivanov and Makarova, 1990). The results revealed a significant and persistent hypotensive effect after 90 min of exposure to DI. The effects were pronounced through two different mechanisms: (a) via a decrease of elevated total peripheral vascular resistance (TPVR) and no changes in cardiac output, or (b) via a decrease of elevated cardiac output. These hemodynamic effects were attributed to sanogenetic mechanisms which optimize blood circulation during hypertension (Evdokimova et al., 1989).

Recent studies (Meigal et al., 2017) have revealed positive effect of short-term, up to 6 h, DI sessions in patients with either Parkinson's disease or parkinsonism: the score of subscale I of Wein's inventory for autonomic disorders has decreased by 45%, and of the subscale II – by 23%. In addition, the time of the choice reaction has significantly improved. These effects were also observed 2 weeks post the course of DI.

The DI is widely used in pediatrics as a component of rehabilitation programs for premature children, to normalize hemodynamics and prevent brain edema. Exposure to DI is also used for the rehabilitation of children with central nervous system disorders, such as perinatal encephalopathy and other perinatal pathology (Kazanskaya, 2008; Khan et al., 2015). The method is also widely used in children undergoing cerebral palsy therapy, since it is effective in lowering hypertonus and hyperactivity, and in reflex facilitation (Yatsuk, 2002; Titarenko et al., 2015).

On the contrary to negative effects of 3–7 days DI on immune system (see above), the positive effect of 40– 50 min DI on immunological parameters were revealed. Namely, promoted normalization of the functional activity of T- helpers and the lower adhesive properties of lymphocytes, as well as reduction in the incidence and severity of neonatal infectious and inflammatory diseases have been demonstrated (Chasha et al., 2008).

The diagnostic value of DI in the early detection of neurological pathologies under normal conditions was suggested in joint Russian–Austrian experiments. After 24, 48, and 72 h of DI, some healthy participants showed neurological symptoms (such as cerebellar disintegration, deterioration in the function of peripheral nerves, posterior tract, as well as pyramid and extrapyramidal systems) (Gerstenbrand and Kozlovskaya, 1990; Berger et al., 2001). This phenomenon may be explained by the fact that some people have latent neurological deficiency (congenital or acquired as a result of injuries, infections, etc.), but the central nervous system's high plasticity typically provides compensation for these conditions. The sensory conflict created by DI (changing the support and proprioceptive afferent inflow) removes compensatory mechanisms that are ineffective in the new conditions, and hidden neurological disorders are revealed.

Due to its relaxation effects, DI was also used in sports medicine for recovery of athletes after high intensity training (Radzievskiy and Radziyevskaya, 2007).

There are also medical contraindications to the use of DI. According to various experts, they include myocardial infarction,

severe cardiac arrhythmia, chronic respiratory diseases with pulmonary heart decompensation, thrombophlebitis, severe shortness of breath at rest (Ivanov and Makarova, 1990), acute inflammatory processes and generalized lesions of the skin (Yatsuk, 2002). However, these considerations are mostly theoretical and no quantitative results have been reported.

# CONCLUSION

Dry immersion reproduces to some extent many of the changes associated with spaceflight. DI can serve as a valid ground-based model for simulation of physiological effects of weightlessness for some systems. The magnitude and rate of physiological changes during DI makes this method advantageous as compared with other ground-based microgravity models. DI is instrumental in the design of countermeasure programs for use in space, especially, for passive countermeasures which do not require intensive motor activity of the subjects. DI may also have a range of clinical applications that warrant further study.

# REFERENCES


# AUTHOR CONTRIBUTIONS

ET wrote the draft of the manuscript and made its revisions. TS prepared the part on DI effects in sensory-motor system and designed the figures. DS contributed in the global revision and reorganizing of the manuscript. IR prepared the part on back pain phenomenon and implementation of the DI model in Earth medicine. IK made the revision of the manuscript and prepared the part of use of DI for countermeasures development.

# FUNDING

This work was supported by the Russian Academy of Sciences (63.1) and grant RFBR 16-29-08320.

# ACKNOWLEDGMENTS

The authors thank the volunteers who participated in Dry Immersion experiments and organizer team of the experiments.




in human terms 7-day immersion hypokinesia. Space Biol. Aerospace Med. 3, 48–51.




microgravity simulation different from bed rest studies. Pain Res. Manag. 2017:9602131. doi: 10.1155/2017/9602131


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Tomilovskaya, Shigueva, Sayenko, Rukavishnikov and Kozlovskaya. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

digital media

of impactful research

article's readership