This randomized controlled study was conducted at the German Aerospace Center (DLR, Cologne, Germany). The longitudinal study involved 15 days of familiarization, including baseline data collection (BDC), 60 days of 6° head-down tilt bed-rest for 24 h/day (HDT) and 90 days of recovery (Figure 1). Physical activity during the familiarization (BDC) and recovery phases was restricted to free movement in the facility, and reeducation training during the recovery phase. The control of posture, gait and functional mobility was tested for all subjects at seven time points: 1 day after arrival at the facility (BDC-14), 1 day before (BDC-1) and after bed-rest (R+0), as well as during recovery (R+7, R+13, R+28, and R+90). For details about the study design, schedule, diet, subject recruitment and eligibility criteria see Kramer et al. (2017b).

The 24 volunteers were selected from a large group of actively recruited males. A priori, the sample size was estimated by means of a power analysis based on the results of previous bed-rest studies (f = 0.4; alpha = 0.05; power = 0.9) with a margin of two dropouts (Dupui et al., 1992). The study was performed in two campaigns (autumn 2015 and spring 2016). Recruitment started approximately 6 months before the first campaign directly after approval from the ethics committee and was finalized early in 2016 with a total recruitment time of 8 months. Inclusion criteria were as follows: male, age between 20 and 45 years, body mass index between 20 and 26 kg/m², non-smoking, no medication, no competitive athlete, no history of bone fractures and medical issues documented in detail in Kramer et al. (2017b). The participants gave written informed consent for the experimental procedure, which was approved by the ethics committee of the Northern Rhine Medical Association (No. 2014105, Dusseldorf, Germany) and the Federal Office for Radiation Protection (Berlin, Germany). The study was designed according to the most recent iteration of the Declaration of Helsinki. All subjects were in good health and were randomly allocated to either a jump exercise group (JUMP) or to a control group (CTRL) using dice roll of each pair of participants in the morning of the first day of HDT. The mean ± SD age, height and body mass were 30 ± 7 years, 181 ± 7 cm and 77 ± 7 kg for JUMP (n = 12) and 28 ± 6 years, 181 ± 5 cm and 76 ± 8 kg for CTRL (n = 11). Of the 24 healthy male subjects that were enrolled in the study, one subject discontinued the study on BDC-4 for medical reasons unrelated to the study. The subject could not be replacement due to time constrains leaving a total of 23 study participants. One participant started in the training group, but was reallocated to the control group after three training sessions due to a possible medial tibia stress syndrome. Two of the 23 subjects that completed the study (one CTRL, one JUMP) were re-ambulated after respectively, 49 and 50 instead of 60 days of HDT due to medical reasons, but completed the recovery phase with all the scheduled measurements.

While subjects in the CTRL group were inactive, subjects in the JUMP group participated in the training intervention during strict 6° head down tilt bed-rest for 60 days, executed under medical supervision. The subjects trained in a sledge jump system (Novotec Medical GmbH, Pforzheim, Germany) allowing natural jumps in the horizontal plane with different acceleration levels (Kramer et al., 2012). The training protocol for the JUMP group comprised 48 training sessions with an effective training duration of approximately 3 min; each session was varied, but contained on average 4 × 12 countermovement jumps and 2 × 15 repetitive hops (Figure 1). A warm-up prior to the training consisted of 6 squats, 6 heel raises, 3 submaximal countermovement jumps, and 10 submaximal hops. All sessions were supervised; peak forces and peak power were documented. The subjects underwent nine 30 min familiarization sessions during BDC to get accustomed to the device and develop the correct jumping technique. Further details about the countermeasure, training procedures and familiarization sessions can be found in Kramer et al. (2017b).

Four protocols served to assess changes in posture, gait control, and functional mobility in response to bed-rest (Figure 2A). They were executed in the same order for each subject, but were randomized among the participants (subjects and therapists were not blinded; assessors were blinded).

A repeated measures analyses of variance (ANOVA), with time [BDC-1, R+0, R+7, R+13, R+28, R+90] as the repeated measure and group [JUMP vs. CTRL] as the inter-subject factor was used to test for adaptations in response to bed-rest. The normality of the data was evaluated with a Kolmogorov-Smirnov test; the data followed a normal distribution. If the assumption of sphericity established by Mauchly's test was violated, the Greenhouse-Geisser correction was used. The level of significance was set to p < 0.05. To compare each time point in the recovery period with BDC-1, Student's t-tests were used. The false discovery rate was controlled according to the Benjamini-Hochberg-Yekutieli method, a less conservative but still stringent statistical approach conceptualizing the rate of type I errors(Benjamini and Hochberg, 1995; Benjamini and Yekutieli, 2005). Partial Eta squared (ηp2) was also used as an estimate of the effect size for the ANOVA (η²_p< 0.04 small, 0.4 ≤ η²_p< 0.14 medium, 0.14 ≤ ηp2 large effect size; Cohen, 1988).

Non-inferiority statistics were used to verify the similarity with baseline values (Walker and Nowacki, 2010). For that purpose, 90% confidence intervals were calculated for the differences between baseline values (BDC-1) and values collected after bed-rest (R+0, R+7, R+13, R+28, and R+90) according to Piaggio et al. (2012). Note that for non-inferiority testing, an alpha level of 0.05 corresponds to the 90% confidence interval. The acceptable bounds were determined for each parameter separately, based on the differences observed between single trials assessed during BDC-1 or between trials assessed during BDC-14 and BDC-1. If the results were statistically non-inferior to baseline the respective parameter was marked with a ≈ symbol. One-tailed Pearson's correlation coefficients were calculated for CTRL, between percentage change in the CoF path length and the CCIs for R+0. One-tailed Pearson's correlation coefficients were calculated for the respective percentage changes for R+0 to determine the strength of linear relations between the degradation in posture control, gait and TUG. Statistical tests were executed with SPSS 25.0 (SPSS, Inc., Chicago, IL, USA). Group data is presented as mean value ± standard deviation.

Most of measured gait parameters were non-inferior after bed-rest for JUMP compared to baseline values, whereas CTRL showed significant changes in spatiotemporal, kinematic and variance characteristics. This was true for the locomotor tests performed with maximal (Table 2, Figures 2B, 4) and preferred gait speed (Table 3). The ANOVA showed significant group^*time interaction effects, with between-group effect sizes ranging from medium to large (Tables 1, 2). Plantarflexion during push off, the minimal foot clearance and maximal knee flexion during the swing phase were significantly reduced in CTRL. Concomitantly, gait speed (maximal gait speed F_{(1, 21)} = 10.3, P = 0.01, ηp2 = 0.33), step length and cadence were significantly reduced, whereas step time, COM displacement, double limb support (maximal gait speed F_{(1, 21)} = 3.3, P = 0.06, ηp2 = 0.14) and single limb support, stance phase and swing phase normalized to total gait cycle time were significantly increased. The CV of step length, step time and stance time was significantly increased after bed-rest. All the deteriorations that were only observed in the inactive CTRL were recovered between seven to 90 days after re-ambulation. Non-inferiority was shown for most of the spatiotemporal, kinematic and variability parameter for R+0, R+7, R+13, R+28, and R+90 compared to baseline values in JUMP, indicating that the gait pattern remained stable throughout bed-rest in the countermeasure group.

The time to complete a right and left turn in the TUG test remained constant after the 60 days of bed-rest for JUMP, whereas the CTRL showed significant changes (20–40%; Figure 2B). Values for CTRL returned to baseline 14 days after re-ambulation. The ANOVA revealed significant group^*time interaction effects with large between-group effect sizes (left turn F_{(1, 21)} = 14.9, P = 0.002, ηp2 = 0.42 and right turn F_{(1, 21)} = 29.3, P < 0.001, ηp2 = 0.58).

Non-inferiority was shown for parameters associated with time and power for R+0, R+7, R+13, R+28, and R+90 compared to baseline values in JUMP, whereas the inactive CTRL showed significant adaptations ranging between 20 and 80% (Figure 2B). The adaptations observed in CTRL were recovered 14 to 28 days after the end of bed-rest. The ANOVA revealed significant group^*time interaction effects with medium and large between-group effect sizes (time per iteration F_{(1, 21)} = 20.1, P = 0.01, ηp2 = 0.49 and maximal power F_{(1, 21)} = 7.2, P = 0.04, ηp2 = 0.26).

We detected a significant positive correlation between the CoF sway path and CCI of SOL_TA, GM_TA, VM_BF and RF_BF (Figure 4), indicating that an increased sway path was associated with a higher antagonistic co-contraction in the proximal and distal limb segments. CoF sway path was also positively correlated to the time needed to accomplish the gait test at maximal gait speed and TUG (Figure 5). No significant interrelations were observed for CoF path length and chair-rising time (Figure 5).

Complete inactivity during bed-rest, in contrast, led to postural and locomotor deficits and a reduction in functional mobility with a persistence of 2 to 4 weeks after re-ambulation, as confirmed by other studies (Haines, 1974; Dupui et al., 1992; Viguier et al., 2009; Muir et al., 2011). Considering the massive decline in postural equilibrium and its negative impact on gait dynamics, and the time to accomplish TUG predicting 69–82% of its variability, it is not surprising that after bed-rest the inactive control group showed locomotor inconsistencies and abnormalities in muscle coordination far from the normative locomotor values of a healthy human (Kerrigan et al., 1995; Whittle, 1996; Lai et al., 2012; Gouelle and Megrot, 2018).

First, an increased sway path and 90% standard ellipse area was manifested concomitant with an increased co-contraction reflected in simultaneously activated antagonistic muscle groups encompassing the ankle and knee joint. Augmented co-contractions are related to a rigid articular stiffening and have been postulated as a safety strategy to enhance security during single limb support to narrow the risk of falling and injury (Hortobágyi et al., 2009; Nagai et al., 2011; Sayenko et al., 2012) while restricting the ability to react precisely to sudden postural perturbations (Allum et al., 2002; Tucker et al., 2008). It may therefore be assumed that the increased co-contraction is a protective mechanism utilized in difficult postural tasks by individuals suffering from muscle weakness and fragility after chronic bed-rest (Kramer et al., 2017a,b), however, leading to an augmented postural sway (Hortobágyi et al., 2009; Nagai et al., 2011) (Figure 4). The condition with eyes open and closed, and the frontal and sagittal trajectories, were similarly affected by bed-rest. The scope of the finding is seen in its transfer effects extending to dynamic movement: mastering body equilibrium is apparently a fundamental prerequisite for various daily movements and its proper degradation overlaps with complex cyclic movement, as indicated by the correlation between the increased sway path and reduced gait speed or time to accomplish TUG (Figure 5).

Second, the bipedal locomotor pattern of the inactive control group after bed-rest was of pathological significance. Findings were independent of the modalities preferred or maximal gait speed. The gait deficits included spatiotemporal, kinematic and variability adaptations that have been empirically identified as predisposing a person to a greater gait instability (Malatesta et al., 2003) and subsequently to an increased risk of fall (Whittle, 1996; Maki, 1997; Gouelle and Megrot, 2018). With an emphasis on gait anomalies, the findings of the current study are of clinical relevance: (i) a decline in pace characterized by gait speed, step time and length is associated with reduced executive function and performance (Watson et al., 2010); (ii) rhythm changes characterized by cadence, swing stance time, and double and single limb support, are related to higher fall rates (Verghese et al., 2009); (iii) reduced foot-to-ground clearance and knee flexion during the leg swing phase can cause tripping and an increased fall incidence (Lai et al., 2012); and an increased vertical COM excursion caused by a smaller plantarflexion at push off has been linked to energetic inefficiency (Kerrigan et al., 1995). An increase in the spatiotemporal variability domain—which was also observed in our study—has been identified as the best predictor of future falls (Whittle, 1996; Gouelle and Megrot, 2018). The aforementioned gait abnormalities detected after 60 days of bed-rest in the inactive control group are typically known in the elderly population (Hollman et al., 2011) or in patients with neurological disorders (Gouelle and Megrot, 2018).

Functional mobility assessed by TUG and repeated chair-rises is impaired. The time required to accomplish the test was increased after bed-rest for both testing modalities. Chair rises were also executed with reduced peak power, indicting a reduced power-generating capacity and underscoring the results of Kramer et al. (2017b), who similarly demonstrated these degradations for countermovement jumps and ballistic movement.

The impact of the jump exercise during bed-rest was of major significance in the weeks after re-ambulation. Interaction effects demonstrate significant differences between both groups after the end of bed-rest and its benefits equally affect posture control, gait, and functional mobility: while the entire recovery phase is characterized by comparable values over time that differ only marginally to baseline in JUMP, great differences in comparison with the inactive CTRL were established for R+0 and R+7 in the tests, including posture control, gait and functional mobility. Concomitantly, neuromuscular control of the skeletal muscle with an emphasis on antagonistic coordination could be preserved throughout the high-intensity jump exercise performed during bed-rest. Other bed-rest studies validating countermeasures such as strength training (Haines, 1974; Kouzaki et al., 2007), flywheel (Viguier et al., 2009), and treadmill (Macaulay et al., 2016), lower-body negative-pressure (Dupui et al., 1992), mechanical stimulation (Muir et al., 2011) or centrifugation (Vernikos et al., 1996), found plyometric jump training to be advantageous compared to past alternatives. Despite differing HDT periods ranging from 5 days to 12 weeks, none of these countermeasures succeeded in entirely preserving gait, posture and functional mobility during bed-rest (Haines, 1974; Dupui et al., 1992; Vernikos et al., 1996; Kouzaki et al., 2007; Macaulay et al., 2016) despite those that permitted daily upright stance, and used standing and walking as exercise modes (Mulder et al., 2014).

A reasonable explanation as to why the countermeasure proved advantageous above other interventions applied during long-term bed-rest (Dupui et al., 1992; Koppelmans et al., 2015; Paloski et al., 2017), with significant treatment effects beyond the actual exercise mode (horizontal jumps), may involve the specific attributes of the countermeasure associated with the preservation of muscle mass and function (Kramer et al., 2017a,b). First, jumps are whole-body movements, which have been shown to elicit larger performance improvements in the lower extremities than segmental single-articular movements (Blackburn and Morrissey, 1998; Stone et al., 2002). Second, jumping is an exercise mode where each repetition requires maximal effort, resulting in exceedingly high forces which are multiples of those occurring during common physical activity such as cycling, stepping or strength training (Komi, 1984). To achieve these peak forces, agonistic muscles need to be contracted entirely while antagonists should be inhibited (Kellis et al., 2003) to reduce co-contraction. Third, jumping relies on the stretch-shortening cycle, which is a natural type of muscle action found in everyday activities such as running, walking and skipping (Taube et al., 2012). An overlap with locomotor movement such as gait and TUG, including the particular neural pattern of synergists and antagonists, may have caused positive effects in the locomotor and mobility test established in our study.

Areas of application for JUMP may range from space-related operations for Astronauts during long-term space missions (Kramer et al., 2017a) to the interface of geriatrics (McGregor et al., 2014), clinical orthopedics (Bugbee et al., 2016), and neurodegeneration (Azizi et al., 2017). Importantly, JUMP can be executed with a wide range of impact loads induced by the sledge's acceleration profile as the equivalent to gravitation within boundaries of 0.5 g up to 1.3 g (Kramer et al., 2017b). Thus, the trainings intensity can be adjusted to the individuals' health and fitness status which allows many different application options to counteract deconditioning induced by inactivity.

Long-term recovery after 60 days bed-rest differed between JUMP and CTRL as indicated by group^*time interaction effects. The countermeasure successfully maintained the neuromuscular system's ability to safely control postural equilibrium and perform complex locomotor movements, and thus, values were mostly comparable to baseline at any point in the recovery phase (Figure 1). In contrast, analysis of the follow-up performance measurements in CTRL during the recovery period showed that even though the decline in functional mobility, posture and gait control was high, detrimental effects were reversible and the recovery was almost complete 1 month after re-ambulation. This is valid for a bed-rest period of 2 months, but would certainly differ for shorter or longer periods (Pavy-Le Traon et al., 2007). Importantly, the range and timescale of the neuromuscular parameters were almost identical to those manifested for the functional deficits. Coupled with positive correlations for changes in posture control with antagonistic co-contraction of the limb musculature, these findings support our hypothesis that neuromuscular mechanisms may underlie functional degradations in addition to the loss in muscle mass (Kramer et al., 2017a,b), and may explain the potential benefit of the selected countermeasure. It is worth discussing the delayed recovery characteristics of the lateral CoF displacement and COM trajectories in gait, which take up to 3 months to return to pre-bed-rest levels. Both factors are associated with the control of the trunk which is the body segment with the largest mass and highest moment of inertia, and are crucial for COM reposition above the base of support (Stapley et al., 1999; You et al., 2001). It is therefore assumed that disregarding the different test paradigms individuals experience great and prolonged difficulties in mastering trunk movement so as to diminish disturbing torques and destabilizing the human body (Kerrigan et al., 1995; Stapley et al., 1999) after bed-rest.

For a conclusive statement, it is crucial to consider the limitations of the study. Two aspects are of substantial importance; the first deals with the study design, and the second with the methodological approach. Although this study provides scientific evidence for exercise prescription in bed-ridden individuals, the application and transferability of the protocol to the clinical context, nursing homes or in spaceflight needs to be specified. The sample size was restricted to 24 young and healthy individuals due to restrained temporal, logistical and financial resources. The limited number of participants may have resulted in non-significant findings regarding the effect of the exercise protocol. Further, in an effort to reduce inter-subject variability, only the male gender was considered for participation in this bed-rest study. Therefore, with reference to Viguier et al. (2009), we are therefore not certain whether our findings are applicable to women or to people with illness suffering from significant health impairments. The methodology used electromyograms of the relevant leg muscles normalized to pMVCs in order to control for changes in fluid shift, skin impedance, electrode positioning or lean or fat mass that would have interfered with changes in muscle activation in response to physical activity and jump training. pMVCs were executed manually in less controlled conditions than recommended in the current literature, however, and thus they should be seen as a limitation (Burden, 2010).

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.