# MESOTHELIAL PHYSIOLOGY AND PATHOPHYSIOLOGY

EDITED BY : Sotirios G. Zarogiannis PUBLISHED IN : Frontiers in Physiology

#### Frontiers Copyright Statement

© Copyright 2007-2018 Frontiers Media SA. All rights reserved. All content included on this site, such as text, graphics, logos, button icons, images, video/audio clips, downloads, data compilations and software, is the property of or is licensed to Frontiers Media SA ("Frontiers") or its licensees and/or subcontractors. The copyright in the text of individual articles is the property of their respective authors, subject to a license granted to Frontiers.

The compilation of articles constituting this e-book, wherever published, as well as the compilation of all other content on this site, is the exclusive property of Frontiers. For the conditions for downloading and copying of e-books from Frontiers' website, please see the Terms for Website Use. If purchasing Frontiers e-books from other websites or sources, the conditions of the website concerned apply.

Images and graphics not forming part of user-contributed materials may not be downloaded or copied without permission.

Individual articles may be downloaded and reproduced in accordance with the principles of the CC-BY licence subject to any copyright or other notices. They may not be re-sold as an e-book.

As author or other contributor you grant a CC-BY licence to others to reproduce your articles, including any graphics and third-party materials supplied by you, in accordance with the Conditions for Website Use and subject to any copyright notices which you include in connection with your articles and materials.

All copyright, and all rights therein, are protected by national and international copyright laws.

The above represents a summary only. For the full conditions see the Conditions for Authors and the Conditions for Website Use. ISSN 1664-8714 ISBN 978-2-88945-541-6 DOI 10.3389/978-2-88945-541-6

## 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

# MESOTHELIAL PHYSIOLOGY AND PATHOPHYSIOLOGY

Topic Editor: Sotirios G. Zarogiannis, University of Thessaly, Greece

The mesothelium is composed by a single layer of mesothelial cells that vest the serosal cavities (pleural, peritoneal and pericardial) and internal organs of the body. The mesothelial cells have a mixed phenotype of epithelial cells and fibroblasts rendering them remarkable plasticity. Besides providing a slippery surface for the frictionless movement of internal organs, the mesothelium participates in a wide range of physiological and pathophysiological processes. Some of its functions include lung development, trans-cellular and para-cellular transport of ions and water, secretion of glycoproteins (mainly hyaluronan), secretion of cytokines and growth factors, wound healing, response to inflammatory stimuli and induction of inflammation, mesothelial to mesenchymal transition and formation of tunneling nanotubes. Many of these functions are pivotal to physiological conditions such as respiratory development, maintenance of steady volume of serosal fluids and serosal permeability, cell-to-cell communication, re-mesotheliazation of serosal membranes after mechanical (e.g. by asbestos or nanoparticles) or inflammatory injury and participation in immune responses. Deviation from the physiological threshold of these functions results in the development of serosal effusions, induction of serosal and lung fibrosis, induction of mesothelial tumorigenesis, leading thus to devastating pathologies. Treatment of pathologies like mesothelioma, pleural and peritoneal fibrosis (in cases of patients under Peritoneal Dialysis) or lung fibrosis still pose a great challenge for researchers.

Citation: Zarogiannis, S. G., ed (2018). Mesothelial Physiology and Pathophysiology. Lausanne: Frontiers Media. doi: 10.3389/978-2-88945-541-6

# Table of Contents


Zia Moinuddin, Angela Summers, David Van Dellen, Titus Augustine and Sarah E. Herrick


William Tse, Cerasela-Zoica Dinu and Yon Rojanasakul

## Editorial: Mesothelial Physiology and Pathophysiology

#### Sotirios G. Zarogiannis\*

*Department of Physiology, Faculty of Medicine, BIOPOLIS, University of Thessaly, Larissa, Greece*

Keywords: mesothelium, pleura, peritoneum, pericardium, physiology, serosal membranes, mesothelial cells

## **Editorial on the Research Topic**

### **Mesothelial Physiology and Pathophysiology**

The Research Topic "Mesothelial Physiology and Pathophysiology" aimed at providing a forum for investigations conducted in all fields of serosal membranes research. The physiology and pathophysiology of mesothelial cells and membranes is a research field with a growing community given that many of the clinical entities that involve serosal contribution, require the detailed understanding of the underlying biological processes in order to provide effective treatments. The research production volume in the field of mesothelial physiology and pathophysiology is a function of the frequency of the clinical entities that they are involved into. Thus, the most active area is pleural research followed by peritoneal and lastly pericardial. Pleural effusions are a common entity that involves the abnormal accumulation of pleural fluid in the pleural cavity due to abnormal turnover, and the underlying diseases stem from congestive heart failure to infectious lung diseases and several intra- and extra- thoracic malignancies (Zarogiannis and Kalomenidis, 2015). It is therefore easily conceived that the water and solute transport systems of the pleural membrane comprising the transcellular and paracellular mesothelial permeability can be altered in pathophysiological conditions. These mechanisms are thoroughly reviewed in this Research Topic with a special focus on the tight junction component that regulates the paracellular permeability of the pleural mesothelium (Markov and Amasheh). The authors provide a detailed description of the electrophysiological studies that yield the knowledge around the transmesothelial permeability and the set of studies on paracellular permeability that provide significant evidence of tight junction switch in the inflamed human pleural mesothelium. During inflammation a protein expression change from sealing to pore forming claudins in the tight junctions occurs, a finding that could be prone to therapeutic intervention. In the same context an experimental study demonstrated the calcium dependent effects of increased paracellular permeability induced by bradykinin, histamine and thrombin in rat primary mesothelial monolayers (Kuwahara). This study provides further evidence of paracellular permeability changes due to the effects of inflammation related molecules attributable to reorganization of mesothelial cell F-actin cytoskeleton and increased actin polymerization. These aspects of pleural membrane permeability changes can supplement the understanding of the pathophysiology of pleural effusion dynamics.

A Perspective on the pleural mesothelium in development and disease provided an overview on the mesothelial contribution of lung development and also highlighted the ability of mesothelial cells to differentiate into several types of cells depending on the underlying stimuli (Batra and Antony). This process renders new roles to pleural mesothelial cells in the context of lung injury and repair mechanisms that can contribute to the development of lung mesenchymal pathologies such as interstitial pulmonary fibrosis. The authors also focus on the therapeutic potential of modulation pleural mesothelial activation and thus parenchymal disease progression. From a different angle the pleural mesothelium is sensitive to the effects of environmental or engineered nanoparticles that through inhalation reach the pleural cavity by disrupting the lung parenchyma and transmigration

#### Edited by:

*Mario Diaz, Universidad de La Laguna, Spain*

#### Reviewed by:

*Kausar Begam Riaz Ahmed, United States Food and Drug Administration, United States*

> \*Correspondence: *Sotirios G. Zarogiannis szarog@med.uth.gr*

#### Specialty section:

*This article was submitted to Membrane Physiology and Membrane Biophysics, a section of the journal Frontiers in Physiology*

> Received: *05 February 2018* Accepted: *07 March 2018* Published: *22 March 2018*

#### Citation:

*Zarogiannis SG (2018) Editorial: Mesothelial Physiology and Pathophysiology. Front. Physiol. 9:263. doi: 10.3389/fphys.2018.00263* or in the case of smaller particles through the circulation. A well characterized such pathology is the malignant pleural mesothelioma (MPM) due to asbestos fibers exposure. Currently there is increasing evidence that engineered nanoparticles such as carbon nanotubes (CNTs) can cause a similar pleural pathology. Lohcharoenkal et al. demonstrated that single-walled CNTs induced neoplastic-like transformation in human mesothelial cells after prolonged exposure. The observed effects were due to H-Ras upregulation and ERK1/2 activation that led to increased expression of cortactin, a protein implicated in cell motility and neoplastic development, and thus a more aggressive transformation phenotype (Lohcharoenkal et al.). In another experimental study Ady et al. investigated the role of Tunneling Nanotubes (TnTs) in MPM cell communication (Ady et al.). The authors in very well designed studies demonstrated TnT occurrence in ex vivo preparations of MPM patient tumor tissues as well as in 2D and 3D MPM cell cultures. Furthermore they provided some interesting preclinical evidence of decreased tumor growth and survival of immunodeficient mice implanted with TnT-primed human MPM cells.

TnTs between peritoneal mesothelial cells in physiological and pathophysiological conditions in the peritoneal cavity has also been the theme of a Mini Review article (Ranzinger et al.). The research in the peritoneal mesothelium has stemmed from questions that have risen due to its use as a dialyzing membrane during the Peritoneal Dialysis (PD) modality in End Stage Renal Disease patients. In such a modality the peritoneal membrane is in a chronic state of inflammation caused by the PD fluids that are daily introduced in the peritoneal cavity. The authors discuss the influence of several insults to the mesothelial cells that are due to the effects of PD fluids, such as increased oxidative stress or increased osmotic stress, that can lead to differences in the mesothelial cells propensity to exhibit intercellular communication via TnTs and highlight the open questions in the field of PD and mesothelial cells TnTs.

## REFERENCES

Zarogiannis, S.G., and Kalomenidis, I. (2015). Can pharmacologic agents speed the rate of resorption of pleural fluid? Curr. Opin. Pulm. Med. 21, 372–375. doi: 10.1097/MCP.0000000000000170

**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.

Pertinent to this is also the Review article of Moinuddin et al., regarding a severe form of peritoneal related disease which is the Encapsulating Peritoneal Sclerosis (EPS). The authors provide a global presentation of EPS in terms of etiology, risk factors, diagnosis and the underlying pathophysiology along with all the available treatment and prevention options. Another manuscript regarding PD was a Mini Review on the available in vivo and ex vivo animal models for the study of PD effects on the peritoneal mesothelial membrane (Nikitidou et al.). Nikitidou et al. critically discuss the advantages and disadvantages of each model and present the parameters studied in each one concluding in the need for establishment of standardized protocols in order to yield clinically relevant and applicable results.

Pericardium is the most neglected research field of mesothelial membranes. A Mini Review that filled an important gap in the literature was centered on the physiology of pericardial fluid production and drainage (Vogiatzidis et al.). The authors provide detailed information on the anatomy and histology of the pericardium, the composition of the pericardial fluid and the physiology underlying its dynamic turnover in the pericardial cavity.

## AUTHOR CONTRIBUTIONS

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

## ACKNOWLEDGMENTS

Since this is the first collection of articles published in a physiology journal that focuses on Mesothelial Physiology and Pathophysiology, I would like to deeply thank all the contributing authors, the reviewers of the manuscripts and the staff of Frontiers in Physiology for their excellent work and dedication.

Copyright © 2018 Zarogiannis. 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.

## Animal models in peritoneal dialysis

Olga Nikitidou<sup>1</sup> , Vasiliki I. Peppa<sup>2</sup> , Konstantinos Leivaditis <sup>1</sup> , Theodoros Eleftheriadis <sup>1</sup> , Sotirios G. Zarogiannis <sup>2</sup> and Vassilios Liakopoulos <sup>1</sup> \*

*<sup>1</sup> Division of Nephrology and Hypertension, 1st Department of Internal Medicine, School of Medicine, American Hellenic Educational Progressive Association Hospital, Aristotle University of Thessaloniki, Thessaloniki, Greece, <sup>2</sup> Department of Physiology, Faculty of Medicine, BIOPOLIS, University of Thessaly, Larissa, Greece*

Peritoneal dialysis (PD) has been extensively used over the past years as a method of kidney replacement therapy for patients with end stage renal disease (ESRD). In an attempt to better understand the properties of the peritoneal membrane and the mechanisms involved in major complications associated with PD, such as inflammation, peritonitis and peritoneal injury, both *in vivo* and *ex vivo* animal models have been used. The aim of the present review is to briefly describe the animal models that have been used, and comment on the main problems encountered while working with these models. Moreover, the differences characterizing these animal models, as well as, the differences with humans are highlighted. Finally, it is suggested that the use of standardized protocols is a necessity in order to take full advantage of animal models, extrapolate their results in humans, overcome the problems related to PD and help promote its use.

#### Edited by:

*Ali Mobasheri, University of Surrey, UK*

Reviewed by:

*Martin Diener, Giessen University, Germany David Marples, University of Leeds, UK*

#### \*Correspondence:

*Vassilios Liakopoulos, Division of Nephrology and Hypertension, 1st Department of Internal Medicine, School of Medicine, American Hellenic Educational Progressive Association Hospital, Aristotle University of Thessaloniki, 1 St. Kyriakidi street, 54636, Thessaloniki, Greece liakopul@otenet.gr*

#### Specialty section:

*This article was submitted to Membrane Physiology and Membrane Biophysics, a section of the journal Frontiers in Physiology*

> Received: *21 June 2015* Accepted: *17 August 2015* Published: *01 September 2015*

#### Citation:

*Nikitidou O, Peppa VI, Leivaditis K, Eleftheriadis T, Zarogiannis SG and Liakopoulos V (2015) Animal models in peritoneal dialysis. Front. Physiol. 6:244. doi: 10.3389/fphys.2015.00244* Keywords: animal models, ion and water transport, membrane physiology, peritoneal dialysis, ussing chamber

Peritoneal dialysis (PD) is a well-established method of kidney replacement therapy. It is currently estimated that the end stage renal disease (ESRD) patients receiving PD worldwide are approximately 200,000 and that they represent 11% of the total number of patients receiving any modality of dialysis (Jain et al., 2012). It is worth mentioning that Putman in 1923 made the first attempt for PD in a canine model (Putnam, 1923).

The extended use of PD over the years makes the in depth understanding of the physiology of the peritoneal membrane, the transport of solute and water across it and the mechanisms influencing inflammation, peritonitis, peritoneal injury and encapsulating peritonitis, a necessity. Moreover, many therapeutic interventions, new PD solutions, and pharmaceutical agents need to be tested before their implementation in the daily clinical practice. Research on humans could probably provide the necessary answers to the aforementioned topics, but it is complicated with many technical problems and ethical concerns (Stojimirovic et al., 2007). The scientific community in order to overcome this problem has exploited the observed similarities among humans and animal models. More precisely, the similar transport properties of solute and water across the peritoneal membrane in humans and animal models was an important foundation for the conduction of series of experiments in a variety of animal models.

Small animals such as rats, rabbits, and genetically modified mice have mostly been used for experimental models, although larger animals such as dogs, sheep, or even kangaroos have been used, as well. The use of each animal model comes with different advantages and disadvantages (**Table 1**). More precisely, rats are easy to breed and affordable, but have a short life expectancy, small size that increases the risk of complications during the insertion of a peritoneal catheter and large size of peritoneal membrane when at the same time only small quantities of dialysate can be introduced in their peritoneal cavity (Lameire et al., 1998). On the other hand, rabbits have a longer life expectancy, can survive longer on PD and the insertion of the peritoneal catheter is easier,

**6**


but are very delicate animals and difficult to breed (Garosi and Di Paolo, 2001). Regarding larger animals, their life expectancy is long and could approximate PD on humans, but their breeding is difficult and expensive and a large time frame is needed for results to be obtained (Van Biesen et al., 2006).

As for the genetically modified mice, both knockout and transgenic models have been used over the past few years in order to provide important information in many different molecular pathways that are significant for PD (Nishino et al., 2007). The ability to modify the genome of mice enabled the scientists to explore the expression of different genes and the results of the modification of their expression. The role of single proteins such as aquaporin-1 in the transport of water across the peritoneal membrane (Yang et al., 1999; Ni et al., 2006) or NOS (Nitric oxide synthase) isoforms in peritonitis (Ni et al., 2010), as well as, IL-6 in inflammation (Hurst et al., 2001; McLoughlin et al., 2004) and TGF-β in encapsulating peritonitis (Park et al., 2008) were explored using genetically modified mice. The extremely small size of these animal makes manipulations difficult (Ni et al., 2003), but over the past few years the development of new molecular techniques provide significant tools to overcome this disadvantage (Devuyst et al., 2010).

Acute and chronic animal models have been developed (Hoff, 2005; Pawlaczyk et al., 2015) (**Table 2**). Acute models describe experiments that have short duration and can study the impact of a single dwell on the peritoneal membrane. They provide crucial information regarding mainly the transport characteristics of the peritoneal membrane and its function. Its permeability to water and solute, the interaction with different dialysis solutions, certain substances, therapeutic agents or even inflammation mediators can be extensively studied this way. On the other hand, chronic models can evaluate the chronic impact of PD resembling the impact of the method on humans due to the longer duration of the experiments. They can focus on changes of the structure and the function of the peritoneal membrane. For example, the effect of glucose used as an osmotic agent in dialysis solutions or the effect of the newer biocompatible solutions with lower GDPs' (Glucose degradation products) concentration, neutral pH or replacement of glucose with icodextrin on the morphology of the peritoneal membrane can be analyzed. Fibrosis, encapsulating peritoneal sclerosis, repeated episodes of peritonitis, and morphological alterations can be in depth studied using the chronic models. In this way, an important background for targeted therapeutic interventions is provided.

In the acute model the animal is anesthetized, a temporary catheter is inserted in the peritoneal cavity and during a 4-h dwell samplings of the peritoneal fluid and blood are obtained (Lameire et al., 1998). In the chronic model, on the other hand, three are the most common methods used (Mortier et al., 2005). The first method includes the direct blind injection of the dialysate into the peritoneal cavity with or without anesthesia. The repeated injections are related to increased risk of intraperitoneal bleeding and infection that could probably interfere with the results of the experiment. The second method is called "open system" and is characterized by the subcutaneous insertion of a permanent indwelling peritoneal catheter from the neck to the peritoneal cavity of the animal. This catheter can be used for the introduction of dialysate into the peritoneal cavity and the drain of the dwell directly or indirectly as it can be used as the tunnel through which another sterile catheter is inserted during every exchange. In this method anesthesia is not required but the rate of peritonitis episodes remains high and the malfunction of the catheter as a result of omental wrapping, adhesions and fibrosis is frequent. Finally, the third method is called "closed system." In this method, as in the "open-system," a permanent indwelling peritoneal catheter is inserted subcutaneously from the neck to the peritoneal cavity of the animal. The catheter is attached to a subcutaneous reservoir in the neck of the animal and the dialysate remains in the peritoneal cavity until it is fully absorbed. This method results in lower rates of peritonitis,

TABLE 2 | The main studied parameters in acute and chronic animal models.


*PM, peritoneal membrane.*

although the malfunction of the catheter still remains a problem.

In any case, even if it is tempting, before applying the extrapolated results of animal models in humans we should point out the differences between these models and humans and the related technical problems. The most important differences concern the visceral, parietal and diaphragm surfaces. It seems that the diaphragm, which is directly related to the lymphatic drainage responsible for the clearance of macromolecules, extends to a larger area in humans than in animal models (Pawlaczyk et al., 2015). On the other hand, the parietal peritoneum that is directly related to the solute transport is larger in rats than in humans, while in rabbits it is similar to humans (Pawlaczyk et al., 1996). Moreover, it is known that the intraperitoneal levels of amylase that are elevated in rats can locally degrade icodextrin, an osmotic agent used in several dialysis fluids, rapidly (de Waart et al., 2001). To these differences one must add the higher ratio of peritoneal surface area to dialysate that is observed in rats, while in rabbits it is similar to the one observed in humans (Stojimirovic et al., 2007). Finally, in rats changes in the permeability and the surface area of the peritoneal membrane are encountered as they grow up, resulting in changes in the kinetics of the peritoneal membrane (Kuzlan et al., 1997). As Van Biesen et al highlights multiple experiments in different animal models and comparison of the results obtained could overcome those differences and provide a more realistic approach of the mechanisms that characterize PD in humans (Van Biesen et al., 2006).

Among the technical problems that have to be faced when handling animal models, tissue-sampling problems must be highlighted. The peritoneal tissue is quite fragile, dries quickly when exposed to the air and can easily be damaged even after delicate contact with surgical gloves or instruments (Duman and Sen, 2009). Furthermore, in patients on PD the visceral peritoneum exhibits more pronounced alterations than the parietal peritoneum, while in many of the studies conducted in animal models the biopsy samples are obtained by the parietal peritoneum. This inconsistency could be misleading if extrapolating results of animal studies to humans.

Catheter obstruction, peritonitis, and anesthesia use represent other major problems. In an attempt to minimize catheter dysfunction, omentectomy, and heparin have been used, but both methods have important pitfalls. On one hand the omentum is considered to be an important immune defense organ of the peritoneal cavity (Beelen, 1992), while on the other heparin is known to have many pleiotropic actions besides being an anticoagulant (De Vriese et al., 2001; Gozdzikiewicz et al., 2009). Therefore, both of these interventions could result in undesirable functional and morphological changes in the peritoneum (Mortier et al., 2005). The use of heparin coated catheters that has been suggested seems to be a good solution, since it reduces the rate of malfunction without inducing changes in the peritoneum (De Vriese et al., 2002). Regarding peritonitis, many investigators have administered antibiotics in order to prevent infections, an intervention that has been proven successful and simultaneously not leading to morphological and function changes in the peritoneum (Mortier et al., 2003; Choi et al., 2006). Finally, the use of anesthesia when handling animals has been related to changes in the peritoneum kinetics, probably through a direct reduction of the lymphatic drainage (Tran et al., 1993) and an increase of ultrafiltration and decrease of the absorption rates from the peritoneal membrane (Shin et al., 2006).

There are some important differences in animal model studies (Mortier et al., 2005; Topley, 2005). There are differences related to the volume of dialysate instilled and the time it remains in the peritoneal cavity, the frequency of the exchanges, and the total time the animals undergo PD before the termination of the study. The drainage of the dialysate is also an issue. More precisely, there are animal models characterized by the presence of the dialysate in the peritoneal cavity until it is fully absorbed, while in other animal models drainage after every dwell is performed. The last model resembles chronic PD in humans and offers the ability to better understand transportation through the peritoneal membrane and at the same time examine the dialysate for the presence of cells and substances (Pawlaczyk et al., 2015). Moreover, differences in the use of anesthesia, heparin or even the definition of peritonitis have been described. A common finding in most studies is the use of high concentration glucose solutions although in humans other lower glucose concentration solutions may also be used. Finally, the high rate of dropout due to catheter malfunction must be highlighted. It is obvious that there are many limitations when analyzing the results of the studies conducted on animal models, nevertheless the information they provide are very important and necessary to better understand this method of renal replacement therapy.

One field of special interest for nephrologists is encapsulating peritoneal sclerosis (EPS) which is a rare but life threatening complication of PD. Several animal models have been used in order to understand the underlying pathophysiological mechanisms that lead to EPS (Hoff, 2005; Park et al., 2008; Hirahara et al., 2015). Chlorexidine and acidic dialysis solutions, povidone iodine, formaldehyde, and even bleach and talc are some of the agents that have been used in an attempt to induce peritoneal injury and development of EPS in animal models. At the same time many different inhibitors of inflammation, angiogenesis, and fibrosis, as well as, inhibitors of the renin angiotensin system have been studied as potential therapeutic interventions in an attempt to resolve EPS (Kitamura et al., 2014).

The in vivo animal models provide useful information for future PD solutions that are yet to be tested or promote our understanding of their effects on the peritoneal membrane of ESRD patients undergoing PD. To this end an ex vivo sheep and human model of the peritoneal ionic and water permeability can further provide useful information that can be applied in the PD solution improvement and development area (Liakopoulos et al., 2006, 2009). In these studies the membrane specimens are mounted in the Ussing System that measures the transmembrane ionic permeability (Zarogiannis et al., 2004; Stefanidis et al., 2005). The majority of such investigations involved electrophysiological experiments in sheep visceral or parietal peritoneum specimens and some of the results have been tested in human specimens yielding identical findings, thus rendering this model a powerful tool for PD solution assessment as to the ion and water transport drug induced alterations (Zarogiannis et al., 2005, 2007; Stefanidis et al., 2007). The focus of these electrophysiological investigations was the identification of ion channels, as well as, hormonal receptors present in the mesothelial compartment of the peritoneum (Kourti et al., 2007, 2013; Zarogiannis et al., 2007; Karioti et al., 2008, 2009). Most of these findings remain yet to be investigated in human specimens and subsequently tested in in vivo animal models.

In conclusion, in vivo and ex vivo animal models have been extensively used for the understanding of the properties of the peritoneal membrane and results of these experiments have been extrapolated in humans. For the time being, the fundamental differences in the animal models used, as already described, remains the main problem when it comes to comparing results of different models and especially using these results in order to evaluate the use of PD in humans. At the same time, the necessity of in depth understanding of PD brings into perspective the importance of establishing standardized protocols that could overcome the aforementioned differences and obstacles.

## References


switch in the pattern of leukocyte recruitment seen during acute inflammation. Immunity 14, 705–714. doi: 10.1016/S1074-7613(01)00151-0


**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 © 2015 Nikitidou, Peppa, Leivaditis, Eleftheriadis, Zarogiannis and Liakopoulos. 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.

## Physiology of pericardial fluid production and drainage

Konstantinos Vogiatzidis <sup>1</sup> , Sotirios G. Zarogiannis <sup>1</sup> \*, Isaac Aidonidis <sup>1</sup> , Evgeniy I. Solenov <sup>2</sup> , Paschalis-Adam Molyvdas <sup>1</sup> , Konstantinos I. Gourgoulianis <sup>3</sup> and Chrissi Hatzoglou<sup>1</sup>

<sup>1</sup> Department of Physiology, Faculty of Medicine, School of Health Sciences, University of Thessaly, Larissa, Greece, <sup>2</sup> Laboratory of Physiological Genetics, Institute of Cytology and Cytogenetics, Siberian Branch of the Russian Academy of Sciences, Novosibirsk, Russia, <sup>3</sup> Department of Respiratory Medicine, Faculty of Medicine, School of Health Sciences, University of Thessaly, Larissa, Greece

The pericardium is one of the serosal cavities of the mammals. It consists of two anatomical structures closely connected, an external sac of fibrous connective tissue, that is called fibrous pericardium and an internal that is called serous pericardium coating the internal surface of the fibrous pericardium (parietal layer) and the heart (visceral layer) forming the pericardial space. Between these two layers a small amount of fluid exists that is called pericardial fluid. The pericardial fluid is a product of ultrafiltration and is considered to be drained by lymphatic capillary bed mainly. Under normal conditions it provides lubrication during heart beating while the mesothelial cells that line the membrane may also have a role in the absorption of the pericardial fluid along with the pericardial lymphatics. Here, we provide a review of the the current literature regarding the physiology of the pericardial space and the regulation of pericardial fluid turnover and highlight the areas that need to be further investigated.

#### Edited by:

Ali Mobasheri, University of Surrey, UK

#### Reviewed by:

Martin Diener, University Giessen, Germany Salah Amasheh, Freie Universität Berlin, Germany

#### \*Correspondence:

Sotirios G. Zarogiannis, Department of Physiology, Faculty of Medicine, School of Health Sciences, University of Thessaly, 3rd Panapistimiou Str., BIOPOLIS, 41500 Larissa, Greece szarog@med.uth.gr

#### Specialty section:

This article was submitted to Membrane Physiology and Membrane Biophysics, a section of the journal Frontiers in Physiology

> Received: 01 December 2014 Accepted: 16 February 2015 Published: 18 March 2015

#### Citation:

Vogiatzidis K, Zarogiannis SG, Aidonidis I, Solenov EI, Molyvdas P-A, Gourgoulianis KI and Hatzoglou C (2015) Physiology of pericardial fluid production and drainage. Front. Physiol. 6:62. doi: 10.3389/fphys.2015.00062 Keywords: mesothelium, pericardiac fluid turnover, pericardium, serosal membranes, transmembrane transport

## Introduction

The pericardium is one of the serosal cavities of mammals (Michailova and Usunoff, 2006). It is a fibrous—serosal conical sac enclosing the roots of the aorta and the pulmonary artery (Chinchoy, 2005). In humans, pericardium is located inside the middle mediastinum posteriorly to the sternum and the cartilages of the third to seventh left rib. Normally, it is not in contact with the frontal wall of the thoracic cavity (Frick et al., 1985; Chinchoy, 2005). Laterally, it is held together with the mediastinal parietal pleura. Pericardium isolates the heart from the adjacent tissues, allowing it's free movement within the boundaries of the pericardial cavity and is filled with a small amount of fluid which is called pericardial fluid (Chinchoy, 2005).

## Anatomy and Histology

The pericardium consists of an external sac of fibrous connective tissue, called fibrous pericardium and an internal called serous pericardium. The latter coats the internal surface of the fibrous pericardium and the heart. Arterial branches from thoracic aorta, right and left pericardiophrenic artery (internal mammary artery branches), are responsible for the blood supply of the whole pericardium while the venous drainage is accomplished through the venae pericardiales which drain to the azygos vein, to the superior vena cava or to the brachiocephalic (Chinchoy, 2005). The pericardium is innervated by the two phrenic nerves, each one giving an afferent branch (pericardial branch) (Randall and Ardell, 1985; Ardell and Randall, 1986; Chiou et al., 1997).

The thickness of the pericardium increases proportionally to the size of the heart and the pericardial cavity, with the exception of humans who have considerably thicker pericardium compared to the mammals with the same heart size (human 1–3.5 mm, sheep 0.32 ± 0.01 mm, pig 0.20 ± 0.01 mm) (D'Avila, 2003). The serous pericardial membrane covers the outside of the heart (visceral pericardium), extending a short distance beyond the atria and ventricles on the great vessels and lines the inside of the fibrous sac (parietal pericardium). The visceral pericardium is also refered to as epicardium and is in continuance with the parietal pericardium (Chinchoy, 2005; Jöbsis et al., 2007).

The parietal lamina of the serous pericardium is composed of a monolayer of flattened, squamous-like, mesothelial cells. Mesothelial cells rest on a thin basement membrane supported by connective tissue stroma in a narrow submesothelial space. The connective tissue stroma contains variously oriented layers of collagen fibrils and small elastic fibers (Ishihara et al., 1980; Mutsaers, 2002). The luminal surface of the mesothelial cells has well developed microvillous border with occasional cilia. The latter bear friction and increase the surface area for fluid transport. There are junctional complexes between adjacent mesothelial cells that consist of desmosomes, which reinforce intercellular adhesion and zonulae occludentes. All of these morphological characteristics form permeability barriers. Actin-like filaments are present in microvilli and in immediate subjacent regions of the cells. These filaments mediate changes on cell shape. Intermediate filaments are associated with desmosomes and form bundles in the perinuclear regions, which provide structural support to the cytoplasm (Ishihara et al., 1980).

Between the mesothelial cells of the parietal pericardium, not the epicardium, there are milky spots, similar to those in the omentum and mediastinal pleura, bulging toward the pericardial cavity. These structures are enclosed with cuboidal mesothelial cells (Takada et al., 1991; Mutsaers, 2002; Michailova and Usunoff, 2006). These openings provide direct access to the underlying submesothelial lymphatic system allowing rapid removal of fluid and cells from the pericardial cavity (Takada et al., 1991). Inside the pericardial cavity and submesothelial layers of the pericardium, there are resident macrophages, readily available in case of immunological response (Ishihara et al., 1980; Mutsaers, 2002; Michailova and Usunoff, 2006).

Histological studies on parietal pericardium of rodents have shown the presence of circular fenestrations ("pores"; diameter up to 50µm), which connect the pericardial cavity with the adjacent mediastinal pleural cavities, except the positions that are held by the adipose tissue (Nakatani et al., 1988; Mohrman and Heller, 2006). Both their stomata and lumen are covered with mesothelial cells and below them elastin and collagen fibers (Fukuo et al., 1988; Nakatani et al., 1988). The functional significance of the pericardial "pores" remains unclear, however labeled erythrocytes have shown to pass through them from one cavity to the other (Fukuo et al., 1988).

The epicardium has common morphological features with the parietal pericardium but there are some functional differences. On the laminal surface there is a monolayer of mesothelial cells, lying on a thin basal membrane (Mutsaers, 2002; Jöbsis et al., 2007). The proportion of cuboidal cells is greater in the epicardium than in the parietal pericardium (Michailova and Usunoff, 2006). Underneath the basal membrane there is a dense network of collagen and elastic fibers, full of hydroxyproline, which embryologically stems from extracardiac tissue and doesn't infiltrate the underlying myocardium. The fibers are parallel to each other, in multiple layers and crossing the heart surface from diagonally to vertically, compared to the myocardial cells (Simionescu et al., 1993; Jöbsis et al., 2007). These mechanical properties play a role in the residual stress and passive stiffness of the heart (Przyklenk et al., 1987; Jöbsis et al., 2007).

## Pericardial Fluid

The composition of the normal human pericardial fluid is difficult to define. All available data have been obtained either by cardiothoracic surgery patients or from animals. This probably compromises the data validity (Ben-Horin et al., 2005). However, the pericardial fluid is a plasma ultrafiltrate having specific characteristics just like the pleura fluid (Mauer et al., 1940; Holt, 1970). Volumetric studies have shown that the pericardial fluid volume is directly analogous to the animal size: in rabbits 0.4–1.9 mL, in dogs 0.5–2.5 mL and in adult humans about 20–60 mL (average 15–35 mL) (Vesely and Cahill, 1986; Ben-Horin et al., 2005).

Pericardial fluid coloring studies report that the fluid distribution inside the cavity is heterogeneous. The largest amount is inside the atrioventricular and the intraventricular sulcus, the superior and the transversal sinus, especially on the supine position (D'Avila, 2003). Nevertheless, there are some pharmacokinetic studies that show that the pericardial fluid is stirring up constantly and thus the supplement's composition is the same regardless the position (Chinchoy, 2005).

Regarding the cell population, studies in human normal pericardial fluid have shown the presence of a heterogenous cell population. There are mesothelial cells, lymphocytes (53%), glanulocytes (31%), macrophages (12%), eosinophils (1.7%), and basophils (1.2%). This means that the pericardial fluid "lymphocytosis" should always be under critical consideration and characterized as pathological only when it exceeds 60% of the whole cell population (Gibson and Segal, 1978a; Benhaiem-Sigaux et al., 1985).

The pericardial fluid is considered to be a plasma ultrafiltration product, like other serosal cavity fluids. Studies in greyhounds showed that the concentrations (in mmole kgr H2O−<sup>1</sup> ) of Na<sup>+</sup> (150.5 ± 0.72), Cl<sup>−</sup> (123.2 ± 0.71), Ca2<sup>+</sup> (1.92 ± 0.04), and Mg2<sup>+</sup> (0.85 ± 0.09) were lower in the pericardial fluid than in the plasma. On the contrary, the concentration of K<sup>+</sup> (3.81 ± 0.07) was higher than the plasma, which was attributed to the K<sup>+</sup> leakage from the myocardial interstitium toward the pericardial cavity, during systole (Holt, 1970; Gibson and Segal, 1978b). The protein concentration was also lower with different proportion of protein fractions; from higher to lower concentration being albumin, globulins, macroglobulins, and fibrinogen. Finally, the pericardial fluid osmomolarity was lower than the plasma (Gibson and Segal, 1978b). Given the net filtration gradients of the substances above, it is obvious that the normal pericardial fluid is a transudate (Holt, 1970; Gibson and Segal, 1978b). However the relatively high protein and LDH concentrations raise caution as far as the applicability of Light's criteria is concerned (Gibson and Segal, 1978b; Meyers et al., 1997; Burgess et al., 2002; Ben-Horin et al., 2005).

## Physiology

The normal pericardium contributes in important functions. In is necessary for: (1) lubricating the moving surfaces of the heart, (2) stabilizing the heart anatomic position, (3) isolating the heart from the adjacent anatomical structures, prohibiting the adhesion formation, the inflammatory or neoplastic extention, (4) limiting heart dilatation during diastole, reducing the endomyocardial tension, (5) preventing cardiac hypertrophy in pressure overload conditions, (6) reducing the right ventricular impulse work in left ventricular overload conditions, (7) the ventriculoatrial blood retrogression prevention during high enddiastolic ventricular pressures, (8) the preservation of the negative endothoracic pressure, which is crucial for the atria blood filling, (9) the nervous stimulation response and regulation of the cardiac frequency and arterial blood pressure, (10) the formation of a hydrostatic compensation system ensuring that enddiastolic pressure remains the same at all hydrostatic levels and the Frank–Starling mechanism is functional (Holt, 1970; Goto and LeWinter, 1990; Cinca and Rodriguez-Sinovas, 2000).

Many studies in both parietal and visceral pericardium (human, canine, pig) have shown the same mechanical properties, only with quantitative differences. In vitro studies in canine parietal pericardium demonstrated the presence of viscoelastic response, mainly attributed to the presence and the arrangement of the elastic and collagen fibers (Lee and Boughner, 1981, 1985). These parietal pericardium properties are responsible for (1) its participation to ventricular volume modulation, (2) the intraventricular interaction, and (3) the participation to the ventricular diastolic pressure/volume relatioships (Lee and Boughner, 1981; Maruyama et al., 1982; Takata et al., 1997; Gibbons Kroeker et al., 2003).

In vitro epicardial studies (humans, pigs), proved the presence of elastic properties that have been attributed to the composition and orientation of the connective tissue. Specifically, the human epicardium undergoes an oblong and circumferential systolic shortening, therebe the energy gathering during the diastolic period may contribute to the passive mechanical properties of the myocardium (Lorenz et al., 2000; Jöbsis et al., 2007). In this way, the epicardium seems to be participating in the ventricular end-diastolic volume control (Jöbsis et al., 2007).

## Hydrodynamics

Under normal conditions, the human pericardial cavity contains 20–60 ml of fluid (Chinchoy, 2005). Moreover, it has been determined that the whole amount of pericardial fluid drains through the lymphatic capillary bed every 5–7 h, in sheep (Yuan et al., 2000). The pericardial fluid volume is determined by the equilibrium between production and drainage. There is strong evidence that the pericardial fluid is derived by plasma ultrafiltration through the epicardial capillaries (and probably the parietal's pericardium), as well as a small amount of interstitial fluid from the underlying myocardium, during the cardiac circle (Stewart et al., 1997). The fluid drainage is mainly accomplished through the parietal pericardium lymphatic capillary bed (Yuan et al., 2000). Nevertheless, the whole procedure is not fully elucidated because of the difficulty to study pericardial fluid dynamics under normal conditions (Shabetai, 2003).

The fluid movement through the pericardial laminae is a hydrostatic/osmotic pressure equilibrium, between the microvasculature and the cavity (Yuan et al., 2000). As shown in **Figure 1**, according to the Starling equilibrium:

Fluid Movement = L × S[(Pcap − Pp)−σ × (5cap − 5p)]

P and 5 stand for hydrostatic and osmotic pressures inside the capillaries (cap) and pericardial cavity (p), respectively. L is a membrane conductance constant for liquids, S is the surface area and σ is the osmotic constant of the endothelium/interstitial tissue barrier for proteins (Yuan et al., 2000).

The P<sup>p</sup> measurement is challenging. The main problems for the direct measurement of the P<sup>p</sup> were the small cavity size (<0.34 mm) as well as the sac deformation that provoved by the catheter insertion. The P<sup>p</sup> depends on the pericardial fluid volume and accumulation velocity, the cardiac and respiration circle phase and the measurement position. The large amount and the fluid rapid accumulation results in rapid and great increase in Pp. This is not present during the gradual fluid accumulation, which allows greater ammounts of fluid without hemodynamic instability. This phenomenon is due to the viscoelastic properties of the parietal pericardium (Holt, 1970). During the diastole (laminae approach) the P<sup>p</sup> increases and during the systole it decreases (Smiseth et al., 1985; DeVries et al., 2001; Hamilton et al., 2004). The P<sup>p</sup> becomes more negative during inspiration and less during expiration (in humans ∼ 3 mmHg endexpiration and −5 mmHg

#### FIGURE 1 | Production of pericardial fluid as a net pressure phenomenon. Epicardial and pericardial cavity hydrostatic (P) and colloidosmotic (5) pressures. There is a net pressure difference of 2–10 mmHg that drives the fluid from the epicardium toward the pericardial cavity.

endinspiration) (Kenner and Wood, 1966; Spodick, 1997). Finally, the P<sup>p</sup> is about 20 mmHg above the free ventricular wall and almost zero inside the great sulcus (atrioventricular, intraventricular), during the enddiastole (Traboulsi et al., 1992). This ensures the constant fluid movement and homogenous composition, regardless the gathering position (Santamore et al., 1990). The pericardial capillaries stem from the systematic circulation. The venous drainage is accomplished through the superior vena cava (high pressure system). Thus, the Pcap is about 25 mmHg, the 5cap is about 25–28 mmHg (Mohrman and Heller, 2006) and the 5<sup>p</sup> is calculated near 5.9–8 mmHg by using the van't Hoff equation (Ben-Horin et al., 2005).

## Pericardial Lymph Drainage

The lympatics have high absorbing capacity due to the smooth myocytes, placed circumferentially around the lumen that transforms them into a pump. This pump has a diastolic and systolic period that is controlled by the Frank–Starling mechanism, like the cardiac pump, increasing the absorbing capacity under increased "preload" conditions (Yuan et al., 2000; Quick et al., 2007).

The initial studies on the role of lympatic vessels during pericardial fluid drainage were controversial (Hollenberg and Dougherty, 1969; Miller et al., 1988). Finally, the leading role of the lymphatics was established (Boulanger et al., 1999). The thoracic duct ligation decreased the fluid drainage, without completely blocking it. That was attributed to the complicated structure of the lymphatic capillary bed (Eliskova et al., 1995; Boulanger et al., 1999).

In vivo studies in sheep support the notion that the fluid drainage through the lympatics increases proportionally to the volume or pressure increase, as much as four times. This property has been related to the effect of external factors on the lymphatics function as well as to the functional alterations due to the neurohormonal stimulation. These characteristics are extremely important under conditions of fluid accumulation like cardiac tamponade (Miserocchi, 1989; Yuan et al., 2000). Also, in rabbits and mice, but not in sheep, there is a proven communication between the pericardial and the pleural cavity, through "pores" (diameter up to 50µm), enabling fluid leakage toward the pleural cavity in certain conditions (Nakatani et al., 1988; Boulanger et al., 1999).

## Water, Electrolyte, and Protein Transport

The common mesodermal origin and the simplicity of its isolation established the parietal pericardium as a surrogate tissue for pleural mesothelial tissue studies in small animals (Ishihara et al., 1980; Zocchi et al., 1998). Early studies showed data consistent with the passive diffusion of water, Na+, Cl<sup>−</sup> and small molecules (sucrose, mannitol) through "small" pores and later it was shown that there is a passive paracellular diffusion of macromolecules (albumin, dextranes) through "large" pores (either like permanent channels or like transient cystic formations). In both cases the main morphological and functional barrier proved to be the mesothelium. However, by that time active ion trasport through the parietal pericardium was thought to be minimal (Zocchi et al., 1998; Bodega et al., 2000).

However, a later study that examined this aspect showed that the electrical resistance of the rabbit parietal pericardium is measurable and attributed to the mesothelial barrier since it had greater resistance (10.1 ± 0.9 × cm<sup>2</sup> ) compared to the underlying connective tissue alone (1 ± 0.2 × cm<sup>2</sup> ) (Bodega et al., 2001). This established the mesothelium as the main barrier for molecular transfer. Furthermore, another study shed more light to the active electrolyte transport through the parietal pericardium showing that it is directed from the basal to the apical surface with a net driving pressure of ∼3 cmH2O, which is 5-fold higher than the one through the pleural lamenae (Tang and Lai-Fook, 2005). In this study it was also shown that the diffusion constant of the rabbit parietal pericardium for albumin (0.26–0.96 ×10−<sup>8</sup> cm<sup>2</sup> /s), is independent from its concentration. This finding contradicted the previous data regarding rabbit pericardium, pleura and omentum where the reference values were higher and directly proportional to albumin concentration (Parameswaran et al., 1999a,b; Bodega et al., 2002). This discrepancy was attributed to the experimental conditions and the tissue differences. The pericardium contains a larger proportion of collagen fibers compared to the elastic ones and a higher concentration of hyaluronic acid that render it more stiff as well as less permeable than the other serosal membranes (Tang and Lai-Fook, 2005). Moreover, the characteristics above are related to the higher values of hydraulic and electrical resistance of the pericardium and mainly the mesothelial layer (Bodega et al., 2002; Tang and Lai-Fook, 2005). There are in vitro data that the hydraulic permeability of the parietal pericardium is independent of the hydrostatic pressure over the range from 6 to 15 cmH2O and directly proportional to the membrane thickness, among the species (Fingerote et al., 1980).

Variance among species seems to be the case also in terms of transmesothelial electrical resistance, an index of ion transport. As mentioned above rabbit parietal pericardium had values in the range of 10.1 ± 0.9 × cm<sup>2</sup> , while in sheep these values were nearly double (22.83 ± 0.4 × cm<sup>2</sup> ) (Vogiatzidis et al., 2006). Moreover, in the last study the effects of morphine on the pericardium were assessed and it was shown that the electrical resistance of the pericardium is increased by the application of morphine. The same results were found in the pleura and the peritoneum indicating a common opoidergic influence of the ionic transport capacity of the three serosal membranes (Vogiatzidis et al., 2006; Zarogiannis et al., 2007).

## Conclusions

The study of pericardial space physiology is an area with many things to be discovered. The mechanism of pericardial fluid production is straightforward in physiological conditions, however it needs to be identified what is the exact role of the mesothelial cells both in the recycling of the pericardial fluid as well as with respect to the paracrine function that they possess. Another important challenge would be to dissect the exact contribution and magnitude of each mechanism regulating the recycling of the pericardial fluid. Finally, few things are known about the interplay of mesothelial cells and pericardial fluid. These areas will increase our understanding of the physiology of the pericardial space once

References


explored as well as they will provide us with new insights regarding drug development in the context of pericardial effusions.


**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 © 2015 Vogiatzidis, Zarogiannis, Aidonidis, Solenov, Molyvdas, Gourgoulianis and Hatzoglou. 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.

## Encapsulating peritoneal sclerosis—a rare but devastating peritoneal disease

#### *Zia Moinuddin1,2, Angela Summers 1, David Van Dellen1, Titus Augustine1 and Sarah E. Herrick2 \**

*<sup>1</sup> Department of Transplantation, Manchester Royal Infirmary, Manchester, UK*

*<sup>2</sup> Faculty of Medical and Human Sciences, Institute of Inflammation and Repair, University of Manchester, Manchester Academic Health Science Centre, Manchester, UK*

#### *Edited by:*

*Sotirios G. Zarogiannis, University of Thessaly, Greece*

#### *Reviewed by:*

*Theodoros Eleftheriadis, University of Thessaly, Greece Antonios Helias Tzamaloukas, U.S. Department of Veterans Affairs, USA*

#### *\*Correspondence:*

*Sarah E. Herrick, Faculty of Medical and Human Sciences, Institute of Inflammation and Repair, University of Manchester, Room 3.106 Stopford Building, Oxford Road, Manchester M13 9PT, UK e-mail: sarah.herrick@ manchester.ac.uk*

Encapsulating peritoneal sclerosis (EPS) is a devastating but, fortunately, rare complication of long-term peritoneal dialysis. The disease is associated with extensive thickening and fibrosis of the peritoneum resulting in the formation of a fibrous cocoon encapsulating the bowel leading to intestinal obstruction. The incidence of EPS ranges between 0.7 and 3.3% and increases with duration of peritoneal dialysis therapy. Dialysis fluid is hyperosmotic, hyperglycemic, and acidic causing chronic injury and inflammation in the peritoneum with loss of mesothelium and extensive tissue fibrosis. The pathogenesis of EPS, however, still remains uncertain, although a widely accepted hypothesis is the "twohit theory," where, the first hit is chronic peritoneal membrane injury from long standing peritoneal dialysis followed by a second hit such as an episode of peritonitis, genetic predisposition and/or acute cessation of peritoneal dialysis, leading to EPS. Recently, EPS has been reported in patients shortly after transplantation suggesting that this procedure may also act as a possible second insult. The process of epithelial–mesenchymal transition of mesothelial cells is proposed to play a central role in the development of peritoneal sclerosis, a common characteristic of patients on dialysis, however, its importance in EPS is less clear. There is no established treatment for EPS although evidence from small case studies suggests that corticosteroids and tamoxifen may be beneficial. Nutritional support is essential and surgical intervention (peritonectomy and enterolysis) is recommended in later stages to relieve bowel obstruction.

**Keywords: encapsulating peritoneal sclerosis, peritoneal dialysis, mesothelium, epithelial–mesenchymal transition, fibrosis**

## **INTRODUCTION**

Encapsulating peritoneal sclerosis (EPS) is a chronic clinical syndrome of insidious onset, presenting late as chronic malnourishment with signs and symptoms of intermittent, acute or sub-acute gastrointestinal obstruction (Augustine et al., 2009). It appears to be a multifactorial disease with several initiating factors that are significant at the different stages of the disease. Diagnosis is confirmed by macroscopic and/or radiological observation of sclerosis, calcification, peritoneal thickening or encapsulation of the intestines (Kawaguchi et al., 2000). EPS can be fatal but, fortunately, it is a rare complication predominantly of long-term peritoneal dialysis (PD). Since the first case of EPS was reported in 1980 (Gandhi et al., 1980), there has been a steady rise in the incidence of the disease from 0.9% in 1996 to 3.3% in 2005 as patients stay on PD for longer and awareness of this complication has increased. The data obtained over the years from several retrospective studies performed to investigate this potentially fatal condition estimate the worldwide incidence of EPS at 0.7–3.3% in patients on PD (Brown et al., 2009). The mortality rate for patients with EPS is great at 25–55%, predominately in the year after diagnosis and is directly proportional to the duration of PD treatment (Rigby and Hawley, 1998; Kawanishi et al., 2004; Kawanishi and Moriishi, 2005; Brown et al., 2009).

## **AETIOLOGY**

EPS is usually seen in end-stage renal disease (ESRD) patients who have been on long-term PD therapy. Dialysis fluid is damaging to the peritoneum due to the high glucose concentration and acidic pH. A high glucose concentration facilitates the osmosis and diffusion gradient across the peritoneum and the low pH acts to prevent the formation of glucose degradation products (GDPs) which are damaging agents (Jorres et al., 1992). Heat sterilization of PD fluid leads to the formation of GDPs (Wieslander, 1996) and these in the presence of glucose cause the formation of advanced glycation end products (AGEs) (Mortier et al., 2002). More biocompatible solutions are now being used that contain less GDPs which results in reduced peritoneal damage (Boulanger, 2008).

Interestingly, EPS can also develop in patients not on PD but associated with other conditions such as autoimmune diseases, sarcoidosis, peritoneal and intra-abdominal malignancies, chronic peritoneal ascites, intra-peritoneal chemotherapy, intraperitoneal exposure to particulate matter or disinfectant, abdominal surgery, endometriosis, intra-peritoneal infections (tuberculosis), and beta-blocker administration (Pollock, 2001; Kawanishi and Moriishi, 2005) (**Table 1**). Chronic renal failure itself could also induce peritoneal changes including thickening before PD therapy (Williams et al., 2002) and uremia, a contributory factor in all PD patients, is known to lead to a proinflammatory state (Baroni et al., 2012).

## **RISK FACTORS**

The duration of PD therapy seems to be the most important risk factor for the development of EPS. In an Australian survey, the incidence of EPS increased with the duration of PD, being 1.9, 6.4, 10.8, and 19.4% in patients on PD for more than 2, 5, 6, and 8 years respectively (Rigby and Hawley, 1998).

A Japanese prospective study reported similar findings, in which the incidence of EPS was 0.7% after 5 years, 2.1% after 8 years, 5.9% after 10 years, and 17.2% after 15 years of PD therapy (Kawanishi et al., 2004). A more recent Scottish study also reported an increase in incidence of EPS from 2% at 2–3 years to 8.8% at 5–6 years of PD therapy (Brown et al., 2009). However, most cases of EPS are diagnosed after discontinuation of PD therapy. A Japanese prospective study reported that 69% of EPS cases occurred after discontinuation of PD therapy (Kawanishi et al., 2004). This suggests that despite PD being a major risk factor for EPS, lavage of the peritoneal cavity during PD may possibly limit the accumulation of factors that encourage the development of the disease (Yamamoto et al., 2005, 2010).

Interestingly, organ transplantation appears to increase the risk of developing EPS as some studies have reported a high incidence shortly after renal transplantation (Fieren et al., 2007; Balasubramaniam et al., 2009). This could either be due to the acute cessation of PD or due to the profibrotic effect of immunosuppressive medication (calcineurin inhibitors). Calcineurin inhibitors, like Tacrolimus and Cyclosporin, cause up-regulation of transforming growth factor-beta (TGF-β), and other profibrogenic factors potentiating matrix accumulation (Khanna et al., 2002). In experimental rat models, Cyclosporin combined with chronic peritoneal exposure to dialysis solution was associated with increased peritoneal fibrosis and angiogenesis (Van Westrhenen et al., 2007). However, the exact mechanism of post-transplantation EPS still remains unknown.

Peritonitis is a common complication of PD and plays a complex part in the development of EPS with the number of peritonitis episodes linked to incidence of EPS (Yamamoto et al., 2005). Recurrent peritonitis due to bacterial contamination including *Pseudomonas* spp., *Staphylococcus aureus*, and certain fungal organisms have been particularly implicated in the development

**Table 1 | Non-peritoneal dialysis related causes of EPS.**


of EPS (Flanigan et al., 1984; Chew et al., 1997; Rigby and Hawley, 1998). Other risk factors suggested to be involved in EPS onset are the composition of dialysis fluid and generation of GDPs, young age, ultrafiltration failure and the exposure to PD catheter cleaning reagent, chlorhexidine (Pollock, 2001) (**Table 2**).

#### **CLINICAL FEATURES OF EPS**

Patients usually present with abdominal symptoms like early satiety, anorexia, nausea, vomiting, and altered bowel habit (constipation or diarrhea in the early stages of EPS) (Nakamoto, 2005; Augustine et al., 2009). These symptoms may be accompanied by signs of inflammation (pyrexia and raised CRP) and/or blood stained ascites in the early stages (Nakamoto, 2005; Maruyama and Nakayama, 2008). Late stages of EPS are associated with abdominal pain, fullness, overt bowel obstruction and presence of an abdominal mass). This is caused by the development of a fibrous cocoon that gradually covers the intestines and leads to malnutrition, weight loss, bowel obstruction, ischemia and strangulation, infection and death (Kawaguchi et al., 2000).

#### **CLASSIFICATION OF EPS**

Based on the clinical presentation, Nakamoto categorized EPS into four groups (Nakamoto, 2005). The following are the proposed clinical stages:


#### **DIAGNOSIS OF EPS**

The diagnosis of EPS relies on clinical findings, radiological tests and pathological appearance of the diseased tissue. Clinically, the diagnosis of EPS is based on recognizing the signs and symptoms (nausea, anorexia, early satiety, weight loss, altered bowel habit, and ascites) in the patients at risk of developing the condition (Nakamoto, 2005). Blood tests may also reveal high CRP and low

#### **Table 2 | Risk factors for EPS in the peritoneal dialysis (PD) population.**


albumin levels. These presenting symptoms and signs are vague and non-localizing. However, the insidious nature and chronicity of development can be a distinguishing feature of EPS (Kawaguchi et al., 2000). As a consequence, EPS is often not recognized in its early stages and requires a high index of suspicion to pursue a diagnosis.

As the clinical picture of EPS can vary considerably, various investigations are necessary to further evaluate a suspected case (Kawaguchi et al., 2005). Ultrasonography, water-soluble contrast studies and computed tomography (CT) scanning are the most widely used radiological tests to aid the diagnosis of EPS. CT scanning, however, is the investigation of choice in patients with established EPS and helps monitor disease progression. Peritoneal enhancement, peritoneal thickening, calcification, bowel tethering, bowel wall thickening, signs of bowel obstruction, and loculated collections are the most common CT findings of EPS (Vlijm et al., 2009). But, given the rare, complex nature of the disease and with most of the CT scan appearances being non-specific, interpretation and diagnosis can be difficult. Surgery (Laparoscopy/Laparotomy and peritoneal biopsy) therefore may be needed to confirm the diagnosis (Kawaguchi et al., 2005). At surgery, macroscopically, advanced cases of EPS typically exhibit a thickened brownish peritoneum with a cocoon-like encapsulation of the entire intestine by the visceral peritoneum (**Figure 1**). The intestinal loops are adherent to one another and the visceral peritoneum is severely thickened with fibrosis. Adhesions between the visceral and parietal peritoneum are rare, except in cases of severe inflammation (Honda and Oda, 2005). Macroscopic evidence of peritoneal calcification is also a feature but is not necessarily present in all cases of EPS (Park et al., 2008).

Numerous studies have been performed searching for potential biomarkers in PD effluent to aid the early diagnosis of EPS. Low CA125 levels (denoting mesothelial cell loss) and high levels of the inflammatory cytokine Interleukin-6 (IL-6) in PD effluent have been noted several years before the diagnosis of EPS suggesting that these factors could be potential early diagnostic markers (Sampimon et al., 2010). However, further research is necessary to identify definite biomarkers for early diagnosis of EPS. Current imaging techniques are also not useful at reliably detecting EPS in its early stages. CT scan screening of asymptomatic PD patients is not recommended as EPS may occur within a year of a normal CT scan. Also, not all long-term PD patients with peritoneal abnormalities on CT go on to develop established EPS (Goodlad et al., 2011). Dynamic cinematographic magnetic resonance (cine-MR) scanning with advanced image analysis may be useful in early detection of EPS. However, more studies are required to validate the efficacy of this imaging modality (Wright et al., 2011). Laparoscopy performed in suspected cases may have an important role to play in the early diagnosis of EPS. Our current lack of understanding of the pathophysiology limits our ability to detect EPS early or prevent its occurrence.

### **HISTOLOGICAL FEATURES OF ENCAPSULATING PERITONEAL SCLEROSIS**

The two most relevant pathologies of long-term PD are simple sclerosis and EPS. Simple sclerosis is characterized

**FIGURE 1 | Macroscopic appearance of EPS at surgery. (A)** Thickened parietal peritoneum is held up by surgical clips. Visceral peritoneum is thickened forming a fibrous cocoon encapsulating the bowel. **(B)** Visceral peritoneum peeled off the small bowel which has a brownish, tanned leathery appearance.

by the thickening of parietal peritoneum in the absence of encapsulation. The uremic state and continuing PD leads to a gradual loss of mesothelium (mesothelial denudation). Medial sclerosis and hyalinization of the peritoneal vasculature (vasculopathy) is also seen along with neoangiogenesis (Honda et al., 1996). The sub-mesothelial compact zone thickens and is composed of myofibroblasts and fibrous collagen (Mateijsen et al., 1999). In addition, AGEs are found to be present in the mesothelial and sub-mesothelial layer of PD patients (Yamada et al., 1994; Nakayama et al., 1997). Of note the amount of angiogenesis was found to be proportional to the severity of intestinal fibrosis (Mateijsen et al., 1999; Plum et al., 2001) and the level of mesothelial deudation correlated with sub-mesothelial thickening and vasculopathy (Williams et al., 2003). Simple sclerosis is a fairly common finding in long-term PD patients while EPS is rare. Therefore, studies analysing the histological differences in the peritoneum from EPS and simple sclerosis patients are limited. Of these, only one study performed by Garosi et al. (2005) analyzed both visceral and parietal peritoneum, and Braun and colleagues analyzed only visceral peritoneum (Braun et al., 2012) while the remaining studies did not specify the type of peritoneum analyzed (**Table 3**). Garosi and colleagues investigated 180 peritoneal biopsies of PD patients with simple sclerosis and


#### **Table 3 | Comparative histopathology of EPS patients and non-EPS PD patients.**

*(S) and (NS) denote significance and non-significance, respectively. (–) denotes data not being measured or recorded.*

compared morphological findings with those from biopsies of 39 patients with EPS (Garosi et al., 2005). Significant findings in patients with EPS were thickening of the sub-mesothelial cell layer, vasculopathy, arterial occlusion, inflammation, tissue and arterial calcification, and ossification. In a similar study, fibrin deposition, increase in the size of fibroblasts, capillary angiogenesis, and mononuclear cell infiltration were more common features of EPS than simple sclerosis (Honda et al., 2003). Sherif and colleagues compared the peritoneum of EPS with simple sclerosis and showed that the sub-mesothelial compact zone was thinner in early-EPS than late-EPS (Sherif et al., 2008). In a more recent study, fibroblast-like cells, mesothelial denudation, calcification, decreased cellularity and positive iron staining were more common in the peritoneum of EPS patients. Positive immunohistochemical staining for podoplanin, a lymphatic endothelial marker expressed by peritoneal mesothelial cells, was significantly more prevalent in EPS peritoneum (Braun et al., 2012). Up-regulation of vascular endothelial growth factor (VEGF) and down-regulation of mast cells (Alscher et al., 2007; Braun et al., 2009) also appeared to be a feature of EPS peritoneum. However, as the above findings are not always specific to EPS tissue, reliable histological diagnosis is difficult. Further studies are therefore necessary to explore the presence of additional unique immunohistochemical markers to aid the diagnosis.

## **PATHOPHYSIOLOGY OF ENCAPSULATING PERITONEAL SCLEROSIS**

Although a number of risk factors have been identified, there are currently no means of preventing EPS development or recognizing earlier stages of the disease. Furthermore, it is a matter of debate whether EPS is a natural progression of simple sclerosis or a completely separate entity. However, it is clear that having simple sclerosis does not predispose to developing EPS. It is generally now accepted that a two-hit theory explains EPS development. It is well established that chronic exposure to bio-incompatible dialysate causes damage to the peritoneum of all patients on PD. This first hit causes mesothelial disruption which can trigger a fibrotic process which is referred to as "simple sclerosis."

The subsequent exposure to a second insult, in some patients, triggers the development of EPS. This can be either an episode of peritonitis, acute cessation of PD, transplantation, any other acute intra-abdominal event or maybe a genetic predisposition (**Figure 2**). A number of genetic polymorphisms with functional effects have been described in patients on PD and these may partially explain the propensity to develop EPS (Pletinck et al., 2012). A number of rodent models of EPS have been developed using chronic chemical irritation to induce peritoneal sclerosis and bowel encapsulation (Hoff, 2005; Summers et al., 2008). Such models replicate the cycles of tissue injury with fibrin deposition and inflammation leading to subsequent repair with mesothelial cell and fibroblast activation, proliferation and matrix deposition that are proposed to drive the development of EPS. Some of these mechanisms of induction are described in more detail below.

#### **INFLAMMATORY PROFILE**

The PD catheter can itself trigger inflammation either directly as a foreign material or by acting as a site of bacterial biofilm formation leading to PD peritonitis (Flessner et al., 2007). Various characteristics of PD fluid such as low pH, high glucose levels, hyperosmolarity, presence of GDPs, and AGEs promote the release of several growth factors such as Transforming Growth Factor-β (TGF-β) and Platelet Derived Growth Factor (PDGF) and pro-inflammatory cytokines, IL-1, IL-6, IL-18, and Tumor Necrosis Factor-alpha (TNF-α) potentiating fibrosis (Margetts and Bonniaud, 2003; Baroni et al., 2012).

Clinical studies have found elevated levels of IL-1β, IL-6, IL-8, TGF-β1, Hepatocyte Growth Factor (HGF), and PDGF in ascites from EPS patients in comparison with non-EPS controls (Masunaga et al., 2003). As EPS develops, patients also suffer from subclinical bowel ischemia which may cause translocation of bacteria across the bowel wall into the peritoneal cavity leading to infection, further inflammation and fibrosis (Augustine et al., 2009).

### **FIBRIN DEPOSITION AND FIBRINOLYSIS**

Peritoneal inflammation initiates fibrin exudation that can either be lysed or remodeled by invading fibroblasts leading to fibrosis and adhesion formation (Holmdahl, 1997; Sulaiman et al., 2002). Plasmin plays a key role not only in fibrin degradation but also in the turnover of extracellular matrix and activation of matrix metalloproteinases. Plasmin is formed from the inactive zymogen plasminogen under the influence of plasminogen activators: tissue plasminogen activator (tPA) and urokinase-like plasminogen activator (uPA), whereas, plasminogen activation inhibitors (PAI-1 and PAI-2) reduce plasminogen activation. Low levels of plasmin and high levels of PAI were found in serum of patients on PD compared with hemodialysis (HD) patients suggesting a higher degree of hypercoagulation (Tomura et al., 1996). Mesothelial cells express PAI-1 and 2 under the influence of TGF-β (Rougier et al., 1998; Holmdahl et al., 2001) and such an inappropriate balance between fibrin deposition and breakdown increases the probability of fibrous adhesion formation possibly contributing to the development of EPS.

### **EPITHELIAL–MESENCHYMAL TRANSITION (EMT) OF MESOTHELIAL CELLS**

In terms of peritoneal fibrosis, substantial evidence points to mesothelial cells (MCs) as being the principle source of myofibroblasts via a process of epithelial–mesenchymal transition (EMT) (Aroeira et al., 2007). MCs are unique in their expression of both epithelial and mesenchymal markers, and when exposed to injurious agents, they lose cell-cell contact and apical-basal cell polarity and invade the basal lamina changing to a mesenchymal phenotype expressing α-smooth muscle actin and depositing extracellular matrix (Aguilera et al., 2005). EMT occurs during normal embryo development and tumor cell invasion and metastasis and is a complex process that requires the correct spatiotemporal expression, interaction and modification of a multitude of intra-and extracellular factors to allow a change in cell phenotype (Thiery et al., 2009). Loss of cell surface E-cadherin is a prerequisite for EMT and primarily controlled by three main families of transcription factors: zinc finger Snail, basic helix-loop-helix, and ZEB (Kalluri and Weinberg, 2009). Several groups have shown the presence of mesothelial markers (ICAM-1 and cytokeratin) localized with spindle-shaped fibroblast- like cells in the sub-mesothelial layer in peritoneal biopsies from patients undergoing PD suggesting EMT had occurred (Yanez-Mo et al., 2003; Jimenez-Heffernan et al., 2004). Mesothelial cells isolated from dialysis fluid effluents from patients undergoing peritoneal dialysis also undergo a transition from epithelial to mesenchymal with a loss of E-cadherin (Zhang et al., 2013) and induction of snail (Yanez-Mo et al., 2003). Furthermore, AGE products (De Vriese et al., 2006) and dialysate (Selgas et al., 2006) induce human mesothelial EMT in culture. The process of EMT can occur under the influence of a number of pro-inflammatory and profibrotic cytokines, however TGF-β is proposed to be a principle mediator of mesothelial EMT as demonstrated both *in vitro* (Hung et al., 2003; Yang et al., 2003) and *in vivo* (Margetts et al., 2005). In terms of therapeutic strategy, blocking TGF-β1 (Loureiro et al., 2011), or addition of tamoxifen (Loureiro et al., 2013a) bone morphogenic protein-7 (Yu et al., 2009), or microRNA30a (Zhou et al., 2013) have all been found to protect the peritoneum from dialysate—induced damage in experimental models. Although strong evidence suggests that mesothelial EMT is important for the development of peritoneal sclerosis, the role this process plays in EPS, is a matter of debate (Loureiro et al., 2013b).

## **GROWTH FACTORS POTENTIALLY INVOLVED IN THE DEVELOPMENT OF EPS**

The pathways involved in the development of EPS are likely to be complex and involve the interaction of a number of important growth factors leading to sub-mesothelial thickening and cocooning of the bowels. The level of the pro-neoangiogenic growth factor, Vascular Endothelial Growth Factor (VEGF), detected in peritoneal effluent has been found to directly correlate with length of time the patients are on PD (Zweers et al., 2001). Furthermore, treatment with VEGF blocking antibody reduced the thickness of the sub-mesothelial layer and reduced vasculopathy in a rat model of EPS (Io et al., 2004) suggesting a key role for this factor in EPS development. VEGF is produced by cultured human peritoneal mesothelial cells in response to various stimuli such as GDPs known to be present in PD fluid (Mandl-Weber et al., 2002). TGF-β, another important growth factor in wound healing and fibrosis, has also been shown to be present in peritoneal dialysate from patients (Lin et al., 1998; Yanez-Mo et al., 2003) Exposure of human mesothelial cells to TGF-β1 induced procollagen type 1 expression (Hung et al., 2003) and in rodent models, TGF-β1 administration caused peritoneal fibrosis (Margetts et al., 2005), and adhesion formation (Williams et al., 1992; Gorvy et al., 2005). Moreover, studies in a hyperglycemiainduced rodent model showed that AGEs upregulate TGF-β and induce submesothelial fibrosis with interstitial accumulation of collagen (De Vriese et al., 2003). Furthermore, intraperitoneal administration of a first generation adenovirus overexpressing TGF-β1 in mice resulted in submesothelial thickening and angiogenesis up to 10 days after administration (Margetts et al., 2005). However, a helper dependent adenovirus subsequently used with longer term expression of TGF-β1 led to encapsulation of the bowels in a thick cocoon in a similar manner to the final stages of EPS (Liu et al., 2009). Over-expression of TGF-β also increased levels of matrix metalloproteinase-2 (MMP-2) and high amounts of this protease found in PD effluent has been considered a potential marker of peritoneal injury and progression to EPS (Hirahara et al., 2007). In a mouse model, inhibition of MMP-2, TGF-β, and VEGF significantly improved peritoneal fibrosis and angiogenesis (Ro et al., 2007). Many other growth factors including HGF, PDGF, Connective Tissue Growth Factor (CTGF), and Fibroblast Growth Factor (FGF), shown to be involved in the development of fibrosis in other organs, may also influence the onset of EPS (Korte et al., 2011a).

## **THERAPEUTIC MANAGEMENT OF EPS DISCONTINUATION OF PD**

Cessation of PD should be done as soon as the diagnosis of EPS has been established. However, as cessation of PD is a known risk factor for the development and progression of EPS, this approach has been under much debate. Regular peritoneal lavage after discontinuation of PD has been tried in Japan with varying results. A recent retrospective study performed by Yamamoto et al. (2010) has demonstrated that regular peritoneal lavage helps mesothelial cell layer recovery and prevention of EPS. However, given the risk of peritonitis with peritoneal lavage and the increased association of peritonitis and prolonged PD therapy with EPS progression, it is best to stop PD at the time of diagnosis of EPS along with removal of the PD catheter (Habib et al., 2011; Kawanishi, 2012).

#### **NUTRITIONAL SUPPORT**

As most patients with EPS suffer from malnutrition, appropriate nutritional support is imperative in the management of these patients. Studies from our centre have demonstrated poor pre-operative nutritional status and improved survival with peri-operative Total Parenteral Nutrition (TPN) in patients with advanced EPS who underwent surgery (peritonectomy and enterolysis) (De Freitas et al., 2007; Campbell et al., 2014). However, TPN therapy on its own does not have any curative effect as demonstrated by a study from Japan which showed no recovery in patients treated with TPN alone (Kawanishi and Moriishi, 2005). TPN therapy is therefore best used as an adjunct to more definite treatment in EPS patients with malnutrition or peri-operatively until gut function recovers.

#### **MEDICAL TREATMENT**

The evaluation of the efficacy of pharmacological treatments for EPS is difficult due to the lack of good evidence based data. However, the treatment options currently in vogue are as follows:

#### *Immunosuppressive therapy*

Corticosteroids are the drugs of choice in the management of EPS. They are particularly useful in the treatment of EPS in its early inflammatory stages (Kuriyama and Tomonari, 2001; Kawanishi, 2012). Steroids possibly act by suppressing inflammation, preventing fibrin deposition and collagen synthesis and maturation (Habib et al., 2011). Steroids also help in preventing the formation and reducing the accumulation of ascites (Mori et al., 1997; Jung and Cho, 2013). Various case reports and series have reported good prognosis and improved survival with the use of steroids compared to any other immunosuppressive agent (Junor and McMillan, 1993; Rajani, 2002). However, in a study from Japan, the rate of clinical improvement with isolated steroid treatment was only 38.5% (Kawanishi et al., 2004). Nevertheless, steroids are effective in treating the early stages of EPS but the clinical response tends to reduce in the later stages that are associated with increased fibrosis and bowel obstruction. A variety of other immunosuppressive medications like azathioprine, rapamycin, mycophenalate mofetil, sirolimus, and cyclosporine have also been used to treat EPS either alone or in combination with steroids (Rajani, 2002; Wong et al., 2005; Lafrance et al., 2008). However, apart from steroids, evidence regarding the efficacy of any other immunosuppressive therapy in EPS remains weak due to lack of robust randomized clinical trials (Bozkurt et al., 2009).

#### *Tamoxifen*

Tamoxifen is a Selective Estrogen Receptor Modulator (SERM) with antifibrotic properties and has been used in the treatment of various fibrotic disorders like retroperitoenal fibrosis, fibrosing mediastinitis, fibrosing cerivicitis, and desmoid tumors (Cornelis and Oreopoulos, 2011). Various case reports and small series have reported satisfactory outcomes following the use of Tamoxifen in the treatment of EPS (Guest, 2009; Cornelis and Oreopoulos, 2011). A recent large retrospective Dutch study has demonstrated significantly reduced mortality in EPS patients that were treated with Tamoxifen (45.8%) when compared to those that were not treated with Tamoxifen (74.4%) (Korte et al., 2011b). Tamoxifen appears to exert its effect through inhibition and modulation of TGF-β. *In vitro* and animal models showed that Tamoxifen treatment blocked EMT induced by TGF-β, preserved the fibrinolytic activity and reduced the migration capacity of mesothelial cells leading to reduced fibrosis and reduced PD effluent levels of VEGF leading to reduced angiogenesis in the peritoneum (Loureiro et al., 2013a). However, Tamoxifen has almost always been used in combination with corticosteroids, therefore the efficacy and safety of Tamoxifen alone in the treatment of EPS still remains to be evaluated. The potential side-effects of Tamoxifen (deep vein thrombosis, endometrial cancer, and calciphylaxis) also need to be considered (Del Peso et al., 2003). Further prospective trials are therefore necessary to establish the safety and efficacy of Tamoxifen in the treatment of EPS.

#### **SURGERY**

Surgical techniques in the management of EPS have evolved over the last decade. Surgery is routinely performed in patients with advanced EPS in Japan and in the UK at specialist referral centers by surgeons experienced in the management of EPS (Augustine et al., 2009; Kawanishi, 2012). Surgery is usually performed in the late stages of EPS, in patients that present with absolute bowel obstruction or as surgical emergencies with an acute abdomen. Currently favored surgical techniques are peritonectomy and careful enterolysis which involves resection of the peritoneum and fibrous tissue together with division of adhesions to release the bowel (**Figure 1**). Mortality post-surgery ranges from 19 to 34.5% (Summers et al., 2005; Kawanishi, 2012; Latus et al., 2013) and is mainly seen in patients with advanced EPS presenting as surgical emergencies (Augustine et al., 2009). Although surgery is the most successful form of treatment for patients with advanced disease, recurrence rates post-surgery tend to be high at around 25%. However, in Japan, the recent introduction of Noble plication (suturing of the intestines to each other to prevent obstruction) along with routine enterolysis has reduced the recurrence rate to 12.3% (Kawanishi, 2012). Tamoxifen and steroids may also be continued post-operatively as they may have a role in preventing recurrence of EPS (Lo and Kawanishi, 2009).

## **PREVENTION**

Although various studies have demonstrated possible mechanisms in the development of EPS, we still have no strategy to prevent the occurrence of this condition. As the risk of developing EPS increases with the duration of PD, there has been much debate on an "expiry date" for PD in patients. Studies from Japan have suggested 8 years to be a safe period to continue PD beyond which patients should be switched to hemodialysis (Kawanishi et al., 2004). However, setting an expiry date is not recommended as it could have a negative impact on the quality of life and could increase the risk of complications of tunneled lines used for hemodialysis in patients who were symptom free on PD (Garosi et al., 2005). A more rational approach would be to assess peritoneal membrane function. The commonly used method to assess peritoneal deterioration is the Peritoneal Equilibration Test (PET) with a high transport of solutes across the peritoneal membrane being indicative of failing function. There has been a suggestion of cessation of PD in all patients that show a high transport status (Kawanishi, 2012). However, not all patients that are high transporters develop EPS. Therefore, this approach has been under much debate. Peritoneal lavage could potentially be used in patients who have been on longterm PD (*>*8 years), with high transport status and increased levels of markers of inflammation (IL-6) and low levels of Ca-125 in their PD effluent in order to delay the development of EPS (Moriishi et al., 2002). However, this approach is not universally practiced due to the associated risk of infective peritonitis. Other proposed methods of possible prevention are by using more bio-compatible PD fluid and taking care to reduce the incidence of infective peritonitis. Addition of therapeutic agents such as, Tamoxifen and Angiotensin Converting Enzyme (ACE) inhibitors, may show promise to ameliorate peritoneal membrane function and fibrosis (Kolesnyk et al., 2007). Further studies are therefore necessary to investigate possible therapeutic strategies to prevent the exacerbated fibrotic process which culminates in EPS.

## **CONCLUSION**

Despite the increase in incidence and awareness over the last decade, EPS still remains a rare but much feared complication of long-term PD. Prolonged PD therapy is the most important risk factor in the development of EPS. Uremia, inflammation, EMT and loss of mesothelial fibrinolytic response are possible mechanisms of peritoneal fibrosis which could influence the development of EPS. However, the aetiopathogenesis still remains poorly understood. Currently a high index of suspicion in at risk groups is required to make the diagnosis of EPS. CT scan is the investigation of choice to aid the diagnosis, but the use of newer imaging modalities like cine-MRI, that shows promising signs, still needs to be validated. Low levels of Ca-125 and high levels of IL-6 in PD effluent have been suggested as early markers of development of EPS. However, further research needs to be done to validate this claim and look for other potential bio-markers to detect EPS early. The treatment of EPS in its early stages is largely medical with corticosteroids, tamoxifen and nutritional support. However, surgery (peritonectomy and enterolysis) is the treatment of choice for advanced cases presenting with overt bowel obstruction. Peri-operative nutritional support with TPN and post-operative steroids with or without Tamoxifen is recommended to prevent recurrence. Improvement in surgical technique has vastly improved survival of patients with EPS. However, despite successful surgery these patients remain at increased risk of mortality due to ongoing renal replacement therapy. Many clinicians are reticent toward transplantation in these patients due to a perception that transplantation might not be feasible given the complex surgical history. Although, the risk of developing recurrent disease post transplantation exists, the chances of survival are much improved with a functioning kidney transplant. These patients should therefore be assessed and worked up for renal transplantation in order to improve long-term survival.

## **REFERENCES**


mesothelial cell–its role in the response of the peritoneum to dialysis. *Nephrol. Dial. Transplant*. 21(Suppl. 2), ii2–ii7. doi: 10.1093/ndt/gfl183


**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.

*Received: 15 September 2014; accepted: 16 November 2014; published online: 05 January 2015.*

*Citation: Moinuddin Z, Summers A, Van Dellen D, Augustine T and Herrick SE (2015) Encapsulating peritoneal sclerosis—a rare but devastating peritoneal disease. Front. Physiol. 5:470. doi: 10.3389/fphys.2014.00470*

*This article was submitted to Membrane Physiology and Membrane Biophysics, a section of the journal Frontiers in Physiology.*

*Copyright © 2015 Moinuddin, Summers, Van Dellen, Augustine and Herrick. 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.*

## Intercellular communication in malignant pleural mesothelioma: properties of tunneling nanotubes

#### *Justin W. Ady1, Snider Desir 2,3, Venugopal Thayanithy2, Rachel I. Vogel 4, André L. Moreira5, Robert J. Downey1, Yuman Fong1,†, Katia Manova-Todorova6, Malcolm A. S. Moore7 and Emil Lou2 \**

#### *Edited by:*

*Sotirios G. Zarogiannis, University of Thessaly, Greece*

#### *Reviewed by:*

*Chiara Zurzolo, Pasteur Institute, France Xiang Wang, University of Bergen, Norway*

#### *\*Correspondence:*

*Emil Lou, Division of Hematology, Oncology and Transplantation, University of Minnesota, Mayo Mail Code 480, 420 Delaware Street SE, Minneapolis, MN 55455, USA e-mail: emil-lou@umn.edu*

#### *†Present address:*

*Yuman Fong, Department of Surgery, City of Hope, Duarte, CA, USA*

Malignant pleural mesothelioma is a particularly aggressive and locally invasive malignancy with a poor prognosis despite advances in understanding of cancer cell biology and development of new therapies. At the cellular level, cultured mesothelioma cells present a mesenchymal appearance and a strong capacity for local cellular invasion. One important but underexplored area of mesothelioma cell biology is intercellular communication. Our group has previously characterized in multiple histological subtypes of mesothelioma a unique cellular protrusion known as tunneling nanotubes (TnTs). TnTs are long, actin filament-based, narrow cytoplasmic extensions that are non-adherent when cultured *in vitro* and are capable of shuttling cellular cargo between connected cells. Our prior work confirmed the presence of nanotube structures in tumors resected from patients with human mesothelioma. In our current study, we quantified the number of TnTs/cell among various mesothelioma subtypes and normal mesothelial cells using confocal microscopic techniques. We also examined changes in TnT length over time in comparison to cell proliferation. We further examined potential approaches to the *in vivo* study of TnTs in animal models of cancer. We have developed novel approaches to study TnTs in aggressive solid tumor malignancies and define fundamental characteristics of TnTs in malignant mesothelioma. There is mounting evidence that TnTs play an important role in intercellular communication in mesothelioma and thus merit further investigation of their role *in vivo*.

#### **Keywords: tunneling nanotubes, malignant pleural mesothelioma, intercellular transfer, intercellular communication**

## **INTRODUCTION**

Malignant pleural mesothelioma (MPM) is a clinically devastating and locally invasive malignancy. Patients with this disease uniformly carry a poor prognosis despite advances in understanding of cancer cell biology and development of new therapies. Unlike other solid tumor malignancies, mesothelioma is highly refractory to all forms of current treatment including surgery, radiation, and chemotherapy. Treatment of mesothelioma and other invasive solid tumor malignancies such as cancers of the colon, pancreas, and ovaries is limited by an inadequate understanding of the modes and functions of intercellular communication in the tumor microenvironment (Axelrod et al., 2006; Ruckert et al.,

**Abbreviations:** MPM, malignant pleural mesothelioma; TnTs, tunneling nanotubes; nm, nanometers; EMT, epithelial-to-mesenchymal transition; IF, immunofluorescence; FBS, fetal bovine serum; HA, hyaluronic acid; GFP, green fluorescent protein; EM, electron microscopy; TNFaip2, tumor necrosis factoralpha-induced protein 2 (also called M-Sec), MSKCC, Memorial Sloan-Kettering Cancer Center.

2012). Intercellular communication is critical to tumor formation, organization, and treatment resistance (Kenny et al., 2007; Bissell and Hines, 2011; Ruckert et al., 2012). Mounting evidence suggests that tumor-stromal cell interactions are important to the invasive phenotype. Stromal cells, once seen as passive structural components of the tumor infrastructure, are now viewed as dynamic components of tumor initiation, progression, and invasion (Mueller and Fusenig, 2004; Tlsty and Coussens, 2006; Pietras and Ostman, 2010). Invasive tumors are composed of a large proportion of stroma; in MPM this proportion can be as high as 34–45% depending on the histologic subtype (Motta et al., 2006). The proportion is highest in biphasic and sarcomatoid tumors, the latter of which is associated with even worse prognosis than other subtypes (Motta et al., 2006). This tumorstroma balance creates a heterogeneous microenvironment composed of, among other things, malignant cells, cancer-associated fibroblasts, vascular endothelial cells, macrophages, and other inflammatory infiltrates. In a study of MPM, inflammatory or

*<sup>1</sup> Department of Surgery, Memorial Sloan-Kettering Cancer Center, New York, NY, USA*

*<sup>2</sup> Division of Hematology, Oncology and Transplantation, University of Minnesota, Minneapolis, MN, USA*

*<sup>3</sup> Integrative Biology and Physiology Program, University of Minnesota, Minneapolis, Minnesota, USA*

*<sup>4</sup> Department of Biostatistics and Bioinformatics, Masonic Cancer Center, University of Minnesota, Minneapolis, MN, USA*

*<sup>5</sup> Department of Pathology, Memorial Sloan-Kettering Cancer Center, New York, NY, USA*

*<sup>6</sup> Molecular Cytology, Memorial Sloan-Kettering Cancer Center, New York, NY, USA*

*<sup>7</sup> Department of Cell Biology, Sloan-Kettering Institute, Memorial Sloan-Kettering Cancer Center, New York, NY, USA*

desmoplastic stroma types correlated with worse patient prognosis, as compared with fibrous or myxoid forms of stroma (Cerruto et al., 2006).

The most commonly studied avenues of cellular transfer among cancer cells include gap junctions, chemokines, cytokine messengers, and microvesicles/exosomes (Bissell and Radisky, 2001; Hegmans et al., 2004; Cottin et al., 2010; Naus and Laird, 2010; Bobrie et al., 2011; Pap et al., 2011; Strassburg et al., 2012). These forms of intercellular communication are most effective over short distances. Furthermore, cell-cell junctions are disrupted upon epithelial-mesenchymal transition, a precursor to metastasis (Lamorte et al., 2002; Huber et al., 2005), making intercellular communication via these junctions impossible for separated cells. Additionally, effective cell-cell "cross-talk" via diffusible factors could be difficult to achieve because of an increase in interstitial fluid pressure in the tumor microenvironment. There remains considerable uncertainty regarding how tumorstroma exchange of cellular information takes place and how distant cells that are not in close proximity are able to communicate within a three-dimensional matrix composed of a significant proportion of stromatous material. A better understanding of the mechanisms and cellular structures that underlie intercellular communication among distant cells in the tumor matrix of malignant tumors is expected to lead to new targeted treatments that block progression of mesothelioma and other invasive solid tumor malignancies.

Our group has investigated tunneling nanotubes (TnTs), a previously underexplored form of cellular protrusions that are distinctly unique from other filamentous cellular extensions. TnTs are long, narrow, actin-based cytoplasmic extensions that form *de novo in vitro*. Nano-sized in width (50–800 nm), TnTs can stretch the length of several cell diameters or longer (as long as several hundred microns) to form direct cell-to-cell cytoplasmic connections. TnTs display non-adherence to the substratum when cultivated *in vitro* (Rustom et al., 2004). These characteristics differentiate TnTs from other, well-known actin-based cytoplasmic extensions including lamellopodia, filopodia, and invadopodia (Rustom et al., 2004). TnTs are open-ended "intercellular bridges" whose walls consist of a contiguous lipid bilayer that can establish a direct connection between the cytoplasm of connected cells, or in some cases interface with gap junctions in plasma membranes (Wang et al., 2010). TnT formation is largely generated by actin-driven membranous protrusions extending to outlying cells. They have been noted to form either by one cell extending a tubular cytoplasmic connection to another cell located at some distance (in contrast with gap junctions, which connect cells in immediate proximity) or to form between cells in close proximity that then move apart via usual mechanisms of cell motility, allowing for continuation of intercellular communication even as the cells move in different directions (Veranic et al., 2008). At least one study has suggested that TnTs interface with gap junctions to connect cells and mediate intercellular cross-talk (Wang et al., 2010). Uniquely, TnTs serve as conduits for intercellular shuttling of cellular organelles and other cargo between connected, non-adjacent cells (Lou et al., 2012a,b). *In vitro* studies have shown that TnTs have the ability to directly mediate cell-tocell communication by serving as long-range conduits between connected cells for intercellular transfer of proteins, mitochondria, Golgi vesicles, and even viruses (Koyanagi et al., 2005; Onfelt et al., 2005, 2006; Sherer et al., 2007; Davis and Sowinski, 2008; Sherer and Mothes, 2008; Plotnikov et al., 2010; Yasuda et al., 2010; He et al., 2011; Kadiu and Gendelman, 2011; Wang et al., 2011; Lou et al., 2012b) (For an example of time-lapse imaging we use in our work, please see **Movie S1** demonstrating intercellular transfer of mitochondria between mesothelioma cells connected via nanotube). The importance of intercellular transfer of genetic material is also a topic of growing interest. Our group recently demonstrated that TnTs can also transport oncogenic microRNAs between malignant cells, as well as between malignant and stromal cells, introducing a new aspect of tumor-stromal cross-talk that warrants further study (Thayanithy et al., 2014a).

TnTs have been studied in a wide variety of non-cancer cell types including dendritic cells and monocytes (Watkins and Salter, 2005; Salter and Watkins, 2006), mature macrophages (Eugenin et al., 2009; Hase et al., 2009), T cells (Sowinski et al., 2008, 2011; Rudnicka et al., 2009), B cells (Xu et al., 2009), neutrophils (Galkina et al., 2010), neuronal cells (Gousset et al., 2009), kidney cells (Gurke et al., 2008), endothelial progenitor cells (Yasuda et al., 2010), mesothelial cells (Ranzinger et al., 2011; Lou et al., 2012b), cardiomyocytes (Koyanagi et al., 2005), and mesenchymal stromal cells (Cselenyak et al., 2010; Plotnikov et al., 2010). Our group focuses on investigation of TnTs in the context of invasive forms of cancer (Lou et al., 2012a,b). To investigate TnTs as a physiologically relevant structure in human solid tumor malignancies, our initial work successfully visualized TnTs in solid tumors resected from patients with mesothelioma and lung adenocarcinomas (Lou et al., 2012b), providing the first evidence of the potential *in vivo* relevance of these cellular structures in cancer. We subsequently performed high-resolution microscopy and 3-dimensional reconstructions to confirm that nanotube structures are present in other invasive malignancies as well, including a murine model of osteosarcoma and human ovarian adenocarcinoma (Thayanithy et al., 2014a). In our *in vitro* work in mesothelioma, we used modified wound-healing assays and demonstrated TnT formation along the leading invasive edge of mesothelioma cells *in vitro*; time-lapse imaging revealed regular formation of TnTs by proliferating and migrating mesothelioma cells advancing to fill the gap (Lou et al., 2012b). This finding introduces the possibility that TnTs facilitate intercellular communication and the progression of malignancy at the leading edge of invasive mesothelioma tumors. More recently, we showed that exosomes and TnTs may work synergistically by demonstrating that exogenous tumor exosomes induced an increased rate of TnT formation (Thayanithy et al., 2014b). Electron microscopy revealed that exosomes were located at the base of TnTs and in the extracellular environment. Our subsequent studies identified enrichment of lipid rafts, small intra-cytoplasmic cholesterol microdomains, in mesothelioma cells connected via nanotubes (Thayanithy et al., 2014b). These findings implicate exosomes as potential chemotactic stimuli for TnT formation and lipid rafts as a potential biomarker for TnTs. The effects of TnT formation and TnT-mediated transport of cellular cargo on malignant cell behavior are currently unknown.

**29**

In the current study, we sought to further characterize the properties of TnTs in mesothelioma, including differences in formation of TnTs between malignant mesothelioma cells and non-malignant mesothelial cells; quantitative differences in TnT length in relation to cell proliferation; properties of TnT formation in clinically relevant models, such as between non-adherent cells, mimicking the scenario of mesothelioma cells floating in peritoneal or thoracic effusions as a hallmark of malignant progression; and structural components of TnTs in mesothelioma cells. Finally, we also sought to develop new approaches to 3 dimensional *in vitro* and *in vivo* modeling for the study of TnTs in tumor propagation and resistance to therapy.

## **MATERIALS AND METHODS**

### **CELL LINES AND CULTURE MEDIA**

MSTO-211H cells were derived from a patient with biphasic mesothelioma (ATCC no. CRL-2081); VAMT is a sarcomatoid mesothelioma cell line; and H2052 is a mesothelioma cell line of epithelioid histology. All three mesothelioma cell lines (MSTO-211H, VAMT, and H2052) were passaged using 10% fetal bovine serum (FBS) in RPMI-1640 with 25 mM glucose, supplemented with 1% penicillin-streptomycin (P-S) and 2% L-glutamine, at normal pH (7.6). The normal immortalized human mesothelium cell line MeT5A was passaged in 10% FBS in M199/MCDB105 (1:1) with 100 U/ml penicillin-streptomycin and 2% L-glutamine. All cell lines used in this study were authenticated by the Core Fragment Analysis Facility at Johns Hopkins University using short tandem repeat profiling. Cells were passaged in 75 cm<sup>2</sup> tissue culture flasks (Falcon, Becton Dickson, Oxnard, CA) at 37◦C in 5% CO2. Nanotube formation was stimulated by growing cells in 2.5% FBS in RPMI-1640 containing 50 mM glucose, supplemented with 1% P-S and 2% L-glutamine as described previously (Lou et al., 2012b); we refer to this combination throughout the text as "TnT medium." For 3-dimensional *in vitro* modeling, we mixed MSTO-211H cells in 5% FBS in RPMI medium containing 100 mM glucose in a 1:1 ratio with 2% agarose to compose a final medium composed of 1% agarose and TnT medium for further culture of cells. These cells were then added to 6-well culture plates for microscopic examination.

#### **QUANTIFICATION OF TnTs PER CELL**

Three MPM cell lines (H2052, VAMT, and MSTO-211H) and one benign mesothelial cell line (MeT5A) were plated at a density of 6 <sup>×</sup> <sup>10</sup><sup>4</sup> cells/well in 6-well adherent tissue culture plates (Fisher Scientific, Pittsburgh, PA) at 37◦C in 5% CO2 with TnTinducing medium (described above). TnTs were identified using the parameters described by Rustom et al. (2004) as well as in our previous publications (Lou et al., 2012b; Thayanithy et al., 2014b). Briefly, these parameters included (i) lack of adherence to the substratum of tissue culture plates, including visualization of TnTs passing over adherent cells; (ii) TnTs connecting two cells or extending from one cell were counted if the width of the extension was appropriately narrow and estimated to be *<*1000 nm in width; and (iii) a narrow base at the site of extrusion from the plasma membrane. Cellular extensions not clearly consistent with the above parameters were not included in the final count. An Olympus IX70 inverted microscope (Olympus Corporation) with 20× objective lens was used to count the number of TnTs and cells in 10 randomly chosen fields of each 6-well plate at 24, 48, and 72 h. A single representative image was taken at all time points for each well for analysis of TnT length. Experiments were performed in duplicate for each cell line. The number of TnTs per cell (TnTs/cell) was counted to exclude the possibility that increases in TnTs were due to increased cell proliferation. Means were calculated and compared using two-sided, two-tailed *t*-tests assuming unequal variances. *P*-values *<*0.05 were considered statistically significant.

### **QUANTIFICATION OF TnT LENGTH**

Representative images taken from the previous experiment were analyzed using ImageJ software. TnT length was measured by normalizing the 200 micron scale bar from the images to the number of pixels. The length of TnTs from each cancer cell line was measured at 24, 48, and 72 h. TnT lengths were not normally distributed; therefore, Wilcoxon Rank Sum tests were used to compare TnT lengths for each combination of time measurements within each cell line. *P*-values *<*0.05 were considered statistically significant.

#### **TnT TETHERING ASSAYS**

Pleural effusion or ascites specimens from cancer patients with MPM and lung adenocarcinoma were obtained via a Memorial Sloan-Kettering Cancer Center (MSKCC) Institutional Review Board (IRB)-approved protocol. Informed written consent was obtained from all patients, and patient identifiers were removed to ensure anonymity. Malignant cells were histologically confirmed by an experienced MSKCC pathologist and seeded in standard tissue culture-treated plates using a clonal dilution assay. Cells were seeded in 24-well non-adherent culture (Nunc Non-Treated Multidishes) and adherent treated tissue culture plates (Fisher Scientific, Pittsburgh, PA) using 10% FBS RPMI. Microscopic imaging was used to confirm the presence of single cells per well, and these wells were marked and monitored daily by microscopic imaging. We additionally performed similar experiments with mesothelioma cell lines VAMT, H2052, and MSTO-211H using an identical approach.

#### **FIXATION AND SAMPLE PREPARATION**

To prepare cells for IF staining, cells were cultured in oneor two-well sterile tissue culture-treated chamber slides (Lab-Tek II Chamber Slide™ system, Nunc, Rochester, NY) or on sterile poly-L-lysine (1 mg/ml; Sigma) coated glass coverslips (VWR VistaVision, catalog no. 16004-312) for 48–72 h using TnT medium to stimulate TnT formation. As TnTs are highly sensitive to movement and to light, we have modified existing protocols for cell fixation and analysis. To perform fixation and prevent disruption of existing nanotubes, 16% w/v paraformaldehyde (PFA) (Alfa Aesar, Ward Hill, MA) was added along the sides of the chambers or the dishes with glass coverslips, keeping the overlying culture medium intact to a final w/v concentration of 4%, After incubation at 4◦C for 1–2 h, the fixative and chambers were removed, and slides were allowed to air dry. We have found this combination provides optimal preservation of intact cells with TnTs to allow for more accurate study of these thin structures. Immunofluorescent staining was then performed to detect expression of the noted proteins.

## **IMMUNOFLUORESCENT STAINING**

The primary antibodies and their working concentrations are as follows: cdc42 (Santa Cruz Biotech, 200 ug/ml, rabbit polyclonal IgG; catalog no. SC-87); NF2/merlin (Santa Cruz Biotech, 200 ug/ml, rabbit polyclonal IgG; catalog no. SC-332), p-selectin (CD62P) (BD Biosciences, 5μg/ml; catalog no. 556087), betatubulin (Sigma–Aldrich, monoclonal anti-acetylated tubulin, clone 6-11B-1; catalog no. T6793-0.2ML); monoclonal anti-β-Tubulin IV (Sigma–Aldrich, catalog no. T7941, clone ONS.1A6); vimentin Alexa Fluor 488 (BD Pharmingen, human IgG; catalog no. 562338), Akt (Sigma–Aldrich, rabbit polyclonal IgG; catalog no. AAB4300259-100UG). Slides were first blocked with blocking solution (10% normal goat serum/2% BSA in PBS) or mouse IgG blocking agent from Vector Labs (catalog no. MKB-2213) for 30 min. Primary antibody incubation lasted 3–7 h at room temperature, followed by 30 min incubation with biotinylated secondary antibodies (Vector Labs, MOM Kit BMK-2202; 1:200 dilution). Detection was performed with Streptavidin-HRP D (Ventana Medical Systems) followed by Tyramide-Alexa Fluor 488 (Invitrogen, catalog no. T20922).

## **DRUG TREATMENT OF CELLS WITH MIGRASTATIN**

Migrastatin core ether is a synthetic analog of migrastatin obtained courtesy of Dr. Samuel Danishefsky; it was used at 100 nM. MSTO-211H cells were prepared as above (i.e., 1 <sup>×</sup> <sup>10</sup><sup>5</sup> cells per well in 24-well tissue culture plates). The number of TnTs was counted in 10 fields per medium condition, at regular time intervals (0, 24, 48, and 72 h) using a 20× objective lens on a Nikon Eclipse Ti inverted microscope (Nikon Instruments, Inc.). Experiments were performed in duplicate. Means were calculated and compared using two-sided, two-tailed *t*-tests assuming unequal variances. *P*-values *<*0.05 were considered statistically significant.

### **TREATMENT OF CELLS WITH OTHER DRUGS (TUNICAMYCIN, DEXTRAN SULFATE, 4-METHYLUMBELLIFERONE) TO ASSESS POTENTIAL ASSOCIATION OF HYALURONAN WITH TnTS**

Tunicamycin (Sigma, catalog no. T7765-1MG, lot no. CAS 11089- 659) was used at a final concentration of 5μg/ml; Poly I:C (Polyinosinic–polycytidylic acid sodium salt; Sigma, catalog no. P0913-10MG, lot no. CAS 42424-50-0) was used at a final concentration of 10μg/ml; dextran sulfate (Sigma, catalog no. D8906-5G) stock solution was made by first dissolving in 2% FBS in phosphate-buffered saline (PBS) to create a stock solution of 100μg/ml, which was then added to culture medium to final concentration 10μg/ml; 4-methylumbelliferone (4-MU; Sigma, catalog no. M 1381) was prepared as a stock solution 20 mg/ml concentration in DMSO, then added to culture medium to final concentration 1.0 mM; Hyaluronidase (Sigma, H3884) was prepared as a stock solution of 10 mg/ml and used at a final concentration of 13μM. For preparation of each drug, stock solution was added to 10% FBS RPMI-1640 medium to obtain final concentrations as listed; this medium was used in cell culture by adding to 1 <sup>×</sup> <sup>10</sup><sup>4</sup> MSTO-211H cells per well of 6-well tissue-culture treated plates (Fisher Scientific, Pittsburgh, PA). Following 48 h of incubation at 37◦C (5% CO2), exclusion assays were performing by adding either U937 mononuclear cells at 4◦C for 1 h or red blood cells (erythrocytes) as noted in the Results Section and per standard protocols (DiCorleto and de la Motte, 1985; Rilla et al., 2008), followed by microscopic imaging. For fluorescent imaging, MSTO-211H cells transfected with a lentivirus expressing green fluorescent protein (GFP) were used along with U937 cells transfected with a lentivirus expressing Tomato Red.

## *IN VIVO* **GROWTH OF MESOTHELIOMA CELLS PRECONDITIONED WITH TnT MEDIUM**

GFP-luciferase expressing MSTO cells were grown in either of two conditions: normal RPMI (10% FBS RPMI, 25 mM glucose) or TnT-inducing medium, which consists of low serum and high glucose RPMI (2.5% FBS RPMI, 50 mM glucose), for 7 days. Both sets of media were supplemented with 1% penicillin/streptoymycin and 2% L-glutamine. All cells were cultured in 6-well adherent tissue culture plates (Fisher Scientific, Pittsburgh, PA) at 37◦C in 5% CO2. Cells (2*.*<sup>9</sup> <sup>×</sup> <sup>10</sup>5) were then suspended in 100μL of RPMI and injected into the peritoneum of each mouse. Ten NOG (NOD/Shi-scid/IL-2RGnull) immunodeficient mice were used for each group. Each mouse was concurrently injected intraperitoneally with 1 ml of thioglycolate as a co-stimulatory inflammatory agent; the rationale for this is that inflammation is known to elicit formation of nanotubes in *in vivo* animal models (Seyed-Razavi et al., 2013). On days 7, 14, 21, and 31 post-tumor inoculation, mice were anesthetized with isoflurane and injected intraperitoneally with 150μL of luciferin (15 mg/mL). Mice were then imaged with an Ivis 200 optical imaging system (Caliper Life Sciences, Hopkinton, MA) 5 min after injection. Capture time was 40 s. Living Image software version 2.5 was used to quantify average radiance (p/s/cm2/sr). Means were calculated and compared using two-sided, two-tailed *t*-tests assuming unequal variances. Overall survival (OS) of the mice was calculated from date of tumor inoculation to date of death, or censored at 40 days for those still alive at the end of the experiment. OS was summarized using a Kaplan-Meier curve and a comparison between groups was made using a Log-Rank test. *P*-values *<*0.05 were considered statistically significant.

## **SECTIONING, STAINING, AND IMAGING HUMAN TUMOR SAMPLES**

Tumor specimens from patients with MPM were obtained via a MSKCC IRB-approved protocol. Informed written consent was obtained from all patients, and patient identifiers were removed to ensure anonymity. Resected intact tumor specimens were placed in PBS. Vibratome sections (100–300 mm thick) were cut and stained using Hoechst 33342 (10 mg/ml) and MitoTracker Red dyes (500 nM) using protocols we developed and have described previously (Lou et al., 2012b). Stained sections were mounted between two glass coverslips and imaged on a confocal microscope.

## **OPTICAL IMAGING OF HUMAN TUMOR SAMPLES AND IMAGE PROCESSING**

Confocal imaging of samples was performed using Zeiss LSM 5Live line-scanning or Leica SP2 point-scanning microscopes using Zeiss oil 406/1.3NA Plan-Neoflur, Zeiss oil 636/1.4NA Plan-Apochromat or Leica water 636/1.2NA HCX PL APO CS objectives. Serial z-stack images were obtained at optimal step size and maximum intensity projection images were produced. The Imaris Viewer program (Bitplane Scientific Software, Inc.) was used to construct and visualize 3-dimensional images of tumor samples. Metamorph (Molecular Devices) image analysis software was used to create still images and movies.

### **ELECTRON-MICROSCOPIC IMAGING OF NANOTUBES**

To perform scanning and transmission EM, 1–3 <sup>×</sup> <sup>10</sup><sup>6</sup> MSTO-211H cells were cultured on Thermanox plastic tissue culture 25 mm cover slips (Lux Scientific Corporation). The fixative— 2.5% glutaraldehyde/2% paraformaldehyde in 0.075 M sodium cacodylate buffer (pH 7.5; 10 ml, Electron Microscopy Sciences, Hatfield, PA)—was added directly to the overlying medium.

### **CELL CULTURE AND RNA ISOLATION**

MSTO-211H cells (8 <sup>×</sup> <sup>10</sup>5) were seeded in 150 cm2 flasks and grown for 7 days using passage medium or TnT medium at 37◦C in a standard humidified chamber with 5% CO2 as already described in the text. After 7 days, the cells were harvested separately following trypsinization and subjected to RNA isolation using mirVana™ total RNA isolation protocol following the protocol of the manufacturer (Life Technologies, Carlsbad, CA). RNA preparation was quantified on a Nanodrop spectrophotometer (Thermo Fisher, Wilmington, DE, USA), and RNA quality was confirmed by resolving on a denaturing 1.2% agarose gel following standard electrophoresis protocol.

#### **cDNA SYNTHESIS AND qRT-PCR**

RNA was subjected to first strand cDNA using miScript II Reverse Transcription kit (Qiagen, Valencia, CA). Total RNA was reverse transcribed for 2 h at 37◦C, and reverse transcriptase was heat inactivated by boiling the reaction mix at 95◦C for 5 min. cDNA (5.0 ng) was diluted and amplified with 10μl of miScript SYBR green PCR mix following the protocol of the manufacturer using gene-specific forward and reverse primers. PCR primers were purchased from a commercial vendor (IDT, Coralville, IA). The nucleotide sequence of the primers used are listed in Supplementary Table 1. The samples were run in triplicate in a Roche Light Cycler II (Roche GmbH, Germany), and values were normalized to the endogenous expression of 18S rRNA. Fold gene expression was calculated by comparative C(T) method (Schmittgen and Livak, 2008) and mean fold expression values relative to expression in control medium were compared using two-sided two-sample *t*-tests assuming unequal variances. *P*-values *<*0.05 were considered statistically significant.

## **TIME-LAPSE IMAGING OF CELLS IN CULTURE**

Time-lapse imaging experiments were performed on Perkin Elmer UltraView ERS spinning-disk confocal microscope or Zeiss LSM 5Live line-scanning confocal microscope. Both microscopes were enclosed in environmental chambers that were maintained at 37◦C with 5% CO2 level. Viable Staining of Cell Lines for time-lapse imaging was performed as we have described previously (Lou et al., 2012b). Briefly, in order to assess the ability of mitochondria to be transmitted between mesothelioma cells via TnTs, we used MitoTracker Red to stain MSTO-211H cells which were then cultured in hyperglycemic, low-serum ("TnT") medium. The cells were cultured in clear-bottomed delta-T culture dishes (Bioptechs Inc., Butler, PA). MitoTracker Red CMX Ros (Invitrogen, M-7512, 50μg/vial) was used at 500 nM to stain mitochondria, per manufacturer's protocols. Stained cells were re-suspended and added to a non-confluent culture of adherent, unstained MSTO-211H cells grown in another dish. Incubation was performed in high glucose medium for 5 h to stimulate formation of TnTs prior to imaging.

## **RESULTS**

## **MALIGNANT MESOTHELIOMA CELLS FORM MORE TnTs THAN BENIGN MESOTHELIAL CELLS**

We propose that TnTs create intercellular networks with the capability of transmitting signals that stimulate proliferation of invasive cancers. To determine whether TnT formation occurs at a higher rate in malignant mesothelioma cells than in benign cells, we cultured the MPM cell lines H2052, VAMT, and MSTO-211H and the benign mesothelial cell line MeT5A. Cells were cultured in medium that we previously demonstrated induces TnT formation (Lou et al., 2012b). Equivalent numbers of cells were added to culture wells and visualized using inverted microscopy at 24, 48, and 72 h; representative images are shown in the accompanying figures (**Figure 1A**; also Supplemental Figure S1 for composite panel of representative images at 24, 48, and 72 h). At each time point, we randomly selected 10 fields of view using the 20× objective and counted the number of TnTs per field. We also counted the number of cells present in each selected field to control for cellular proliferation. For all three malignant mesothelioma cell lines, the average number of TnTs/cell was significantly higher than that seen for the benign mesothelial MeT5A cell line (**Figure 1B**). No evidence of TnT formation was evident in the mesothelial cell line (MeT5A) at 24 h, and thus a ratio could not be reported. As expected, cell proliferation was higher in malignant mesothelioma cell lines as compared to normal MeT5A cells (**Figure 1C**); however, among malignant cells proliferation appeared to be inversely associated with the rate of TnT formation. The ratio of malignant:mesothelial TnTs/cell doubled or tripled from 48 to 72 h for all malignant cell lines (26.73–78.16 for H2052, 9.80–42.66 for VAMT, and 9.80–18.84 for MSTO) (**Figure 1D**). Taken together, these *in vitro* data show that TnTs formed at a markedly higher rate among malignant mesothelioma cell lines than among normal mesothelial cells in inverse proportion to the rate of cell proliferation, ranging from nearly 20-fold to 80-fold higher by 72 h of *in vitro* culture. Moreover, these data support the use of a "nanotube index" to quantitatively assess TnT formation in future studies of the biological role of TnTs in aggressive malignancies. This markedly higher rate of TnT formation in mesothelioma, and likely in other cancers as well, provides evidence to support TnTs as a potential novel target for selective therapy of such cancers.

## **OVERALL TnT LENGTH DECREASES WITH TIME AND WITH PROLIFERATION OF MESOTHELIOMA CELLS**

In the context of a heterogeneous tumor matrix, TnTs may play an important role in long-distance cellular communication.

To accomplish this, TnTs would need to extend to variable lengths depending on the distance of targeted cells. As more cells accumulate, this distance would become shorter. We hypothesized that TnT length would decrease as cells proliferate and accumulate over time in *in vitro* culture. We cultured the MPM cell lines H2052, VAMT, and MSTO-211H and the mesothelial cell line MeT5A. TnT length decreased over time among all malignant cell lines (**Figure 2**); these changes were statistically significant at most time points (**Table 1**). TnT length decreased slightly from day 2 to day 3 among the non-malignant MeT5A cells (**Figure 2**); however, this change was not statistically significant. We depict the data in the form of box plots in order to demonstrate the median and the wide range of lengths we observed in the malignant mesothelioma cell lines (**Figure 2**). Since TnT formation between MeT5A cells was rare, it was not possible to construct box plots for the distribution of TnT length over the three-day period for this non-malignant cell line. The decrease in TnT length observed among malignant cells was most noticeable for H2052 and VAMT cells, but was less dramatic for MSTO cells. This difference could be due to the relatively steady rate of proliferation of H2052 and VAMT cells and high proliferation rate of MSTO cells (**Figure 1C**). In addition, this finding is consistent with our prior study showing that TnTs are most prominent in sub-confluent cultures; in fully confluent cultures, cells are in close proximity, making it either difficult to discern any present TnTs and/or decreasing the number of TnTs that form

**Table 1 |** *P***-values from comparisons of TnTs Length in mesothelioma cells by cell line and time.**


*P-value* ≤ *0.05 was considered significant.*

in conditions that do not require long-distance connections (Lou et al., 2012b).

### **TnT TETHERING OF MESOTHELIOMA CELLS IN** *IN VITRO* **MODEL OF PLEURAL EFFUSIONS**

Advanced thoracic malignancy is frequently associated with accumulation of malignant fluid in the pleural or peritoneal cavities. These effusions often contain detached, free-floating suspended malignant cells that are capable of undergoing epithelial-tomesenchymal transition (EMT), thus increasing their invasive capability in some cancer types, such as lung adenocarcinoma (Chen et al., 2013; Chunhacha et al., 2013). In our prior study, we demonstrated that EMT effectively stimulates TnT formation in mesothelioma (Lou et al., 2012b). We hypothesized that non-adherent viable cells in culture in essence behave similarly to suspended effusion cells in mesothelioma patients. We thus next obtained pleural effusion specimens from 5 patients diagnosed with MPM or lung adenocarcinoma. After isolating malignant cells via centrifugation of pleural effusions, we confirmed the presence of malignant cells by morphology and TnT formation

**FIGURE 3 | TnT tethering of mesothelioma cells in** *in vitro* **models of pleural effusions. (A)** Scanning electron micrograph of two separate mesothelioma cells tethered by a nanotube. **(B,C)** TnTs connecting primary malignant cells from pleural effusions. 40× magnification. **(D,E)** Clonal dilution assay using VAMT (sarcomatoid mesothelioma) cells grown in a low-adherence culture plate using 10% FCS RPMI medium; cells are shown on day 24 of culture in separate wells from the same plate. Arrowheads indicate nanotubes.

among these cells by inverted as well as by scanning electron microscopy (**Figure 3A**). We then cultured cells *in vitro* in standard tissue culture-treated plates and confirmed formation of TnTs connecting these patient-derived, primary malignant cells (**Figures 3B,C**).

We noted that malignant cells presented in effusions as dispersed single cells or as spheroid aggregates that could be disassembled into single cells through trypsinization or by physical separation with vigorous pipetting. We hypothesized that clusters of cells derived from a single parent cell could form TnTs in suspension and without full cell adherence to the substratum of the culture plate. To investigate this possibility, we performed separate clonal dilution assays using VAMT and H2052 cell lines cultured in 24-well non-adherent culture plates. We visually confirmed the presence of single cells and marked these wells for further daily follow-up. We performed daily microscopic imaging and reproducibly detected growth of groups of cells derived from parent cells forming mesh-like syncytial networks of TnTs connecting daughter cells. Cell aggregates formed prominently under non-adherent culture conditions while maintaining extensive TnT connections; these aggregates were in many instances connected to each other by nanotube structures while remaining suspended in culture medium (**Figures 3D,E**; also Supplemental Figure S2). As TnTs are in essence 3-dimensional structures (i.e., non-adherent to the substratum), they may "oscillate" upon movement of the culture plate. Other actin-based structures, such as invadopodia, do not demonstrate this trait. Using fine focus to evaluate TnTs in live cell culture also helps to differentiate TnTs from other cell extensions. The observation that primary malignant cells formed a syncytial "network" connected by TnTs indicates that these or other cytoplasmic extensions may play an unrecognized role in facilitating communication among suspended malignant cells and malignant/mesothelial cells adherent to the pleural lining of the thoracic cavity.

These data suggest that TnTs may play a role in tethering suspended, non-adherent cells. As development of pleural effusions or abdominal ascites is a hallmark of a number of aggressive solid tumor malignancies—most especially malignant pleural and peritoneal mesotheliomas, as well as lung adenocarcinomas these results provide potential insight into the cellular behavior of malignant cells at the advanced stages of cellular invasion. They also build upon our work demonstrating TnT formation between primary malignant mesothelioma cells *in vitro* (Lou et al., 2012b) as well as similar work demonstrating TnTs between human peritoneal mesothelial cells in culture (Ranzinger et al., 2011). Further *in vivo* studies will be necessary to clarify whether such TnT connections occur among malignant pleural or peritoneal mesothelioma cells invading the mesothelial lining.

#### **DRUG INHIBITION OF MESOTHELIOMA TnTs USING MIGRASTATIN, AN INHIBITOR OF FASCIN**

We previously demonstrated that fascin localizes at the site of TnT extrusion from the plasma membrane at the leading edges of cells (Lou et al., 2012b). To determine whether fascin inhibition blocks TnT formation, we used migrastatin core ether (ME), a drug derived from migrastatin, a polyketide product initially derived from *Streptomyces* (Oskarsson et al., 2010). Synthetic analogs of migrastatin inhibit migration of cancer cells (Oskarsson et al., 2010) by targeting fascin and thereby blocking tumor progression (Chen et al., 2010). MSTO cells treated with ME exhibited a statistically significant difference with fewer TnTs, compared with the Control group without drug treatment, at 24 h (*p* = 0*.*036) and at 72 h (*p* = 0*.*010) (**Figure 4**). There was not significant difference at the 48-h timepoint.

### **THREE-DIMENSIONAL** *IN VITRO* **MODEL OF MESOTHELIOMA TUMOR MICROENVIRONMENT**

Routine use of 3-dimensional modeling both *in vitro* and eventually *in vivo* will be critical to advancing the field of TnT biology, in cancer and in other diseases. To develop a 3-dimensional *in vitro* model of the tumor microenvironment, we used 1% agarose to culture mesothelioma cells to simulate suspension of cells within a 3-D viscous matrix. TnTs visualized in the 3D matrix were readily seen forming TnTs vertically and horizontally within the agarose matrix as compared to 2D tissue culture (standard tissue culture in **Figure 5A**; with TnT medium in 1% agarose, **Figure 5B**). Additionally, z-stacked confocal imaging of TnTs connecting cells stained with immunofluorescent antibodies can be used to visualize TnTs 3-dimensionally. Using this technology, we depicted a representative TnT stained with an immunofluorescent antibody to vimentin (**Figure 5C**; **Movie S2**, depicting rotating

**FIGURE 4 | Effect of Migrastatin on TnT formation in MSTO cells.** Migrastatin was used at 100 nM. TnTs were counted in 10 fields per timepoint per condition using 20× objective lens, and the results averaged. Experiments were performed in duplicate. Comparisons between Migrastatin and the Control were statistically significant at 24 h (*p* = 0*.*036) and at 72 h (*p* = 0*.*010). Means ± standard deviations are presented. Asterisk indicates statistical significance (*p <* 0*.*05).

**FIGURE 5 | 3-dimensional modeling of TnTs** *in vitro***. (A)** MSTO cells forming TnTs in regular tissue culture treated plate; **(B)** MSTO cells cultured in 1% agarose, forming TnTs vertically as well as horizontally within the agarose matrix. **(C)** 3-dimensional modeling of TnT connecting cells using confocal imaging with z-stacking (IF staining performed using fluorescent vimentin-specific antibody); also see **Movie S2**. Arrowheads indicate nanotubes.

3-dimensional model of this image). Immunofluorescence staining indicated the presence of vimentin along the length of TnTs.

In the case of solid tumor malignancies, including mesothelioma, standard and conventional evaluation of tumors involves histopathologic analysis of extremely thin 2-dimensional tumor

**FIGURE 6 | Comparison of 2-dimensional histopathology section vs. 3-dimensional confocal imaging of primary human malignant mesothelioma tumors** *ex vivo***, with specific examination for TnTs. (A)** Hematoxylin and eosin (H&E) staining of sectioned primary human mesothelioma tumor; arrowheads indicate putative nanotube-like

structures; **(B,C)** compiled confocal image produced by combining individual z-stacked images of representative portions of tumor tissue resected from a human patient. Tumors were stained with MitoTracker and Hoechst 334 dye; arrowheads indicate nanotube structures. Please also see **Movie S5**.

sections (**Figure 6A**). While slides prepared in this manner may potentially yield views of putative nanotube-like structures, advances in microscopic imaging are required to more effectively study 3-dimensional tumors. Advances in microscopic imaging that allow for layered z-stacked images of cells *in vitro* and methods we have developed to image *ex vivo* tumors provide an enhanced approach that allows for more advanced analysis of nanotubes connecting malignant cells in the stroma-rich tumor microenvironment (**Figures 6B,C**; Also see **Movie S5**). These visually detailed 3D images of mesothelioma cells provide further impetus for studying TnTs in this manner *in vitro*.

#### **HYALURONIC ACID IS NOT ASSOCIATED WITH TnTs**

Hyaluronic acid (HA), or hyaluronan, is a well-studied glycosaminoglycan that is secreted into the extracellular matrix; increased production of HA induces increased cell motility and an invasive phenotype in mesothelioma (Li and Heldin, 2001). Hyaluronan receptors are expressed preferentially on malignant mesothelioma cells but not on normal mesothelium (Asplund and Heldin, 1994). However, normal mesothelial cells and malignant cells derived from several organ sites synthesize relatively high quantities of hyaluronan, whose pericellular coat comprises bunches of short, adherent membranous protrusions consistent with actin-based stress fibers and microvilli (Kultti et al., 2006; Rilla et al., 2008). These coats create zones that have been well-described and are readily visualized microscopically using erythrocyte exclusion assays or the equivalent (McBride and Bard, 1979; Rilla et al., 2008). Conditions of cellular stress induced by either hyperglycemia, viral mimetic agent poly I:C, tunicamycin, or dextran sulfate, to name just several examples, have been shown to induce increased hyaluronan production and hyaluronan-based cellular "cables" that induce monocyte adhesion *in vitro* (Kultti et al., 2006); tunicamycin and dextran sulfate in particular induce endoplasmic reticulum-related metabolic stress that leads to increased production of hyaluronic acid, which in turn attracts and leads to increased adhesion of leukocytes via surface binding of CD44 (de La Motte et al., 1999; Majors et al., 2003; Wang and Hascall, 2004; Lauer et al., 2009). In our earliest studies examining what we later confirmed to be TnTs in mesothelioma, we performed standard exclusion assays using primary red blood cells (erythrocytes) or U937 lymphocyte (mononuclear) cells (Kultti et al., 2006; Rilla et al., 2008), but found no visual evidence of either pericellular zones or monocyte adhesion to TnT structures, indicating that TnTs were unlikely to harbor a significant amount of hyaluronan externally, also demonstrating that these entities are distinct from hyaluronan cables (**Figure 7**; Supplemental Figure S3). For fluorescent imaging and confirmation, we used MSTO-211H cells transfected with a lentivirus expressing GFP which we have used and described previously (Lou et al., 2012b) along with U937 cells transfected with a lentivirus expressing Tomato Red. To convincingly confirm that the extensions connecting cells were indeed TnTs, we performed time-lapse imaging that visibly demonstrated intercellular transfer of GFP (**Movies S3**, **S4**).

We also applied the HA-stimulating drug tunicamycin (final concentration 5μg/ml) to MSTO cells in culture and examined these cells every 24 h up to 96 h. However, this led to no changes in TnT formation or morphology. Likewise, separate addition of dextran sulfate (10μg/ml) to MSTO cells in culture led to cellular aggregation not unlike aggregates seen in patient effusion samples (Supplemental Figure S4). Poly I:C (10μg/ml) induced a transformation of MSTO cells into a more mesenchymal, spindle-cell morphologic appearance without alteration of TnTs. This finding is consistent with similar effects of this drug on stimulating EMT in other cell types (Harada et al., 2009). We further treated MSTO cells with 4-methylumbelliferone (1.0 mM), an inhibitor of hyaluronan synthase and thus of hyaluronan cables (HAS) (Morohashi et al., 2006; Kultti et al., 2009); in separate wells we also assessed potential effects of the enzyme hyaluronidase (13μM); neither drug had any effect on TnTs, consistent with the above data indicating HA does not play a notable role in TnT formation or maintenance (Supplemental Figure S4).

### **DECREASED TUMOR GROWTH IN MICE IMPLANTED WITH TnT-PRIMED MESOTHELIOMA CELLS ALSO CORRESPONDS WITH DECREASED SURVIVAL**

Animal studies have identified nanotubes or similar structures *in vivo* in an inflammatory corneal mouse model (Chinnery et al., 2008; Seyed-Razavi et al., 2013) and *ex vivo* in adult mouse heart tissue (He et al., 2011), mouse alveoli (Islam et al., 2012),

mononuclear cells (expressing Tomato Red) were added to cell culture growing MSTO-211H cells (expressing GFP) with TnTs. **(A)** Brightfield and **(B)** Fluorescent views of the same field of view are shown. Arrowheads indicate nanotubes.

rabbit kidney parenchyma (Minuth and Denk, 2012), and mouse embryo non-neural ectoderm (Pyrgaki et al., 2010). Our group was the first to image TnTs in resected human tumor samples, initially on tumors from patients with MPM and lung adenocarcinoma (Lou et al., 2012b); we have been able to reproduce this finding using human mesothelioma tumors described in the current study (**Movie S5**). Our group has further extended demonstration of nanotube structures in malignant human ovarian tumors (Thayanithy et al., 2014a) as has another group (Pasquier et al., 2013). However, the technical difficulties of imaging nanotubes in the *in vivo* setting remain highly challenging. To assess effects of TnTs on *in vitro* cell proliferation, we used MSTO cells pre-conditioned in culture medium that we previously demonstrated increases the rate of TnT formation *in vitro* (Lou et al., 2012b). We conditioned MSTO cells in either low serum, hyperglycemic (2.5% FBS, 50 mM glucose) RPMI medium (referred to as "TnT medium") or control passage RPMI medium (10% FBS, 25 mM glucose) for 7 days. This experiment demonstrated that proliferation of MSTO cells in low-serum, hyperglycemic medium was approximately half that of cells cultured in passage medium by 72 h (**Figure 8A**). To next examine the effect of TnTs on tumor growth *in vivo*, we used a NOG xenograft mouse model of malignant mesothelioma. We conditioned MSTO cells transfected with a lentivirus expressing luciferase in either TnT medium or control passage RPMI medium (10% FBS, 25 mM glucose) for 7 days. We then injected these cells into the peritoneal cavity of NOG immunocompromised mice. Mice were bioimaged every 7 days up to 31 days and the average radiance was measured; interestingly, by 31 days the mice injected with cells pre-conditioned in low-serum, hyperglycemic medium had developed less tumor burden than mice injected with the same cell line pre-conditioned in passage medium (**Figure 8B**). Thus, this*in vivo* finding mirrored the *in vitro* studies that demonstrated that proliferation of MSTO cells in low-serum, hyperglycemic medium was approximately half that of cells cultured in passage medium. Moreover, none of the NOG mice injected with control medium (*n* = 10) had died by day 31, but 2 mice injected with cells pre-conditioned in low serum, hyperglycemic medium (*n* = 10) had died by day 31 and an additional one died just after imaging (*p* = 0*.*067, **Figure 8C**); in a separate experiment repeating this approach, 5 of 10 mice with injected cells preconditioned in TnT medium died by Day 31, whereas 0 of 10 died by that day (data not shown). Using weight as a surrogate measure for morbidity, mice injected with cells primed with TnT medium displayed a sharper drop in weight over time than did mice injected with cells cultured in passaged medium (data not shown). Bioimages demonstrating the visual differences between the two groups are shown (**Figure 8D**). These *in vivo* findings set the stage for further evaluation of the potential role for TnTs in solid tumor malignancies, possibly by increasing the locoregional but not distant invasive capability of mesothelioma cells, with a mechanism independent of cell proliferation.

#### **GENE EXPRESSION PROFILING OF TnT-PRIMED MESOTHELIOMA CELLS**

Due to the above findings, we next sought to determine relative differences in gene expression between MSTO cells conditioned in low serum, hyperglycemic (2.5 % FBS, 50 mM glucose) RPMI medium (i.e., "TnT medium") or control passage RPMI medium (10% FBS, 25 mM glucose). We first investigated RNA levels of M-Sec (which is also called TNFaip2, or tumor necrosis factorα-induced protein 2) and leukocyte specific transcript 1 (LST1), two gene products that are known to be enriched in TnTs (Hase et al., 2009; Schiller et al., 2013). Both genes were significantly upregulated in MSTO cells cultured in TnT medium compared to control medium (**Figure 9A**). We then investigated whether TnT medium, which is significantly lower in essential nutrients and also includes low percent of added serum (2.5% FBS) relative to passage medium (10% FBS), affects genes that promote cell cycle progression (Bracken et al., 2004; Nalepa et al., 2013). RNA levels of E2F1, its downstream targets CCNA2 and CDC20, and CDKN3 were significantly lower in MSTO cells grown in TnT medium than in cells grown in control medium (**Figure 9B**). This finding is consistent with our observation that cells grown in TnT medium undergo a lower rate of cell division. We next studied whether key genes involved in cellular migration, invasion, and metastasis are altered in mesothelioma cells cultured in TnT medium (Scholler et al., 1999; Rittling and Chambers, 2004; Guttery et al., 2010; Al-Alwan et al., 2011; Servais et al., 2012; Pietras et al., 2014). Relative to the MTSO cells grown in normal medium, tenascin-C, CD44, osteopontin, fascin, and mesothelin were all significantly induced in MSTO cells grown in TnT medium (**Figure 9C**). Further studies will evaluate whether induction of these genes in cells grown in TnT medium induce an adaptive gene expression profile leading to TnT formation and a higher propensity to invade, migrate, and metastasize.

#### **DISCUSSION**

Intercellular communication between cancer cells is crucial to the progression of invasive cancers, but the mechanisms by which communication occurs between distant and proximal cells in a tumor matrix remains poorly understood. TnTs are a novel candidate to explain how this communication process occurs (Lou et al., 2012a). Our prior and current work have consistently demonstrated TnTs in malignant mesothelioma and lung adenocarcinoma tumors from human patients (Lou et al., 2012b). This observation is consistent with the finding of another group

that successfully imaged membrane nanotubes *in vivo* using an inflammatory cornea animal model (Chinnery et al., 2008). We have further demonstrated that TnTs are not exclusive to malignant mesothelioma and lung adenocarcinoma, but can form between malignant cells from a wide variety of histologic cancers, including pancreatic cancer as one example (Lou et al., 2013). In the present study, we describe our approaches to studying the relevance of TnTs in invasive malignancies, specifically in the context of MPM.

## **TnT LENGTH AS A FUNCTION OF THE NEED FOR INTERCOMMUNICATION AMONG MESOTHELIOMA CELLS**

When interpreting our findings of both cell proliferation rate and changes in TnT lengths, we take into account likely differences in natural biology and aggressiveness of the variable cell histologies (i.e., VAMT = sarcomatoid; H2052 = epithelioid, likely less aggressive; MSTO-211H = biphasic, encompassing features of both of VAMT and H2052). It is logical that as cells proliferate and are also motile, with less distance between cells over time, the average TnT length would decrease over time. The most aggressive cell line (VAMT, sarcomatoid) displayed the longest TnT length at 24 h (**Figure 2**), but not the highest proliferation rate by 72 h (**Figure 1C**); conversely, MSTO cells had the highest proliferation rate, but relatively shorter TnTs. Logically, this presents an interesting supposition: That cells proliferating at a low rate can be just as—if not more—aggressive, perhaps by forming longer TnTs and/or more TnTs. Knowing the clinical outcomes are worse with sarcomatoid variants than with other forms of mesothelioma, we postulate that this may hold true in the clinical setting. Further investigation is warranted based on these findings.

A key clinical manifestation of advanced thoracic malignancy (i.e., MPM and lung adenocarcinoma) is accumulation of pleural

in TnT media. Conversely, genes attributed to cellular invasion, migration and metastasis functions are upregulated (∗indicate *p*-values *<*0.05).

fluid that marks development of pleural effusions. Ascites is a similar process marked by accumulation of fluid in the abdominal cavity or peritoneum. In advanced cancers such as pleural or peritoneal mesothelioma, the accumulation of malignant fluid is a diagnostic and prognostic hallmark of aggressive disease. These effusions often contain a significant number of free-floating or suspended malignant cells; diagnosis can be made following cytologic examination of extracted fluid. However, beyond examination as a mere diagnostic marker, little is known about the role this specific cell population plays in the propagation of mesothelioma and disease progression. This population of free-floating, suspended cells is capable of undergoing epithelialto-mesenchymal transition (EMT), thus increasing their invasive capability. In our prior study, we demonstrated that EMT effectively stimulates TnT formation in mesothelioma, as does acidic pH (Lou et al., 2012b) in the context of low-serum, hyperglycemic medium. This is especially important as low pH of pleural fluid derived from malignant effusions, including those derived from malignant mesothelioma, is a poor prognostic factor associated with decreased overall patient survival (Sahn and Good, 1988; Gottehrer et al., 1991).

In the present study, we examined extra-cytoplasmic actin extension of TnTs. We found that as mesothelioma cells proliferated the average TnT length decreased over time, possibly as a result of the low cellular requirement for long-distance connections among confluent cultures. Further, our *in vitro* model of pleural/peritoneal effusions indicated that TnTs may function as tethers that link suspended, non-adherent malignant cells to other suspended cells or adherent cells of the pleural lining of the thoracic cavity. These findings indicate that non-adherent viable cells in culture behave similarly to suspended effusion cells from mesothelioma patients and that these cells are capable of forming TnTs. Relative differences in TnT length may be a function of chemotactic factors promoting their formation and guiding their extension; a precedent has been established for this in correlation of length of cytonemes (actin-based filopodial protrusions similar to TnTs) to length of chemotactic gradients of Hedgehog signals in Drosophila wing disc models (Bischoff et al., 2013).

#### **DEVELOPING** *IN VITRO* **MODELS THAT SIMULATE POTENTIAL TnT ACTIVITY IN AN** *IN VIVO* **TUMOR MICROENVIRONMENT**

To study the formation and function of TnTs, 3-dimensional *in vitro* models are needed to simulate the complex tumor microenvironment. A major challenge to 3-dimensional models of TnT signaling is the presence of other forms of external signaling that may confound study results. Exosomes, microvesicles, and other freely diffusible signals play an established role in intercellular signaling. Indeed, data recently published by our group demonstrated that tumor-derived exosomes stimulate TnT formation in mesothelioma cell culture (Thayanithy et al., 2014b). A previous report showed that other forms of external signaling could be excluded by culturing normal rat kidney cells in a viscous agarose matrix; this approach significantly decreased diffusion of extracellular signals, including from microvesicles or other free particles, while permitting TnT formation (Gurke et al., 2008). For the first time, in the current study we apply use of "TnT medium" (low-serum, hyperglycemic culture medium) to agarose in a relatively viscous microenvironment that not only remains permissive for, but also further induces, TnT formation in a manner that is reliable for further study. We propose using this approach for future studies that aim to minimize potential effect of exosomes and other diffusible factors that act as stimulatory signals and intercellular carriers of cargo, thus isolating examination of TnTs for specific analyses.

#### **TnTs AS A THERAPEUTIC TARGET FOR MESOTHELIOMA AND OTHER TnT-FORMING MALIGNANCIES**

TnTs are not exclusive to cancer and are a cellular entity in "normal," non-malignant cells. For the purposes of studying the potential role of TnTs in cancer, we developed a tool to

**39**

quantitatively assess TnT numbers that could be used as a "nanotube index" to study how increased "intercellular trafficking" of cellular cargo via TnTs is related to cell transformation and tumor formation. We adapted confocal microscopic techniques to visualize narrow nanotube structures. Quantitative assessment of TnTs/cell indicated that TnTs formed at a markedly higher rate among malignant mesothelioma cell lines than among normal mesothelial cells and in inverse proportion to the rate of cell proliferation. In fact, we found that the culture condition that most increased TnT numbers (low-serum, hyperglycemic medium, which we call "TnT medium") did not lead to a corresponding increase in cell number, but rather a notable decrease in cell proliferation by approximately half. We attribute this decreased cell proliferation to both elements, i.e., low serum concentration (2.5% FBS) as well as high glucose, which induces reductions in cell proliferation as well as increased cell apoptosis of pericytes *in vitro* (Beltramo et al., 2006). These data indicate that TnT formation and cell proliferation are distinct processes that may occur during specific stages of malignant growth.

We previously used several inhibitors of pathways that have been implicated in actin-based cell invasion, including latrunculin A, an actin-destabilizing agent that is commonly used in *in vitro* studies and that has been used in multiple TnT studies (Tavi et al., 2010; Lou et al., 2012b; Vallabhaneni et al., 2012). We previously investigated potential metabolic pathways essential for TnT formation in mesothelioma and effectively demonstrated suppression of TnT formation using drugs that are used in the clinical setting for other malignancies, such as an mTOR inhibitor and the widely available drug metformin (Lou et al., 2012b), which stimulates AMP-activated protein kinase (AMPK) and thus indirectly stimulates the mTOR pathway as well (Zhou et al., 2001). This is particularly important in context of our findings in the current study that effusion-derived mesothelioma cells (from both cell lines and from human patients) form aggregate spheroids that are tethered by TnTs. *In vitro* 3-D spheroid models of mesothelioma demonstrate increased resistance to drugs and apoptosis compared to 2-dimensional cultures, and this is at least in part mediated by mTOR; however, inhibitors of mTOR can overcome this acquired resistance (Barbone et al., 2008; Wilson et al., 2008). Considering susceptibility of TnTs to mTOR inhibition, TnTs may play an important role in mediating chemoresistance of spheroids.

In the current study, we demonstrate that an additional rational drug—migrastatin—suppresses TnT formation in mesothelioma. Migrastatin has been found to be potent in blocking migration and metastasis of lung cancer in separate studies (Lecomte et al., 2011), providing impetus for further exploration of this drug as a potential novel therapeutic agent for both lung adenocarcinoma and MPM.

#### **ACCURATE** *IN VIVO* **ASSESSMENT OF TnT FUNCTIONS**

*In vivo* examination of TnTs remains a major barrier to the study of the function of TnTs in cancer. *In vivo* studies of nanotubes have been limited to an inflammatory corneal model in mice (Chinnery et al., 2008; Seyed-Razavi et al., 2013), but successful intravital microscopy has been performed visualizing similar cellular extensions called cytonemes in Drosophila (Bischoff et al., 2013). We also developed specific protocols for imaging putative TnTs in human tumors, as our group's main research focus is cancer cell biology of highly invasive malignancies. In previous studies, we obtained 6 fresh intact tumor specimens from patients with either MPM or lung adenocarcinoma immediately following surgical resection. Using a Vibratome, we processed the specimens into thin cuts that were amenable to staining with Hoechst dye to visualize nuclei and MitoTracker Red to visualize the cell body and TnTs, which we had already demonstrated to work *in vitro* (Lou et al., 2012b). Of the six human patient-derived tumors, we detected TnTs in all six using confocal microscopy and 3-dimensional reconstruction using the Imaris software. TnTs have also been reported in ovarian tumor explants by our group and others (Pasquier et al., 2013; Thayanithy et al., 2014a). In the present study, cells primed in TnT medium had a lower proliferation rate *in vitro*; proliferation rate of these cells was similar when injected *in vivo*, as measured by lower detectable radiance. This is also supported by our finding that the level of genes involved in cell proliferation were also expressed at lower levels in cells that were primed in TnT medium compared to the cells primed with normal media (**Figure 9**). Although mice implanted with TnT-primed mesothelioma cells had lower overall disease burden, they also had poorer survival compared to mice implanted with mesothelioma cells pre-conditioned in control medium. Thus we postulate that a higher rate of formation of TnTs *in vivo* is associated with a higher level of local invasion of tumors, leading to lower survival in this animal model. Future work in our lab will follow up on these findings.This *in vivo* model will prove to be especially important to examining TnTs as a potential therapeutic target for treatment of cancer and subsequent studies for drug delivery via TnTs.

#### **CONCLUSIONS**

In Conclusion, the mechanisms by which cells communicate with one another in the tumor microenvironment are not well understood (Bissell and Radisky, 2001; Pietras and Ostman, 2010; Bissell and Hines, 2011). Our works challenges the current paradigm that gap junctions, exosomes, or cytokines and other diffusible chemical signals are exclusive modes by which cells mediate intercellular communication in mesothelioma. Tunneling nanotubes are a novel biologic conduit for intercellular signaling and transport of cellular cargo. At this time, there appear to be more questions than answers in terms of what the mechanisms, functions, and role of these nano-sized structures are in various cell types. TnTs have attracted the interest of researchers across a spectrum of fields including neuroscience, immunology, infectious diseases, and cancer as described here. What remains unknown is how these diseases use TnTs to coordinate "social networking" among connected cells, which characteristics of TnTs are universal across cell types (e.g., cancer vs. non-cancer cells), and which aspects may be unique to the cell type studied. Researchers with interest in cellular communication may adapt their approach to the study of TnTs according to their research objectives. In this paper, we describe our approach to studying the relevance of TnTs in cancer, specifically in the context of the solid tumor matrix of aggressive malignancies.

## **ACKNOWLEDGMENTS**

We wish to thank Tao Tong, Sho Fujisawa, Ph.D, Yevgeniy Romin, Afsar Barlas, M.D., Sanghoon Oh, Ph.D. and Jiang-Cheng Wang, Ph.D. of the Molecular Cytology Core Facility at MSKCC for their helpful suggestions, expertise, and assistance with imaging and immunofluorescence; Mesruh Turkekul and Ning Fan for assistance with immunofluorescent staining; Nina Lampen for assistance with electron microscopy; Samuel Danishefsky, Ph.D for providing the migrastatin core ether compound; Valerie Rusch, M.D. of the Thoracic Surgery Service at MSKCC for providing primary pleural effusion and ascites specimens; and Ms. Katherine DeBeer for administrative assistance and support. We also thank Michael Franklin, M.S. for excellent editorial suggestions and critical review of the manuscript. This research was kindly funded and supported in part by the Baker Street Foundation, Minnesota Masonic Charities, Minnesota Medical Foundation, Institutional Research Grant #118198-IRG-58-001- 52-IRG94 from the American Cancer Society, the Deborah Powell Center for Women's Health at the University of Minnesota, the National Pancreas Foundation, and the UMN Clinical and Translational Science Institute KL2 Scholar Award (to Emil Lou, grant number 5KL2TR113). Additional funding was provided by the National Center for Advancing Translational Sciences of the National Institutes of Health Award Number UL1TR000114. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

### **SUPPLEMENTARY MATERIAL**

The Supplementary Material for this article can be found online at: http://www.frontiersin.org/journal/10.3389/fphys. 2014.00400/abstract

#### **Movie S1 | Time-lapse microscopy imaging of intercellular transfer of**

**mitochondria between mesothelioma cells connected via a TnT.** We used MitoTracker Red to stain MSTO-211H cells which were then cultured in hyperglycemic, low-serum ("TnT") medium. Cells were allowed to adhere and form TnTs, which were then identified for analysis. Images were taken every 20 min for 24 h using a Perkin–Elmer spinning-disk confocal microscope.

**Movie S2 | 3-dimensional modeling of a nanotube.** Cells were stained with vimentin (red) and nuclear staining with Hoechst 33342 dye (blue) as described. Images were were taken as a z-stack at 40× and compiled using Imaris Viewer software.

#### **Movie S3 | Intercellular transfer of GFP between mesothelioma cells**

**connected via TnTs.** GFP-expressing (green) MSTO-211H cells were co-cultured with MSTO cells stained with the lipophilic dye DiI (red) as we have reported previously (Lou et al., 2012b) and cultured in 37◦C, 5% CO2 for time-lapse imaging. Images were taken every 15 min for 5 h. In this sequence, the middle cell (green) is connected to two cells simultaneously via TnTs, which facilitate transfer of GFP both to and from that cell.

#### **Movie S4 | Intercellular transfer of GFP via TnT connecting two**

**MSTO-211H cells.** Higher-magnification view and time-lapse microscopy demonstrating bidirectional transfer of GFP between connected cells.

**Movie S5 | 3-dimensional reconstruction of a tumor surgically resected from a human patient with malignant pleural mesothelioma.**

3-dimensional imaging was performed using the Imaris Viewer.

## **REFERENCES**


**41**


co-culture of mesenchymal stem cells and nucleus pulposus cells. *PLoS ONE* 7:e33739. doi: 10.1371/journal.pone.0033739


**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.

*Received: 31 May 2014; accepted: 26 September 2014; published online: 31 October 2014.*

*Citation: Ady JW, Desir S, Thayanithy V, Vogel RI, Moreira AL, Downey RJ, Fong Y, Manova-Todorova K, Moore MAS and Lou E (2014) Intercellular communication in malignant pleural mesothelioma: properties of tunneling nanotubes. Front. Physiol. 5:400. doi: 10.3389/fphys.2014.00400*

*This article was submitted to Membrane Physiology and Membrane Biophysics, a section of the journal Frontiers in Physiology.*

*Copyright © 2014 Ady, Desir, Thayanithy, Vogel, Moreira, Downey, Fong, Manova-Todorova, Moore and Lou. 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.*

## Membrane nanotubes between peritoneal mesothelial cells: functional connectivity and crucial participation during inflammatory reactions

#### *Julia Ranzinger <sup>1</sup> \*, Amin Rustom2 and Vedat Schwenger <sup>1</sup>*

*<sup>1</sup> Department of Nephrology, University of Heidelberg, Heidelberg, Germany*

*<sup>2</sup> Department of New Materials and Biosystems, Max Planck Institute for Intelligent Systems, Stuttgart, Germany*

#### *Edited by:*

*Sotirios G. Zarogiannis, University of Thessaly, Greece*

#### *Reviewed by:*

*Martyn P. Mahaut-Smith, University of Leicester, UK Abelardo I. Aguilera, Hospital Universitario de la Princesa, Spain*

#### *\*Correspondence:*

*Julia Ranzinger, Department of Nephrology, University of Heidelberg, Im Neuenheimer Feld 162, Heidelberg, 69120, Germany e-mail: julia.ranzinger@ med.uni-heidelberg.de*

Peritoneal dialysis (PD) has attained increased relevance as continuous renal replacement therapy over the past years. During this treatment, the peritoneum functions as dialysis membrane to eliminate diffusible waste products from the blood-stream. Success and efficacy of this treatment is dependent on the integrity of the peritoneal membrane. Chronic inflammatory conditions within the peritoneal cavity coincide with elevated levels of proinflammatory cytokines leading to the impairment of tissue integrity. High glucose concentrations and glucose metabolites in PD solutions contribute to structural and functional reorganization processes of the peritoneal membrane during long-term PD. The subsequent loss of ultrafiltration is causal for the treatment failure over time. It was shown that peritoneal mesothelial cells are functionally connected via Nanotubes (NTs) and that a correlation of NT-occurrence and defined pathophysiological conditions exists. Additionally, an important participation of NTs during inflammatory reactions was shown. Here, we will summarize recent developments of NT-related research and provide new insights into NT-mediated cellular interactions under physiological as well as pathophysiological conditions.

**Keywords: peritoneal dialysis, peritoneal mesothelial cells, inflammation, nanotubes, oxidative stress**

## **SECTION**

Peritoneal dialysis (PD) is an accepted alternative to hemodialysis in the treatment of end-stage renal disease. However, when compared internationally, PD-treatment is still underrepresented in Germany (Lameire and Van Biesen, 2010). In this treatment, the peritoneal membrane functions as semipermeable membrane allowing for ultrafiltration and diffusion (Krediet, 1999). In order to produce osmotically induced ultrafiltration, heat sterilized glucose containing dialysis solutions are used to ensure hyperosmolarity. The heat sterilization leads to the formation of glucose degradation products (GDPs) (Wieslander et al., 1995) which mainly contribute to the formation of advanced glycation endproducts (AGEs) including methylglyoxal (MG) (Nakayama et al., 1997; Muller-Krebs et al., 2008). The implantation and presence of an indwelling catheter, high glucose concentrations, and GDPs in the dialysis solutions as well as peritonitis - a known complication of PD - coincide with high levels of proinflammatory cytokines within the peritoneal cavity and subsequent induce chronic inflammatory conditions leading to structural and functional changes of the peritoneal membrane (Witowski et al., 2000; Flessner et al., 2007, 2010).

Moreover, a strong induction of the expression of the receptor for advanced glycation end-products (RAGE) in the peritoneal membrane of uremic patients was demonstrated which further increased after PD treatment (Kihm et al., 2008), showing AGE-RAGE interactions being crucial in peritoneal damage due to inflammatory conditions, uremia, and PD. The resulting loss of ultrafiltration, as well as progressive fibrosis, angiogenesis, and vascular degeneration limit long-term PD-treatment (Margetts and Churchill, 2002; Devuyst et al., 2010).

It is now clear that peritoneal mesothelial cells, specialized epithelial cells that line the peritoneal cavity, not only function as non-adhesive surface to facilitate intracoelomic movement. These cells are crucial for the maintenance of peritoneal homeostasis, transport processes across the peritoneal membrane and tissue repair (Mutsaers, 2004; Yung and Chan, 2007). Furthermore, they provide defense against bacterial insult and are essentially exposed to the bioincompatible dialysis solutions during PD-treatment. In response to Tumor-Necrosis-Factor (TNF) and Interleukin-1 (IL-1) secreted by peritoneal macrophages, peritoneal mesothelial cells synthesize various cytokines, including IL-1, IL-6, and IL-8, thus enhance the inflammatory signal and recruit leukocytes in the peritoneal cavity (Douvdevani et al., 1994; Topley, 1995; Li et al., 1998). In the immune system, peritoneal mesothelial cells bear an effective antigen-presenting function for T cells and thereby play a relevant role during the immune response in the peritoneal cavity e.g., during peritonitis (Valle et al., 1995; Hausmann et al., 2000). However, with increasing PDtreatment, peritoneal mesothelial cells undergo a progressive loss of their epithelial phenotype toward a myofibroblast-like phenotype (Yanez-Mo et al., 2003). Thereby, a loss of characteristic cell-cell junctions, apical-basal polarity as well as reorganization of the cytoskeleton and reprogramming of the gene expression take place (Lamouille et al., 2014). This differentiation process, known as epithelial-mesenchymal transition (EMT), reflects the enormous plasticity of mesothelial cells (Yung and Chan, 2012).

In the context of inflammatory immune reactions, intercellular communication plays a crucial role. In 2004, the discovery of Nanotubes (NTs) mediating membrane continuity has extended the understanding of cell-to-cell communication (Rustom et al., 2004). These NTs were initially characterized as thin intercellular membrane channels, formed between cultivated pheochromocytoma (PC12) cells at their nearest distance and without contact to the substratum (**Figure 1A**), displaying diameters from 50 to 200 nm and lengths of up to several cell diameters (Rustom et al., 2004). NTs contain F-actin and/or microtubule backbones and facilitate the intercellular transmission of various cellular components, including organelles as well as plasma membrane constituents or the transfer of electric signals (Hurtig et al., 2010; Wang et al., 2010; Wang and Gerdes, 2012). Meanwhile, NTs have been found to be present in cultures of different cell types including for example mesothelial (**Figure 1B**) and epithelial cells, fibroblasts, immune cells, and neurons (Vidulescu et al., 2004; Castro et al., 2005; Watkins and Salter, 2005; Davis and Sowinski, 2008; Gerdes and Carvalho, 2008; Pontes et al., 2008; Ranzinger et al., 2011). Moreover, recent research demonstrates the existence of NTs in human primary tumors (Lou et al., 2012).

During the last 10 years, a lot of knowledge concerning formation, function, and biological implications of NTs in health and disease has been gained. Aside from mediating functional connectivity between various cell types, it is now clear that NTs participate in several pathological processes of substantial medical interest. NTs were proposed to be involved in the intercellular spread of prion proteins (Gousset and Zurzolo, 2009; Dunning et al., 2012) and viral proteins, e.g., during HIV infections (Eugenin et al., 2009; Kadiu and Gendelman, 2011; Sowinski et al., 2011), the transfer of drug resistance between cancer cells (Pasquier et al., 2012) or the transfer of Aβ peptides in the context of Alzheimer's disease (Wang et al., 2011). Moreover, it could be demonstrated that the number of NTs in humans is dependent on the individual donor background (Ranzinger et al., 2011) and correlates with defined pathophysiological conditions. In *in vitro* experiments, in which cells were stimulated with TNF, the number of NTs is significantly increased and associated with a remodeling of the actin cytoskeleton. This finding could be confirmed when NT-numbers were investigated between cells from PD-patients (Ranzinger et al., 2011) pointing to an important participation of NTs during inflammatory reactions.

Additionally, a strong correlation of NT occurrence with cellular cholesterol contents and its distribution throughout the cell could be demonstrated. Experiments, in which cellular cholesterol homeostasis in HPMCs was affected by cholesterol depletion via methyl-β-cyclodextrin (MβCyc), revealed that gradual cholesterol depletion results in a strong, non-linear modulation of NT-numbers and lengths with significant peaks at given MβCyc concentrations, pointing to narrow windows of defined

**45**

**FIGURE 1 | Schematic model of molecular mechanisms with implications for NT-mediated cellular interactions during PD. (A)** The cartoon depicts the principle of NT-formation and NT-mediated transfer of organelles between cells. **(B)** The fluorescence microscopy picture shows NTs (arrows) spanned between primary human peritoneal mesothelial cells. **(C)** Illustration of the interdependence of NT-formation and communication in the context of PD-treatment including stimulating/inhibiting factors.

cholesterol contents being beneficial or detrimental, e.g., by affecting NT tensile strength or by influencing the formation process (Ranzinger et al., 2011). The finding that statin-treatment of HPMCs resulted in significantly increased NT-numbers coincides with comparably high numbers of NTs found between cells from a patient undergoing statin treatment (Ranzinger et al., 2013). In a recent study, Thayanithy et al. (2014) explored exosomes and lipid rafts as mediators of NT-formation in mesothelioma cells. Their results provide evidence for exosomes as chemotactic stimuli for NT-formation and lipid raft formation as potential biomarker for NT-forming cells.

To date, aside from inflammatory conditions, several factors are known that lead to the induction of NT-formation, among these oxidative stress as well as several receptor-ligand interactions (Martinez et al., 2002; Zhu et al., 2005; Chinnery et al., 2008; Ranzinger et al., 2011; Wang et al., 2011; Sun et al., 2012). In the context of PD-treatment, the use of dialysis solutions lead to a significant reduction in NT-numbers between peritoneal mesothelial cells (Ranzinger et al., 2011). Observed more closely, oxidative stress caused by both methyglyoxal (MG) and acidified pH-value results in higher NT-numbers whereas alterations in cellular osmolarity due to enhanced glucose concentrations lead to a strong decrease in NT-numbers between the cells (Ranzinger et al., 2014). In this context, by blocking of RAGE, whose expression is upregulated during PD-treatment, it could be shown that this receptor is a strong regulator in NT-formation processes between murine and human peritoneal mesothelial cells *in vitro* and *in vivo* (Ranzinger et al., 2014).

Respective underlying molecular mechanisms involved in the formation of NTs are controversially discussed. Studies from Wang et al. (2011) showed that NT-formation in primary rat hippocampal astrocytes and neurons is dependent on the activation of the tumor suppressor protein p53 through hydrogen peroxide induced cellular stress. Andresen et al. (2013) however showed that p53 is dispensable for NT-formation in SAOS-2 cells and dKO-MSCs. The results of these studies demonstrate that signaling pathways and involved proteins having an impact on the formation of NTs act strongly cell-type dependent. A previous study from Zhu et al. (2005) showed that oxidative stress induced by H2O2 increases the formation of NTs in astrocytes through activation of the p38 mitogen-activated protein kinase (MAPK) pathway. In the context of RAGE being involved in the formation of NTs between peritoneal mesothelial cells, the MAPK signaling cascade, which is addressed upon RAGE activation, was investigated in a recent study from Ranzinger et al. (2014). It could be demonstrated that oxidative stress induced by MG not only induces NT-formation but also increases phosphorylated p38 protein levels. Subsequently, blocking of p38 resulted in reduced NT-numbers between the cells arguing that the action of p38 regulates NT-formation in peritoneal mesothelial cells.

The existence of NTs *in vivo* has been supported by an increasing number of publications (Eugenin et al., 2009; Pyrgaki et al., 2010; Caneparo et al., 2011; Ranzinger et al., 2014). However, their occurrence, architecture, and function in the body is still a matter of considerable debate and may vary in accordance to the respective species, tissue, developmental stage, age, genetic background, and pathophysiological variations. One possible function could be a NT-mediated rescue for cells and/or organs under pathophysiological conditions like oxidative stress (**Figure 1C**).

In this view, several studies showed e.g., the transfer of mitochondria via NTs (Koyanagi et al., 2005; Domhan et al., 2011; Wang et al., 2011). A study from Vallabhaneni et al. (2012) showed that vascular smooth muscle cells initiate proliferation of mesenchymal stem cells through the exchange of mitochondria in co-cultures. Pasquier et al. (2013) demonstrated that cancer cells acquiring mitochondria from endothelial cells display significant chemoresistance. Furthermore, work from Spees et al. (2006) could demonstrate that aerobic respiration in cells with dysfunctional mitochondria could be rescued by mitochondrial transfer. More recently, concerning ischemia/reperfusion (I/R)-induced injury of the kidney, in a RAGE knockout mouse model, it could be shown that under conditions where RAGE is absent, NT-numbers are increased and kidney tissue morphology is improved compared to kidneys from wild-type mice (Ranzinger et al., 2014). Interestingly, when RAGE is blocked and dialysis solution as secondary stimulus has been applied, increased NTnumbers could also be demonstrated in peritoneal mesothelial cells (Ranzinger et al., 2014). In these cases, one could speculate that a protective effect for the respective organ/tissue might be attributed to an increased NT-formation accompanied by frequent exchanges of for example mitochondria.

In the future, further investigations are needed to investigate NT-mediated transport mechanisms within the peritoneal cavity in greater detail. This will have significant impact on the understanding of a variety of processes, such as inflammatory immune reactions. Potentially, this knowledge will allow for the development of improved treatment options during peritoneal infection.

## **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.

*Received: 26 August 2014; accepted: 03 October 2014; published online: 24 October 2014.*

*Citation: Ranzinger J, Rustom A and Schwenger V (2014) Membrane nanotubes between peritoneal mesothelial cells: functional connectivity and crucial participation during inflammatory reactions. Front. Physiol. 5:412. doi: 10.3389/fphys.2014.00412 This article was submitted to Membrane Physiology and Membrane Biophysics, a section of the journal Frontiers in Physiology.*

*Copyright © 2014 Ranzinger, Rustom and Schwenger. 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.*

## *Hitesh Batra\* and Veena B. Antony*

*Division of Pulmonary, Allergy and Critical Care Medicine, Department of Medicine, University of Alabama at Birmingham, Birmingham, AL, USA*

#### *Edited by:*

*Sotirios G. Zarogiannis, University of Thessaly, Greece*

#### *Reviewed by:*

*Vassilios Liakopoulos, Aristotle University of Thessaloniki, Greece K. Onuma, Juntendo University, Japan*

#### *\*Correspondence:*

*Hitesh Batra, Division of Pulmonary, Allergy and Critical Care Medicine, Department of Medicine, University of Alabama at Birmingham, 1900 University Blvd., THT 422, Birmingham, AL 35294-0006, USA e-mail: hiteshbatra921@gmail.com*

The pleural mesothelium, derived from the embryonic mesoderm, is formed by a metabolically active monolayer of cells that blanket the chest wall and lungs on the parietal and visceral surfaces, respectively. The pleura and lungs are formed as a result of an intricate relationship between the mesoderm and the endoderm during development. Mesenchymal signaling pathways such as Wnt/B-catenin, Bmp4, and sonic hedgehog appear to be quintessential for lung development. Pleural Mesothelial Cells (PMCs) are known to express Wilms tumor-1 (Wt1) gene and in lineage labeling studies of the developing embryo, PMCs were found to track into the lung parenchyma and undergo mesothelial-mesenchymal transition (MMT) to form α-smooth muscle actin (α-SMA)-positive cells of the mesenchyme and vasculature. There is definite evidence that mesothelial cells can differentiate and this seems to play an important role in pleural and parenchymal pathologies. Mesothelial cells can differentiate into adipocytes, chondrocytes, and osteoblasts; and have been shown to clonally generate fibroblasts and smooth muscle cells in murine models. This supports the possibility that they may also modulate lung injury-repair by re-activation of developmental programs in the adult reflecting an altered recapitulation of development, with implications for regenerative biology of the lung. In a mouse model of lung fibrosis using lineage-tracing studies, PMCs lost their polarity and cell-cell junctional complexes, migrated into lung parenchyma, and underwent phenotypic transition into myofibroblasts in response to the pro-fibrotic mediator, transforming growth factor-β1 (TGF-β1). However, intra-pleural heme-oxygenase-1 (HO-1) induction inhibited PMC migration after intra-tracheal fibrogenic injury. Intra-pleural fluorescein isothiocyanate labeled nanoparticles decorated with a surface antibody to mesothelin, a surface marker of mesothelial cells, migrate into the lung parenchyma with PMCs supporting a potential role for pleural based therapies to modulate pleural mesothelial activation and parenchymal disease progression.

**Keywords: pleural mesothelium, pleural mesothelial cells, idiopathic pulmonary fibrosis (IPF), Wilms tumor-1 (WT1), epithelial-mesenchymal transition (EMT)**

## **INTRODUCTION**

The lung is a complex, integrated structure of airways, vasculature and interstitium, surrounded by the pleural mesothelium (Jantz and Antony, 2006). The pleural mesothelium, derived from the embryonic mesoderm, is formed by a layer of cells that blanket the chest wall and lungs on the parietal and visceral surfaces, respectively. A complex interplay of signaling pathways such as Wnt/βcatenin, Bone morphogenetic protein 4 (Bmp4), sonic hedgehog, and Fibroblast growth factor 10 (Fgf10) between the developing endoderm and mesoderm is essential for development.

Wilms tumor-1 (Wt1), a zinc finger transcription factor (Haber et al., 1990), is expressed in certain mesoderm-derived tissues including the pleura (Park et al., 1993) and is known to regulate many functional properties of the developing mesothelium (Ito et al., 2006; Jomgeow et al., 2006). It also regulates pleural mesothelial cell (PMC) plasticity. Wt1 expressing PMCs have been shown to migrate into the lung parenchyma and differentiate into subpopulations of bronchial smooth muscle cells, vascular smooth muscle cells and fibroblasts (Colvin et al., 2001; Dixit et al., 2013).

The presence of Wt1-expressing PMCs on both the pleural surface and in the lung parenchyma of patients with IPF suggests a role for crosstalk between the parenchymal lung injury and activation of the pleural mesothelium in the pathogenesis of IPF. PMCs migrate into the pulmonary parenchyma in IPF and transition into myofibroblasts suggesting the novel hypothesis that IPF is an altered, recapitulation of development, with implications for regenerative biology of the lung.

PMCs could represent a new cellular therapeutic target in a disease where we currently lack treatment modalities. Intra-pleural delivery of compounds is a lung targeted, innovative

**Abbreviations:** PMCs, Pleural Mesothelial Cells (PMCs); Wt1, Wilms tumor-1 (Wt1); MMT, Mesothelial-mesenchymal transition (MMT); α-SMA, α-smooth muscle actin; Bmp4, Bone morphogenetic protein 4; Shh, Sonic hedgehog; Fgf10, Fibroblast growth factor 10; IPF, Idiopathic Pulmonary Fibrosis; TGFβ1, Transforming growth factor-β1; HO-1, Heme-oxygenase-1 (HO-1); EMT, Epithelial-mesenchymal transition; NOX-4, NADPH oxidase-4; UIP, Usual interstitial pneumonia; RNA, Ribonucleic acid; BAL, Bronchoalveolar lavage; NP, Nanoparticle; FITC, Fluorescein isothiocyanate; PLGA, Poly-lactic-co-glycolic acid.

therapeutic modality that can be refined to deliver drugs, minimizing systemic toxicity.

## **DEVELOPMENT**

The lungs arise from the anterior foregut endoderm via a complex and highly regulated process comprising of the embryonic stage, pseudo-glandular stage, branching morphogenesis, canalicular stage, saccular stage, and the alveolarization stage. During all stages of endodermal development, the lung mesoderm interacts with the lung endoderm to generate the various lineages within the lung (Herriges and Morrisey, 2014) and plays a central role in regulating morphogenesis of the lung, including branching, lung size, and vascular development (Que et al., 2008).

#### **ROLE OF MESODERM DURING DEVELOPMENT**

The lung mesoderm is an important source of paracrine signals such as Fgf10 and Wnt2 (Bellusci et al., 1997; Sekine et al., 1999; Weaver et al., 2000; Goss et al., 2009), which are essential for multiple processes during lung development, including the patterning of early endoderm progenitors, epithelial proliferation, and differentiation. The mature lung contains many mesodermal derivatives, including airway smooth muscle, vascular smooth muscle, endothelial and mesothelial cells, as well as multiple less well-understood cell types, such as pericytes, alveolar fibroblasts, and lipofibroblasts (Herriges and Morrisey, 2014). In lineage labeling studies of the developing embryo, PMCs were found to track into the lung parenchyma and undergo mesothelial-mesenchymal transition (MMT) to form α-smooth muscle actin (α-SMA) positive cells of the mesenchyme and vasculature (Que et al., 2008).

## **SIGNALING PATHWAYS**

During development, Wnt/β-catenin signaling is responsible for specifying Nkx2.1+ respiratory endoderm progenitors, which mark the specification of the respiratory system in the anterior foregut endoderm (Herriges and Morrisey, 2014). This, in turn, is dependent upon active Bmp signaling that represses the transcription factor Sox2, thereby allowing for expression of Nkx2.1. Loss of Bmp signaling through inactivation of the Bmp receptors, Bmpr1a and Bmpr1b, leads to tracheal agenesis with retention of the branching region of the lungs (Domyan et al., 2011). Moreover, signaling between the developing endoderm and mesoderm appears to be essential for branching morphogenesis and the loss of Fgf10 signaling to Fgfr2 in the developing endoderm leads to disruption of branching (Sekine et al., 1999; Ohuchi et al., 2000). Fgf10 expression is regulated by several signaling pathways such as Bmp4 and sonic hedgehog (Shh), suggesting a complex interplay of signaling molecules during development (Bellusci et al., 1997; Pepicelli et al., 1998; Weaver et al., 2000).

## **Wt1**

The Wt1 gene encodes a 49–52 kDa protein with an N-terminal domain that is involved in protein-RNA interactions critical for its transcriptional regulatory function (Call et al., 1990).

#### **Wt1 IN PRENATAL DEVELOPMENT**

PMCs are known to express Wt1 gene and migrate into the lung parenchyma to form smooth muscle cells of the vascular wall, as well as other cells of the lung mesenchyme during development (Que et al., 2008; Zhou et al., 2010, 2011).

Recent cell lineage labeling studies in the developing heart provide evidence that the surface epicardial mesothelium undergoes epithelial-mesenchymal transition (EMT) and migrates into the myocardium where it differentiates into various cell types, including endothelium, smooth muscle cells, and cardiomyocytes (Mikawa and Gourdie, 1996; Dettman et al., 1998; Cai et al., 2008; Zhou et al., 2010). In addition, lineage tracing and other studies show that the serosal mesothelium of the gut also contributes the majority of vascular smooth muscle cells (Wilm et al., 2005; Kawaguchi et al., 2007). A major role of Wt1 and PMCs in mesenchymal differentiation and development was demonstrated by lineage tracing studies in the embryonic mouse (Colvin et al., 2001). PMCs were shown to readily migrate into the lung parenchyma and express α-SMA. Another recent study employing Wt1CreERT2*/*<sup>+</sup> mice visualized Wt1<sup>+</sup> mesothelial cell entry into the lung by live imaging, identified their progenies in subpopulations of bronchial smooth muscle cells, vascular smooth muscle cells, and desmin+ fibroblasts by lineage tagging; and demonstrated that mesothelial cell movement into the lung requires the direct action of sonic hedgehog (shh) signaling (Dixit et al., 2013). Together, these studies demonstrate that the mesothelium lining of serosal surfaces of multiple organ systems plays a critical role during development and organogenesis, supporting the possibility that they may also modulate lung injury-repair by re-activation of developmental programs in the adult.

#### **Wt1 IN POSTNATAL DISEASE**

Wt1 can induce a morphological transition from an epithelial phenotype to a mesenchymal phenotype (Burwell et al., 2007). It is a potent transcription factor that can function as a tumor suppressor (Zhang et al., 2003), or as an oncogene depending on the cell type (Loeb et al., 2001; Oji et al., 2002; Ueda et al., 2003). It was in fact initially discovered as a tumor suppressor gene in Wilms tumor of the kidney (Haber et al., 1990). Wt1 demonstrates tissue specific responses and is recognized to regulate TGF-β1 in the kidney. In other cell types such as hematological cells it is reported to confer oncogenic properties (Licciulli and Kissil, 2010). Wt1 is expressed in normal pleural mesothelium, and is over-expressed in malignant mesothelioma (Amin et al., 1995). However, there is limited information of the role of Wt1 in non-malignant lung diseases.

#### **Wt1 IN IDIOPATHIC PULMONARY FIBROSIS (IPF)**

Several studies support a role for Wt1 in mesothelial-tomesenchymal transition (MMT) (Burwell et al., 2007; Bax et al., 2011a,b). Hecker et al. have evaluated the responses of PMCs to TGF-β1, their expression of MMT markers such as e-cadherin, α-SMA, vimentin; their contractile ability using a collagen gel contraction assay (Hecker et al., 2009); and also biological functional assays such as migration and proliferation (Hecker et al., 2009). Studies indicate a role for TGF-β1 in mediating MMT (Nasreen et al., 2009). For example, treatment of mouse PMCs with TGF-β1 increased expression of the myofibroblast markers α-SMA and NADPH oxidase-4 (NOX-4) (Zolak et al., 2013). Wt1-expressing PMCs are present both on the pleural surface and in the lung parenchyma of patients with IPF suggesting their role in the pathogenesis of IPF. However, the precise relationship between Wt1 and TGF-β1 remains unclear. Also, the role of Wt1 in regulating PMC plasticity needs to be analyzed.

## **PLURIPOTENCY OF MESOTHELIAL CELLS**

Although mesenchymal in origin, mesothelial cells exhibit characteristics such as a polygonal cell shape, expression of surface microvilli, epithelial cytokeratins, and tight junctions, which are typical of epithelial cells (Mutsaers, 2002). During development, endothelium and vascular smooth muscle cells of the vascular system, heart, liver and gut are derived from the differentiation of mesothelium via a process of epithelial-to-mesenchymal transition (EMT) (Munoz-Chapuli et al., 1999; Perez-Pomares et al., 2002, 2004). Increasing evidence suggests existence of a population of progenitor-like mesothelial cells with the capacity to differentiate into cells of different phenotypes (Herrick and Mutsaers, 2004). By targeting mesothelin and using a genetic eage to trunk fibroblasts, smooth muscle cells, and vasculature (Rinkevich et al., 2012). lineage tracing approach, Rinkevich et al. demonstrated that embryonic and adult mesothelium represents a common lin-

Recently, primary rat and human mesothelial cells, when maintained in osteogenic or adipogenic media, were shown to differentiate into osteoblast- and adipocyte-like cells via epithelial-to-mesenchymal transition as suggested by the changes in mRNA expression of these cells. This supports mesothelial cell differentiation as the potential source of different tissue types in malignant mesothelioma and other serosal pathologies (Lansley et al., 2011). The term *mesodermoma* was introduced by Donna and Betta to define neoplasms arising from undifferentiated and multipotential mesoderm (Donna and Betta, 1981).

Mesothelial cells appear to retain the ability to produce mesenchyme, including smooth muscle cells, in response to transforming growth factor-β<sup>1</sup> (TGF-β1) and platelet derived growth factor (Wada et al., 2003; Kawaguchi et al., 2007). Moreover, mesothelial cells have been shown to adopt a myofibroblast phenotype *in vitro*, in response to TGF-β<sup>1</sup> (Yang et al., 2003). Transfection of the peritoneum and pleura of rats with an adenovirus expressing TGF-β1induced mesothelial cells to undergo EMT with subsequent fibrotic changes (Margetts et al., 2005; Decologne et al., 2007).

The significance of EMT in malignant pleural mesothelioma (MPM) was demonstrated by Schramm et al. in their study of a retrospective cohort of 352 patients. Immunohistochemistry of a tissue microarray showed that the activation of periostintriggered EMT is associated with the sarcomatoid histotype of malignant mesothelioma and has an impact on shorter survival of patients (Schramm et al., 2010). Periostin secretion by MPM cells has in turn been shown to be upregulated by CD26 (Komiya et al., 2014), the expression of which is increased in various cancers (Havre et al., 2008).

## **IDIOPATHIC PULMONARY FIBROSIS (IPF)**

#### **SUB-PLEURAL DISTRIBUTION**

IPF is a progressive lung disease process that appears to begin in the sub-pleural regions, and then extend centrally. This subpleural and basilar distribution, which is the defining feature of idiopathic pulmonary fibrosis (IPF) has not been adequately explained (King et al., 2011). However, this pattern of distribution suggests pleural involvement in the disease process.

#### **MYOFIBROBLASTS AND FIBROBLASTIC FOCI**

The presence of myofibroblasts in fibrobastic foci in lungs of patients with IPF is well-established. The fibroblast and myofibroblast foci secrete excessive amounts of extracellular matrix, mainly collagens, resulting in scarring and destruction of the lung architecture. The profusion of fibroblastic foci is predictive of survival in IPF patients (King et al., 2001a,b). A recent study has shown that the hallmark lesion in usual interstitial pneumonia (UIP), fibroblastic foci, on three-dimensional histo-pathological reconstruction are part of a complex and highly interconnected reticulum of fibrous tissue that extends from the pleura into the lung parenchyma (Cool et al., 2006). Neither the origin of these myofibroblasts nor the molecular mechanisms involved in the formation of fibroblastic foci have been well-defined (Phan, 2008).

While a number of cellular source(s) and progenitors of tissue myofibroblasts have been proposed, none of these hypotheses provide a indisputable explanation for the histo-pathological pattern of usual interstitial pneumonia (UIP) and its peripheral localization (Raghu et al., 2011). However, evidence suggests that abnormal recapitulation of developmental pathways and epigenetic changes may play a role in the pathogenesis of IPF (King et al., 2011). IPF is a devastating disease with an inexorable course and its pathogenesis remains unclear. There are no therapies that directly impact the disease progression or mortality, and we urgently need innovative new ideas that will impact therapeutics in this disease (King et al., 2011; Raghu et al., 2011). PMCs serve as myofibroblast progenitors in animal models of fibrosis, and therapeutic targeting of these cells may be effective as an anti-fibrotic approach (Que et al., 2008).

#### **PLEURAL MESOTHELIAL CELLS IN IPF**

PMCs are metabolically active cells (Antony, 2003; Mohammed et al., 2007), responsive to their microenvironment (Antony, 2003), and are recognized to demonstrate plasticity of their phenotype (Nasreen et al., 2009).The presence of calretinin and mesothelin (markers of mesothelial cells) (Mubarak et al., 2012) and Wt-1 expressing PMCs (Zolak et al., 2013) was recently demonstrated in parenchymal cells of explanted lung tissues from 16 patients with IPF supporting a role for PMC differentiation and their trafficking into the lung as contributors to the myofibroblast population in lung fibrosis. It is not known if the number or profusion of PMCs may be predictive of the severity and/or progression of IPF.

The Ashcroft score is a score for histo-pathological grading of pulmonary fibrosis (Hagiwara et al., 2000; Matsuoka et al., 2002; Simler et al., 2002; Murakami et al., 2006). There exists a correlation between the number of calretinin-positive cells and the degree of fibrotic change in the parenchyma, as measured by the Ashcroft score. Whether the number of calretinin-positive cells was measured as a raw number, or as percentage of the total nucleated cells seen in a photomicrograph, the correlation with the degree of fibrosis was highly significant (Mubarak et al., 2012). These studies suggest that PMCs may represent a novel cellular biomarker of disease activity, and more importantly, play a role in the tissue remodeling responses seen in patients with IPF.

Recent studies demonstrate that PMCs cells lose their polarity and cell-cell junctional complexes, migrate into lung parenchyma, and undergo phenotypic transition into myofibroblasts in response to the pro-fibrotic mediator, TGF-β1 (Nasreen et al., 2009; Mubarak et al., 2012). The transition of PMC to myofibroblasts is dependent on smad-2 signaling and knockdown of smad-2 gene by silencing small interfering RNA significantly suppresses the transition of PMCs to myofibroblasts and inhibits PMC haptotaxis (Nasreen et al., 2009). We have demonstrated in a murine model, that TGF-β1 induces PMC trafficking into the lung and differentiation into myofibroblasts. Moreover, carbon monoxide or the induction of heme oxygenase-1 (HO-1) inhibits the expression of myofibroblast markers, contractility, and haptotaxis in PMCs treated with TGF-β1. These findings support a potential role for pleural-based therapies to modulate pleural mesothelial activation and parenchymal fibrosis progression (Zolak et al., 2013).

## **PLEURAL MESOTHELIAL CELLS AS A POTENTIAL THERAPEUTIC TARGET**

PMCs migrate into the pulmonary parenchyma in IPF and transition into myofibroblasts. This invokes a novel, alternative hypothesis for the origin and source of the myofibroblasts in the lungs of patients with IPF, and provides a rational explanation for the spatio-temporal distribution of fibrosis in IPF. PMCs could represent a new cellular therapeutic target in a disease where we currently lack treatment options. Intra-pleural delivery of compounds is a lung targeted, innovative therapeutic modality that can be refined to deliver drugs, minimizing systemic toxicity.

Direct delivery of the small molecule inhibitors to the pleura could potentially have several advantages: (Jantz and Antony, 2006) It can potentially provide a direct, high concentration of the compound to target the pro-fibrogenic activities of PMCs; (Haber et al., 1990) Intra-pleural delivery might delay systemic absorption, protecting against systemic toxicity while increasing efficacy; (Park et al., 1993) Intra-pleural delivery may result in higher, sustained levels in the Bronchoalveolar lavage (BAL) when compared with serum levels (Mubarak et al., 2012); and (Jomgeow et al., 2006) innovative, less invasive methods of accessing the pleural space such as tunneled pleural catheters could make intra-pleural delivery a viable option in patients.

There are several methods that can be utilized for sitedirected delivery of therapeutic agents. These include liposomal drug delivery, nanoparticle (NP) delivery of proteins, and gene therapy (Watanabe et al., 2010). Several of these methods have been utilized to target the pleura (Perez-Soler et al., 1997; Liu et al., 2006; Watanabe et al., 2010). We have demonstrated that PMCs migrate into the parenchyma (Zolak et al., 2013). Biodegradable fluorescein isothiocyanate (FITC) labeled PLGA (poly-lactic-co-glycolic acid) nanoparticles (which can carry therapeutic compounds conjugated to PLGA) decorated with antibody targeted to mesothelin (a PMC marker) localize to the pleural surface in control mice, but diffuse into the lung parenchyma in mice given intra-tracheal bleomycin (**Figure 1**). It appears that intra-pleural delivery of molecules to the lung is feasible, albeit, with the recognition that refinement of techniques for minimal lung injury will be required.

## **CONCLUSION**

The pleural mesothelium is a derivative of the embryonic mesoderm. Several complex pathways of interplay between the mesoderm and endoderm lead to the development of the lungs and pleura. Lineage labeling and other studies have shown the ability of the PMCs to undergo MMT to form other mesenchymal structures and have therefore established their pluripotency during development. Moreover, recent studies have shown that the plasticity of mesothelial cells is retained postnatally and gives them the ability to migrate and differentiate into other cell types. This supports the possibility that PMCs may modulate lung injury-repair by reactivation of developmental programs in the adult, reflecting an altered recapitulation of development with implications for regenerative biology of the lung. We have shown that PMCs are present in the lung parenchyma in lungs with IPF. Also, PMC migration into lung parenchyma and phenotypic transition into myofibroblasts can be induced by the pro-fibrotic mediator, TGF-β1 and is inhibited by intra-pleural heme-oxygenase-1 (HO-1). The possible role of PMCs in disease pathogenesis and/or progression of IPF makes them an attractive cellular target for potential therapeutic interventions in a disease where we currently lack treatment modalities. Intra-pleural delivery of compounds is a lung targeted, innovative therapeutic modality that can be refined to deliver drugs, thereby minimizing systemic toxicity. Intra-pleural fluorescein isothiocyanate labeled nanoparticles decorated with a surface antibody to mesothelin, a surface marker of mesothelial cells, migrate into the lung parenchyma with PMCs supporting a potential role for pleural based therapies to modulate pleural mesothelial activation and parenchymal disease progression. However, further studies are

**FIGURE 1 | Intra-pleural delivery of PLGA NP.** Biodegradable FITC labeled PLGA nanoparticles (which can carry therapeutic compounds conjugated to PLGA) decorated with antibody targeted to mesothelin (a PMC marker) localize to the pleural surface in control mice treated with saline **(A)**, but diffuse into the lung parenchyma in mice given intra-tracheal bleomycin **(B)**. The arrow point to FITC labeled PLGA nanoparticles. PLGA, Poly Lactic-co-Glycolic Acid; NP, Nanoparticles; FITC, Fluorescein isothiocyanate.

needed to develop such therapies and effective pleural-based drug delivery systems.

#### **REFERENCES**


from embryonic mesothelial cells? A model on the origin of the vertebrate cardiovascular system. *Differentiation* 64, 133–141. doi: 10.1007/s0025800 50268


malignant pleural mesothelioma. *Eur. J. Cardiothorac. Surg*. 37, 566–572. doi: 10.1016/j.ejcts.2009.08.027


**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.

*Received: 27 May 2014; accepted: 10 July 2014; published online: 01 August 2014. Citation: Batra H and Antony VB (2014) The pleural mesothelium in development and disease. Front. Physiol. 5:284. doi: 10.3389/fphys.2014.00284*

*This article was submitted to Membrane Physiology and Membrane Biophysics, a section of the journal Frontiers in Physiology.*

*Copyright © 2014 Batra and Antony. 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.*

## *Masayoshi Kuwahara\**

*Laboratory of Veterinary Pathophysiology and Animal Health, Department of Veterinary Medical Sciences, Graduate School of Agricultural and Life Sciences, The University of Tokyo, Tokyo, Japan*

#### *Edited by:*

*Sotirios G. Zarogiannis, University of Thessaly, Greece*

#### *Reviewed by:*

*Irena Levitan, University of Illinois at Chicago, USA Huazhi Eric Han, University of Illinois at Chicago, USA Kinji Asahina, University of Southern California, USA*

#### *\*Correspondence:*

*Masayoshi Kuwahara, Laboratory of Veterinary Pathophysiology and Animal Health, Department of Veterinary Medical Sciences, Graduate School of Agricultural and Life Sciences, The University of Tokyo, 1-1-1 Yayoi, Bunkyo-ku, Tokyo 113-8657, Japan e-mail: akuwam@ mail.ecc.u-tokyo.ac.jp*

The mesothelial layer acts as a biological barrier between the organ and the enveloping serous cavity and may have functions of transport, equilibrium maintenance, and protection. However, the role of the mesothelial cells in regulation of pleural permeability remains essentially undefined. The present study was designed to clarify the effects of bradykinin, histamine, and thrombin on permeability in pleural mesothelial cells. Rat pleural mesothelial cells were cultured *in vitro*, and the permeability of mesothelial monolayers was evaluated by transmesothelial albumin diffusion and electrical resistance measurements. Furthermore, the temporal relationship between changes in the levels of [Ca2+]i and the mesothelial permeability was examined. Bradykinin (10μM), histamine (1 mM), and thrombin (10 U) caused albumin diffusion within 5 min. The electrical resistance of mesothelial monolayer began falling within 5 min of adding each agent. Time and concentration dependency of changes in electrical resistance were almost the same as that in albumin diffusion. Each agent also induced a biphasic elevation of [Ca2+]i in pleural mesothelial cells. The concentration-dependency of the [Ca2+]i responses were almost similar to that noted for each agent induced albumin diffusion and electrical resistance fall. The increase in permeability occurred with reorganization of F-actin cytoskeleton and increased actin polymerization. These results suggest that the Ca2+- dependency of increases induced by these agents in mesothelial permeability have been related to the regulatory role of Ca2<sup>+</sup> in the F-actin cytoskeletal reorganization in pleural mesothelial cells.

**Keywords: actin cytoskeleton, bradykinin, cytosolic calcium, cytochalasin B, F-actin, histamine, pleura, thrombin**

## **INTRODUCTION**

The visceral and parietal pleura are lined by a unicellular layer of mesothelial cells that overlie a discontinuous basal lamina. This mesothelial layer acts as a biological barrier between the organ and the enveloping serous cavity and may have functions of transport, equilibrium maintenance, and protection (Wang, 1985). Changes in pleural permeability, the influx of phagocytic cells, and the leak of serum proteins into the pleural space lead to the development of an inflammatory exudate (Robbins and Kumar, 1987). However, the role of the mesothelial cells in regulation of pleural permeability remains essentially undefined.

We have shown that pleural and pericardial mesothelial cells can respond to histamine (Ito et al., 1995). Histamine elevates intracellular calcium concentration ([Ca2+]i) and induces F-actin cytoskeletal reorganization in mesothelial cells. Because F-actinstaining studies have shown the regulatory role of [Ca2+]i in cytoskeleton assembly in mesothelial cells, we have suggested that changes in [Ca2+]i may have relevance to the regulation of mesothelial permeability. Therefore, the present study was designed to clarify the effects of bradykinin, histamine, and thrombin on permeability in pleural mesothelial cells. For this purpose, rat pleural mesothelial cells were cultured *in vitro*, and the permeability of mesothelial monolayers was evaluated by transmesothelial albumin diffusion and electrical resistance measurements. Furthermore, the temporal relationship between changes in the levels of [Ca2+]i and the mesothelial permeability was examined.

## **MATERIALS AND METHODS REAGENTS, MEDIA, AND BUFFERS**

For the Ca2<sup>+</sup> measurements we used a N-2-hydroxyethylpiperazine-N- -2-ethanesulfonic acid (HEPES) buffer containing (in mM) 10 HEPES, 136.9 NaCl, 5.4 KCl, 1.0 MgCl2, 1.5 CaCl2, 0.001 EDTA, and 5.5 glucose (HEPES-buffered solution). Dulbecco's modified eagles medium (DMEM), Hanks' balanced salt solution (HBSS), fetal bovine serum (FBS), and trypsin-EDTA were purchased from GIBCO, Grand Island, NY. Bradykinin, thrombin, ionomycin, albumin, cytochalasin B and 1,2-bis(2-aminophenoxy) ethane-N,N,N',N'-tetraacetic acid (BAPTA) were purchased from Sigma Chemical Co., St. Louis, MO. Fura 2-acetoxymethyl ester (fura 2-AM) was obtained from Dojindo Laboratories, Kumamoto, Japan. Histamine was purchased from Wako, Tokyo, Japan. Other materials and chemicals were obtained from commercial sources.

#### **RAT PLEURAL MESOTHELIAL CELL CULTURES**

Rat pleural mesothelial cells were established and maintained, as described previously (Ito et al., 1995). Briefly, adult Wistar Kyoto rats were anesthetized with sodium pentobarbital (50 mg/kg ip) and were immediately killed by exsanguination from a severed abdominal aorta. The complete thoracic wall was removed under sterile conditions and immersed in petri dish for 20 min in HBSS. The parietal pleural surface was scraped repeatedly with cell scrapers. The cells were then seeded into 55-cm2 tissueculture dishes (Corning, Wexford, PA) in DMEM with 10% FBS, 10<sup>5</sup> U/l penicillin, and 100 mg/l streptomycin. Subsequently, the mesothelial cell cultures were maintained for up to 15 passages at 37◦C in a humidified environment containing 5% CO2. The cultured cells exhibited the characteristic features of mesothelial cells: a polyhedral, cobblestone morphologic pattern and positive immunohistochemical staining for cytokeratin and vimentin (Kuwahara et al., 1991).

#### **TRANSMESOTHELIAL ALBUMIN PERMEABILITY**

The transmesothelial albumin flux across cultured monolayers of mesothelial cells was measured using an *in vitro* system as described by Rotrosen and Gallin (1986). Monolayers grown on membranes in Millicell-HA inserts (Millipore, Bedford, MA) were gently washed with HBSS and replaced in 24-well culture plates (Costar Cambridge, MA). The monolayers were covered immediately with DMEM containing a trypan blue-albumin complex [36 mg trypan blue and 800 mg bovine serum albumin were dissolved in 100 ml DMEM to yield a stable complex (trypan blue *>* 99% protein bound as determined by TCA precipitation) with absorption maximum at 590 nm]. To avoid hydrostatic pressure across the monolayer, the fluid levels inside and outside the culture well were equalized. The monolayers were then incubated during the defined time with stimulant at 37◦C, 5% CO2. At the end of the incubation the monolayer-inserts were carefully removed, and albumin diffusion across the monolayer was quantified by measuring absorbance at 590 nm of the bottom well fluid. Stimulated albumin diffusion was compared with simultaneous controls and expressed as:

% change vs. control = 100 × [Absorbance (test)-Absorbance (control)]/Absorbance (control).

To show the diffusion rate in control (without stimulant), % change albumin diffusion in control was calculated by dividing the absorbance at defined time point by starting (0 time) absorbance.

#### **TRANSMESOTHELIAL ELECTRICAL RESISTANCE MEASUREMENTS**

Transmesothelial electrical resistance was determined using a Millicell-ERS (Millipore, Bedford, MA). Pleural mesothelial cell monolayers grown on membranes in Millicell-HA inserts (Millipore, Bedford, MA) were gently washed with HBSS and replaced in 24-well culture plates (Costar Cambridge, MA). The monolayers were incubated with DMEM at 37◦C, 5% CO2 for at least 30 min before the start of each experiment. The initial transmesothelial electrical resistance was then determined, and the monolayers were incubated during the defined time with stimulant at 37◦C, 5% CO2. The resistance of each sample and blank (without monolayers) was measured using Ag/AgCl electrodes placed into both inside and outside Millicell insert. Transmesothelial electrical resistance (TER) was calculated as:

TER (Ohms <sup>×</sup> cm2) <sup>=</sup> [Resistance (test)-Resistance (blank)]/Effective membrane area (in this study 0.6 cm2).

## **MEASUREMENT OF [CA2+]<sup>I</sup>**

Changes in [Ca2+]i were determined as previously reported (Ito et al., 1995). Pleural mesothelial cells were incubated on 25 mm glass coverslips (Matsunami, Tokyo, Japan) in DMEM with 10% FBS. After reaching confluence, the cells were cultured further in serum-free culture medium for 12 h, and then the mesothelial cell monolayers were loaded with fura 2 by incubating them with 2μM fura 2-AM for 30 min at 37◦C in HEPESbuffered solution. Loaded cells were washed in HEPES-buffered solution and maintained in this solution for 20 min at room temperature to allow for complete hydrolysis of fura 2 to the acid form.

The glass coverslip was placed horizontally in a temperaturecontrolled (37◦C) bath that was mounted on Intracellular Ion Analyzer (CAF-110, Japan Spectroscopic, Tokyo, Japan). Fluorescence excitation was obtained from a xenon highpressure lamp (150 W). Ultraviolet light of alternating 340 and 380 nm (10 nm bandwidth) was obtained with a monochromator equipped with a chopping wheel (400 Hz) placed in front of the monochromator. Fura 2 fluorescence from the cells was imaged with a Nikon UV-Fluor objective lens (× 10). The dichroic mirror was used as a beam splitter to transmit emitted fluorescence (500 nm) into the photomultiplier. The fluorescence signals (340 and 380 nm) and their ratio (340:380 nm) were continuously recorded on a chart recorder. At the end of experimental run, background autofluorescence (the inherent fluorescence emitted from cells, coverslip, and bath at 340 and 380 nm) was obtained by the method of Hallam et al. (1988).

After autofluorescence was subtracted, the changes in [Ca2+]i were determined quantitatively by using the following equation: [Ca2+]i <sup>=</sup> *Kd*[(*R*-*R*min)/(*R*max-*R*)](*Sf* 2/*Sb*2) where the dissociation constant *Kd* has a value of 224 nM (Grynkiewicz et al., 1985), *R* is the fluorescence ratio within the cells, *R*max is the maximal fluorescence ratio after addition of 40μM ionomycin in the presence of 1.5 mM CaCl2, *R*min is the minimal ratio determined by the subsequent addition of 5 mM EGTA, and *Sf* 2/*Sb*<sup>2</sup> is the ratio of fluorescence values at 380-nm excitation determined at *R*min and *R*max, respectively.

#### **F-ACTIN STAINING**

Pleural mesothelial cells were fixed in 3% phosphate-buffered saline (PBS)-formalin for 10 min, and permeabilized with 1% Triton X-100 in PBS for 10 min at room temperature. After two washes with PBS, cells were stained with fluorescein isothiocyanate (FITC)-labeled phalloidin (5 unit/ml in PBS) to localize F-actin for 20 min in a dark room at room temperature. Cells were washed with PBS twice and maintained in PBS. Dishes were mounted on the stage of Leica TCS/NT confocal laser scanning microscope equipped with an Ar–Kr laser. The excitation and emission wavelengths for FITC-phalloidin were 490 and 525 nm. To standardize the fluorescence intensity measurements among experiments, the time of image capture, the image intensity gain, the image enhancement, and the image black level in both channels were optimally adjusted at the outset and kept constant for all experiments.

#### **QUANTIFICATION OF F-ACTIN CONTENT**

Mesothelial cells were stained with nitrobenzoxadiazole (NBD)-phallacidin (Molecular Probes, Junction City, OR) and analyzed with a FACScan (Becton Dickinson Immunocytometry System, Mountain View, CA) (Howard and Oresajo, 1985) with following some modifications. In all cases a two-step stain procedure was used. Mesothelial cells suspensions (1 <sup>×</sup> <sup>10</sup><sup>6</sup> cells/ml) were incubated at the desired time in HBSS with and without agents; fixed with formalin (3.7% vol/vol) for 15 min at 25◦C; and then exposed to a final concentration of 100μg/ml lysophosphatidyl choline and 1.65 <sup>×</sup> <sup>10</sup>−<sup>7</sup> M NBD-phallacidin. Stained cells were analyzed by FACScan within 1 h of staining. In all instances the fluorescence histogram of cells yielded a normal distribution, and the fluorescence was recorded as the peak fluorescence channel number. The relative F-actin content is expressed as the ratio of the agents treated peak channel number to the control peak channel number.

### **Ca2<sup>+</sup> CHELATION AND STABILIZATION OF THE ACTIN CYTOSKELETON**

To show the role of Ca2<sup>+</sup> and F-actin on the mesothelial permeability, effects of Ca2<sup>+</sup> chelation by BAPTA (5 mM) and stabilization of the actin cytoskeleton by cytochalasin B (10 μg/ml) were determined for some experiments. The dose of cytochalasin B was selected according to previous study (Kuwahara et al., 1994).

#### **STATISTICAL ANALYSIS**

Results were expressed as means ± *SD*. Statistical comparisons were made with the use of the Student's *t*-test or analysis of variance. A value of *P <* 0*.*05 was considered significant.

## **RESULTS**

### **CHANGES IN ALBUMIN DIFFUSION AND ELECTRICAL RESISTANCE**

The effects of histamine (1 mM), bradykinin (10 μM), and thrombin (10 U) on time course of trypan blue-albumin diffusion across the mesothelial cell monolayers are shown in **Figure 1A**. All of these agents caused albumin diffusion within 5 min. Albumin diffusion was gradually increased after 5 min exposure. The magnitude of increase in bradykinin and thrombin to albumin diffusion was almost the same, but larger than that in histamine. Albumin diffusion was not induced in control group. The electrical resistance of mesothelial monolayer began falling within 5 min of adding each agent and leveled off after 30 min as shown in **Figure 1B**. Time dependency of changes in electrical resistance was almost similar to that in albumin diffusion. The effects of Ca2<sup>+</sup> chelation by BAPTA and stabilization of the actin cytoskeleton by cytochalasin B on albumin diffusion across the mesothelial cell monolayers are shown in **Figure 1C**. The increase in agents induced albumin diffusion after 10 min incubation was significantly reduced by BAPTA and cytochalasin B treatments. Concentration dependency of changes in both albumin diffusion and electrical resistance of mesothelial monolayer after 30 min incubation is shown in **Figure 1D**. Each agent induced albumin diffusion and electrical resistance fall in a concentration-dependent manner.

## **CHANGES IN [Ca2+]<sup>I</sup> IN MESOTHELIAL CELLS**

As shown in **Figure 2A**, histamine (1 mM), bradykinin (10μM), and thrombin (10 U) induced a biphasic elevation of [Ca2+]i in pleural mesothelial cells that consisted of an initial transient component and a following sustained component in the presence of 1.5 mM extracellular Ca2+. The peak of initial transient component was appeared within 30 s after stimulation and sustained component was followed. The characteristics of the response were similar for each agent. Each agent elicited the elevation of [Ca2+]i in a dose-dependent manner. Concentration-response curves were obtained from the peak values of each initial component of the [Ca2+]i response as shown in **Figure 2B**. These responses of [Ca2+]i in the concentration-dependency were almost similar to that noted for each agent induced albumin diffusion and electrical resistance fall.

## **F-ACTIN CYTOSKELETAL ORGANIZATION**

The F-actin cytoskeletal organization was studied after 5 min of histamine (1 mM), bradykinin (10μM), and thrombin (10 U) challenge. The representative F-actin patterns are shown in **Figure 3**. In the control group, mesothelial cells formed a cobblestone mosaic pattern composed of polyhedral cells (**Figure 3A**). After 5 min of agent exposure, the cells became elongated compared with the control cells. The density of stress fibers spanning the cells increased in histamine (**Figure 3B**), bradykinin (**Figure 3C**), and thrombin (**Figure 3D**) exposure.

## **F-ACTIN CONTENT**

**Figure 4A** shows the time course of change in relative F-actin content during 30 min after agent stimulation. During the first 1 min after stimulation there was a dramatic increase in relative F-actin content that was maximal at 3–5 min and reflects agents induced actin polymerization. After maximal F-actin content was observed there was a decline in relative F-actin content during the next 30 min. These agents induced actin polymerization after 5 min incubation were significantly inhibited by Ca2<sup>+</sup> chelation or cytochalasin B treatment (**Figure 4B**).

## **DISCUSSION**

The present study has demonstrated that bradykinin, histamine, and thrombin increased transmesothelial permeability in cultured pleural mesothelial cell monolayers. The magnitude of increase in mesothelial permeability in response to histamine was smaller than that to bradykinin and thrombin. The Ca2+ dependency of the bradykinin-, histamine-, and thrombininduced increases in mesothelial permeability have been related to the regulatory role of Ca2<sup>+</sup> in the F-actin cytoskeletal reorganization in pleural mesothelial cells.

We have shown that the mesothelial cells respond to histamine and elevate [Ca2+]i via H1 receptors pathway (Ito et al., 1995). Hott et al. (1992) have reported that thrombin induces proliferation and chemotaxis of rat pleural mesothelial cells. Bradykinin has also induced [Ca2+]i via BK2 receptors pathway in human mesothelial cells (Andre et al., 1998), and bradykinin is generated

by the serine proteases, kallikreins, at sites of tissue injury and responds in a variety of tissues (Regoli and Barabe, 1980). Furthermore, these agents are known to be potent permeabilityincreasing agents for endothelial cells (Rotrosen and Gallin, 1986; Alexander et al., 1988; Lum et al., 1989, 1992). Therefore, we have chosen these agents to examine the effects on mesothelial permeability.

Bradykinin, histamine, and thrombin-induced increases in endothelial permeability may be dependent on [Ca2+]i elevation and related to the regulatory role of Ca2<sup>+</sup> in cytoskeleton assembly, structure, and contractility (Berridge, 1987). Rotrosen and Gallin (1986) have reported that concentrations of histamine required to augment monolayer permeability are of the same order of magnitude as those shown to elevate endothelial [Ca2+]i. Because responses of [Ca<sup>2</sup>+]i in the concentrationdependency were almost similar to that noted for each agent induced mesothelial permeability, our results are in accordance with those observations in endothelial cells. However, because circulating physiological concentrations of these agents may be lower than experimental concentrations in this study, the local paracrine action of these agents in diseased states are more important than their systemic effects in healthy states.

Although the magnitude of the initial transient increase in [Ca2+]i induced by higher dosage was almost the same among agents, the magnitude of increase in mesothelial permeability in response to histamine was smaller than that to bradykinin and thrombin. The signal transduction pathways to these agents in mesothelial cells are not fully understood. Three receptor subtypes (BK1, BK2, and BK3) have been classified according to their affinities for bradykinin and to the relative potencies of their agonists and antagonists (Regoli and Barabe, 1980; Farmer et al., 1989; Steranka et al., 1989). Because only the activation of BK2 receptors mediates the bradykinin-induced increase in [Ca2+]i both endothelial (Morgan-Boyd et al., 1987) and mesothelial (Andre et al., 1998) cells, and phosphoinositide turnover in endothelial cells (Derian and Moskowitz, 1986), it seems that bradykinin-induced elevations of [Ca2+]i and permeability in mesothelial cells are dependent on BK2 receptors. It is generally thought H1 receptor couples G*q/*<sup>11</sup> and mediated by protein kinase C. We suggest that angiotensin II-induced actin reorganization in pleural mesothelial cells is dependent on the angiotensin AT1 receptor coupled with pertussis toxin-insensitive heterotrimeric G proteins, Rho GTPases and tyrosine phosphorylation pathways (Kuwahara and Kuwahara, 2002). Evidence has suggested G protein activation is a critical step in relaying signals from the receptors to the endothelial contractile machinery (Brock and Capasso, 1989; Voyno-Yasenetskaya et al., 1989; Garcia et al., 1991; Tkachuk and Voyno-Yasenetskaya, 1991).

Therefore, one possible explanation of the differences in the magnitude of permeability is the contribution of specific G protein subtypes in activating mesothelial signaling pathways in response to these agents. Another possibility is that selective activation of protein kinase C isoforms or the presence of different degrees of negative feedback pathways may contribute these differences. However, further studies will be necessary to clarify the signal transduction pathways to these agents in mesothelial cells.

The bradykinin, histamine, and thrombin-induced increase of F-actin polymerization occurred at similar time points as that of the permeability increases. The majority of mesothelial cells showed an increased number of centralized actin stress fibers within 5 min of stimulation. Our results show that immediately after stimulation there is a rapid polymerization of actin (Factin content) followed by a slower depolymerization of actin. Furthermore, agents-induced polymerization is inhibited by in the presence of BAPTA or cytochalasin B. The temporal relationship between cytoskeletal reorganization and permeability increase supports the hypothesis that reorganization of F-actin microfilament is involved in mediating the permeability increase. These findings have important implications for our understanding of the intracellular mechanisms that control the state of actin polymerization and permeability of the mesothelial cells.

Changes in pleural permeability, the influx of phagocytic cells, and the leak of serum proteins into the pleural space lead to the development of an inflammatory exudate. The pleural mesothelial cells produce chemoattractants for fibroblasts, fibronectin (Kuwahara et al., 1991), and for neutrophils, interleukin-8 (Boylan et al., 1992). Fibrinogen and fibrin may serve as chemotaxins (Sueishi et al., 1981), and promote fibroblast

**FIGURE 3 | The effects of histamine (1 mM), bradykinin (10 µM), and thrombin (10 U) on F-actin cytoplasmic distribution in rat pleural mesothelial cells.** Control **(A)**, after 5 min exposure histamine **(B)**, bradykinin **(C)**, and thrombin **(D)**.

adherence, proliferation, and collagen production (Senior et al., 1986). Thrombin may participate in macrophage chemotaxis (Bar-Shavit et al., 1983). Therefore, these factors are also important to maintain the pleural space in healthy and diseased states.

control.

In summary, the effects of bradykinin, histamine and thrombin on permeability in mesothelial cells were studied. The Ca2+-dependency of these agents-induced increases in mesothelial permeability have been related to the regulatory role of Ca2<sup>+</sup> in the F-actin cytoskeletal reorganization in pleural mesothelial cells. This temporal relationship is consistent with the hypothesis that these agents-induced mobilization of Ca2<sup>+</sup> signals increase in mesothelial permeability.

by cytochalasin B (10μg/ml) on histamine (1 mM), bradykinin (10 μM),

#### **ACKNOWLEDGMENT**

This work was supported in part by Grant-In-Aid from the Ministry of Education, Science, Sports and Culture, Japan.

### **REFERENCES**


Grynkiewicz, G., Poenie, M., and Tsien, R. Y. (1985). A new generation of Ca2<sup>+</sup> indicators with greatly improved fluorescence properties. *J. Biol. Chem*. 260,


Steranka, L. R., Farmer, S. G., and Burch, R. M. (1989). Antagonists of B2 bradykinin receptors. *FASEB J*. 3, 2019–2025.


**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.

*Received: 12 March 2014; accepted: 02 June 2014; published online: 30 June 2014. Citation: Kuwahara M (2014) Role of [Ca2*+*]i and F-actin on mesothelial barrier function. Front. Physiol. 5:232. doi: 10.3389/fphys.2014.00232*

*This article was submitted to Membrane Physiology and Membrane Biophysics, a section of the journal Frontiers in Physiology.*

*Copyright © 2014 Kuwahara. 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.*

**REVIEW ARTICLE** published: 24 June 2014 doi: 10.3389/fphys.2014.00221

## Tight junction physiology of pleural mesothelium

#### *Alexander G. Markov1 and Salah Amasheh2 \**

*<sup>1</sup> Department of General Physiology, St. Petersburg State University, St. Petersburg, Russia*

*<sup>2</sup> Department of Veterinary Medicine, Institute of Veterinary Physiology, Freie Universität Berlin, Berlin, Germany*

#### *Edited by:*

*Sotirios G. Zarogiannis, University of Thessaly, Greece*

#### *Reviewed by:*

*Alexi Alekov, Medizinische Hochschule Hannover, Germany Jose L. Reyes, National Polytechnic Institute, Mexico*

#### *\*Correspondence:*

*Salah Amasheh, Institute of Veterinary Physiology, Oertzenweg 19b, 14163 Berlin, Germany e-mail: salah.amasheh@fu-berlin.de* Pleura consists of visceral and parietal cell layers, producing a fluid, which is necessary for lubrication of the pleural space. Function of both mesothelial cell layers is necessary for the regulation of a constant pleural fluid volume and composition to facilitate lung movement during breathing. Recent studies have demonstrated that pleural mesothelial cells show a distinct expression pattern of tight junction proteins which are known to ubiquitously determine paracellular permeability. Most tight junction proteins provide a sealing function to epithelia, but some have been shown to have a paracellular channel function or ambiguous properties. Here we provide an in-depth review of the current knowledge concerning specific functional contribution of these proteins determining transport and barrier function of pleural mesothelium.

**Keywords: pleura, tissue barrier, mesothelial cells, tight junctions, claudins**

## **INTRODUCTION**

Pleura, peritoneum and pericardium are squamous epithelia of mesothelial origin, which line lungs, chest cavity, abdomen, and heart, respectively. Pleura can be divided in the pleura visceralis or pleura pulmonalis covering lungs, and pleura parietalis covering the ribs. Both are delimitated by a single layer of epithelial cells which attach to a lamina propria, and have different functions regarding transport and barrier properties. Between the two layers, the pleura cavity is located (**Figure 1**). The pleura provides a fluid layer for lung movement, and therefore is necessary for breathing This small volume of pleural liquid is a result of filtration and absorptive processes, and its volume and composition is tightly regulated (Zocchi, 2002). An imbalance of this regulation can result in an accumulation of fluid between the parietal and the visceral pleura, a condition defined as pleural effusion (Light, 1997).

The difference between hydrostatic and colloid-osmotic pressure of pleural liquid and capillary blood results in filtration of pleural liquid in the pleural cavity. In caudal parietal pleura, the elimination of liquid from a pleural cavity occurs through lymphatic vessels. In this context, further mechanisms related to mesothelial cell activity have been discussed to be more relevant, including a solute-coupled absorption of liquid from the pleural space (Zocchi, 2002). However, vectorial transport across epithelia depends on a concerted action of transcellular transport mechanisms in combination with paracellular channel and barrier functions, and several transcellular transport mechanisms have been identified to participate in the formation of pleural liquid. For analysis, electrophysiological measurements have been performed in different species including man (Hatzoglou et al., 2001; Sarkos et al., 2002; Kouritas et al., 2008; Markov et al., 2011). Moreover, the molecular correlate of paracellular channel and barrier function has been discovered in recent years, namely the tight junction (TJ) which shows a tissue specific expression of tight junction proteins determining the functional properties of the tissues (for review, see e.g., Amasheh et al., 2011; Rosenthal et al., 2012). Neighboring epithelial cells are linked together by these structures, which show a distinct strand pattern in freeze fracture electron micrographs (Staehelin, 1973). Typically tight junctions are located between the apical and the basolateral membrane of epithelial cells, providing not only a *gate* function describing the role for paracellular permeability, but also a *fence* function which is a prerequisite for polarity of the cells, as it delimits diffusion of basolateral and apically located integral membrane molecules to the opposite membrane compartment, respectively (Tsukita et al., 2001). Taken together, both transcellular and paracellular determinants of transport and barrier function can be regarded to contribute to the generation of different ionic composition of pleural fluid compared to plasma (**Table 1**).

Furthermore, inflammation was identified as one major mechanism perturbing barrier integrity, as shown previously in inflammatory bowel diseases (Amasheh et al., 2009a). In this context, also an effect of pleura inflammation on tight junction proteins was observed (Markov et al., 2011).

## **TRANSPORT AND BARRIER MECHANISMS IN PLEURA MESOTHELIUM**

#### **ELECTROPHYSIOLOGICAL PROPERTIES OF PLEURA MESOTHELIUM**

In Ussing chambers, transmesothelial potential (*Vm*), and transmesothelial resistance (*Rm*) have been reported. The experiments revealed differences concerning absolute values, which might be attributed to different specimens and preparative protocols. However, these differences were rather marginal and in comparison with studies focusing on established epithelial models of kidney and intestine (Amasheh et al., 2002; Markov et al., 2010), these values indicated rather leaky properties of both, visceral and parietal pleura (**Table 2**).

The existence of electrophysiological parameters indicates transport and barrier mechanisms, as transmesothelial transport

**Table 1 | Comparison of ionic composition of pleural fluid and plasma.**


*\*Corrected according to the concentration of electrolytes per liter of serum water or of pleural liquid water.*

implies a transmesothelial barrier function to limit the paracellular pathway during active transport mechanisms, restricting back leak and therefore allowing vectorial transport and the generation of gradients, as e.g., also demonstrated in human gastrointestinal tract before (Amasheh et al., 2009b). This functional interplay of transport and barrier function can be identified by transmesothelial potential measurements; **Figure 2**.

There are specific regional functional differences in organs with large epithelial surface, as shown in detail for intestine and kidney (Markov et al., 2010; Amasheh et al., 2011). In intestine, this segment specific correlation of barrier properties was analyzed by a combination of electrophysiological and molecular analyses (Markov et al., 2010). In analogy the segmentation of different pleura areas may also be associated with differential barrier properties. Electrophysiological experiments employing paracellular flux markers of different masses, and protein expression profiles allow a detailed analysis of these variations. Currently, **Table 2 | Comparison of transmesothelial potential (***Vm***) and resistance (***Rm***) values measured in different species including human tissue specimens.**


*– not provided.*

different electrophysiological properties are known concerning parts of parietal pleura, namely mediastinal, costal and diaphragmal pleura. In sheep, transmesothelial resistance of costal pleura showed a lower transmesothelial resistance than diaphragmal pleura (Zarogiannis et al., 2007) whereas in human pleura, a higher resistance of cranial pleura compared to caudal pleura was reported (Kouritas et al., 2008; **Table 2**).

For analyses of pleural transport function, ouabain and amiloride have been employed. Whereas ouabain is a selective blocker of the—typically basolaterally located—Na+/K+-ATPase, application of amiloride inhibits the epithelial sodium channel (ENaC), typically located in the apical membrane. Application of ouabain and amiloride had different effects on human parietal pleura. Whereas in the caudal region, an increase of transmesothelial resistance was reported, in the cranial and mediastinal pleura no significant ouabain effect was observed (Kouritas et al., 2008). This apparent contradiction might indicate that there is no generally uniform mapping, as some studies refer to the cranial-caudal axis, and others referring to single rib areas (Kouritas et al., 2008).

#### **PLEURAL TRANSMESOTHELIAL TRANSPORT SYSTEMS**

Transport properties have been analyzed in pleura in detail by inducing a hydrothorax (Agostoni and Zocchi, 1990). This model includes an injection of Ringer solution into the pleural cavity to replace single ions, and the application of a variety of blockers including ouabain and amiloride. With this technique, a solute-coupled absorption of liquid from the pleural space was observed (Agostoni and Zocchi, 1990; Zocchi et al., 1991). Moreover, a general epithelial phenotype was identified, which included the expression of the Na+/K+-ATPase. Interestingly, the distinct localization of the Na+/K+-ATPase is not resolved yet: Western blots revealed an expression of the Na+/K+-ATPase a1 subunit in both, visceral and parietal pleura of rabbits, and mesothelial primary cell culture of rabbits (Sironi et al., 2007, 2008). Existence of this subunit is characteristic for epithelia and endothelium (Krenek et al., 2006; Gupta et al., 2012). Employing the hydrothorax model, injection of Ringer's solution with the Na+/K+-ATPase blocker ouabain resulted in a reduction of water absorption, and a decrease of ion absorption from the pleural cavity (Agostoni and Zocchi, 1990; Zocchi et al., 1991). These findings indicated that the Na+/K+-ATPase was located in the apical membrane of pleural mesothelium, but a discrimination between visceral and parietal pleura was approached only in later experiments employing the Ussing chamber technique. In these studies, analysis of transmesothelial resistance of different pleural specimens with ouabain revealed differences of Na+/K+ ATPase localization in visceral and parietal pleura (Hatzoglou et al., 2001), as ouabain induced an increase of transmesothelial resistance only in visceral pleura when added to the apical compartment, whereas in parietal pleura, an effect of ouabain on both, the apical and basolateral side of the epithelium, was observed.

Further evidence for the localization of Na+/K+-ATPase of parietal pleura was found in human tissue. In Ussing chamber experiments employing parietal pleura specimens, apical addition of ouabain showed an effect (Kouritas et al., 2008).

As until now no immunohistochemical detection of the localization has been published, these results cannot fully provide an evidence for localization of Na+/K+-ATPase in pleura mesothelium, though. This still makes it difficult to define the transcellular absorption from and secretion into the pleural cavity in detail. Another indicator of typical epithelial properties is provided by apical expression of the epithelial sodium channel (ENaC) and exchangers, which also contribute to vectorial transport. ENaC is the limiting factor for epithelial sodium absorption in a variety of tissues and organs, including the distal nephron and colon epithelium (Amasheh et al., 2000). For proper function in epithelia, functional co-regulation of the tight junction to limit paracellular back leak has been reported (Amasheh et al., 2009b).

Evidence of ENaC in parietal and visceral pleura was provided by the hydrothorax model employing amiloride injection, which decreased water absorption, and transport of Na+ and Cl− from the pleural cavity (Agostoni and Zocchi, 1990; Zocchi et al., 1991). Terbutaline, an activator of ENaC, as well as the specific ENaC inhibitor amiloride, were injected in mice pleural cavity, leading to an increase, and decrease of water absorption from the pleural cavity, respectively (Jiang et al., 2003).

Application of amiloride from the apical side of human and sheep parietal pleura increases transmesothelial resistance (Sarkos et al., 2002; Kouritas et al., 2008). In contrast, a different group reported that amiloride from the apical side of parietal pleura did not change transmesothelial resistance, but increased the value when added to the basolateral side (Hatzoglou et al., 2001).

This apparent contradiction could be resolved by Nie et al. (2009); when ENaC was apically detected in primary culture of human mesothelial cells and murine parietal pleura (Nie et al., 2009). Although these studies did not include a mapping of the expression in pleura as shown e.g., on functional level (Kouritas et al., 2008), they might become important for the understanding of pathophysiological regulatory mechanisms of transcellular pleural transport, shown for histamine and prostaglandins, recently (Kouritas et al., 2011, 2012).

In visceral pleura, Ussing chamber studies on sheep tissue specimens did not reveal effects of amiloride, when applied to the apical or basolateral side, indicating no functional ENaC expression at all (Hatzoglou et al., 2001). Therefore, it remains an open question if ENaC is localized in the apical membrane of visceral pleura.

In pleural fluid, concentration of glucose is comparable with blood levels. As glucose is present in the pleural space, a transport mechanism for glucose can be expected. In accordance, Western blotting experiments revealed expression of the sodium-glucose linked transporter-1 (SGLT1) in visceral and parietal pleura of rabbit and sheep, and human primary cells (Sironi et al., 2007, 2008). Functional properties were analyzed by experiments using the SGLT-blocker phloricin in the hydrothorax model, which reduced water absorption from the pleural space (Zocchi et al., 1996). Na+-Glucose cotransport can be molecular candidate for solute-coupled absorption of liquid from the pleural space, indicating an epithelial phenotype as also found e.g., in intestine and kidney.

Taken together the analyses of transport function, a clear indication of a typical epithelial phenotype is not given. To discriminate an epithelial phenotype in more detail, a lateral diffusion of molecules has to be restricted by tight junctions, though. As epithelial cells typically show a tissue- and organ-specific expression pattern of claudins, these proteins may provide more insight regarding the epithelial functions of pleura.

## **PHYSIOLOGICAL CONTRIBUTION OF TIGHT JUNCTION PROTEINS**

The ignition of molecular tight junction protein research occurred in 1993, when the group of Shoichiro Tsukita described the first discovery of a molecular correlate of barrier function, namely occludin (Furuse et al., 1993). Before this time point, it was even not clear if the tight junction, which had been visualized in electron microscopic images for decades before, was of protein or lipoid origin. This finding soon catalyzed the establishment of a new research field, resulting in a broad variety of novel basic physiological as well as clinically related findings. Soon after Tsukita's pioneer work in the field of molecular tight junction analysis, a second landmark paper was published, this time presenting the first members of the family of claudins, namely claudin-1 and -2. Paralleled by the finding that occludin might not be the most crucial factor for barrier integrity, as the complete knock out did not show a perturbation of barrier properties (Furuse et al., 1998; Schulzke et al., 2005), a series of further publications highlighted the functional contribution of single claudins to major epithelial functions (Amasheh et al., 2011). These studies led to a classification of single members of the claudin family in barrier building (sealing), pore-forming (paracellular channel function) and intermediate claudins, which still is valid and can be regarded as a general outcome of tight junction research. Finally, a third type of tight junction protein was described, which led to a novel classification of non-claudin tight junction proteins as members of the TAMP family, namely tricellulin, MarvelD3, and occludin (Ikenouchi et al., 2005; Steed et al., 2009).

Among the variety of tetraspan TJ proteins that have been reported to be localized within TJ strands of different organs, occludin (Furuse et al., 1993) and claudins (Furuse et al., 1993) have been identified to primarily provide a barrier against the paracellular passage of ions. Although occludin was the first tetraspan membrane protein that was detected within TJ strands (Furuse et al., 1993), its contribution to barrier properties is still discussed, though (Schulzke et al., 2005). At least, ubiquitous detection in immunofluorescent stainings, established occludin as a general marker of TJ localization, which has been evaluated in many studies (Amasheh et al., 2002, 2005, 2009a,b; Dittmann et al., 2014; Markov et al., 2014). Moreover, occludin can also be visualized in pleura (**Figure 3**).

In contrast, members of the claudin family have been attributed to the organ-specific properties of epithelia, as e. g. shown in intestine, mammary gland, kidney and lungs (Kaarteenaho et al., 2010; Kirk et al., 2010; Markov et al., 2010, 2012). The single contribution of claudins to barrier properties varies, though. As e.g., claudin-1, -3, and -5, decrease tight junctional ion permeability, other claudins have been demonstrated to specifically mediate paracellular permeability, as it has been shown in detail for claudin-2 which forms a paracellular channel (Amasheh et al., 2002, 2005; Furuse et al., 2002; Milatz et al., 2010).

Information on the molecular correlate of barrier function in the pleura reaches back to the pre-molecular era of tight junction research, when desmosomes and TJs have been discovered as intercellular contacts by electron microscopy (Wang, 1974), and a final breakthrough was achieved by detection of claudins and the correlation with functional barrier properties (Markov et al., 2011).

Recently, for the first time a combined analysis of pleura mesothelial barrier function and expression of TJ proteins was performed. In this study, claudin-1, -3, -5, and -7, were detected in visceral pleura. In parietal pleura, the same TJ proteins were detected, except claudin-7 (Markov et al., 2011; **Figure 4**). Moreover, in inflamed pleura, claudin-2 was induced, indicating a typical pathophysiological mechanism (see paragraph 4).

According to the functional contribution to barrier properties, claudins can be divided in three groups, namely (i) sealing tight junction proteins, (ii) claudins mediating paracellular permeability and (iii) claudins with ambiguous function. Among the claudins detected in pleura mesothelium, claudin-1, -3, and -5 belong to the first group, whereas claudin-2 belongs to the second, and claudin-7 to the third group.

**FIGURE 3 | Detection of tight junction proteins in pleural cells.** Immunostaining of occludin with anti-occludin (green), detected by confocal laser scanning microscopy reveals a honeycomb-like distribution of tight junctions in pleural cells (visceral pleura, typical experiment, nuclei stained in blue, bar: 5μm).

## **PLEURA PROTEINS WITH PORE FORMING, SEALING, AND AMBIGUOUS FUNCTION**

Claudin-1-deficient mice die within hours after birth because of dehydration. These animals show a severe weight loss due to evaporation of water through the skin (Furuse et al., 2002). Therefore, claudin-1 is regarded as one major barrier-building TJ protein.

The functional contribution of claudin-3 has been analyzed in detail by analysis of stably transfected cells, recently (Milatz et al., 2010). This study demonstrates a strong sealing effect on the paracellular pathway regarding the passage of cations, anions, and uncharged solutes. These findings are in accordance with the literature, which reports a ubiquitous presence of claudin-3 in many epithelia including kidney and intestine (Kiuchi-Saishin et al., 2002; Markov et al., 2010).

Sealing properties of claudin-5 have been analyzed in detail in both knock out experiments, and stable transfection of epithelial cells (Nitta et al., 2003; Amasheh et al., 2005). Moreover, recent findings suggest that apart from a tightening of the paracellular barrier against the passage of ions, also the passage of uncharged molecules up to a size of 330 Da is restricted by claudin-5 (Dittmann et al., 2014). Taken together a combination of claudin-1, -3, and -5 can also be found in other epithelia with distinct barrier properties such as the gastrointestinal tract (Markov et al., 2010) and airway epithelium (Coyne et al., 2003).

In contrast to clearly defined claudin properties outlined above, the contribution of claudin-7 to barrier properties is still discussed, as cell type specific differences were observed concerning Cl− and Na+ permeability (Alexandre et al., 2005; Hou et al., 2006).

The obtained results indicate that barrier properties of both parietal and visceral pleura mesothelium participate in the formation and determination pleural liquid ionic composition, and different expression levels of occludin, claudin-3, -5, and -7, reflect different extent of functional contributions, respectively.

In contrast to the sealing properties provided by the majority of claudins, claudin-2 has been identified to form a paracellular channel selective for small cations and water (Amasheh et al., 2002; Rosenthal et al., 2010). As a TJ protein mediating paracellular permeability, increased claudin-2 expression is discussed to sustain and aggravate inflammation (Amasheh et al., 2009a,b).

## **PLEURAL TIGHT JUNCTIONS IN HEALTH AND DISEASE**

The expression of claudin-1, -3, -5, and -7 in human pleura indicates an important role of specific barrier properties of the mesothelial cell layers (Markov et al., 2011). These findings are in accordance with previous studies, which have underlined specific contributions of single members of the tight junction protein family of claudins to barrier function of other leaky epithelia such as small intestine and proximal tubule (Markov et al., 2010; Amasheh et al., 2011).

Expression of claudins in pleura mesothelium has been shown to be altered in several pathophysiological conditions. These changes can be found in inflammatory events and in cancer. In previous studies, the paracellular channel claudin-2 has been reported to be induced by tumor necrosis factor α (TNFα), which explains e.g., the pathomechanism of inflammatory bowel diseases (Amasheh et al., 2009a, 2010). The two main forms of inflammatory bowel diseases, namely Crohn's disease and ulcerative colitis, as well as collagenous colitis and pouchitis, all typically show a perturbation of TJ protein expression and localization. This mechanism is regarded to be an important factor for the sustained and aggravated course of the diseases mediated by TNFα (Bürgel et al., 2002; Heller et al., 2005; Zeissig et al., 2007; Amasheh et al., 2009a).

In inflamed pleura, a general reduction of tightening TJ proteins and an increase of permeability mediating TJ proteins was reported, recently (Markov et al., 2011). Claudin-2 was reported to be elevated in inflamed pleura, which has also been shown for Crohn's disease, ulcerative colitis, and pouchitis (Heller et al., 2005; Zeissig et al., 2007; Amasheh et al., 2009a). Moreover, in inflamed pleura, a decrease of occludin, claudin-1, -3, -5, and -7 was detected, which also shows parallels with mechanisms reported in inflammatory bowel diseases (Heller et al., 2005; Zeissig et al., 2007; Amasheh et al., 2009a).

In accordance, pleural exudates showed a marked increase of TNFα levels in a mouse lung inflammation model (Mazzon and Cuzzocrea, 2007). Moreover, TNFα is also elevated in other lung pathologies, e.g., tuberculosis and cancer (Qian et al., 2012).

TNFα at higher levels was reported to induce a downregulation of the tight junction scaffolding protein zonula occludens protein 1 (ZO-1), and an opening of the paracellular barrier (Ma et al., 2004). As a prerequisite of the pathomechanism of TNFα altering tight junction protein expression in lungs, the TNFα receptor TNFR1 was identified (Mazzon and Cuzzocrea, 2007). If, apart from claudin-2, a number of tightening tight junction proteins are reduced due to a general perturbation of ZO-1, this results in a decrease of barrier properties despite reduction of the paracellular pore claudin-2 (Schneeberger and Lynch, 1992; Fink and Delude, 2005; Jacob et al., 2005), though. Taken together, whereas the increase of claudin-2 in pleural inflammation indicates a separate regulatory pathway which has been reported for inflammatory bowel diseases, the general reduction of tightening proteins in inflamed pleura also points at a mechanism reported in a mouse lung inflammation model, which is not even based on an increase of claudin-2, but a universal decrease of tight junction proteins which may less specifically perturb the epithelial barrier. Both regulatory pathways however appear to be primarily targeted by TNFα.

These alterations in claudin expression are in contrast to findings reported in analyses of pleura cancer and mesothelioma, which show an alteration of claudin-4. In this context, Claudin 4 was identified to be an important marker for carcinoma vs. mesothelioma diagnosis in pleural and peritoneal biopsies and effusions, as it is detected in metastatic tumor cells but not in benign forms of mesotheliomas (Facchetti et al., 2007; Lonardi et al., 2011).

## **CONCLUSIONS**

Transport and barrier function of pleura mesothelium indicate typical epithelial characteristics. Expression of the major sealing tight junction proteins-1, -3, and -5 indicate a physiological role of the mesothelial barrier function for pleural liquid formation. Moreover, the effects observed in inflamed pleura, namely an induction of the paracellular pore claudin-2 and the decrease of tightening tight junction proteins are in accordance with regulatory mechanisms observed in inflamed epithelia, as shown e.g., for inflammatory bowel diseases. These findings may contribute to future therapeutic and preventive approaches regarding mesothelia.

## **AUTHOR CONTRIBUTIONS**

Alexander G. Markov and Salah Amasheh conceived the review, conducted literature survey and wrote the review.

## **ACKNOWLEDGMENTS**

The study has been supported by the Deutsche Forschungsgemeinschaft (DFG), the Sonnenfeld-Stiftung Berlin, the Partnership Program FU Berlin—University St. Petersburg and the Grant of St. Petersburg State University, Grant No. 1.38.231.2014.

## **REFERENCES**


deficient mice. *Biochim. Biophys. Acta Biomembr*. 1669, 34–42. doi: 10.1016/j.bbamem.2005.01.008


**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.

*Received: 21 March 2014; accepted: 27 May 2014; published online: 24 June 2014. Citation: Markov AG and Amasheh S (2014) Tight junction physiology of pleural mesothelium. Front. Physiol. 5:221. doi: 10.3389/fphys.2014.00221*

*This article was submitted to Membrane Physiology and Membrane Biophysics, a section of the journal Frontiers in Physiology.*

*Copyright © 2014 Markov and Amasheh. 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.*

## Role of H-Ras/ERK signaling in carbon nanotube-induced neoplastic-like transformation of human mesothelial cells

## *Warangkana Lohcharoenkal 1, Liying Wang2, Todd A. Stueckle2, Jino Park3, William Tse3, Cerasela-Zoica Dinu4 and Yon Rojanasakul 1,3\**

*<sup>1</sup> Department of Pharmaceutical Sciences, West Virginia University, Morgantown, WV, USA*

*<sup>2</sup> Health Effects Laboratory Division, Pathology and Physiology Research Branch, National Institute for Occupational Safety and Health, Morgantown, WV, USA*

*<sup>3</sup> Department of Medicine, Mary Babb Randolph Cancer Center, West Virginia University, Morgantown, WV, USA*

*<sup>4</sup> Department of Chemical Engineering, West Virginia University, Morgantown, WV, USA*

#### *Edited by:*

*Sotirios G. Zarogiannis, University of Thessaly, Greece*

#### *Reviewed by:*

*Dimitra Mangoura, Biomedical Research Foundation of the Academy of Athens, Greece Georgios T. Stathopoulos, University of Patras, Greece*

#### *\*Correspondence:*

*Yon Rojanasakul, Department of Pharmaceutical Sciences and Mary Babb Randolph Cancer Center, West Virginia University, 5706 Medical Center Dr., Morgantown, WV 26506, USA e-mail: yrojan@hsc.wvu.edu*

Rapid development and deployment of engineered nanomaterials such as carbon nanotubes (CNTs) in various commercial and biomedical applications have raised concerns about their potential adverse health effects, especially their long-term effects which have not been well addressed. We demonstrated here that prolonged exposure of human mesothelial cells to single-walled CNT (SWCNT) induced neoplastic-like transformation as indicated by anchorage-independent cell growth and increased cell invasiveness. Such transformation was associated with an up-regulation of H-Ras and activation of ERK1/2. Downregulation of H-Ras by siRNA or inactivation of ERK by chemical inhibitor effectively inhibited the aggressive phenotype of SWCNT-exposed cells. Integrin alpha V and cortactin, but not epithelial-mesenchymal transition (EMT) transcriptional regulators, were up-regulated in the SWCNT-exposed cells, suggesting their role in the aggressive phenotype. Cortactin expression was shown to be controlled by the H-Ras/ERK signaling. Thus, our results indicate a novel role of H-Ras/ERK signaling and cortactin in the aggressive transformation of human mesothelial cells by SWCNT.

**Keywords: mesothelial cells, H-Ras, SWCNT, ERK, cortactin, cell transformation**

## **INTRODUCTION**

CNTs share some common properties with the carcinogenic asbestos fibers, including high aspect ratio, biopersistence and route of exposure, which raises a major concern about their potential carcinogenicity. A recent study by Sargent et al. indicated the tumor-promoting effect of CNTs (Sargent et al., 2014), while several other studies suggested the potential mesothelioma pathogenicity of this nanomaterial (Takagi et al., 2008, 2012; Sakamoto et al., 2009). At the cellular level, CNTs were shown to induce malignant transformation of human lung cells upon chronic exposure (Wang et al., 2011, 2014; Lohcharoenkal et al., 2013). In a recent study, we reported neoplastic transformation of human pleural mesothelial cells by chronic CNT exposure and demonstrated the role of matrix metalloproteinase (MMP)-2 in the process (Lohcharoenkal et al., 2013). Since our study has indicated the induction of proto-oncogenes by CNTs (Wang et al., 2014) and since *H-Ras* oncogene has been demonstrated to be involved in the DNA damage signaling induced by CNTs (Tong et al., 2011), we asked if H-Ras plays a role in

**Abbreviations:** SWCNT, single-walled carbon nanotubes; ERK, extracellular signal-regulated kinases; MAPK, mitogen-activated protein kinases; PI3K, phosphatidylinositol 3-kinase; MMP, matrix metalloproteinase; EGF, epidermal growth factor; GSN, gene signaling network; IPA, ingenuity pathway analysis; E-cad, Ecadherin; N-cad, N-cadherin; FN, fibronectin; VIM, vimentin; FAK, focal adhesion kinase; uPA, urokinase-plasminogen activator; BEAS-2B, human bronchial epithelial cells; SAEC, human small airway epithelial cells; MeT-5A, immortalized human pleural mesothelial cells; LP-9, normal human peritoneal mesothelial cells.

the invasive transformation of CNT-exposed human mesothelial cells.

*Ras* oncogene family has been extensively studied during the past few decades. Ras protein is a major regulator of various pathological and physiological processes that control cell proliferation, differentiation and survival. Ras GTPase family proteins are critical players in many signaling networks, connecting a great variety of upstream signals to an even wider set of downstream effector pathways that control numerous cellular functions including cell cycle progression, growth, migration, cytoskeleton dynamic, apoptosis and senescence (Tong et al., 2011). Ras is a membrane-associated guanine nucleotide-binding protein that is normally activated in response to the binding of extracellular signals such as growth factors, receptor tyrosine kinases, T-cell receptors, and phorbol myristate acetate. It acts as a binary signal switch between ON and OFF states. In the resting state, Ras is tightly bound to guanosine diphosphate (GDP), which is exchanged for guanosine triphosphate (GTP) upon binding of extracellular stimuli to cell membrane receptors. In the GTP-bound form, Ras interacts specifically with effector proteins and initiates downstream cascades. To return to the inactive OFF state, Ras cleaves off the terminal phosphate moiety by the intrinsic GTPase reaction and the remaining GDPbound Ras is no longer able to interact with effectors (Kolch, 2002). Reactive free radicals and cellular redox stress have also been proposed to directly activate Ras. Nitric oxide promotes the direct post-translational modification of Ras by S-nitrosylation at Cys118. This modification results in stimulation of guanine nucleotide exchange, possibly by destabilization associated with other effectors, leading to transduction of Ras mediated signals (Diaz-Meco et al., 1994). Three members of the Ras family, Harvey-Ras (H-Ras), Kirsten-Ras (K-Ras), and N-Ras, are known to be activated in human tumors (Lowy and Willumsen, 1993; Stites and Ravichandran, 2009). The amino-terminal 85 amino acids are identical and the middle 80 amino acids exhibit an 85% homology between the Ras proteins, whereas the carboxylterminal sequence is highly divergent (Barbacid, 1987; Boguski and McCormick, 1993). Up to about 30% of all human tumors carry some forms of alteration in the canonical *Ras* genes. The biological effects of Ras proteins are mediated through the activation of several downstream effectors, including Raf, Rac, phosphatidylinositol 3-kinase (PI3K) and Ral (Marshall, 1996). Ras stimulates serine/threonine kinase Raf, followed by activation of the downstream kinase MAPK/ERK kinase (MEK), which in turn phosphorylates extracellular signal-regulated kinases (ERKs) (Kyriakis et al., 1992). In addition to the Ras/Raf/ERK pathway, the small GTPase Rac and PI3K are involved in the mitogenic and oncogenic effects of Ras (Joneson et al., 1996). PI3K is activated by G-protein-coupled receptors in response to extracellular stimuli or by direct interaction with Ras (Kapeller and Cantley, 1994; Rodriguez-Viciana et al., 1994).

Although these Ras proteins share many common signaling pathways leading to similar cellular responses, studies have clearly demonstrated the unique roles of Ras family members in physiological and pathological conditions. Moon *et al*. reported H-Ras, but not N-Ras, induced invasive and migrative phenotypes by activating p38 and ERK signaling pathways, whereas both induced transformed phenotype in human breast epithelial cells through an up-regulation of MMP-2 (Moon et al., 2000). In laryngeal carcinoma, a weaker tumorigenic effect of N-Ras vs. H-Ras and K-Ras was reported (Kiaris and Soandidos, 1995). In this study, we investigated the role of H-Ras in SWCNT-induced neoplastic transformation of human mesothelial cells and evaluated the downstream targets of H-Ras signaling in the transformed cells.

## **MATERIALS AND METHODS**

#### **CELL CULTURE AND SWCNT EXPOSURE**

Human pleural mesothelial MeT-5A (ATCC® CRL9444) cells were acquired from American Type Culture Collection (Manassas, VA) and maintained in M199 medium (Life Technologies, Grand Island, NY) with 5% fetal bovine serum (FBS), 2 mM L-glutamine, 100 U/mL penicillin/streptomycin, 1μg/mL EGF and 50μg/mL hydrocortisone. Human peritoneal mesothelial LP-9 cells were obtained from NIA Aging Cell repository (Camden, NJ) at passage 5 and maintained in a medium containing a 1:1 ratio of M199 and Ham's F-12 nutrient mixture (Life Technologies, Grand Island, NY), supplemented with 10% FBS, 10 ng/mL EGF, 0.4μg/mL hydrocortisone, 100 U/mL penicillin/streptomycin and 2 mM L-glutamine. Cell cultures were performed in a humidified atmosphere of 5% CO2 at 37◦C. SWCNT, synthesized by using a high-pressure carbon monoxide disproportionate process (HiPCO), were obtained from Carbon Nanotechnology (CNI, Houston, TX). Elemental analysis of the supplied CNT by nitric acid dissolution and inductively coupled plasma-atomic emission spectrometry (ICP-AES, NMAM #7300) showed that SWCNT were 99% elemental carbon and contained less than 1% w/w of contaminants.

Cells were continuously exposed to SWCNT at various surface area concentrations of 0.02, 0.06, and 0.2μg/cm2 for 2 months according to the method previously described (Sargent et al., 2014). Briefly, 0.1 mg/mL stocks of SWCNT in phosphate buffer saline (PBS) containing 1% FBS were sonicated and diluted in media (0.1μg/mL) prior to cell exposure. Cultured MeT-5A and LP-9 cells were exposed to the dispersed CNT every 3 days following a PBS wash and passaged once per week.

### **SOFT AGAR COLONY FORMATION ASSAY**

Soft agar assay was performed as previously described (Ottestad et al., 1988). SWCNT-exposed cells at 3 <sup>×</sup> 104 cells were mixed with the culture medium containing 0.5% agar to the final concentration of 0.33% agar. Cell suspension was immediately plated onto the dish coated with 0.5% agar in culture medium. Colonies were examined under a light microscope (Leica DM, IL) after 2 and 4 weeks.

### **CELL INVASION ASSAY**

Cell invasion was determined in BD Matrigel® invasion chamber (BD Biosciences, NJ). Briefly, cells at the density of 3 <sup>×</sup> 104 cells per well were seeded into the upper chamber of the Transwell® unit in serum-free medium. The lower chamber of the unit was added with a normal growth medium containing 5% FBS. The unit was incubated at 37◦C in a 5% CO2 atmosphere for 48 h. The non-invading cells were removed from the inside of insert with a cotton swab. Cells that invaded to the lower side of the membrane were fixed and stained with Diff-Quik® (Dade Behring, Newark, DE). Inserts were visualized under a light microscope. The experiment was performed three times independently and the representative data of one experiment are shown.

#### **WESTERN BLOT ANALYSIS**

Cells at the density of 2.<sup>5</sup> <sup>×</sup> 105 cells were seeded into each well of 6-well plates and cultured to confluence. They were washed twice with ice-cold PBS and incubated in lysis buffer containing 20 mM Tris-HCl (pH 7.5), 1% Triton X-100, 150 mM NaCl, 10% glycerol, 1 mM Na3VO4, 50 mM NaF, 100 mM phenylmethylsulfonyl fluoride, and a commercial protease inhibitor mixture (Roche Molecular Biochemicals, Indianapolis, IN) at 4◦C for 20 min. Cell lysates were collected and analyzed for protein content using the BCA protein assay kit (Pierce Biotechnology, Rockford, IL). Samples containing 50μg of cell lysate proteins per lane were resolved under denaturing conditions by 10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) along with EZ-run pre-stained protein ladder (Fisher Scientific, Pittsburgh, PA) and transferred onto PVDF membranes (Invitrogen, Carlsbad, CA). The transferred membranes were blocked for 1 h in 5% nonfat dry milk in TBST (25 mM Tris-HCl, pH 7.4, 125 mM NaCl, 0.05% Tween 20) and incubated with the appropriate primary antibodies (Cell Signaling Technology, Danvers, MA or Santa Cruz Biotechnology, Dallas, TX) at 4◦C overnight. Membranes were washed twice with TBST

**70**

for 10 min and incubated with horseradish peroxidase-coupled isotype-specific secondary antibodies (Cell Signaling Technology, Danvers, MA) for 1.5 h at room temperature. The immune complexes were detected by enhanced chemiluminescence detection system (Amersham Biosciences, Piscataway, NJ) and quantified using analyst/PC densitometry software (Bio-Rad Laboratories, Hercules, CA).

#### **H-Ras siRNA TRANSFECTION**

Cells were transfected with pre-designed human H-Ras siRNA (Santa Cruz Biotechnology, Dallas, TX: sc-29340) or control siRNA (Santa Cruz Biotechnology, Dallas, TX: sc-37007), according to the manufacturer's protocol. H-Ras expression in the H-Ras and control siRNA-transfected cells was determined by Western blotting as described above.

### **INHIBITION OF ERK MAPK SIGNALING PATHWAY**

SWCNT-exposed cells at 2 <sup>×</sup> 105 cells were seeded into each well of 6-well plates and cultured for 48 h. Cells were then pre-incubated with the ERK inhibitor U0126 (Cell Signaling Technology, Danvers, MA) at different concentrations for 2 h prior to subjecting to cell invasion or Western blot assays as described above.

### **INGENUITY PATHWAY ANALYSIS**

H-Ras-ERK invasion signaling network was generated using Ingenuity Pathway Analysis (IPA, version Fall 2012; Redwood City, CA) from whole genome expression data of SWCNTtreated mesothelial cells previously deposited on NCBI's Gene Expression Omnibus (GenBank ID: GSE48855) (Lohcharoenkal et al., 2013). Gene signaling networks (GSN) associated with cell invasion were created and mapped. Only genes that have the first order relationship with ERK were kept in the network. Genes were included in the GSN if they promoted invasion and were overexpressed or if they inhibited invasion and were underexpressed.

#### **QUANTITATIVE REAL-TIME PCR OF EPITHELIAL-MESENCHYMAL TRANSITION (EMT) TRANSCRIPTIONAL REGULATORS**

The expression of EMT transcriptional regulators including SnaI, Twist, E-cad, N-cad, FN1, and VIM was analyzed in control and SWCNT (0.2μg/cm2)-exposed MeT-5A cells. Briefly, total RNA was isolated from cells using RNeasy mini kit (Qiagen, Valencia, CA), according to the manufacturer's instructions. The extracted RNA was then reverse transcribed into cDNA by high capacity RNA to cDNA kit (Applied Biosystems, Carlsbad, CA). After the reverse transcription reaction was finished, 10 μL of diluted cDNA product (final cDNA quantity 100 ng) was mixed with 10μL of Taqman® master mix (Applied Biosystems) and transferred into Taqman® array plate (Applied Biosystems). Quantification of the PCR products was performed by NFQ-FAM® method using the Applied Biosystems 7500 Real-Time PCR system with the following profile: 1 cycle at 94◦C for 2 min, 40 cycles at 94◦C for 15 s, 60◦C for 1 min, 72◦C for 1 min. Data analysis was performed using the ABI sequence detection software (Applied Biosystems) by relative quantification. The threshold cycle (Ct), which is defined as the cycle at which PCR amplification reaches a significant value, is given as the mean value. The relative expression of each mRNA was calculated by the -Ct method, where -Ct is the value obtained by subtracting the Ct value of the housekeeping gene *18S* mRNA from the Ct value of the target mRNA. The amount of the target relative to *18S* mRNA was expressed as 2−-*Ct*.

## **IMMUNOFLUORESCENCE STAINING**

Cellular cortactin expression was visualized by immunofluorescence microscopy (Zeiss LSM 510 Axiovert 100 M, Zeiss, Thornwood, NY). Briefly, cells were cultured to confluence on glass cover slips and fixed in 4% paraformaldehyde in PBS. The samples were rinsed three times, permeabilized with 1.2% Triton X-100 for 5 min, rinsed three times and blocked with 1% bovine serum albumin (BSA) in PBS for 1 h before staining with 1:100 cortactin primary antibody (Cell Signaling Technology, Danvers, MA) followed by Alexa Fluor-conjugated secondary antibody (Invitrogen, Carlsbad, CA). The stained cells were mounted with ProLong® gold antifade reagent with DAPI (Invitrogen, Carlsbad, CA) and visualized by fluorescence microscopy. All microscopic exposure conditions were set the same between samples for fluorescence intensity comparison.

### **STATISTICAL ANALYSIS**

All experiments were performed in triplicate. Difference between groups was assessed by One-Way analysis of variance (ANOVA). Differences were considered significant if *P*-values were < 0.05.

## **RESULTS**

#### **SWCNT EXPOSURE INDUCES AGGRESSIVE NEOPLASTIC-LIKE PHENOTYPE OF MESOTHELIAL CELLS**

Cancer hallmark phenotypes were determined in SWCNTexposed and control MeT-5A and LP-9 cells by established methods. Anchorage-independent cell growth was determined by assessing the size and number of isolated colonies on soft agar. Increased colony size was observed in the MeT-5A cells exposed to all doses of SWCNT with the highest dose (0.2μg/cm2) inducing the biggest colonies (**Figure 1A**). At the same dose range, SWCNT had minimal effect on colony formation in LP-9 cells (**Figure 1A**), possibly due to their decreased susceptibility and limited lifespan under non-adherent conditions.

The invasiveness of SWCNT-exposed mesothelial cells was assessed by Transwell® invasion assay. SWCNT was able to increase the invasiveness of both MeT-5A and LP-9 cells in a dosedependent manner as compared to their passage-matched control cells (**Figure 1B**). At the highest exposure dose (0.2μg/cm2), significant increase in the number of invading cells was observed in both cell types. These results indicate the invasion-promoting activity of SWCNT in mesothelial cells.

#### **H-Ras OVEREXPRESSION IN SWCNT-EXPOSED MESOTHELIAL CELLS**

We hypothesized that H-Ras expression and its downstream targets may be activated and play a role in the neoplastic transformation of SWCNT-exposed mesothelial cells. To test this possibility, we analyzed H-Ras (also known as transforming protein p21) expression in SWCNT-exposed and control cells by Western blotting. Increased expression of H-Ras was observed in

the SWCNT-exposed MeT-5A and LP-9 cells as compared to control cells (**Figure 2A**). The importance of H-Ras overexpression on the aggressive behavior of SWCNT-exposed cells was evaluated by siRNA silencing experiments. SWCNT-exposed MeT-5A and LP-9 cells were transfected with siRNA against H-Ras (siH-Ras) or control siRNA (siControl), and subjected to cell invasion assays. **Figures 2B,C** show that the siH-Ras treatment resulted in a substantial reduction of the H-Ras expression and a parallel decrease in cell invasion activity as compared to the siControl treatment.

#### **ACTIVATION OF DOWNSTREAM SIGNALING PATHWAY OF H-Ras**

To dissect the signaling pathway downstream of H-Ras that may be involved in the neoplastic phenotype of SWCNTexposed cells, the activation of several known effectors of *Ras* including ERK, JNK, Akt, and NF-κB was investigated. Among

these, p44/42 (ERK1/2) was shown to be prominently activated in the SWCNT-exposed cells, suggesting the possible role of ERK signaling in the aggressive phenotype of these cells (**Figure 3A**).

To assess the functional importance of activated p44/42 in the aggressive phenotype, SWCNT-exposed cells were treated with non-cytotoxic doses of U0126, a selective ERK kinase inhibitor, and analyzed for cell invasiveness by Transwell® invasion assay.

The results showed that the ERK inhibitor effectively inhibited the invasive activity of the cells in a dose-dependent manner (**Figure 3B**). These results support the role of ERK signaling downstream of H-Ras activation in SWCNT-induced cell invasiveness.

#### **POTENTIAL GENES ASSOCIATED WITH ERK SIGNALING IN SWCNT-EXPOSED MESOTHELIAL CELLS**

To determine the possible effectors of H-Ras-ERK signaling in the tested cells, whole genome expression data of SWCNT-exposed mesothelial cells from our previous study was subjected to IPA. Invasion GSN with first order relationship to ERK and a cross-section diagram for location of each gene within the cells are shown in **Figure 4**. Genes of potential interest to the SWCNT-induced cell invasiveness include some proteolytic enzyme-encoding genes (*MMP2*, *PLAU*), *AKT*, cyclin D1, integrin, several inflammatory genes, and EMT related genes (*TWIST*, *FN1*).

### **INTEGRIN ALPHA V AND CORTACTIN OVEREXPRESSION IN SWCNT-EXPOSED MESOTHELIAL CELLS**

Further investigations on the genes from the invasion GSN were performed using Western blotting and quantitative real-time PCR. Western blot analysis showed a dose-dependent increase in integrin alpha V expression in SWCNT-exposed MeT-5A and LP-9 cells (**Figure 5A**). Cortactin, a protein known to be involved in cell motility and important in neoplasia development, was also overexpressed in the SWCNT-exposed cells. PCR analysis of EMT transcriptional regulators showed a repression of all EMT regulators examined in the exposed cells (**Figure 5B**). The relationship between cortactin and H-Ras-ERK activation was further evaluated by immunofluorescence staining. The results showed that cortactin was substantially induced in the SWCNT-exposed cells,

**FIGURE 4 | Whole genome analysis of cell invasion network in SWCNT-exposed mesothelial cells.** H-Ras-ERK invasion signaling network was created from whole genome expression data previously deposited on NCBI's Gene Expression Omnibus (GenBank ID: GSE48855) using IPA. Gene signaling networks (GSN) associated with cell invasion were created and mapped. Genes were included in the GSN if they promoted invasion and were overexpressed or if they

inhibited invasion and were underexpressed. Invasion GSN with first order relationship to ERK **(A)** and a cross-section diagram for location of each gene within the cells **(B)** are shown. *MMP2*, *PLAU*, *AKT*, *CCND1*, *ITG*, several inflammatory genes and EMT-related genes (*TWIST*, *FN1*) were found to be the up-regulated genes with first relationship with ERK in the GSN. Most of these genes were reported to involve in cell motility and metastasis process.

consistent with the H-Ras/ERK data (**Figure 5C**). Overexpression of H-Ras in the MeT-5A cells also resulted in the up-regulation of cortactin (Supplementary Figure 1A), whereas siRNA downregulation of H-Ras decreased the cortactin expression in SWCNTexposed cells (**Figure 5D**). In addition, the ERK kinase inhibitor U0126 mitigated the cortactin expression as analyzed by Western blotting (**Figure 5E**).

## **DISCUSSION**

Rapid increase in CNT production and utility has raised a concern over the potential adverse effects of CNTs on human health and the environment. Both *in vivo* and *in vitro* studies have provided useful information on the biological and toxicological effects of CNTs; however, most of these studies have focused on short-term acute effects of the nanomaterials, which is due in part to the lack of appropriate experimental models for long-term studies. We have developed a chronic cellular exposure model to study the long-term biological effects of nanomaterials in various lung cell types including human bronchial epithelial cells (BEAS-2B), human small airway epithelial cells (SAEC), and human pleural mesothelial cells (MeT-5A) (Wang et al., 2011, 2014; Lohcharoenkal et al., 2013). Neoplastic transformation of these cells was demonstrated upon chronic exposure to low-dose CNTs as demonstrated by their anchorage-independent cell growth, apoptosis resistance, increased cell motility and angiogenesis. The induction of proto-oncogenes and cell division anomalies including centrosome fragmentation, mitotic spindle disruption, and aneuploidy has also been reported in CNT-exposed cells (Sargent et al., 2009; Wang et al., 2014). Despite its established importance in lung carcinogenesis, the role of H-Ras in CNTinduced carcinogenesis has not been reported, although a recent study indicated the involvement of H-Ras signaling in DNA damage caused by CNTs (Tong et al., 2011). Thus, the role of H-Ras in the neoplastic transformation of human mesothelial cells was focused in this study.

Two types of human mesothelial cells (MeT-5A and LP-9) were continuously exposed to different concentrations (0.02, 0.06, and 0.2μg/cm2) of SWCNT for 2 months and analyzed for cancer hallmark phenotypes. As shown in **Figure 1**, SWCNT induced colony formation and cell invasion in MeT-5A cells, whereas it induced mainly cell invasion with minimal effect on colony formation in LP-9 cells. Anchorage-independent cell growth is the most commonly used *in vitro* indicator of malignant transformation and it correlates well with *in vivo* tumorigenicity (Risser and Pollack, 1974; Shin et al., 1975). This indicates malignant transformation of SWCNT-exposed MeT-5A cells, although the control MeT-5A cells also formed slow-growing colonies on soft agar. This could be due to the fact that the cells were immortalized by transfection with the pRSV-T plasmid (an SV40 oriconstruct containing the SV40 early region and sarcoma virus long terminal repeat). ATCC indicates that this cell line can form colonies in a semi-solid medium but is non-tumorigenic

in immunocompromised mice. Therefore, the colony formation observed in the control cells is due to indigenous properties of the cells, and not because of their tumorigenicity. For normal cells, spontaneous colony formation was concentration dependent and the threshold for colony formation varied between cell types and some exogenous source of colony stimulating factor (CSF) may be needed. In case of normal mesothelial cells such as LP-9 cells, high concentrations of epidermal growth factor (EGF) and hydrocortisone in addition to fetal calf serum were required to induce colony formation in semi-solid medium (La Rocca and Rheinwald, 1985). Cell invasion is a crucial step in many physiological processes and its impairment has been implicated in many pathological disorders such as tumor growth and metastasis. It has been used to assess the aggressive and malignant phenotypes of cells (Cho and Klemke, 2000). Our results demonstrated the induction of aggressive cancer phenotypes by SWCNT in mesothelial cells.

H-Ras expression was examined in SWCNT-exposed cells and was found to be elevated in both the SWCNT-exposed MeT-5A and LP-9 cells (**Figure 2A**). The functional importance of H-Ras overexpression was demonstrated by colony formation assay in H-Ras-transfected MeT-5A cells (Supplementary Figure 1B). Since overexpression of H-Ras has been shown to be important in cancer cell invasion and progression (Theodorescu et al., 1990), we hypothesized that H-Ras overexpression and the activation of its downstream targets may be crucial to the invasive transformation of SWCNT-exposed mesothelial cells. The results of our H-Ras knockdown experiments support this notion (**Figures 2B,C**).

Downstream effectors of *Ras* include members of the MAPK family comprising of JNK, ERK and stress-activated protein kinase-2 (p38). MAPKs are among the major kinases that transduce extracellular signaling into cellular responses and play a pivotal role in the regulation of cell proliferation, apoptosis, differentiation, cytoskeleton remodeling, and cell cycle regulation (Brunet and Pouyssegur, 1996; Foltz et al., 1997). Key effectors of Ras including ERK, JNK, AKT, and NF-κB were examined in SWCNT-exposed MeT-5A and LP-9 cells. An activation of ERK1/2 (p44/42) was prominent in these cells (**Figure 3A**), suggesting the possible role of H-Ras-ERK signaling in the invasiveness of SWCNT-exposed mesothelial cells. Our results on the inhibitory effect of U0126 ERK inhibitor on the invasivity of SWCNT-exposed MeT-5A cells support this notion (**Figure 3B**). Consistent with this finding, several studies have indicated the involvement of ERK in the migration and invasion of various forms of cancer including glioblastoma, pancreatic carcinoma and lung cancer (Lakka et al., 2000; Lu et al., 2011).

Invasion GSN obtained from the whole genome expression data of SWCNT-exposed mesothelial cells was created and filtered for ERK-related genes as shown in **Figure 4**. *MMP2*, *PLAU*, *AKT*, *CCND1*, *ITG*, several inflammatory genes and EMT-related genes (*TWIST*, *FN1*) were all up-regulated and exhibited the first relationship with ERK. *MMPs* are signature invasion marker genes that encode proteins involved in the degradation of ECM and are typically highly active during cancer development and progression (Passlick et al., 2000). MMPs belong to a family of zinc-dependent endopeptidases which are divided into different classes. MMP-2 (gelatinase A) has an important role in basement membrane turnover due to its specific activity to collagen type IV or gelatin. Its degradation plays a role in cell invasion of the vasculature and is considered to have a key role in metastasis. *MMP-2* expression has been associated with the invasiveness of many cancer cell lines and is elevated in high-grade tumors, specifically at the invasive front and in vascular invasion (Birkedal-Hansen et al., 1993; Coussens and Werb, 1996). Our previous study has demonstrated the importance of MMP-2 in cell invasion induced by chronic CNT and asbestos exposure (Lohcharoenkal et al., 2013). Urokinase-plasminogen activator or *PLAU* gene is known to encode uPA, the protease which degrades ECM and plays critical roles in cell migration, tissue remodeling, angiogenesis, tumor invasion, and metastasis (Suzuki et al., 2004). Plasminogen activators convert plasminogen to plasmin, which works efficiently in proteolysis of the fibrin. In addition, plasminogen activators activate several MMPs and could also facilitate the MMP activities. A positive feedback loop between the binding of uPA to uPA receptor (uPAR) and Ras-ERK signaling pathway activation has been reported in many cell types and implicated in cell migration and progression of cancer (Ma et al., 2001). AKT (also known as protein kinase B or PKB) is a serine/threonine kinase that is involved in mediating various biological responses, such as inhibition of apoptosis and stimulation of cell proliferation. There are three highly related isoforms of AKT (AKT1, AKT2, and AKT3) and these represent the major signaling arm of PI3K (Park et al., 2001). MEK/ERK and PI3K/AKT pathways are often concurrently activated by separate genetic alterations in cancer cells but it was reported that ERK and AKT signaling cooperate to translationally regulate metastatic progression of certain types of cancer, e.g., colorectal cancer (Ye et al., 2013). Integrins are cell surface receptors that interact with ECM and mediate intracellular signals that regulate many cellular processes including cell shape, mobility and progression through the cell cycle (Hynes, 2001). Integrins play an important role in cell signaling by affecting the cell signaling pathways of protein kinases including Ras-Raf-MEK-ERK pathway (Schlaepfer et al., 1994). Numerous studies have shown that integrin expression profiles are subject to change during cancer growth and progression and that such change contributes to the aggressive behavior of cancer cells (Danen and Sonnenberg, 2003; Danen, 2005). The enhanced expression of integrin is reported to be associated with EMT and poor prognosis of cancer (Bates et al., 2005).

Western blot and real-time PCR studies demonstrated the up-regulation of integrin alpha V, but not EMT transcriptional regulators, in SWCNT-exposed mesothelial cells (**Figures 5A,B**). Since integrin/focal adhesion kinase (FAK)/cortactin has been reported to regulate cell motility and proliferation, cortactin expression in the SWCNT-exposed cells was examined and found to be upregulated in the SWCNT-exposed MeT-5A and LP-9 cells (**Figure 5A**). Cortactin is a multidomain adapter protein, essentially contributing to cortical actin regulation. Regulation of this pool of actin is controlled by a variety of actin regulatory proteins at integrin or cadherin adhesion sites and is important in many normal and pathological cellular processes, such as adhesion, migration, morphogenesis, tumor progression and metastasis (Weed and Parsons, 2001; Clark et al., 2007). Cortactin functions in actin assembly via interaction with actinrelated protein-2/3 (Arp2/3) complex, which is dependent on Src-mediated phosphorylation of cortactin (Schubert and Dotti, 2007).

Besides tyrosine phosphorylation, cortactin is a target for multiple serine/threonine kinases (Martin et al., 2006). Stimulation of tumor cells with EGF leads to phosphorylation of serine residues 405 and 418, coincident with a characteristic shift in cortactin electrophoretic mobility from 80 to 85 kDa in SDS-PAGE. The mobility shift and phosphorylation of S405/S418 are impaired by pharmacologic inhibition of mitogen activated protein/extracellular signal regulated kinase (MEK)1/2. Biochemical evidence indicates that the MEK effector kinases ERK1/2 directly phosphorylate cortactin at these sites (Campbell et al., 1999). Thus, the relationship between cortactin and H-Ras-ERK activation in SWCNT-exposed cells was evaluated and the positive correlation between H-Ras and cortactin expression was observed (**Figure 5C** and Supplementary Figure 1A). In good agreement, knockdown of H-Ras or chemical inhibition of ERK kinase decreased the level of cortactin expression in the SWCNT-exposed cells (**Figures 5D,E**).

In summary, we demonstrated that prolonged exposure of human mesothelial cells to SWCNT induced aggressive neoplastic-like transformation in concomitant with H-Ras upregulation. knockdown and overexpression studies indicated the relationship between H-Ras expression and the invasive phenotype of SWCNT-exposed cells. ERK1/2 was identified as an important effector for the aggressive phenotype of the cells as indicated by their reduced invasiveness by ERK kinase inhibitor. Whole genome microarray and Western blot analyses indicated the possible involvement of integrin alpha V in the H-Ras-ERK invasion signaling. Additionally, cortactin was shown to be a downstream target of H-Ras-ERK signaling in the SWCNTexposed cells.

## **AUTHOR CONTRIBUTIONS**

Warangkana Lohcharoenkal designed and performed cellular and molecular studies, and prepared the manuscript. Liying Wang prepared nanoparticle preparations and performed chronic exposure. Todd A. Stueckle performed microarray and Ingenuity Pathway Analysis. Jino Park performed real-time PCR experiments. William Tse participated in the design of the study and provided research reagents. Cerasela-Zoica Dinu characterized nanomaterials. Yon Rojanasakul designed and coordinated the project, and prepared the manuscript. All authors read and approved the final manuscript.

## **DISCLAIMER**

The findings and conclusions in this report are those of the authors and do not necessarily represent the views of the National Institute for Occupational Safety and Health.

## **ACKNOWLEDGMENTS**

This work was supported by grants from the National Institutes of Health (R01-HL095579 and R01-ES022968) and National Science Foundation (EPS-1003907). Imaging experiments were performed in the West Virginia University Imaging Facility, which is supported in part by the Mary Babb Randolph Cancer Center and NIH grants P20 RR016440, P30 RR032138/GM103488, and P20 RR016477.

#### **SUPPLEMENTARY MATERIAL**

The Supplementary Material for this article can be found online at: http://www.frontiersin.org/journal/10.3389/fphys. 2014.00222/abstract

## **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.

*Received: 09 March 2014; accepted: 27 May 2014; published online: 12 June 2014. Citation: Lohcharoenkal W, Wang L, Stueckle TA, Park J, Tse W, Dinu C-Z and Rojanasakul Y (2014) Role of H-Ras/ERK signaling in carbon nanotube-induced neoplastic-like transformation of human mesothelial cells. Front. Physiol. 5:222. doi: 10.3389/fphys.2014.00222*

*This article was submitted to Membrane Physiology and Membrane Biophysics, a section of the journal Frontiers in Physiology.*

*Copyright © 2014 Lohcharoenkal, Wang, Stueckle, Park, Tse, Dinu and Rojanasakul. 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.*

digital media

of impactful research

article's readership