IntroductionMesothelium lines the peritoneal, pleural and pericardial cavities,with visceral and parietal surfaces covering the internal organsand body wall, respectively. Mesothelium comprises amonolayer of predominantly flattened epithelial-like cells restingon a basement membrane supported by connective tissue thatcontains blood vessels, lymphatics and subserosal mesenchymalcells (Wang, 1974). Ultrastructural analysis of polarisedmesothelial cells demonstrates tight junctions (zonula occludens)located towards their luminal aspect (Baradi and Hope, 1964;Kluge and Hovig, 1967). Injury to this layer can lead to theformation of adhesions, bands of fibrous tissue that form betweenserosal surfaces. Adhesions occur in up to 95% of patientsundergoing abdominal surgery (Menzies and Ellis, 1990),therefore understanding the mechanisms regulating normalserosal repair may lead to novel strategies to prevent thiscommon surgical disorder.

The mechanisms involved in mesothelial regenerationfollowing injury are controversial. Hertzler (Hertzler, 1919) wasthe first to observe that small and large peritoneal injuries healedat the same rate and concluded that the mesothelium could notregenerate solely by centripetal migration of cells at the woundedge as occurs for healing of squamous epithelia. Subsequently,

several hypotheses have been proposed for the origin of the cellsin regenerating mesothelium; these cells include subserosalmesenchymal precursors (Ellis et al., 1965; Raftery, 1973; Bolenet al., 1986), bone marrow-derived precursors (Wagner et al.,1982), free-floating macrophages (Eskeland and Kjærheim,1966; Ryan et al., 1973) and free-floating mesothelial cells(Cameron et al., 1957; Watters and Buck, 1973; Whitaker andPapdimitriou, 1985). To date, the most accepted proposal is thatrepopulating mesothelial cells originate from a pool ofpluripotent subserosal fibroblast-like cells, which migrate to theserosal surface, divide and differentiate into mesothelial cells(Ellis et al., 1965; Raftery, 1973). However, irradiation studieshave demonstrated impaired local mesothelial regeneration,which was recoverable by addition of peritoneal lavage cells(Whitaker and Papadimitriou, 1985), suggesting that subserosalfibroblasts are not the source of regenerating mesothelial cells. Inaddition, studies of the kinetics of serosal repair demonstratedthat subserosal cells were not essential for mesothelial healingand that the regenerating cells were likely to originate from thesurrounding uninjured serosal surface (Mutsaers et al., 2000).

In 1957, Cameron and colleagues proposed that mesothelialhealing involved attachment of free-floating mesothelial cells tothe injured surface. Peritoneal lavage fluid recovered from

1383

Regeneration of the mesothelium is unlike that of otherepithelial-like surfaces, as healing does not occur solely bycentripetal migration of cells from the wound edge. Themechanism of repair of mesothelium is controversial, but itis widely accepted, without compelling evidence, thatpluripotent cells beneath the mesothelium migrate to thesurface and differentiate into mesothelial cells. In this studywe examined an alternative hypothesis, using in vivo cell-tracking studies, that repair involves implantation,proliferation and incorporation of free-floating mesothelialcells into the regenerating mesothelium. Culturedmesothelial cells, fibroblasts and peritoneal lavage cellswere DiI- or PKH26-PCL-labelled and injected into ratsimmediately following mesothelial injury. Implantation oflabelled cells was assessed on mesothelial imprints usingconfocal microscopy, and cell proliferation was determinedby proliferating cell nuclear antigen immunolabelling.Incorporation of labelled cells, assessed by the formation ofapical junctional complexes, was shown by confocal

imaging of zonula occludens-1 protein. Labelled culturedmesothelial and peritoneal lavage cells, but not culturedfibroblasts, implanted onto the wound surface 3, 5 and 8days after injury. These cells proliferated and incorporatedinto the regenerated mesothelium, as demonstrated bynuclear proliferating cell nuclear antigen staining andmembrane-localised zonula occludens-1 expression,respectively. Furthermore, immunolocalisation of themesothelial cell marker HBME-1 demonstrated that theincorporated, labelled lavage-derived cells weremesothelial cells and not macrophages as it had previouslybeen suggested. This study has clearly shown that serosalhealing involves implantation, proliferation andincorporation of free-floating mesothelial cells into theregenerating mesothelium.

Key words: Mesothelium, Wound healing, Tight junction,Fluorescent dyes, Confocal microscopy

Summary

Evidence for incorporation of free-floating mesothelialcells as a mechanism of serosal healingAdam J. Foley-Comer, Sarah E. Herrick, Talib Al-Mishlab, Cecilia M. Prêle, Geoffrey J. Laurentand Steven E. Mutsaers*Department of Medicine, Royal Free and University College Medical School, The Rayne Institute, London, WC1E 6JJ, UK*Present address: University Department of Surgery, University of Western Australia, Royal Perth Hospital, Perth, Western Australia, 6000, AustraliaAuthor for correspondence (e-mail: a.foley-comer@ucl.ac.uk)

Accepted 8 January 2002Journal of Cell Science 115, 1383-1389 (2002) © The Company of Biologists Ltd

Research Article

1384

experimental animals following injury to the mesothelium wasfound to contain a significantly higher number of viable free-floating mesothelial cells two days post injury than the controls(Whitaker and Papadimitriou, 1985). The increased free-floatingcell population was thought to be caused by the proliferation ofmesothelial cells adjacent to (Johnson and Whitting, 1962;Mutsaers et al., 2000) and opposing (Watters and Buck, 1973;Fotev et al., 1987) the serosal injury.

In this study, we have conclusively demonstrated that serosalhealing involves incorporation of free-floating mesothelial cellsinto the regenerating mesothelium. Fluorescently labelled celltracking confirmed implantation of cultured and peritoneallavage-derived mesothelial cells onto the denuded wound surfacein a well characterised rodent model of normal serosal repair(Fotev et al., 1987; Mutsaers et al., 1997). Furthermore,proliferation and incorporation of fluorescently labelledmesothelial cells was demonstrated by immunolocalisation ofproliferating cell nuclear antigen (PCNA) and the tight junction-associated protein, zonula occludens-1 (ZO-1), respectively.

Materials and MethodsCell isolation and characterisationNormal mesothelial cells were isolated from the anterior peritonealwall of male Lewis rats (Harlan, Bicester, UK) as previously described(Stylianou et al., 1990). Briefly, peritoneal tissue was incubated in asolution of 0.25% trypsin and 0.02% EDTA in Dulbecco’s modifiedEagle’s medium (DMEM; Gibco, Paisley, UK) for 30 minutes at37°C. The intact tissue was discarded and the remaining cellsuspension centrifuged at 1000 rpm for 5 minutes. The cell pellet wasre-suspended and the cells maintained in DMEM supplemented with15% foetal calf serum (FCS; CSL UK Ltd, Andover, UK), 4 mM L-glutamine (Gibco, Paisley, UK), 5 ng/ml epidermal growth factor(Roche Diagnostics, Lewes, UK), 0.4 µg/ml hydrocortisone (SigmaAldrich, Poole, UK) and antibiotics (penicillin, 100,000 units/l andstreptomycin, 50 mg/l; Gibco, Paisley, UK).

Primary fibroblasts, used as a control in cell tracking studies, wereisolated from the lungs of male Lewis rats. Peritoneal fibroblasts werenot used owing to possible contamination with surface mesothelial cells.Diced lung parenchyma (carefully avoiding the serosal surface) wasincubated in 1 mg/ml of type II collagenase (Worthington BiochemicalCorp, Lakewood, New Jersey, USA) in DMEM for 2 hours at 37°C. Thecell suspension was centrifuged at 1000 rpm for 5 minutes, resuspendedand cells maintained in DMEM supplemented with 10% FCS, 4mM L-glutamine and antibiotics (penicillin, 100,000 units/l and streptomycin,50 mg/l). All cells were grown in a humidified atmosphere of 10% CO2in air at 37°C.

Mesothelial cells, which are embryologically derived from themesoderm, share characteristics of both epithelial and mesenchymalcells (Whitaker et al., 1982) and so were distinguished fromfibroblasts using monoclonal antibodies directed against human pan-cytokeratin (dilution 1:20) and human vimentin (dilution 1:400; DakoLtd, Ely, UK).

Fluorescence labelling of cultured cellsTo examine the role of free-floating cell populations in vivo, culturedmesothelial cells and fibroblasts were fluorescently labelled with the cell-tracking probe DiI (1,1′-dioctadecyl-3,3,3′,3′-tetramethylindo-carbacyanine perchlorate; Molecular Probes Inc., Eugene, Oregon, USA).Briefly, subconfluent cultures, up to passage three, were incubated inserum-free DMEM containing 10 µM DiI for 20 minutes under standardconditions, washed with phosphate buffered saline (PBS; pH 7.3) andthen incubated in standard supplemented medium for a further 30 minutesaccording to the manufacturers instructions. In order to perform studies

examining cell proliferation, the chloromethylbenzamido derivative ofDiI, CM-DiI, was used for cell labelling because of its ability to withstandhistological tissue processing (Andrade et al., 1996).

Previous studies have demonstrated negligible transfer of DiIbetween adjacent membranes (Honig and Hume, 1986), an importantproperty for cell tracking. To confirm this finding, six replicatesuspensions containing equal numbers (2.5×105) of unlabelled andDiI-labelled mesothelial cells were plated and incubated understandard conditions for 3 days. The proportion of cells labelled withDiI was then determined by direct cell counting using a Zeiss MC80DX fluorescent microscope.

Fluorescence labelling of peritoneal lavage cellsTo collect peritoneal lavage cells, male Lewis rats received a widespreadabrasion injury to the anterior peritoneal wall with a sterile gauze swab.Animals were sacrificed 2 days post injury, when the free-floatingmesothelial cell population was maximal (Whitaker and Papadimitriou,1985), and the peritoneal cavity was lavaged with 20 ml serum-freeDMEM. The lavage fluid was centrifuged at 1000 rpm for 5 minutes,the cell pellet re-suspended in serum-free DMEM containing 10 µM DiIand incubated for 30 minutes at 37°C, before washing with PBS and re-suspending the labelled cells in serum-free DMEM.

Peritoneal free-floating macrophages have been suggested as apotential source of regenerating mesothelial cells (Eskeland andKjærheim, 1966; Ryan et al., 1973). Therefore, peritoneal free-floatingcells were labelled using the red fluorescent dye, PKH26-PCL (SigmaAldrich, Poole, UK), which is specifically taken up by phagocytic cellsand remains within the cells for more than 21 days in vivo (Melnicoff etal., 1989). Rats were injected intraperitoneally (i.p.) with 0.5 µM PKH26-PCL 2 days post injury, killed 2 hours later and the peritoneal cavitylavaged with 20 ml serum-free DMEM to retrieve labelled lavage cells.

Mesothelial healing model7 to 9 week old male Lewis rats (Harlan, Bicester, UK) weighing 160-170 g were used throughout this study (n=3 for each experimentaltreatment). Animals were housed in groups of five and fed on acommercial diet and water ad libitum. A testicular thermal injury model(Fotev et al., 1987; Mutsaers et al., 1997) was used to examine normalserosal healing. Briefly, a metal probe, consisting of a mica-coated brassrod with a 1 cm diameter tip heated to 60°C, was applied to a standardsite on both testicular serosal surfaces for 3 seconds. The tunica vaginalisand scrotal skin were closed using 4-0 silk sutures.

In vivo cell trackingAn equal number of labelled and unlabelled cells, suspended inserum-free DMEM at a concentration of 1×106 cells/ml were used forall in vivo cell tracking studies. The inclusion of unlabelled cellsallowed clear distinction of DiI-labelled cells. Aliquots (1 ml)containing labelled and unlabelled, cultured or lavage-derived cellswere injected i.p. immediately following serosal injury. To assesswhether mesothelial cell implantation was restricted to the wound site,an additional set of uninjured animals was injected i.p. with DiI-labelled cultured mesothelial cells.

At 3, 5 and 8 days post injury, animals were sacrificed and bothtestes excised. In order to assess cell implantation and expression ofthe tight-junction-associated protein, ZO-1, and the mesothelial cellsurface marker HBME-1, the testes were washed with PBS, the serosalsurface dried with compressed air and mesothelial monolayer imprintsobtained on 5% gelatin coated microscope slides (Mutsaers et al.,1997). This technique removed almost all cells from the regeneratingsurface, although occasional cells failed to adhere to the gelatin.Whole testes from animals injected with CM-DiI labelled culturedmesothelial cells were removed 4 days post injury, when there ismaximal mitotic activity on the wound surface (Watters and Buck,

Journal of Cell Science 115 (7)

1385Origin of regenerating mesothelial cells

1973; Mutsaers et al., 2000), and processed for histology and PCNAimmunohistochemistry.

Cell proliferationParaffin-wax-embedded sections (3 µm) of whole testes weremicrowaved in 10 mM citrate buffer, pH 6.4, for 10 minutes to allownuclear antigen retrieval. Endogenous peroxidase activity and non-specific binding sites were blocked by incubating sections in 1%hydrogen peroxide and 1.5% normal rabbit serum, respectively. Sectionswere incubated with a monoclonal antibody directed against PCNA(dilution 1:75; Dako Ltd, Ely, UK) for 2 hours in a humidified chamberat room temperature. Negative controls were treated with isotype-specific mouse IgG2a antibody (PharMingen, San Diego, California,USA). Sections were then incubated with biotinylated rabbit anti-mouseantisera (dilution 1:100; Dako Ltd) for 1 hour followed by streptavidin-HRP (dilution 1:200; Dako Ltd) for 30 minutes with subsequentdetection using the chromogenic substrate 3,3′-diaminobenzidine (DAB;Sigma Aldrich, Poole, UK). Sections were mounted with DPX (BDH,Poole, UK) and consecutive sections were examined by confocal laserscanning microscopy for the presence of CM-DiI labelled cells.

HBME-1 and ZO-1 localisationImmunolocalisation of the mesothelial cell surface marker HBME-1 andthe tight-junction-associated protein ZO-1 on mesothelial imprints wasperformed to demonstrate incorporation of labelled mesothelial cells intothe reconstituted serosal surface. Imprints were fixed with 4% (w/v)paraformaldehyde, pH 7.4, for 5 minutes and permeabilised for 5minutes in PBS, pH 7.0, containing 20 mM HEPES (N-2-hydroxyethylpiperazine-N′-2-ethane sulfonic acid), 300 mM sucrose, 50mM sodium chloride, 3 mM magnesium chloride and 0.5% Triton X-100. Non-specific staining was blocked with 5% newborn calf serumbefore incubating the imprints with monoclonal antibodies directedagainst ZO-1 (dilution 1:25; Zymed Laboratories Inc, San Francisco,California, USA) or HBME-1 (dilution 1:100; Dako Ltd, Ely, UK) for1 hour at room temperature in a humidified chamber. Negative controlswere treated with isotype-specific antisera (Dako Ltd). Imprints werethen incubated with rabbit anti-mouse fluorescein isothiocyanate(FITC)-conjugated antisera (dilution 1:40; Dako Ltd) for 1 hour at roomtemperature in a humidified chamber before being washed, mounted inImmu-mount (Shandon, Runcorn, UK) and examined using confocallaser scanning microscopy.

Microscopy and imagingGelatin imprints and tissue sections of regenerating mesothelium wereexamined by confocal laser scanning microscopy using the Leica TCSNT system. Photomultiplier tube voltage thresholds for confocalmicroscopy were set to gate out background fluorescence produced byisotype-specific negative controls. Fluorescent images were sequentiallycollected through regenerating mesothelial imprints and tissue sectionsfor FITC and TRITC (tetrarhodamine isothiocyanate) fluorochromes at488 and 568 nm emission wavelengths, respectively. Tissue sectionsstained for PCNA were examined using an Olympus BX40 lightmicroscope, and images were captured using KS300 image analysissoftware.

ResultsCell culture and characterisationCultured peritoneal mesothelial cells were bipolar or multipolarin appearance but at confluence they adopted a polygonalconfiguration and became increasingly fibroblast-like withrepeated passage. Cultured fibroblasts were more elongated andformed a whorl-like pattern at confluence. Both mesothelial and

fibroblast cell cultures expressed the mesenchymal intermediatefilament, vimentin, in a diffuse filamentous distribution, whereasonly mesothelial cells expressed the epithelial intermediatefilament, cytokeratin, confirming the isolation of pure cultures ofmesothelial cells (data not shown).

Transfer of DiI between cultured cellsEqual numbers of unlabelled and DiI-labelled mesothelial cellswere cultured for 3 days, and the proportion of DiI-labelled cellswere determined. Cell counts demonstrated no significant changein the proportion of DiI-labelled mesothelial cells between thetime of plating (50%) and 3 days (45.8±2.7%; Student’s two-tailed t-test: p > 0.05).

Implantation of DiI- and PKH26 PCL-labelled cells onto adenuded serosal surfaceDiI-labelled cultured mesothelial cells, which demonstrated redfluorescence localised to both the plasma membrane and vesicle-like structures within the cell cytoplasm (Fig. 1A; inset), wereinjected i.p. into injured rats to determine whether these cellswere capable of implanting onto a denuded serosal surface. 5days post injury, imprints comprised islands of predominantlyrounded cells, but after 8 days the wound surface was completelycovered with cells, which had assumed a more polygonalconfiguration. Imprints of regenerating mesothelium at 5 and 8days post injury demonstrated the presence of DiI-labelled cells(Fig. 1A), which were most numerous at the wound centre andleast in number at the periphery. However, labelled cells wereabsent on imprints following transplantation of DiI-labelledfibroblasts at all time points examined (Fig. 1B). In addition,imprints of uninjured mesothelium, taken from animals injectedwith DiI-labelled cultured mesothelial cells, did not demonstrateany incorporation of DiI-labelled cells (data not shown).

When animals were injected with DiI-labelled peritoneallavage cells, labelled cells were found on imprints of regeneratingmesothelium at both 5 and 8 days post injury (Fig. 1C) anddisplayed a similar distribution to that found using culturedmesothelial cells. However, although PKH26-PCL-labelledperitoneal lavage phagocytes (Fig. 1D inset) were present onimprints 3 and 5 days post injury (Fig. 1D,E), they werecompletely absent by 8 days (Fig. 1F).

Proliferation of implanted DiI-labelled cultured mesothelialcells on the regenerating serosal surfaceFour days post injury, transverse sections of healing serosa,immunostained for the proliferation marker PCNA, demonstratedpositive cells within the seminiferous tubules (Fig. 2A) and theregenerating mesothelium (Fig. 2B). Isotype-specific negativecontrols did not reveal any non-specific staining (data not shown).Adjacent sections revealed multiple layers of CM-DiI-labelledcultured mesothelial cells at the wound site (Fig. 2C), whichcorresponds to the position of PCNA-positive cells.

Incorporation of fluorescence-labelled cells into theregenerating mesotheliumTo confirm the identity of implanted DiI-labelled peritoneallavage cells, imprints were examined for expression of the

1386

mesothelial cell surface marker HBME-1. Normal mesotheliumshowed a plasma membrane distribution of HBME-1 expression(Fig. 3A), and this was not present on isotype-specific-treatednegative controls (data not shown). Dual colocalisation of DiI andHBME-1 demonstrated that implanted DiI-labelled lavage cellsexpressed the mesothelial cell surface marker at both 5 and 8 dayspost injury (Fig. 3B).

The incorporation of DiI-labelled cultured mesothelial andperitoneal lavage cells into regenerating mesothelium wasexamined by immunolocalisation of the tight-junction-associatedprotein ZO-1 on mesothelial imprints. In normal mesothelium,ZO-1 immunoreactivity was detected at the plasma membranewith weak staining observed within the cell cytoplasm (Fig. 4A).In regenerating mesothelium, cells on the wound surface 5 dayspost injury demonstrated ZO-1 staininglocalised towards the plasma membraneand displayed a punctate distribution atsites of cell-to-cell contact (Fig. 4B), whichwas increased at day 8 (Fig. 4C). Confocaloverlay images revealed cell membranelocalised expression of ZO-1 in implantedDiI-labelled cultured mesothelial cells(Fig. 4B) and lavage-derived cells (Fig.4C) at both 5 and 8 days post injury. Weakcytoplasmic expression of ZO-1 was alsoobserved in regenerating mesothelium atall time points examined in both DiI-labelled and unlabelled cells. Negativecontrols treated with specific isotypesdemonstrated negligible FITC staining(data not shown).

DiscussionThis study clearly demonstrates for the firsttime that cultured and lavage-derivedmesothelial cells attach to injured serosalsurfaces, proliferate and incorporate intothe regenerating mesothelium. Thesestudies were performed using the lipophiliccell tracking dye, DiI, to determine the fateof mesothelial cells transplanted into theperitoneal cavity following mesothelialinjury. DiI has been used for long-term celltracking studies in vivo and its presence incell membranes does not affect cell

viability, proliferation or other physiological processes (Honigand Hume, 1986; Kuffler, 1990). Furthermore, it was confirmedin this study that DiI is not transferred between neighbouringcells. Using this technique, DiI-labelled cells were clearlydemonstrated on the injured serosal surface and within thereconstituted mesothelium. PKH26-PCL-labelled peritonealmacrophages also attached to the wound surface but did notincorporate into the regenerated mesothelium; however DiI-labelled fibroblasts failed to attach, demonstrating the selectiveimplantation of certain cell types. Furthermore, labelled culturedmesothelial cells failed to attach to uninjured mesothelium,suggesting that mesothelial cells adhere to exposed extracellularmatrix (ECM) substrates following injury (Tietze et al., 1999;Leavesley et al., 1999).

Journal of Cell Science 115 (7)

Fig. 2.Proliferation of CM-DiI-labelledcultured mesothelial cells in regeneratingmesothelium. Immunolocalisation of PCNA4 days post injury shows proliferating cells inseminiferous tubules (A) and regeneratingmesothelium (boxed area and B). Theadjacent tissue section shows CM-DiI-labelled mesothelial cells in multiple layersat the wound site (C), some corresponding toPCNA-positive cells (arrowed). Bar, 10 µm.

Fig. 1.Implantation of fluorescence-labelled cultured mesothelial cells and peritoneal lavage-derived cells on representative areas of serosal injury. DiI-labelled cultured mesothelial cells,characterised by intense, punctate, red fluorescence in cell suspension (inset A), were present onimprints of regenerating mesothelium 5 days post injury (A), whereas DiI-labelled culturedfibroblasts were absent at the same time point (B). DiI-labelled peritoneal lavage cells were alsopresent 8 days post injury (C). DiI-labelled cells (arrowed) were clearly distinguished fromadjacent unlabelled cells (arrowheads). PKH26-PCL-labelled phagocytic cells were characterisedby intense cytoplasmic red fluorescence in peritoneal lavage fluid obtained 2 days after peritonealabrasion injury (inset D). Labelled macrophages (arrowed) were implanted onto the wound surfaceat 3 and 5 days post injury (D and E, respectively) but were absent by 8 days (F). Bars, 10 µm.

1387Origin of regenerating mesothelial cells

It has previously been shown that there is an increase in thenumber of free-floating mesothelial cells in peritoneal fluidfollowing serosal injury (Whitaker and Papadimitriou, 1985).Peritoneal lavage fluid also contains inflammatory exudate cells,predominantly macrophages, and it has been proposed thatmacrophages can transform into mesothelial cells to reconstitutethe mesothelium (Eskeland and Kjærheim, 1966; Ryan et al.,1973). To assess whether macrophages are a source ofmesothelial cells, free-floating phagocytic cells were labelledwith PKH26-PCL and transplanted into theperitoneal cavity following injury. Labelledcells were present on the wound surface at 3and 5 days but were absent at 8 days,demonstrating that macrophagetransformation into mesothelial cells does notoccur. This supports previous studies in whichperitoneal macrophages labelled with trypanblue (Ellis et al., 1965) or loaded withpolystyrene spheres (Raftery, 1973) were notidentified within healed mesothelium.

Although new mesothelial cells do notoriginate from inflammatory cells, previousstudies have suggested that macrophagessecrete mitogens, which stimulate mesothelialproliferation and initiate healing (Fotev et al.,1987; Rodgers and diZerega, 1992).Mesothelial cells surrounding a serosal lesionproliferate between 24 and 48 hours afterinjury, when collections of macrophages arepresent on the wounded area. Maximal cellproliferation at the centre of the wound occurs4 days post injury following attachment offree-floating mesothelial cells and migrationof cells at the edge of the lesion towards thewound centre (Whitaker and Papadimitriou,1985; Watters and Buck, 1973; Mutsaers et al.,2000). Fotev and coworkers (Fotev et al.,1987) demonstrated mitogenic activity formesothelial cells in wound lavages frominjured serosal tissue and conditioned mediumfrom macrophage cultures. Subsequently,Mutsaers et al. (Mutsaers et al., 1997)identified fibroblast growth factor-2, tumournecrosis factor-α and platelet-derived growthfactor as cytokines with significantmesothelial cell mitogenic potency in vivo.Our studies demonstrated numerous CM-DiI-labelled mesothelial cells and correspondingPCNA-positive nuclei on the healing serosalsurface 4 days post-injury, confirmingproliferation of free-floating mesothelial cellsonce implanted onto the wound. It is likely thatthese cells proliferate in response to mediatorssecreted by macrophages present on thewound surface early in the process ofregeneration.

To show that implanted mesothelial cellsbecome incorporated into the reconstitutedmesothelium, we examined the formation ofapical junctional complexes betweenmesothelial cells in the healing monolayer.

Mesothelial cells form a number of junctional complexesincluding tight junctions (Baradi and Hope, 1964; Kluge andHovig, 1967). ZO-1, a plaque protein associated with apicaljunctions, links the cadherin-catenin complex with the actin-based cytoskeleton (Itoh et al., 1997). In uninjuredmesothelium, ZO-1 expression localised towards the plasmamembrane at sites of cell-to-cell contact. During mesothelialregeneration, ZO-1 expression was predominantlycytoplasmic, owing to the loss of intercellular communicating

Fig. 4.Incorporation of labelled cells intoregenerating mesothelium. ZO-1 expressionin normal mesothelium is localised to theplasma membrane (A). Dual localisation ofDiI and ZO-1 on imprints of regeneratingmesothelium, 5 days following injection ofDiI-labelled cultured mesothelial cells (B,arrowed) and 8 days following injection ofDiI-labelled peritoneal lavage cells (C,arrowed), demonstrating relocation of ZO-1to the plasma membrane at points of cell-to-cell contact. Bar, 10 µm.

Fig. 3.Identification of mesothelial cells by HBME-1 immunostaining. HBME-1 expressionin normal mesothelium (A). Dual localisation of DiI and HBME-1 on gelatin imprints ofregenerating mesothelium, 8 days following injection of DiI-labelled peritoneal lavage cells,demonstrating membranous HBME-1 expression on the surface of implanted DiI-labelledcells (B, arrowed). Bar, 10 µm.

1388

junctions. By 5 days, however, ZO-1 was detected towards sitesof cell-to-cell contact, implying the reformation of apicaljunctional complexes. Mobilisation of ZO-1 from thecytoplasm has previously been demonstrated in Madin-Darbycanine kidney cells in which low calcium levels preventedapical junction formation, whereas switching to normalcalcium levels resulted in a re-distribution of ZO-1 to the cellsurface (Rajasekaran et al., 1996). The extent and intensity ofZO-1 staining at the cell membrane appeared to have increasedby 8 days after injury, which coincided with the re-establishment of an intact mesothelial monolayer. Whether theintensity of staining was due to an upregulation of ZO-1 was notdetermined.

To show that mesothelial cell implantation and incorporationoccurs on other serosal surfaces in different models of injury, werepeated these studies using an abrasion injury to the peritonealwall. All results were consistent with the testicular injury model(data not shown), suggesting that free-floating mesothelial cellsare a source of new mesothelium on all serosal surfaces.

In summary, the following model is proposed. Mesothelialregeneration requires recruitment of inflammatory cells to thewound surface and release of mitogenic cytokines to activate andstimulate mesothelial cell proliferation surrounding the wound(Fotev et al., 1987; Mutsaers et al., 1997). Activated mesothelialcells break their cell-to-cell contacts and migrate onto the woundsurface (Whitaker and Papadimitriou, 1985). Recent evidencesuggests that this may be induced by hepatocyte growth factor,which is secreted by mesothelial cells (Warn et al., 2001) andsurrounding fibroblasts (Yashiro et al., 1996). Additionalmesothelial cells detach and become free-floating (Whitaker andPapadimitriou, 1985), accounting for a 12-fold increase inperitoneal lavage mesothelial cell counts 2.5 days after serosalinjury (Fotev et al., 1987). However the mechanisms by whichthese cells become detached from the basement membrane andremain viable in the serosal fluid is not known. Free-floatingmesothelial cells move down chemotactic gradients, attach toECM components exposed beneath the mesothelium or aredeposited from the serosal fluid, then proliferate and reconstitutean intact mesothelial monolayer.

Our findings complement the studies of Bertram et al.(Bertram et al., 1999) who demonstrated a significant reductionin adhesion formation following the intraperitonealtransplantation of autologous mesothelial cells, suggesting thatthe implanted cells may enhance serosal healing. This may beof therapeutic significance to certain subgroups of patients athigh risk of peritoneal adhesion formation, for example thosereceiving continuous ambulatory peritoneal dialysis as theyhave ready access to peritoneal lavage fluid. In conclusion, wehave shown that cultured and lavage-derived mesothelial cells,but not cultured lung fibroblasts or macrophages, implant ontoareas of serosal injury, proliferate and become incorporated intothe reconstituted mesothelium, conclusively demonstrating thatfree-floating mesothelial cells are an origin of the regeneratingmesothelium.

This work was supported by the Middlesex Hospital SpecialTrustees, the Wellcome Trust (061566) and Johnson and JohnsonMedical/COSAT. We are indebted to Geoffrey Bellingan for hisadvice during the study and Michael Horton for access to the confocallaser scanning microscope facilities. We also appreciate the criticalcomments from Grenham Ireland (University of Manchester,Manchester, UK) in the preparation of this manuscript.

ReferencesAndrade, W., Seabrook, T. J., Johnston, M. G. and Hay, J. B.(1996). The

use of the lipophilic fluorochrome CM-DiI for tracking the migration oflymphocytes. J. Immunol. Methods194, 181-189.

Baradi, A. F. and Hope, J.(1964). Observations on ultrastructure of rabbitmesothelium. Exp. Cell Res. 34, 33-44.

Bertram, P., Tietze, L., Hoopmann, M., Treutner, K.-H., Mittermayer, C.and Schumpelick, V. (1999). Intraperitoneal transplantation of isologousmesothelial cells for prevention of adhesions. Eur. J. Surg. 165, 705-709.

Bolen, J. W., Hammar, S. P. and McNutt, M. A.(1986). Reactive andneoplastic serosal tissue. A light microscopic, ultrastructural andimmunocytochemical study. Am. J. Surg. Pathol. 10, 34-47.

Cameron, G. R., Hassan, S. M. and De, S. N.(1957). Repair of Glisson’scapsule after tangential wounds of the liver. J. Path. Bact. 73, 1-10.

Ellis, H., Harrison, W. and Hugh, T. B. (1965). The healing of peritoneumunder normal and pathological conditions. Br. J. Surg. 52, 471-476.

Eskeland, G. and Kjærheim, Å.(1966). Regeneration of parietal peritoneumin rats. An electron microscopical study. Acta. Pathol. Microbiol. Scand. 68,379-395.

Fotev, Z., Whitaker, D. and Papadimitriou, J. M. (1987). Role ofmacrophages in mesothelial healing. J. Path. 151, 209-219.

Hertzler, A. E. (1919). In The peritoneum vol. 1, (ed. A.E. Hertzler), pp. 264-265. St Louis, CV Mosby Company.

Honig, M. G. and Hume, R. I.(1986). Fluorescent carbocyanine dyes allowliving neurons of identified origin to be studied in long-term cultures. J. CellBiol. 103, 171-187.

Itoh, M., Nagafuchi, A., Moroi, S. and Tsukita, S.(1997). Involvement ofZO-1 in cadherin-based cell adhesion through its direct binding to α cateninand actin filaments. J. Cell Biol. 138, 181-192.

Johnson, F. R. and Whitting, H. W.(1962). Repair of parietal peritoneum.Br. J. Surg. 49, 653-660.

Kluge, T. and Hovig, T. (1967). The ultrastructure of human and ratpericardium. 1. Parietal and visceral mesothelium. Acta. Path. Microbiol.Scand. 71, 529-546.

Kuffler, D. P. (1990). Long-term survival and sprouting in culture bymotoneurons isolated from the spinal cord of adult frogs. J. Comp. Neurol.302, 729-738.

Leavesley, D. I., Stanley, J. M. and Faull, R. J.(1999). Epidermal growthfactor modifies the expression and function of extracellular matrix adhesionreceptors expressed by peritoneal mesothelial cells from patients on CAPD.Nephrol. Dial. Transplant. 14, 1208-1216.

Melnicoff, M. J., Horan, P. K. and Morahan, P. S.(1989). Kinetics ofchanges in peritoneal cell populations following acute inflammation. CellImmunol. 118, 178-191.

Menzies, D. and Ellis, H.(1990). Intestinal obstruction from adhesions: howbig is the problem? Ann. R. Coll. Surg. Engl. 72, 60-63.

Mutsaers, S. E., McAnulty, R. J., Laurent, G. J., Versnel, M. A., Whitaker,D. and Papadimitriou, J. M. (1997). Cytokine regulation of mesothelialcell proliferation in vitro and in vivo. Eur. J. Cell Biol. 72, 24-29.

Mutsaers, S. E., Whitaker, D. and Papadimitriou, J. M.(2000). Mesothelialregeneration is not dependent on subserosal cells. J. Path. 190, 86-92.

Raftery, A. T. (1973). Regeneration of parietal and visceral peritoneum. Alight microscopical study. Br. J. Surg. 60, 293-299.

Rajasekaran, A. K., Hojo, M., Huima, T. and Rodriguez-Boulan, E.(1996).Catenins and zonula occludens-1 form a complex during early stages in theassembly of tight junctions. J. Cell Biol. 132, 451-463.

Rodgers, K. E. and diZerega, G. S.(1992). Modulation of peritoneal re-epithelialization by postsurgical macrophages. J. Surg. Res. 53, 542-548.

Ryan, G. B., Grobéty, J. and Majno, G.(1973). Mesothelial injury andrecovery. Am. J. Path. 71, 93-112.

Stylianou, E., Jenner, L. A., Davies, M., Coles, G. A. and Williams, J. D.(1990). Isolation, culture and characterization of human peritonealmesothelial cells. Kidney Int. 37, 1563-1570.

Tietze, L., Bornträeger, J., Klosterhalfen, B., Amo-Takyi, B., Handt, S.,Günther, K. and Merkelbach-Bruse, S.(1999). Expression and functionof β1 and β3 integrins of human mesothelial cells in vitro. Exp. Mol. Path.66, 131-139.

Wagner, J. C., Johnson, N. F., Brown, D. G. and Wagner, M. F.(1982).Histology and ultrastructure of serially transplanted rat mesotheliomas. Br.J. Cancer46, 294-299.

Wang, N.-S. (1974). The regional difference of pleural mesothelial cells inrabbits. Am. Rev. Respir. Dis. 110, 623-633.

Warn, R., Harvey, P., Warn, A., Foley-Comer, A., Heldin, P., Versnel, M.,Arakaki, N., Daikuhara, Y., Laurent, G. J., Herrick, S. E. and Mutsaers,

Journal of Cell Science 115 (7)

1389Origin of regenerating mesothelial cells

S. E. (2001). HGF/SF induces mesothelial cell migration and proliferationby autocrine and paracrine pathways. Exp. Cell Res. 267, 258-266.

Watters, W. B. and Buck, R. C.(1973). Mitotic activity of peritoneum incontact with a regenerating area of peritoneum. Virchows Arch. (CellPathol.)13, 48-54.

Whitaker, D., Papadimitriou, J. M. and Walters, N.-I. (1982). Themesothelium and its reactions: a review. CRC Crit. Rev. Toxicol. 10, 81-144.

Whitaker, D. and Papadimitriou, J. M. (1985). Mesothelial healing.Morphological and kinetic investigations. J. Path. 145, 159-175.

Yashiro, M., Chung, Y. S., Inoue, T., Nishimura, S., Matsuoka, T.,Fujihara, T. and Sowa, M. (1996). Hepatocyte growth factor (HGF)produced by peritoneal fibroblasts may affect mesothelial cellmorphology and promote peritoneal dissemination. Int. J. Cancer67,289-293.