Mesothelial Cells: A Cellular Surrogate for Tissue Engineering of Corneal Endothelium

Abstract

Purpose.: To evaluate whether mouse adipose tissue mesothelial cells (ATMCs) share morphologic and biochemical characteristics with mouse corneal endothelial cells (CECs) and to evaluate their capacity to adhere to the decellularized basal membrane of human anterior lens capsules (HALCs) as a potential tissue-engineered surrogate for corneal endothelium replacement.

Methods.: Adipose tissue mesothelial cells were isolated from the visceral adipose tissue of adult mice, and their expression of several corneal endothelium markers was determined with quantitative RT-PCR, immunofluorescence, and Western blotting. Adipose tissue mesothelial cells were cultured in a mesothelial retaining phenotype medium (MRPM) and further seeded and cultured on top of the decellularized basal membrane of HALCs. ATMC-HALC composites were evaluated by optical microscopy, immunofluorescence, and transmission electron microscopy.

Results.: Mesothelial retaining phenotype medium–cultured ATMCs express the corneal endothelium markers COL4A2, COL8A2, SLC4A4, CAR2, sodium- and potassium-dependent adenosine triphosphatase (Na⁺/K⁺-ATPase), β-catenin, zona occludens-1, and N-cadherin in a pattern similar to that in mouse CECs. Furthermore, ATMCs displayed strong adhesion capacity onto the basal membrane of HALCs and formed a confluent monolayer within 72 hours of culture in MRPM. Ultrastructural morphologic and marker characteristics displayed by ATMC monolayer on HALCs clearly indicated that ATMCs retained their original phenotype of squamous epithelial-like cells.

Conclusions.: Corneal epithelial cells and ATMCs share morphologic (structural) and marker (functional) similarities. The ATMCs adhered and formed structures mimicking focal adhesion complexes with the HALC basal membrane. Monolayer structure and achieved density of ATMCs support the proposal to use adult human mesothelial cells (MCs) as a possible surrogate for damaged corneal endothelium.

Introduction

Tissue engineering is a complex interdisciplinary field, gathering principles and methods of bioengineering, materials science, and life sciences aimed at building biological surrogates that may be able to substitute for the native tissue functions lost after disease or traumatic processes. ^1–3 Electing a suitable cellular phenotype displaying the functions of the cells to repair, together with an adequate biological matrix or scaffold finely mimicking the biophysical properties of the tissue, reflects two critical determinants for engineering successful tissue biomimetics. 

The cornea is an extraordinary example of fine natural engineering, with its sophisticated disposition of collagen lamellae and cells with a lack of blood vessels in order to shape a totally clear lens. ⁴ In this disposition, the corneal endothelium with its single layer of flat hexagonal cells attached firmly to Descemet's membrane plays an important role in regulating the state of corneal stromal hydration by a sodium- and potassium-dependent adenosine triphosphatase (Na⁺/K⁺-ATPase) endothelial pump and focal tight junctions that allow permeability to nutrients and other molecules from the aqueous humor. Furthermore, corneal endothelial cells (CECs) are metabolically very active, with large numbers of mitochondria to provide the high amount of energy required to pump water efficiently. ⁴  

Cumulative evidence suggests that the postnatal corneal endothelium lacks regenerative capacities and compensates for its gradual loss of cellularity over the life span through hypertrophy of preexisting cells. ^4–6 A critical loss of corneal endothelium cellularity caused by either accidents, surgical trauma, or diseases may no longer ensure proper regulation of stromal hydration, leading to severe corneal swelling, loss of stromal transparency, and severe visual impairment. The only effective treatment so far is corneal transplantation to restore normal vision. Descemet stripping automated endothelial keratoplasty (DSAEK) allows selective substitution for the damaged corneal endothelium that achieves very good results. ^7,8 Unfortunately, development of DSAEK is largely constrained by the limited availability of donor corneas. 

Several studies have reported on tissue engineering of corneal endothelium biosubstitutes as an alternative to corneal transplantation: cultured human CECs transplanted onto chitosan-based membranes, ^9,10 Descemet's membrane, ^11–21 collagen matrix, ²² human corneal stromal discs, ^23,24 gelatin hydrogel discs, ^25,26 acellular porcine corneal matrix, ²⁷ pericellular matrix prepared from human decidua-derived mesenchymal cells, ²⁸ and plastic compressed collagen. ²⁹ Alternatively, the lack of sources of human CECs and their limited proliferation capacity in vitro led several researchers to evaluate the generation in vitro of CEC-like cells from distinct cell types such as embryonic neural crest cells, ¹⁰ corneal stroma stem cells, ³⁰ umbilical cord blood mesenchymal stem cells, ³¹ or human embryonic stem cells. ³² Despite interesting results highlighting the differing potential of these stem cells in regeneration of the corneal endothelium, their clinical application faces problems related to immune rejection, ethical considerations, or limited tissue accessibility. The need to overcome these limitations points toward further efforts to identify novel sources of autologous cells that are phenotypically and functionally as close as possible to native CECs and that should, in addition, be preferentially isolated from a source that is extraocular, abundant, and clinically accessible. 

The mesothelium is the outermost tissue layer lining the parietal surface of coelomic cavities (pleural, pericardial, and peritoneal) and the visceral organs where they are housed. ³³ It was first described by Bichat Xavier in 1827 as a tissue displaying features of simple squamous epithelium. ³⁴ Although mesothelial cells (MCs) originate from the embryonic mesoderm, ^33,35 they display morphologic and biochemical characteristics consistent with simple squamous epithelial cells. ^33,36 Among their main biological functions described so far is secretion of glycosaminoglycans and lubricants to provide a protective and slippery surface for optimal sliding of visceral organs inside coelomic cavities, such as the beating heart or expanding lungs. ^31,36–38 In addition, MCs play a central role in a variety of intraserosal and submesothelial processes, including the transport of water and solutes, inflammation, host response, angiogenesis, tissue repair, and extracellular matrix remodeling. ^31,36–40  

Careful examination and comparison of the morphologic and biochemical hallmarks specific to the mesothelium and corneal endothelium lead to the conclusion that the two tissues share many similarities. They are both composed of a single monolayer of flattened cells tightly compacted together, anchored onto a basement membrane that functions as a scaffold to maintain orderly tissue structures. ^4,37,41 Developmentally, the corneal endothelium is derived from the cranial neural crest. ^42,43 In contrast to vascular endothelial cells, human CECs lack significant expression of the vascular endothelial cell markers von Willebrand factor (factor VIII) and CD31. ^44,45 Phenotypically, human CECs rather resemble human MCs, and as such, both cell types constitutively express cytokeratin 18, ⁴⁶ an intermediate filament that is absent in vascular endothelial cells. ⁴⁷ Furthermore, human CECs also express significant levels of the mesothelial proteins HBME1, mesothelin, and calbindin 2. ⁴⁸ Functionally, the corneal endothelium and mesothelium are both semipermeable membranes involved in electrolyte and water transport, mechanisms that are mainly mediated through their significant Na⁺/K⁺-ATPase pump activity. ^4,49,50  

Taking into account these findings, we hypothesized that the adult mesothelium should represent a valuable source with the capacity to substitute structurally and biochemically for damaged corneal endothelium. Working on this basis, we first performed a comparative analysis of corneal endothelium markers between mouse adipose tissue mesothelial cells (ATMCs) and CECs to evaluate the extent to which they share marker similarities. In a next step, we devised a methodology to achieve a full mesothelialization of the decellularized basal membrane of human anterior lens capsules (HALCs) using mouse ATMCs as a potential biomimetic to substitute for a damaged corneal endothelium. 

Materials and Methods

Isolation of Mouse ATMCs

Adipose tissue mesothelial cells were isolated from nonpregnant CD1 adult female mice (6–20 weeks old) as similarly performed previously. ⁵¹ Guidelines for the animal research protocols were established and approved by the Animal Experimentation and Ethics Committee of CABIMER (permit no. 19-2010). In a first step, uterine cords and adipose tissue were surgically extracted from the peritoneal cavity and submerged within sufficient cold sterile PBS to avoid drying of surface mesothelium. Uterine cords and connected fat pads were then carefully separated with the aid of a scalpel and fine tweezers under magnification. Fat pads isolated from two to four mice were then washed in PBS, transferred to a conical 50-mL tube containing 10 to 20 mL PBS with 2% BSA (Sigma-Aldrich Corp., St. Louis, MO, USA) and 0.25% trypsin (Gibco, Grand Island, NY, USA), and finally transferred to a water bath at 37°C for 20 minutes. The tube was sealed and left to float during the entire enzymatic digestion period. The release of ATMCs from the outermost mesothelial layer of uterine fat pads was improved by three or four gentle shakes of the tube during the enzymatic digestion procedure. The tube was transferred to a flow hood and placed upright. A 10-mL culture pipette was then introduced into the tube. After complete floating of fat pads, the underlying trypsin solution was gently recovered by pipetting and transferred to 15-mL conical tubes. Fat pads were gently washed with an additional volume of 10 mL PBS. Trypsin solution containing released ATMCs was then centrifuged (300g, 10 minutes) and ATMCs pellet finally resuspended in 1 mL PBS. Numbers of viable ATMCs were then counted after staining with trypan blue (Gibco). 

Culture of ATMCs

Adipose tissue mesothelial cells were seeded (35,000 cells/cm²) in T-25 flasks (Nunc, Roskilde, Denmark) into 5 mL mesothelial retaining phenotype medium (MRPM), which consisted of Dulbecco's modified Eagle's medium (DMEM) low-glucose GlutaMax medium (Gibco) supplemented with 2% heat-inactivated fetal bovine serum (FBS) (Lonza, Basel, Switzerland), 1% B27 supplements (Gibco), 1% penicillin-streptomycin (Gibco), 100 μM β-mercaptoethanol (Gibco), and 1 μg/mL hydrocortisone (Sigma-Aldrich Corp). The ATMCs were cultured for 2 days in MRPM (5% CO₂, 21% O₂, 37°C) and harvested with 0.05% trypsin (Gibco). 

Murine CEC Culture

Primary murine CECs were isolated from corneas of two to three adult CD1 mice (2–6 months old). The corneal endothelium layer was gently stripped off with the aid of a scalpel. Corneal endothelial cell clusters released through stripping were then subjected to primary explant culture during 7 days in an incubator (5% CO₂, 21% O₂, 37°C) in medium consisting of DMEM low-glucose GlutaMax medium (Gibco) containing 5% heat-inactivated FBS (Lonza), 1% penicillin-streptomycin (Gibco), 1% nonessential amino acids (Gibco), 100 μM β-mercaptoethanol (Gibco), 1× insulin transferrin selenium (ITS; Gibco), 10 ng/mL recombinant murine basic fibroblast growth factor (bFGF; PeproTech, Rocky Hill, NJ, USA), 10 ng/mL recombinant murine epidermal growth factor (EGF; PeproTech), and 1 μg/mL hydrocortisone (Sigma-Aldrich Corp.). Primary expanded CECs were further subcultured three times at 5000 cells/cm² during 4 to 5 days in T-75 flasks (Nunc). Corneal endothelial cells in subculture passage 3 were used as control for corneal endothelium marker expression as determined by quantitative PCR (qPCR), immunofluorescence, and Western blot. 

Immunofluorescence

For immunofluorescence, ATMCs and CECs were cultured in μ-Dish (Ibidi GmbH, Martinsried, Germany). For cell-surface antigen detection, cells were fixed with 4% paraformaldehyde (PFA; Sigma-Aldrich Corp.) and blocked in PBS containing 3% BSA (PBS-BSA). For intracellular antigen detection, cells were permeabilized with 0.5% Triton X-100 (Sigma-Aldrich Corp.) or cold methanol (−20°C) and then blocked in PBS-BSA. Antibodies used in this study are described in Supplementary Table S1. Nuclei were counterstained with 1 μg/mL Hoechst 33342 (Sigma-Aldrich Corp.). Fluorescence images were captured with the Olympus IX71 inverted fluorescence microscope (Olympus, Tokyo, Japan). 

Quantitative Reverse Transcription–Polymerase Chain Reaction

Total RNA content was extracted with RNeasy Mini Kit (Qiagen, Hilden, Germany) and reverse transcribed into cDNA by using MMLV reverse transcriptase (Promega, Madison, WI, USA). Quantitative real-time PCR was performed using SYBR Green and detected using an ABI Prism 7500 system (Applied Biosystems, Foster City, CA, USA). Gene expression was normalized to HYWAZ mRNA (TATAA Biocenter AB, Göteborg, Sweden). Stripped corneal endothelium served as the calibrator sample. Primer sequences are in listed in Supplementary Table S2. 

Western Blot

Protein lysates isolated from MRPM-cultured ATMCs, subcultured CECs, and CD1 adult heart (whole) were subjected to electrophoresis separation (40 μg/lane) in 8% SDS-PAGE and transferred to a Hybond-P polyvinylidene difluoride (PVDF) membrane (Amersham, Buckinghamshire, UK). Membranes were blocked in Tris-buffered saline with 2% BSA and 0.2% Tween 20. Blots were incubated overnight with primary antibodies, and immunoreactive bands were detected with horseradish peroxidase–conjugated secondary antibodies (Supplementary Table S1) followed by ECL Plus detection system (Amersham). 

Seeding of HALCs With ATMCs

A total of 34 HALCs of 5-mm diameter on average were obtained after patient informed consent during normal cataract surgery procedures. Human anterior lens capsules were directly stored in distilled water to accomplish a total decellularized basal membrane. Of note, HALCs were invariantly rolled or folded due to their natural convex structure, their inner side corresponding to the decellularized epithelial side. Human anterior lens capsules were further distributed into 35-mm cell culture dishes vented with four inner rings (Greiner CELLSTAR, Carrollton, TX, USA) in 100 μL distilled water. The HALCs were correctly oriented with their outer side facing the plastic surface. Their complete flattening was achieved by gradually eliminating the water with the use of a pipette and needles. 

An initial volume of 60 μL MRPM containing 10⁵ ATMCs was dropped onto the epithelial side of decellularized HALCs. Culture dishes containing the seeded HALCs were then placed in a 140-mm Petri dish containing pieces of tissue paper soaked in water to maintain optimal humidification and were finally transferred to an incubator (5% CO₂, 21% O_2, 37°C). Approximately 80% to 90% of the ATMCs were already firmly adhered after 2 hours of culture. An additional volume of 120 μL MRPM was then carefully added to the rings. After 6 hours, a final volume of 1.2 mL MRPM was carefully added outside the rings. After 24 and 48 hours, MRPM (1.5 mL) was exchanged. After 72 hours, HALCs typically displayed complete cellular coverage and are henceforth termed ATMC-HALC composites. 

Transmission Electronic Microscopy

The ATMC-HALC composites (n = 8) were fixed with 1.6% glutaraldehyde in 0.15 M sodium cacodylate buffer (pH 7.3) for 1 hour at 4°C. Human anterior lens capsules were then postfixed in 1% OsO₄ in the same buffer, dehydrated in ethanol, and embedded in epoxy resin. Thin sections were cut and stained with uranyl acetate and lead citrate to be examined with a CM-10 transmission electron microscope (Philips/FEI Corporation, Eindhoven, The Netherlands) operating at 80 kV. The electron microscopy data were collected after identification of transversal cuts of ATMC-HALC composites at primary magnification ×5000, and pictures were taken at enlargements of ×20,000. 

Immunohistochemistry

Mouse corneas surgically isolated from CD1 adult mice were fixed with cold 4% PFA, then permeabilized and blocked into PBS containing 0.5% Triton X-100 and 3% BSA (PBS-TX-BSA) and finally embedded in Tissue-Tek optimum cutting temperature (OCT) cryostat sectioning compound (Sakura Finetechnical Co., Tokyo, Japan). For Na⁺/K⁺-ATPase and β-catenin detection, corneas were fixed and permeabilized in cold methanol. Sections (15 μm) were mounted onto poly-L-lysine–coated glass slides and incubated with primary and secondary antibodies (Supplementary Table S1). 

MetaMorph-Based Quantification of Cellular Density

Cellular density of ATMC-HALC composites was quantified with Meta Imaging Software MetaMorph Offline version 7.5.1.0 (MDS Analytical Technologies, Sunnyvale, CA, USA). Square areas of 40,000 μm² (n = 12) were extracted from phase-contrast pictures of ATMCs lining HALCs, and numbers of cells in each square were calculated with the “manually count objects” function. For cell density quantification of the adult mouse corneal endothelium, intact corneas were fixed with PFA and stained with Hoechst 33342 before being mounted in a sandwich between thin glass coverslips. Fluorescence pictures of the corneal endothelium layer were taken. Numbers of cells in square areas of 40,000 μm² were determined by calculating numbers of nuclei. Mean numbers of cells in squares were used to deduce the mean ± SD number of cells/mm². 

Statistical Analysis

Values are presented as mean ± SD. Statistical significance was calculated using an unpaired Student's t-test; P < 0.05 was considered significantly different. 

Results

Characterization of ATMCs Cultured in MRPM

A mild trypsinization procedure of uterine fat pads led to an efficient and selective detachment of their outermost MC layer (Fig. 1A). (Trypsin-isolated adipose tissue MCs are termed ATMCs.) Adipose tissue mesothelial cells cultured for 48 hours in MRPM typically generated a cobblestone-like monolayer of flattened polygonal epithelioid cells (Fig. 1B). Consistent with their mesothelial phenotype, ATMCs displayed strong intercellular expression of β-catenin and zona occludens-1 (ZO-1) and also, to a lesser extent, displayed intercellular expression of E-cadherin. Furthermore, they displayed wide nuclear expression of Wilm's tumor protein (WT1) and moderate membrane expression of mesothelin. The MRPM-cultured ATMCs lacked detectable expression of the pan endothelial cell marker CD31. Furthermore, and consistent with their retention of the original mesothelial phenotype, the MRPM-cultured ATMCs were devoid of stress fibers positive for F-actin and alpha-smooth muscle actin (α-SMA), which are early hallmarks in ATMCs undergoing epithelial-to-mesenchymal transition (EMT). ⁵¹ Finally, a subset (20%–25%) of the cultured ATMCs was proliferating as indicated by their nuclei positive for the proliferative marker Ki-67. 

Figure 1

 

Efficient recovery of adipose tissue mesothelial cells (ATMCs) by gentle trypsinization of the adult mouse visceral adipose tissue. (A) Left shows representative aspect of uterine cords and fat pads (AT) after surgical separation. Middle shows isolated fat pads from several mice. Right shows representative aspect of ATMCs released through trypsinization of adipose fat pads. Adipose tissue mesothelial cells were mainly detached as small sheets that rapidly adopted a “grape-like” aspect after full cellular retraction. (B) Immunofluorescence characterization of ATMCs cultured for 48 hours in MRPM (mesothelial retaining phenotype medium). Upper left image shows the typical cobblestone-like morphology adopted by ATMCs cultured for 48 hours in MRPM. Adipose tissue mesothelial cells displayed wide intercellular expression (arrowhead) of β-catenin and ZO-1. E-cadherin expression was more heterogeneous at intercellular contact. Overall, the majority of ATMCs displayed strong nuclear expression of WT1 and moderate surface expression of mesothelin. Furthermore, ATMCs displayed ring-like F-actin staining (arrowhead) and diffuse cytoplasmic expression of α-SMA (alpha-smooth muscle actin) and did not express the vascular endothelial cell marker CD31. A significant number of ATMCs cultured in MRPM displayed nuclear expression of the proliferative marker Ki-67 (arrowheads). Nuclei are counterstained in blue with Hoechst 33342. Scale bars: 50 μm.

Isolation and Characterization of Mouse CECs

The stripping procedure typically resulted in the detachment of large clusters of the corneal endothelium layer from Descemet's membrane (Figs. 2A, 2B). Immunofluorescence analysis of fragments of stripped Descemet's membrane indicated that they were mainly devoid of their corneal endothelium cell layer, thus confirming that large cell clusters corresponded to stripped CECs (Supplementary Fig. S1). Large clusters of CECs were picked up and subjected to explant culture. Corneal endothelial cell clusters could successfully adhere, spread, and proliferate to form a compact cobblestone-type monolayer within 5 to 7 days of culture (Fig. 2C). The establishment of highly homogeneous cultures of polygonal CECs could be achieved through three sequential subculture steps (Figs. 2D–F). 

Figure 2

 

Isolation and establishment of CEC cultures. (A, B) Corneal endothelium stripping procedure typically resulted in the detachment of large fragments of the Descemet membrane (A) and of the corneal endothelium layer (B). (C) Explant culture of large CEC clusters typically generated a compacted cobblestone-type monolayer of polygonal cells (see enlarged spot). (D) Morphology of CECs at the end of their first subculture, with observable large clusters of tightly packaged CECs. (E, F) Representative morphologies of CECs after their second and third subculture steps, respectively. Although CECs increased moderately in size through sequential subculture steps, it is of note that CECs retained their original polygonal morphologies. Scale bars: 200 μm.

ATMCs Share Phenotypic Marker Similarities With CECs

Adipose tissue mesothelial cells cultured for 2 days in MRPM and the subcultured CECs were analyzed by qPCR for their expression of several corneal endothelium markers (Fig. 3). Accordingly, the collagen alpha-2 IV and VIII chains (COL4A2 and COL8A2, respectively), solute carrier family 4 (anion exchanger) member 4 (SLC4A4), carbonic anhydrase II (CAR2), Na⁺/K⁺-ATPase subunit alpha-1 (ATP1A1), and N-cadherin (CDH2) genes were significantly overexpressed in samples of stripped corneal endothelium compared to RNA samples extracted from the whole cornea, thus confirming the high purity of stripped corneal endothelium samples. 

Figure 3

 

Quantitative RT-PCR analysis of corneal endothelium gene expression in ATMCs and CECs. Bar graph shows comparative corneal endothelium gene (COL4A2, COL8A2, SLC4A4, CAR2, Na⁺/K⁺-ATPase, and N-cadherin) expression of intact corneas (cornea), subcultured CECs, freshly isolated ATMCs, and ATMCs cultured for 2 days in MRPM (cultured ATMCs) relative to expression values obtained in the stripped mouse corneal endothelium (calibrator sample, shown as black bars). Values for freshly isolated and cultured ATMCs and CECs were obtained from three independent isolation procedures and cultures. Expression values of the different corneal endothelium genes were normalized to the housekeeping gene YWHAZ. Results are shown as mean fold change ± SD in mRNA expression relative to the calibrator sample (set as 1) calculated from three independent isolations and cultures of CECs and ATMCs. Statistical differences (*P < 0.05 and **P < 0.03) were determined with a Student's t-test.

Interestingly, COL4A2 and N-cadherin were found to be overexpressed in the subcultured CECs compared to the stripped corneal endothelium (freshly isolated CECs), thus evidencing phenotypic changes occurring in CECs through subculture. Supporting their partial dedifferentiation upon subculture, the subcultured CECs displayed downregulation of both COL8A2 and CAR2 genes by comparison to values found in those freshly isolated. 

Our results indicated that COL4A2, SLC4A4, CAR2, and NA⁺/K⁺-ATPase genes were expressed at higher levels in freshly isolated ATMCs than in the stripped corneal endothelium, thus suggesting that mesothelium tissue shares phenotypic and functional similarities with the corneal endothelium. In contrast, COL8A2, a highly specific corneal endothelium marker, ⁵² was expressed to a lower extent in freshly isolated ATMCs (P < 0.03). Similar to values found in the subcultured CECs, the MRPM-cultured ATMCs also displayed strong upregulation of the COL4A2 and N-cadherin genes (P < 0.03), suggesting phenotypic changes in ATMCs in response to culture components. The finding that the corneal endothelium genes COL8A2, SLC4A4, CAR2, and NA⁺/K⁺-ATPase were expressed at almost similar levels between the freshly isolated and MRPM-cultured ATMCs may, however, suggest that the culture conditions described result in phenotypic similarities. 

Comparison of the MRPM-cultured ATMCs and subcultured CECs indicated that they displayed a similar expression pattern for COL4A2, COL8A2, SLC4A4, and N-cadherin genes; NA⁺/K⁺-ATPase, however, was more highly expressed in the MRPM-cultured ATMCs than in the subcultured CECs (P < 0.03). 

The expression pattern of corneal endothelium markers in the MRPM-cultured ATMCs and subcultured CECs was explored by immunofluorescence (Fig. 4). Positive immunoexpression for COL8A2, N-cadherin, ZO-1, β-catenin, NA⁺/K⁺-ATPase, and SLC4A4 could be detected in the corneal endothelium layer of mouse cornea sections and widely in the subcultured CECs (Fig. 4). Of major relevance, the ATMCs cultured in MRPM immunoexpressed the different corneal endothelium markers to a nearly similar extent. 

Figure 4

 

Immunofluorescence analysis of corneal endothelium marker expression patterns in cultured CECs and ATMCs. Left shows representative expression patterns for F-actin, COL8A2, N-cadherin, ZO-1, β-catenin, Na⁺/K⁺-ATPase, and SLC4A4 in mouse cornea section. Middle and right show immunoexpression patterns obtained in subcultured CECs and MRPM-cultured ATMCs, respectively. Nuclei are counterstained in blue with Hoechst 33342. Scale bars: 50 μm.

We next used Western blot to quantitatively compare the expression profile of corneal endothelium markers in the MRPM-cultured ATMCs and the subcultured CECs (Fig. 5). Our results indicated that ATMCs and CECs displayed quite similar expression patterns for mesothelin, NA⁺/K⁺-ATPase, and β-catenin. In contrast, ZO-1, COL8A2, and SLC4A4 were expressed at higher levels in CECs. N-cadherin, a marker strongly expressed in the mouse heart, was in contrast minimally expressed in ATMCs and CECs. Overall, Western blot analysis could confirm the expression of structural and functional corneal endothelium markers in ATMCs, thus confirming phenotypic similarities between ATMCs and CECs. 

Figure 5

 

Western blotting analysis of corneal endothelium markers in cultured ATMCs and CECs. Total protein extracts were isolated from ATMCs cultured for 2 days in MRPM (ATMCs), from mouse CECs subjected to three subculture steps in CEC growth medium (CECs), and from the mouse FVB adult heart (Heart). A total of 40 μg proteins from ATMCs, CECs, and heart was subjected to 8% SDS-PAGE electrophoresis. Figure shows comparative blot expression of N-cadherin, mesothelin, ZO-1, β-catenin, COL8A2, Na⁺/K⁺-ATPase, and SLC4A4 in ATMCs, CECs, and heart. Adipose tissue mesothelial cells and CECs were found to display similar expression of mesothelin, β-catenin, and Na⁺/K⁺-ATPase. In contrast, ZO-1, COL8A2, and SLC4A4 were expressed to a higher extent in CECs than in ATMCs. Strong expression of N-cadherin and Na⁺/K⁺-ATPase was found in the heart. Molecular weight (kDa) of the different proteins studied is shown in the right part of the figure. The housekeeping protein glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was used as loading control.

Characterization of ATMC-HALC Composites

Human anterior lens capsules obtained by capsulorhexis were transparent sheets of approximately 5-mm diameter and 20-μm thickness. Transmission electronic microscope analysis of decellularized HALCs could confirm that their decellularization was fully accomplished in sterile water (Supplementary Fig. S2). 

Adhesion of ATMCs on top of HALCs was mostly achieved within only 2 hours of culture (data not shown). After 24 hours of culture, ATMCs were firmly adhered to the top of HALCs, and subsets were actively proliferating as indicated by the presence of small round and refringent cells (Fig. 6A). Adipose tissue mesothelial cells gradually covered the entire surface of HALCs between 24 and 72 hours of culture in MRPM (Fig. 6B). Remarkably, confluent ATMCs lacked α-SMA- and F-actin-positive stress fibers, indicating that they did not undergo significant EMT (Fig. 6C). Cell density quantification of the ATMC-HALC composites cultured for 72 hours indicated that the monolayer established by ATMCs reached a mean cell density of 1378 ± 187 cells/mm² (Fig. 6D). By comparison, the native adult mouse corneal endothelium displayed a mean cell density of 3188 ± 277 cells/mm² (Fig. 6D, Supplementary Fig. S3). 

Figure 6

 

Morphologic analysis of ATMC-seeded human anterior lens capsules. (A) Left shows a 10× phase-contrast picture of one ATMC-seeded HALC after 24 hours of culture in MRPM. White arrows point to the border of the HALC. Plastic surface with adhered ATMCs (left upper area) is observable as out of focus. Right upper images show enlargement of blue and red spots drawn in the left image. Proliferation at this step was evidenced by the presence of many duplets of small round and refringent proliferating cells (arrowhead). Other cells were in turn more flattened (arrow) and represent mature ATMCs. Scale bar: 200 μm. (B) Representative aspect of ATMC-seeded HALC surface after 72 hours of culture. At this step the HALC surface was fully recovered by ATMCs that established tight contact (see arrow in enlarged black square). Scale bar: 200 μm. (C) Double immunofluorescence labeling of F-actin and α-SMA in ATMC-HALC composites showing ring-like staining pattern of F-actin and absence of α-SMA-positive stress fibers in ATMCs. Scale bar: 100 μm. (D) MetaMorph-based quantification of cellular density in the adult mouse corneal endothelium and ATMC-HALC composites. Data are expressed as the mean ± standard deviation of cell counts performed from distinct areas (n = 12). Statistical significance was determined by using Student's t-test.

Ultrastructural analysis of ATMC-HALC composites with transmission electron microscopy indicated the presence of numerous microvilli on the apical membrane of ATMCs (Fig. 7A), structures that are also particularly abundant in CECs. ²⁹ The ATMCs also displayed abundant mitochondria (Fig. 7A); rough endoplasmic reticulum (Fig. 7B); two organelles that are also particularly abundant in CECs ⁴ ; and electron-dense junctional complexes at apicolateral cell–cell contact, a type of structure consistent with desmosomes (Fig. 7B). Interestingly, the basal membrane of ATMCs was found to be in close contact with the HALC surface and displayed numerous invaginations (Fig. 7C). Other segments of the ATMC basal membrane, in contrast, were rather exvaginated, thicker, and more electron dense; this is a type of structure consistent with the establishment of focal adhesion complexes with the HALC surface. 

Figure 7

 

Transmission electron microscope analysis of ATMC-HALC composites. (A) Left image shows transversal section through an ATMC-HALC complex. A total of four ATMCs (black arrowheads) can be observed adhered on top of the HALC basal membrane side. Scale bar: 20 μm. Right image shows magnification of black spot. Note how ATMCs, like CECs, display typical apical membrane protrusions or microvilli on their apical surface (black arrows) and numerous mitochondria (red stars). (B) Left image is a higher magnification allowing the visualization of tight contact established between ATMCs. Adipose tissue mesothelial cells typically displayed a large rough endoplasmic reticulum (yellow spot). Scale bar: 1 μm. Right image shows an enlargement of black spot drawn in the left image and allows visualization of electron-dense tight junction complexes at apicolateral intercellular contact between ATMCs (black arrowheads). (C) Left image shows representative aspect of ATMC basal membrane in tight contact with HALC surface. Scale bar: 1 μm. Right image shows an enlargement of black spot drawn in the left image and allows visualization of invaginated segments of ATMC basal membrane (red arrows) and of exvaginated membrane segments that are strongly electron dense (black arrowheads) and feature hemidesmosomes.

Discussion

Stem cell–based tissue engineering of corneal endothelium equivalents holds great promise for treatment of corneal endothelial disorders. In recent years, significant approaches have been used to identify suitable scaffolds and source of stem cells to bioengineer functional corneal endothelium equivalents. In most cases, the scaffolds proposed were used in combination with primary human CECs. ^{14,15,17,18,20–27,53} Alternatively, various studies have reported the generation of CEC-like cells with distinct efficiency from corneal stromal cells ³⁰ or extraocular stem cells such as neural crest cells, ¹⁰ umbilical cord blood stem cells, ³¹ and human embryonic stem cells. ³²  

The use of heterologous human CECs to bioengineer functional corneal endothelium equivalents represents so far the most promising approach. ⁷ This technology is, however, strongly limited by the important shortage of suitable donor corneas and the difficulty of achieving their expansion in vitro in sufficient numbers without loss of original phenotypic properties due to their induction of EMT by growth medium components. ⁵⁴ Finally, and no less importantly, corneal endothelium equivalents built with heterologous cells are susceptible to immunologic rejection, principally in patients with highly vascularized corneal beds. ⁵⁵ There is therefore a need to find a new source of autologous CEC-like cells. 

In this study we show that ATMCs are phenotypically quite similar to CECs and could therefore represent a novel cellular surrogate to bioengineer autologous nonimmunogenic corneal endothelium equivalents. In additional support of their therapeutic use, previous studies have already reported the successful cultivation of human ATMCs isolated from clinical samples of the greater omentum, ^44,45,56,57 the largest visceral adipose tissue depot in humans. Of particular interest, it was indicated that each square centimeter of human omental tissue could provide approximately 1 million ATMCs. ⁵⁷ On this basis, and given that the adult human corneal endothelium is roughly composed of approximately 300,000 cells, ⁵⁸ surgical isolation through laparoscopic surgery of reduced pieces of omental tissue should provide sufficient numbers of ATMCs to bioengineer several corneal endothelium equivalents. Alternatively, reports that human peritoneal mesothelial cells (HPMCs) can be isolated from peritoneal fluids collected by needle aspiration ^42,45 suggest that minimally invasive approaches can also be a way to obtain MCs for tissue engineering of corneal endothelium equivalents. 

Our results indicate that the adipose tissue mesothelium and corneal endothelium share morphologic and phenotypic similarities. Of particular relevance, the high expression of functional corneal endothelium markers such as SLC4A4, CAR2, and Na⁺/K⁺-ATPase in freshly isolated ATMCs suggests that the mesothelium is functionally close to the corneal endothelium. In further support of this concept, two previous reports have indicated that HPMCs display significant Na⁺/K⁺-ATPase pump activity, ^49,50 which is required for the transport of fluids across the mesothelium membrane. 

A critical parameter in tissue engineering of corneal endothelium equivalents is the maintenance of the density of replacement cells. Herein, we demonstrate that murine ATMCs could adhere, proliferate, and generate a confluent and compact monolayer of polygonal cells firmly attached on top of the HALC basal membrane. Interestingly, previous studies have indicated that HALCs represent an excellent substrate for the adhesion, growth, and viability of primary human CECs, ^59,60 limbal epithelial stem cells, and human trabecular meshwork (HTM) cells. ^61,62 The use of HALC as a substitute for Descemet's membrane could be additionally supported on the basis that the two structures are quite similar; they are both smooth and transparent basal membranes permeable to water, electrolytes, and nutrients from the aqueous humor. In addition, the HALC and Descemet's membrane display quite similar thickness and elasticity. ^63,64 Furthermore, they are both principally composed through deposition of collagen fibers; the lens capsule is mainly composed of collagen type III and to a lesser extent collagen type I ⁶⁵ , whereas Descemet's membrane is mainly composed of collagen types IV and VIII. ^66–68 However, comparative transplantation experiments on HLAC-based and Descemet's membrane–based corneal endothelium substitutes, to evaluate the performance of HALCs as a substrate surrogate for Descemet's membrane, remain to be performed. 

In summary, we have shown that primary murine ATMCs and CECs display similar morphology, which is consistent in simple squamous epithelial cells. Both share structural and functional markers, thus suggesting that the adipose tissue mesothelium may represent a source of CEC-like cells to bioengineer nonimmunogenic corneal endothelium equivalents. Adipose tissue mesothelial cells could adhere and form adhesion structures with the HALC basal membrane. Monolayer structure and density of the ATMC-HALC composites were better than expected. Further experiments to assess the functionality of human ATMCs as a substitute for human CECs are in progress. For the future, we propose a similar technique, termed Descemet membrane endothelial keratoplasty, that provides fast and high visual rehabilitation. ⁶⁹  

Supplementary Materials

pdf S1 

pdf S2 

pdf S3 

Acknowledgments

Supported by Fondos FEDER, Fundación Progreso y Salud, Consejería de Salud, Junta de Andalucía (Grant PI-0022/2008), INNPACTO Program (INP-2011-1615-900000), and SUDOE Program-BIOREG (Intereg SOE3/P1/E750); Consejería de Innovación Ciencia y Empresa, Junta de Andalucía (Grant CTS-6505); Ministry of Science and Innovation (Red TerCel-FEDER Grant RD12/0019/0028); Instituto de Salud Carlos III Grant PI10/00964); the Ministry of Health and Consumer Affairs Advanced Therapies Program Grant TRA-120 (BS); and Corporación Tecnológica de Andalicía CTA (NBT). CIBERDEM is an initiative of the Instituto de Salud Carlos III. 

Disclosure: C.C. Lachaud, None; F. Soria, None; N. Escacena, None; E. Quesada-Hernández, None; A. Hmadcha, None; J. Alió, None; B. Soria, None 

Footnotes

 AH, JA, and BS are joint senior authors.