Fig. 16.1
Transverse diagram of the human cornea and limbus. The diagram illustrates the anatomic relationships of the tissues and regions discussed in McGowan’s study (outlined in red). The posterior limbus is the region herein identified between the solid vertical lines and consists of the transition zone, the trabecular meshwork, and Schlemm’s canal. The corneal endothelium is not part of the limbus (Reprinted from McGowan et al. [15]. Adapted from Histology of the Human Eye, Hogan, Alvarado, Weddell, The Limbus, p.113, 1971, with permission from Elsevier)
Following injury to the corneal endothelium, the authors showed activation of additional stem cell markers, suggesting that these stem cells could initiate an endothelial wound repair process. These progenitor/stem cells, however, were never isolated for culture purposes, the major difficulty being that the only successful method presently used to isolate corneal endothelial cells is by peeling Descemet’s membrane (see Sect. 16.2.3). Stem cells peripheral to Descemet’s membrane are thus not isolated.
The presence of slow-cycling cells (a characteristic of stem cells) and of cells expressing progenitor markers (such as nestin, leucine-rich repeat-containing G-protein-coupled receptor 5, Sox9, and nerve growth factor receptor p75) was identified in the mature endothelium of murine eyes [16]. These cells were exclusively located at the extreme periphery of the endothelium, close to the transitional zone.
An efficient method to culture stem cells in vitro is by using the sphere-forming assay. This method is based on the principle that stem cells in suspension will continue to proliferate in vitro, forming spheres, contrary to differentiated cells that need adhesion to survive. Sphere-forming assays have been used to isolate human corneal endothelial progenitor cells [17–19]. A higher number of spheres were formed using cells from the periphery [18]. Some of the peripheral cells isolated using a sphere-forming assay expressed the stem cell marker nestin [17]. The number of cells expressing the stem cell-related markers GFAP, OCT3/OCT4, and nestin has been reported to be influenced by the composition of the cell culture media [17]. Progenitor cells have also been reported to be present at the center of Descemet’s membrane. Using an approach different than the sphere-forming assay, Hara et al. showed that cells isolated from the periphery and from the center had the same colony-forming efficiency, suggesting that progenitor cells are present throughout the human corneal endothelium. This result is consistent with the presence of p75NTR-expressing cells both at the center and the periphery [20].
16.2.3 Cell Isolation and Growth of Corneal Endothelial Cells
A common strategy for the isolation of corneal endothelial cells consists of peeling off the Descemet’s membrane (with the endothelial cells attached) and then to detach the cells using ethylenediaminetetraacetic acid (EDTA) [21] or enzymes such as collagenase [22, 23], trypsin [19], or dispase [24]. Alternatively, cells can grow out of Descemet’s membrane using an explant method [25]. The comparison of different isolation methods showed that the explant and the collagenase (0.1 or 0.2 % collagenase II) methods were more successful in generating cultures [12]. Success rates were in the order of 30 % using the explant method, 32 % using 0.1 % collagenase, and 37 % using 0.2 % collagenase, compared to 8 % with the other methods (trypsin/EDTA or 0.1 % dispase II). These observations were consistent with another study in which collagenase A was better than trypsin for the isolation of bovine corneal endothelial cells [26].
Once the P0 cells have reached confluence, cell dissociation reagents (such as trypsin/EDTA) are used to separate the cells from the dish and from each other, which allows to seed them in new cell culture dishes, usually at a 1:3 ratio. The disruption of intercellular junctions that unavoidably happens when using trypsin/EDTA may activate canonical Wnt signaling and promote endothelial-to-mesenchymal transition (EndMT) [27]. Success in avoiding EndMT of cultured corneal endothelial cells could reside in the development of culture conditions in which confluence and gain of apical–basal polarity is rapidly achieved. For example, increasing cell seeding density has been shown to diminish the occurrence of fibroblastic transformation within human corneal endothelial cell culture [28]. The maintenance of a functional endothelial phenotype throughout cell expansion by finding ways to prevent EndMT is currently a research area of high interest [29]. Conditioned media obtained by NIH-3T3 cells or human bone marrow-derived mesenchymal stem cells (MSC) have been shown to help maintain a corneal endothelial phenotype during in vitro expansion [30]. The media conditioned by MSC was superior to the 3T3-conditioned media in terms of increasing proliferation and motility. The authors showed that the effect on proliferation was maximal using 10 % MSC-conditioned media [30]. SB431542, a selective inhibitor of the transforming growth factor-beta (TGF-ß) receptor, has also been shown to counteract the appearance of a fibroblastic phenotype [31]. Cells cultured in the presence of SB431542 generated a contact-inhibited monolayer of polygonal corneal endothelial cells that maintained the expression of function-related proteins [31].
Recently, a novel dual media culture approach has been proposed for the in vitro expansion of corneal endothelial cells. This method using two different culture media, one for proliferation and one for maintenance once cells reach 80–90 % confluence, showed consistent third-passage corneal endothelial cells with homogenous polygonal morphology. Cells expressed high levels of the sodium–potassium pump Na+/K+-ATPase, ZO-1, glypican-4, and CD200 [32]. The authors also combined this dual media approach with the use of 10 μM selective Rho-associated kinase (ROCK) inhibitor, Y-27632 [33]. They showed increased cell attachment of P0 cells when using Y-27632. Also, in the presence of Y-27632, P0 cells retained a smaller cell size while maintaining their polygonal morphology. Finally, the Y-27632 treatment increased cell proliferation, with a 1–6 % overall increase for the 24-h 5-ethynyl-2’-deoxyuridine incorporation assay [33].
16.2.4 Determination of Cell Quality Prior to Tissue Engineering
In the absence of specific markers, the discrimination of cultured corneal endothelial cells from possible keratocyte/corneal stromal fibroblast contaminants is presently performed on the basis of cell morphology (polygonal cells at post-confluence) and the expression of ZO-1 and Na+/K+-ATPase proteins. Other corneal endothelial cell markers that can be used to discriminate between corneal stromal and endothelial cells include glypican-4, CD200, S100A4, S100A6, stress-induced protein immediate early response 3 (IER3), CD166, peroxiredoxin-6, and integrin subunit alpha 3 [34–37]. Cell quality should also be assessed prior to seeding, based on cell size, trans-endothelial resistance, and permeability.
16.2.4.1 Cell Size
Eye banks will only consider donor corneas for transplantation if the endothelial cell density is at least 2,000 cells/mm2 [38, 39]. Therefore, engineering an endothelium with a high cell density is of high importance. Unfortunately, corneal endothelial cell size increases following successive passages. Different approaches have been proposed in order to generate an endothelium of high cell density, including selecting the smallest cells by density-gradient centrifugation [40]. The authors hypothesized that they could separate the enlarged and flattened cells from the smaller more potent cells according to their specific gravity using density-gradient centrifugation. They showed that the smallest cells from the pellet generated a mean cell density of 1,584 cells/mm2, while those of the supernatant yielded a mean cell density of 828 cells/mm2 [40], indicating that this approach could be used to separate the small cells from a culture of cells of heterogeneous sizes. Finding culture conditions that could maintain the small endothelial cell size throughout cell expansion would be ideal. Further research is needed in order to address this point.
16.2.4.2 Trans-endothelial Resistance and Permeability
In the living eye, the corneal endothelium is a leaky selective barrier that allows the diffusion of nutrients from the aqueous humor toward the stroma to nourish the keratocytes while limiting the entrance of fluid in order to maintain corneal stromal deturgescence. In vitro, the barrier integrity of an endothelium formed from expanded corneal endothelial cells can be assessed using trans-endothelial electrical resistance (TER). Its paracellular permeability can be measured using diffusion of sodium fluorescein or FITC–dextran [41, 42]. For example, Singh et al. showed that bovine corneal endothelial cells grown at three different cell densities (<1000, 1000–1999, and >2000 cells/mm2) had TER measurements of 14 ± 5, 13 ± 3, and 14 ± 4 Ωcm2, respectively. They also observed high paracellular permeability for the lowest cell density [41]. Thus, prior to cell seeding on a carrier, a sample of cells can be used to measure the capacity of the expanded endothelial cells to form a leaky barrier in vitro, allowing to select the best populations for transplantation.
16.2.5 Carriers
A monolayer of human endothelial cells can be obtained in vitro with the proper culture conditions (reviewed in [43, 44]). However, a monolayer of endothelial cells is too fragile to be transplanted on its own. The cells need to be seeded on a carrier. The optimal carrier should be thin, flexible, transparent, and strong enough to be manipulated. It should allow the development of a fully functional endothelial phenotype and of interactions between stromal and endothelial cells. Different carriers have been proposed for the engineering of a posterior cornea using cultured cells (reviewed in [11, 45]).
16.2.5.1 Thin Membranes
Thin membranes of the thickness of a Descemet’s membrane offer the advantage of being “ready to use” without additional tissue preparation or stromal dissection prior to grafting. However, they can be difficult to handle as carriers. Because they are usually more fragile than a native Descemet’s membrane, each fold or stretch in the carrier damages the overlying endothelial cells. Scrolling can also be bothersome if a Descemet’s membrane is used as a carrier. In the context of the growing experience with DMEK, however, corneal surgeons are now ready to replace the diseased corneal endothelium by a thin lamellar posterior donor consisting of a tissue-engineered endothelium reconstructed on a thin carrier.
Proposed thin membranes include gelatin membranes [46–48], hydrogel lenses [49, 50], vitrigel (collagen sheets) [51], cross-linked collagen [52–54], isolated Descemet’s membrane [55], and amniotic membranes [24, 56].
A thin carrier can also be tissue engineered in vitro using the self-assembly approach. The technique is based on the capability of mesenchymal cells to secrete and deposit their own extracellular matrix in the presence of ascorbic acid [57, 58]. Corneal stromal fibroblasts produce a “living” corneal stromal substitute, in which cells are embedded in an extracellular matrix composed of collagen types I, V, VI, and XII, as well as lumican and decorin [59, 60]. Human corneal endothelial cells, seeded on top of this substitute, formed a uniform endothelial monolayer [60–62] with a cell density of 966 ± 242 cells/mm2 [60]. This endothelium expressed both the Na+/K+-ATPase and the sodium/bicarbonate cotransporter Na+/HCO3 −. To test the biocompatibility of these stromal carriers, they were transplanted in vivo (without endothelial cells) in the corneal stroma of feline eyes. Four months after transplantation, the stromal substitutes were transparent, functional, and well tolerated by the eye. All grafts remained avascular, with no signs of immune rejection, despite only a short course of low-dose topical steroids. Furthermore, corneal sensitivity returned to preoperative levels [59].
16.2.5.2 Native Stromas
Fresh native corneas denuded of their endothelium have also been used as carriers for the reconstruction of a corneal endothelium [63–80]. Prolonged periods of culture, however, are not possible with fresh corneas, due to the contamination of the endothelial cell culture by the proliferating corneal epithelial cells. Eliminating the native cells from fresh corneas before engineering a new endothelium would prevent contamination. Different techniques of corneal devitalization or decellularization have been described (reviewed in [81]). An endothelium engineered using cells seeded on a decellularized native cornea has been reported [70, 82, 83]. A main advantage of using a native stroma as a carrier is that cells are seeded on a Descemet’s membrane, which is their natural substrate. Proulx et al. reported that endothelial cells seeded on decellularized native stromas and cultured for 2 weeks formed a highly cellular monolayer of normal endothelial morphology [82]. Endothelial cells were well adhered to Descemet’s membrane and expressed the function-related proteins ZO-1, Na+/K+-ATPase, and Na+/HCO3 − (Fig. 16.2). Mean endothelial cell counts were 2272 ± 344 cells/mm2 using feline cells [82].
Fig. 16.2
Function-related protein expression in a feline endothelium engineered on a devitalized cornea. Immunofluorescence labeling showing that the engineered feline endothelium (a, c, e) and the native feline endothelium (b, d, f) expressed Na+/HCO3 − (a, b), Na+/K+-ATPase (c, d) and ZO-1 (e, f). Scale bar, 10 μm (Figure from Proulx et al. [82], with permission from Mary Ann Liebert)
16.2.6 In Vivo Transplantation of a Tissue-Engineered Endothelium on a Carrier
In vivo outcomes of these engineered endothelia have been assessed [83, 84]. Results showed that cultured endothelial cells seeded on a devitalized stromal carrier can recover an active pump function, restore and maintain normal corneal thickness, and yield crystal clear corneal transparency during 7 days after transplantation in the living feline eye. The reconstructed endothelium had a normal morphology and ultrastructure and expressed the function-related proteins Na+/K+-ATPase α1, Na+/HCO3 −, and ZO-1.
16.2.7 Injection of Corneal Endothelial Cells in the Anterior Chamber Without a Carrier
Cell injection in the anterior chamber as a means to replace the corneal endothelium is an old concept that has recently regained interest. Cell adhesion to the Descemet’s membrane represents the first challenge and different strategies have been proposed to address it. Maintenance of an eye-down prone position, allowing cells to reach Descemet’s membrane by gravity, has been shown to be helpful in reforming an endothelium in vivo, especially when the injected cells were precursors obtained by sphere-forming assay [85, 86]. Using a rabbit endothelial deficiency model, the authors showed that corneal transparency was restored when animals were kept in prone position for at least 6 h, contrary to the group positioned for only 1 h, which displayed focal edema and incomplete coverage of the endothelial defects in three out of six rabbits [85].
Another proposed strategy consists of the use of an extraocular magnetic field to attract injected endothelial cells loaded with magnetic particles such as iron [87, 88], magnetite oxide superparamagnetic microspheres [89], or superparamagnetic nanoparticles [90] toward the Descemet’s membrane.
Finally, the addition of the selective ROCK inhibitor Y-27632 has been shown to increase the adherence of corneal endothelial cells to plastic tissue culture plates in vitro [33, 91, 92] and to the Descemet’s membrane after injection in vivo [93]. The injection of cells and ROCK inhibitor Y-27632 in rabbit and monkey eyes resulted in the recovery of a transparent cornea with a monolayer of hexagonal endothelial cells expressing Na+/K+-ATPase and ZO-1 [40, 93]. In their recent paper, Koizumi’s group [40] mentioned that corneal endothelial cell injections, combined with ROCK inhibitor, are presently being performed in bullous keratopathy patients, a first-in-man clinical trial at the Kyoto Prefectural University of Medicine. Clinicians around the world are anticipating the results of this trial, as it will confirm the feasibility of cell injection for the treatment of corneal endotheliopathies in human subjects.
16.3 Tissue Engineering of a Corneal Endothelium from FECD Corneal Endothelial Cells
16.3.1 FECD Endothelial Cell Morphology in Culture
The culture of FECD cells has initially been reported after transduction with the human papilloma virus E6 and E7 oncogenes [94]. Zaniolo et al. then showed that FECD endothelial cells can be successfully isolated and cultured without transduction, with the same isolation and culture conditions used for healthy cells [13]. The Descemet’s membrane and endothelium specimens collected from 29 consenting patients having undergone Descemet’s stripping automated endothelial keratoplasty (DSAEK) were used for this study. Of these 29 specimens, 18 successfully initiated a culture, corresponding to a 62 % success rate. Cell counts between 8.0 × 103 and 1.8 × 105 cells (mean ± SD, 9.0 × 104 ± 4.5 × 104) were obtained. Of these 18 cultures, 12 had an endothelial rounded, slightly elongated cell morphology and six had more of a fibroblastic-like morphology (thin and very elongated cells) (Fig. 16.3). Cultures initially showing an endothelial morphology maintained it after subsequent passages (Fig. 16.3). These differences in cell morphology were also observed with the normal human corneal endothelial cell cultures.
Fig. 16.3
Morphology of FECD and healthy corneal endothelial cells in culture. (a–d) FECD cells. (e–f) Healthy human cells. (a) Endothelial-like morphology of a primary FECD (P0) culture. (b) Mixed morphology of FECD corneal endothelial cells containing many thin and elongated cells growing on top of each other (fibroblast-like morphology) of another primary (P0) culture. (c) Second-passage FECD cells of the population shown in A. (d) Confluent third-passage FECD cells of the same population shown in (a) and (c). The culture formed a monolayer of polygonal cells. (e) Subconfluent culture of normal corneal endothelial cells in P0. (f) Confluent third-passage culture of healthy corneal endothelial cells. Scale bars, 100 μm (Reprinted from Experimental Eye Research, Zaniolo et al. [13], with permission from Elsevier)
16.3.2 Conditions Favorable to FECD Cell Growth
This group also identified factors favoring the success of FECD cell culture [13]. Younger patient’s age was the leading factor, as generally accepted for primary cell culture [12, 21, 95–97]. All specimens from patients aged 60 years or less successfully started a culture of endothelial morphology, while the success rate for patients aged 80 years and over was very low, with only one of six specimens generating a culture and none with an endothelial morphology. A negative correlation was found between age and the number of days to confluence in the group displaying an endothelial morphology.
The absence of a fibrillar layer covering the posterior aspect of the Descemet’s membrane was also associated with a greater culture success rate. Among the 13 Descemet’s membranes studied, none of specimens that allowed the establishment of a culture with an endothelial cell morphology had a fibrillar layer. The endothelial cells isolated from Descemet’s membranes covered with a fibrillar layer (n = 6) either did not generate a culture (n = 4) or generated one with fibroblastic-like cell morphology (n = 2).
The corneal edema, presence of guttae, pigmentation of the endothelium, specimen size, and donor gender were not predictive factors for culture success.
16.3.3 Short-Term In Vivo Functionality of a Tissue-Engineered Corneal Endothelium Using FECD Cells
Haydari et al. [84] were the first study to demonstrate that the diseased endothelial cells of clinically decompensated FECD corneas, when cultured and seeded on a devitalized stromal carrier, can recover active pump function and restore and maintain corneal transparency for 7 days after transplantation in the living feline eye. Fifteen healthy animals underwent full-thickness corneal transplantation, with a corneal endothelium tissue engineered from human FECD (n = 7) or healthy (n = 2) endothelial cells. Controls received a native full-thickness corneal graft (n = 2) or a stromal carrier only (no endothelial cells; n = 4). Seven days after transplantation, six of the seven tissue-engineered FECD grafts, all tissue-engineered healthy grafts, and all healthy native grafts were clear, while all carriers-only grafts were opaque. Tissue-engineered FECD grafts progressively became thinner, with a mean central thickness of 772 ± 102 μm (659–1023 μm) and an average endothelial cell count of 966 ± 165 cells/mm2 on postoperative day 7; after which they were harvested for histopathology analysis. Transmission electron microscopy showed subendothelial loose fibrillar material deposition in all tissue-engineered FECD grafts, but no structures suggestive of guttae were observed. The tissue-engineered endothelium expressed Na+/K+-ATPase and Na+/HCO3−, however, with less intensity than in the normal native grafts.
16.3.3.1 Partial Rehabilitation
The major finding of this paper was that diseased endothelial cells of clinically decompensated FECD corneas were able to recover active pump function in culture and restore corneal transparency for 7 days after transplantation in the living eye. The endothelium tissue engineered from FECD cells performed better in the animal eye than the diseased FECD endothelium in the patient’s eye. Proposed hypotheses for this partial rehabilitation include the beneficial effect of the culture conditions and/or the natural selection of the healthiest cells in culture. Removal of the diseased thickened Descemet’s membrane may also have played a role in the recovery of these endothelial cells.
The partial recovery of these end-stage FECD endothelial cells harvested from the central and most diseased part of the cornea and the demonstration of their in vivo functionality open the door to a new horizon. Further studies are needed to evaluate if these first results can eventually lead to therapeutic strategies capable of delaying the course of the disease in FECD patients.
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
