Corneal endothelial transplantation from bench to bedside: A review of animal models and their translational value for therapeutic development

1. Introduction

Progressive advancements in corneal endothelial cell (CEC) transplantation have facilitated the gradual replacement of full thickness corneal transplantation as the definitive treatment for corneal endotheliopathies. Besides partial thickness transplantation, novel alternative therapies harness tissue engineered grafts as well as injectable cellular therapeutics to replace diseased endothelium. Such strategies further reduce potential for endothelial rejection and provide a minimally invasive option for patients.

Studies on penetrating keratoplasty in various animal models establish a foundation for ongoing innovations. This work dates back to 1837 when Samuel Bigger first performed penetrating keratoplasty on a domesticated gazelle, while Eduard Zirm and peers did so on chickens and rabbits (Armitage et al., 2006; Crawford et al., 2013; B. Rycroft, 1965; P. V Rycroft, 1965). Based on foundational knowledge gained from surgical techniques first developed in animals, Gerrit Melles launched a transplantation program using donor corneal tissue from cadavers (Melles et al., 1998). This program, and its associated transplant patient recipients, provided the impetus and clinical need for studies focused on the immunology of graft rejection in human penetrating keratoplasty. However, limitations to transplantation have prompted investigation of endothelial cell replacement technologies. Scarcity of donor corneas worldwide posed an initial obstacle to accessibility and efficacy for corneal transplantation, especially given distinct indications for penetrating keratoplasty and deep anterior lamellar keratoplasty in developing countries. However, in much of the developed world, the most current and salient motivations for endothelial cell replacement innovations center on ways to improve outcomes and procedures’ technical ease. Corneal transplantation may require intentional specialization and additional training. By contrast, an “off-the-shelf” and injectable cellular therapeutic could revolutionize the endothelial cell replacement space and offer ophthalmologists, cornea and non-cornea trained alike, a treatment tool with dramatically expanded options for administration.

Multiple groups have cultured CECs for delivery on various scaffolds, including collagen sheets, amniotic membranes, and even donor Descemet membrane (Ishino et al., 2004; Joyce and Zhu, 2004; Mimura et al., 2004a; Moysidis et al., 2015; Niu et al., 2014; Ozcelik et al., 2013; Peh et al., 2017; Watanabe et al., 2011). Attempts to deliver cultured cells intracamerally without a scaffold have also been explored; Shigeru Kinoshita’s team has been most successful in facilitating a microenvironment that suitably accomplishes this while achieving long-term corneal transparency (Kayukawa et al., 2020; Kinoshita et al., 2018; Mimura et al., 2003; Okumura et al., 2012; Schlötzer-Schrehardt et al., 2021). Jeffrey Goldberg’s team proposes yet another alternative in magnetic cell delivery, where CECs tagged with magnetic particles can be delivered and directed to the appropriate location using an external magnet (Mimura et al., 2003, 2005a; Morimoto et al., 2019; Moysidis et al., 2015; Patel et al., 2009; Xia et al., 2019).

Importantly, these discoveries—along with their translation into therapeutic options for patients—have been dependent on studies first conducted in an array of animal models. Foundational work on cultured CEC transplantation confirmed that human CECs could be successfully implanted on the Descemet membranes of mice (Joo et al., 2000), rabbits (Gospodarowicz et al., 1979a; Gospodarowicz and Greenburg, 1979; Jumblatt et al., 1978), cattle (Gospodarowicz and Greenburg, 1979; Jumblatt et al., 1978; Lange et al., 1993), cats (Gospodarowicz et al., 1979b), and non-human primates (Koizumi et al., 2007); the resulting grafts could be subsequently transplanted with penetrating keratoplasty. This work provides a foundation for exploration of alternative corneal endotheliopathy treatments, supporting the introduction of xenografts (tissue transplantation from donors of another species to the recipient), and allografts (from a donor of the same species, but not genetically identical). Despite immunogenic concerns surrounding transplantation methods involving xenografts, ongoing efforts to address these barriers in preclinical models pave a way for more cost effective and reproducible grafts with optimal optical and biomechanical properties in close proximation to human cornea (Kumari et al., 2018; Yang et al., 2020). Masking potentially immunogenic factors and reducing the total volume of transplanted tissue will also maximize the potential benefit of these developing treatments. Understanding assets alongside limitations for each animal model remains integral to continued innovation.

Research efforts to date have investigated possibilities for inducing active immunosuppression in the context of CEC transplantation. With the possibility of rejection ever present, studying both partial and full-thickness allograft implementation in animals has provided the basis for multiple advancements in surgical treatments, including Descemet Stripping Automated Endothelial Keratoplasty (DSAEK) (Honda et al., 2009; Mimura et al., 2004b). Challenges both before and after graft transplantation require critical consideration: graft creation demands distinctive growth conditions, while successful graft integration similarly necessitates a unique microenvironment to prevent neovascularization and scarring among other undesirable outcomes (Khodadoust and Silverstein, 1972; Williams et al., 1985). Navigating these obstacles as a collective has bolstered the scientific community’s overall insight regarding corneal endotheliopathies, associated treatments, tissue rejection and best investigative approaches. Here, we highlight landmark studies that have indelibly shaped how both clinicians and scientists alike tackle corneal endotheliopathies, while out-lining a roadmap to further these discoveries from the bench to bedside.

2. The models

The unique immune-privilege of the eye creates a hospitable microenvironment for otherwise host-incompatible grafts (P. V Rycroft, 1965). This enables xenogeneic CEC graft transplantation and overcomes the scarcity problem that limits allotransplantation of human CEC donor tissue. In developing the ideal xenograft tissue and transplantation method, it stands to reason that one model may not be appropriate for every aspect of treatment development. Instead, strategically selecting models for targeted innovation of corneal allotransplantation will likely improve the trajectory of the field. For example, the model most amenable to genetic manipulation and focus on individual molecular players of corneal transplant (i.e., mouse and zebrafish models) may not be the best model for assessing graft carrier design, proof of concept, and surgical implantation (i.e., rabbit, pig or other relatively larger models). Some models prone to corneal endothelial regeneration (i.e., rabbit) may also be less than ideal due to the obscuration of true therapeutic effect by innate healing mechanisms. For this reason, recognizing the strengths and limitations of each model organism is critical.

2.1. Graft rejection in small rodents: Mouse and rat models

Historically, the eye size of small rodent animals, such the mouse and rat, has excluded their significant use in preclinical CEC studies. Size constraints of the anterior chamber, for example, necessitate exceptional surgical skill and caution to limit procedural trauma in these models and reduce activation of the innate immune response. However, extensive knowledge of the mouse genome, alongside increasing availability of tools for genetic manipulation, provide a critical platform for these species’ utility in cornea research. For example, new genetic editing technologies enabled investigation of channels that facilitate corneal wound healing (Kumari et al., 2018). While aquaporins are traditionally recognized as water channels, studies comparing wild-type and aquaporin-5 knockout mice suggest that aquaporin-5 may mediate CEC migration and proliferation to promote corneal epithelial wound healing (Kumari et al., 2018). These data suggest that aquaporin-5 significantly contributes to corneal wound healing alongside its counterparts aquaporin-1 and -3, which are also highly expressed in the cornea (Levin and Verkman, 2006; Ruiz-Ederra and Verkman, 2009; Verkman, 2011). Similar genetic techniques supported work examining paxillin-mutant mice to better understand how this cellular adhesion molecule promotes dysfunction thru corneal angiogenesis (Yang et al., 2020).

Beyond these, numerous immunological studies focused on indirect allorecognition and specific immune players to create a clearer picture of corneal rejection, including both the molecules and cells that promote this process (Hayashi et al., 2009; Sonoda and Streilein, 1992). Studies using both mouse and rat models have expanded understanding of corneal graft rejection (Kumari et al., 2018; Yang et al., 2020). For example, evidence by the Mizuki lab suggests that endothelial cells are not the sole factor inducing allograft rejection, despite that their successful or unsuccessful transplantation strongly correlates with maintenance of corneal transparency (Hayashi et al., 2009). Rather, their work asserts that the non-endothelial component of full-thickness allografts contribute more significantly to instances of rejection (Hayashi et al., 2009). More specifically, they report that allotransplantation of CECs alone did not provoke typical markers of rejection, such as mixed lymphocyte or delayed hypersensitivity reaction (Hayashi et al., 2009). By contrast, allotransplantation of BALB/c mice-derived epithelium and stroma into C3H/HeN mouse recipients correlates with a rejection rate similar to that of full-thickness corneal allograft recipients (Hayashi et al., 2009).

Rat models provide additional insight, particularly given that rats exhibit corneal edema, corneal vascularization, and inflammatory cell invasion as manifestations of corneal graft rejection (Collin and Hoban, 1987; Schwartzkopff et al., 2010). Their smaller mouse counterparts also exhibit corneal opacification, and ongoing efforts seek to establish multiple key indices that will support significantly improved predictions around the extent of rejection (Hasková et al., 1996; He et al., 1991; Hori and Streilein, 2001; Joo et al., 1995; She et al., 1990; Sonoda et al., 2002; Sonoda and Streilein, 1992; Yamagami and Tsuru, 1999; Zhang et al., 1996).

2.2. Graft rejection in the rabbit model

Studies in rabbits have provided additional clarity on the basis of graft rejection. As early as the 1930s, rabbits became a standardized model organism for studying corneal grafts and secondary lymphoid tissue, with multiple reports of successful corneal grafts at the time (Bourne et al., 1976; Castroviejo, 1937; Khodadoust, 1968; Stansbury and Wadsworth, 1947; Thomas, 1930). Given its dimensions and structural qualities, the rabbit eye also served as a suitable platform for pioneering other surgical methods, including posterior lamellar keratoplasty, limbal epithelial graft placement and limbal stem cell graft placement (Hill and Maske, 1988; Williams and Coster, 1989; Castroviejo, 1937). Evidence in rabbits affirms the relationship between vascularization of the host graft junction at the time of transplant and increased rates of rejection; a parallel phenomenon is observed in human transplantation (Hill and Maske, 1988). Unlike in small rodents, the size of the rabbit eye’s enables greater ease of examination with slit lamp microscopy. This enables enhanced detection of signs of rejection such as fibrin deposition or keratic precipitates (Mimura et al., 2004b).

Despite numerous advantages, the rabbit model does have limitations. Rapid and extensive clotting of the rabbit aqueous humor occurs upon entry of the anterior chamber (Khodadoust and Silverstein, 1972). Intracameral heparin injection during surgery, along with post-operative application of topical heparin, help to alleviate this obstacle and decrease the rate of rabbit allograft rejection (Bourne et al., 1976). Further efforts to characterize rejection in rabbits has led to improved understanding of rejection in humans as well. For example, evidence suggests that both the size and precise location of a graft contribute to its rejection (Bourne et al., 1976). Between 7 mm and smaller 5 mm graft counterparts transplanted in rabbits, the larger graft induced significant and widespread rejection (Bourne et al., 1976; Khodadoust and Silverstein, 1972). Regardless of size, graft placement with greater proximity to the limbus also resulted in higher rejection rates (Khodadoust and Silverstein, 1972; Williams and Coster, 1989).

Current knowledge recognizes neovascularization as a possible conduit for a subsequent immune response in rejection (Khodadoust and Silverstein, 1975). A simultaneous mechanism may involve increased cellular traffic to draining lymph nodes, thereby creating a forward feedback loop to drive vasculogenesis and lymphangiogensis (Mimura et al., 2001). Early suture removal, however, helped to limit neovascularization, therefore improving rejection rates as well (Williams et al., 1985). Notably, even larger grafts could be transplanted successfully when supported by early suture removal, topical heparin application, and minimally traumatic surgical technique (Khodadoust, 1968). Although both interrupted and continuous sutures appear to be appropriate, some suggest that continuous sutures pose a lesser nidus for neovascularization (Eliason and McCulley, 1990). Nevertheless, vascularization may cross the host-graft junction as early as postoperative day 10, and evidence suggests that early removal of a nylon or other non-dissolvable suture is advantageous (Bourne et al., 1976).

This work also lays out the landmarks of corneal graft rejection’s progression in the rabbit eye. Along the road to rejection, the authors highlight rejection of epithelium, its neovascularization (Khodadoust and Silverstein, 1969), retrocorneal membrane development, and engulfment of the failing endothelium by inflammatory and spindle shaped fibroblastoid cells (Cho et al., 1998; Cohen et al., 1995).

2.3. Graft rejection in large animals: Sheep models

Larger animal models for endothelial cell transplantation are more limited, but sheep have also been included among studies to explore corneal graft rejection (Williams et al., 1999). In a sheep model, comparisons were made between sheep that received corneal autografts and those that received corneal allografts (Williams et al., 1999). However, there have been some difficulties in overcoming rejection within this model. From one study, allografts led to rejection at a median of 20 days after graft insertion, with immunohistochemical staining revealing up-regulation of major histocompatibility complex class I molecules and infiltration by predominantly mononuclear cells (Williams et al., 1999). CD4⁺ T cells first appeared in graft tissue within 2 days of rejection onset, and CD8⁺ T cells followed several days thereafter (Williams et al., 1999). Allografts proportionately contained more mRNA transcripts for interleukin (IL)-2 and tumor necrosis factor (TNF)-alpha than autograft counterparts (Williams et al., 1999). Compared to other animal models, evidence suggests that sheep exhibit a weaker inflammatory response and do not require heparin to prevent anterior chamber clotting; by contrast, heparinization in cat and rabbit models facilitate ease of endothelial keratoplasty (Williams et al., 1999). In the context of median graft rejection at postoperative day 20, with concurrent keratic precipitation, the sheep model may be suited to studying rejection of CEC grafts, but it remains to be seen whether this model should be pursued for more holistic application of CEC transplantation (Amano et al., 1992).

Combined efforts in rodent species importantly advance our knowledge of rejection in the context of corneal graft transplantation. Small rodent models play a critical role in propelling CEC replacement forward at both the molecular and cellular levels. These discoveries build a necessary foundation for other technologies in the field and ensure the indispensability of small rodent models. The rabbit model has also augmented progress, with the identification of rejection markers alongside concrete actions to mitigate these undesirable outcomes. Even so, divergences in the rodent and human immune systems, in addition to inherent size discrepancies, necessitate the use of larger animal models. Studies in sheep also provide invaluable information, particularly given their similar body and organ size along with certain physiologic similarities (Pinnapureddy et al., 2015). Furthermore, rabbits exhibit natural corneal wound healing abilities, which inevitably limit the generalization of results from this model to human therapies (Yamashita et al., 2018). Suggesting that results observed were not due to rabbits’ natural endothelial healing mechanisms, the authors report sustained corneal endothelial dysfunction for up to six weeks, which contrasts with typical and shorter rabbit corneal recovery time to baseline (Yamashita et al., 2018). Given these features of the rabbit model, comparisons to control and gold standard of care treatments are especially necessary. However, the low cost and similarity of anatomical features between rabbit and human eyes continue to recommend this model. Ultimately, as scientists develop improved knock-in/knock-out non-human primate models, more detailed studies of graft rejection will become possible in these species as well (Kishi and Okano, 2017; Kumita et al., 2019).

2.4. Engineering across species: Graft and carrier design

Extensive work developing surgical techniques in the rabbit established necessary foundations for the design of corneal grafts and carriers to date. Ultimately, the ideal corneal graft carrier would be easily reproduced for a multitude of patients, with a transplantation process that surgeons could also reliably perform the same way each time.

Gospodarowicz and Greenburg developed the first tissue engineered cornea using a rabbit model in the 1970s (Gospodarowicz et al., 1979b; Gospodarowicz and Greenburg, 1979). They successfully transplanted cultured bovine CECs into rabbits with stripped corneal endothelium (Gospodarowicz et al., 1979b; Gospodarowicz and Greenburg, 1979). Even without serum, cultured CECs could be introduced onto denuded rabbit Descemet membrane, with successful attachment and subsequent distribution of the cultured bovine CECs (Gospodarowicz and Greenburg, 1979). Not only this, but they were able to successfully reproduce these results in a cat model, which has extremely limited corneal regenerative capacities compared to rabbits, with maintained clarity of the corneal buttons and no edema (Gospodarowicz et al., 1979b). Here, they observed reorganization of the bovine CECs into a highly uniform cellular monolayer within eight days of introduction to the feline host eye (Gospodarowicz et al., 1979b). These first triumphs paved the way for further exploration of possibilities in CEC transplantation. In more recent years, the Amano team cultured human CECs on a carrier to form full-thickness grafts (Amano, 2003, 2002). Using collagen carriers for human CECs, they later performed DSAEK in rabbits (Mimura et al., 2004b). This collection of work demonstrates that seeding cultured human CECs onto human stroma can also lead to successful DSAEK (Honda et al., 2009). These endeavors underscore the rabbit model’s value in advancement of corneal graft engineering and set a precedence for further exploratory efforts.

2.4.1. Graft carrier transport for human CECs

Multiple teams have dedicated themselves to designing graft carriers for cultured CECs, with subsequent DSAEK-mediated transplantation in a rabbit model. Using human endothelial cells seeded onto a Poly(N-isopropylacrylamide) and gelatin-based carrier, the Hsiue team’s graft achieved uniform corneal transparency within 2 weeks of DSAEK in rabbits (Lai et al., 2007). When Vázquez et al. transplanted artificial silk fibroin endothelial graft carriers into rabbits through a DMEK-like procedure, they also observed excellent results, with normal corneal thickness and transparency by the sixth post-operative week (Vazquez et al., 2017). Using optical coherence tomography, they confirmed that surgically treated corneas and their contralateral control counterparts exhibited comparable corneal thickness. Further analysis demonstrated that the silk fibroin graft properly integrated into the host tissue, maintaining intercellular junctions and cellular pump function, forming an appropriate monolayer, and expressing key functional markers (Vazquez et al., 2017). Another study utilized an amniotic membrane carrier to transplant human CECs through a DSAEK-like lamellar procedure, consisting of DSAEK on a graft performed ex vivo followed by suturing with full thickness penetrating keratoplasty (Ishino et al., 2004). While this method appeared to maintain graft transparency for only one week, the study’s transplantation procedure also incorporated 7 mm full-thickness trephination of the host tissue to make way for implantation of the graft (Ishino et al., 2004).

Minimizing tissue transfer can reduce risk of rejection, and this principle guides some aspects of graft carrier design. Devitalization, for example, removes living cells from the carrier to prevent its endothelial, stromal, and epithelial cells from contaminating the graft. Evidence suggests that this reduces immunogenicity of the carrier itself (Hori and Niederkorn, 2007; Quantock et al., 2005). Liquid nitrogen gas can decellularize the corneal stroma while preserving corneal transparency, and a single free-thaw cycle could denude Descemet membrane without compromising cell adhesion (Amano et al., 2008; Bednarz et al., 2001). Similar results could be achieved by ammonium hydroxide chemical debridement (Proulx et al., 2009a, 2009b). These favorable findings in the rabbit model provide further guidance for graft carrier design and the process for graft carrier production.

Exploration of human CEC transplantation on electron beam-grafted poly(N-isopropylacrylamide) based carriers has also been notable. However, studies revealed that corneal thickness decreases significantly after endothelial stripping and transplantation of these grafts, as compared to control groups (Lai et al., 2007). While decreased corneal thickness can be beneficial to ensure appropriate clarity and transparency for vision, excessive corneal thinning at the opposite end of the spectrum is not desirable. This finding underscores graft fragility, unsurprising given reports of graft tearing during their transfer for implantation (Lai et al., 2007). However, the Hsiue team importantly utilized a circular poly (vinylidene fluoride) membrane to stabilize the graft, thereby enhancing graft adherence to the posterior corneal stroma while maintaining proper tissue orientation (Lai et al., 2007). This technology exemplifies considerations of graft and carrier integrity, while supporting additional efforts to improve practical transferability of corneal grafts from bench to bedside.

2.4.2. Material engineering for grafts

Other attempts to develop carriers include a gelatinous collagen disc, used to facilitate endothelial graft transfer following Descemet stripping (Lai et al., 2007). For this approach, collagen was not used as a cell carrier; rather, collagen discs faced the anterior chamber to provide additional support, with the endothelial graft layered between disc and the recipient stroma (Lai et al., 2007). The disc’s porous nature enables nutrient transfer from the aqueous humor to the graft, while simultaneously permitting graft components to excrete cellular waste products into the anterior chamber (Lai et al., 2013; Lai and Li, 2010). Though additional details remain to be uncovered, the disc may also serve as a physical barrier to circulating immune cells within the anterior chamber and limit direct immune interactions with the graft in the same vein as cellular micro-encapsulation approaches to allo- and xenotransplantation (Barnett et al., 2007). The polymer utilized will influence how the graft will interact with host. For example, if a polymer such as alginate is employed, cells will remain entrapped in the embedding matrix (Barnett et al., 2007). Such an approach can enable therapeutic factors of a low enough molecular weight to pass through the polymer and interact with the host tissue. Other polymers such as collagen, are readily degraded in the host. Lai et al. demonstrated that the eventual dissolution of a collagen disc seeded with endothelial cells, will enable direct contact against the anterior chamber in the proper apical-basal orientation when transplanted (Lai et al., 2007). However, it remains unclear whether collagen use over human stroma is principally responsible for improved outcomes in these studies (Lai et al., 2007, 2013; Lai and Li, 2010; Mimura et al., 2004b).

2.4.3. Advancing graft development through stem cell technologies

As work in higher order animal models continues to mature, additional possibilities for graft development appear. For example, scientists have identified widespread presence of oligopotent stem cells across the entire ocular surface (Altshuler et al., 2021; Collin et al., 2021; Li et al., 2021; Majo et al., 2008). These stem cell populations bear the capacity to generate individual colonies of corneal and conjunctival cells (Majo et al., 2008). Harnessing this discovery and applying these ocular, oligopotent stem cell (OOSC) populations to redirected endothelial cell fates may become a viable treatment option, with benefits including reduced immunogenicity, especially for patients who use their own OOSCs. While further detail on stem cell oriented-strategies are beyond the scope of this review, endothelial cell regeneration remains of strong interest among the cornea research community, and continued efforts seek to develop corneal renewal technologies (Altshuler et al., 2021; Collin et al., 2021; Li et al., 2021; Majo et al., 2008). Altogether, these efforts demonstrate a need for continued pursuit of graft engineering in larger animals as well.

2.5. Bridging the gap for transplantation in large animals: Pigs, cats, and non-human primates

Given both anatomic and physiologic parallels, several larger animal models have also gained traction as ideal candidates for corneal endothelial transplantation models to better approximate human disease and intervention. Perfecting surgical techniques in these models translates more readily to patients.

Increasing regard for porcine models stems in part due to endothelial cell behavior that better mimics that of humans, affordable cost, and easy accessibility of porcine tissue. Compared to the rabbit model, for example, porcine endothelial cells do not readily regenerate, reducing possible confounds from natural healing (Sweatt et al., 1999). Physical accessibility of tissue to surgical manipulation also requires consideration when selecting a model. Dimensions of the porcine model allow for improved manipulation; they have also enabled comparative studies of techniques for donor cornea insertion into the recipient anterior chamber during DSAEK (Hwang and Kim, 2009).

2.5.1. Graft preparation in pig models

To improve the potential for corneal grafts, corneal decellularization methods have also been explored in the porcine model (Isidan et al., 2019). Xenograft preparation requires consideration of each technique’s overall decellularization efficacy, methodological biocompatibility, and ultrastructural effect on the cornea. Reviewing the advantages and disadvantages of various methods, evidence suggests that a combination of the following approaches can be employed for xenograft preparation: (a) sodium dodecyl sulfate, (b) triton X-100, (c) hypertonic saline, (d) human serum with electrophoresis, (e) high hydrostatic pressure, (f) freeze-thaw, (h) nitrogen gas, (h) phospholipase A2, and (i) glycerol with chemical crosslinking methods (Isidan et al., 2019). While sodium dodecyl sulfate use sufficiently removes cytoplasmic and components of a preparation, it may harm the extracellular matrix and lamellar structure, lead to collagen fibril disorganization, and reduce overall corneal transparency (Du and Wu, 2011; González-Andrades et al., 2015; Isidan et al., 2019; Yoeruek et al., 2012; Zhao et al., 2014; Zhou et al., 2011); freeze-thawing leads to similar results (Isidan et al., 2019; Lin et al., 2008; Van den Bogerd et al., 2018). Another chemical approach utilizing triton X-100 application, failed to remove cytoplasmic and nuclear components, while damaging glycosaminoglycans and reducing transparency (González-Andrades et al., 2015; Kelber et al., 2017; Lynch et al., 2016; Sasaki et al., 2009; Wilson et al., 2016). Hypertonic saline has been used in a human clinical trial to successfully remove cellular components while maintaining extracellular matrix integrity, but graft transparency has been more difficult to maintain, and tissues often become edematous (Crapo et al., 2011; Gonzalez-Andrades et al., 2011; Isidan et al., 2019; Lee et al., 2014; Luo et al., 2013; Marisi and Aquavella, 1975; Oh et al., 2009; Zhang et al., 2015). Human serum with electrophoresis maintains extracellular matrix integrity and transparency, but this method fails to remove cellular components sufficiently and homogenously (Isidan et al., 2019; Shao et al., 2012, 2015). Although applying high hydrostatic pressure to the tissue still allows for maintenance of lamellar and collagenous structure and only requires a rapid procedure, it is cost prohibitive, disorganizes collagen fibrils, may cause edema and reduce corneal transparency (Funamoto et al., 2010; Gospodarowicz and Greenburg, 1979; Hashimoto et al., 2010). Tissue maintains transparency with N₂ gas use, but resulting edema remains a concern (Amano et al., 2008; Isidan et al., 2019; Lee et al., 2014). Phospholipase A₂ use also maintains tissue transparency, extracellular matrix integrity, and collagenous structure, but requires supplemental treatment to sufficiently accomplish decellularization (Isidan et al., 2019; Wu et al., 2009). Applying glycerol with chemical crosslinking maintains corneal transparency, but it leaves traces amounts of residual nucleic acids in the decellularization process (Isidan et al., 2019; Oryan et al., 2018). The study duration did not allow for exploration of long-term consequences from residual nucleic acids, but a possibility exists for this material to become immunogenic or to later interfere with the health of transplanted cellular tissue (Isidan et al., 2019; Oryan et al., 2018). Overall, however, these results demonstrate the immense potential of existing decellularization technologies, while hinting at existing barriers in graft preparation that we must overcome.

Despite clear successes in the porcine model, it remains less commonly utilized than small rodent counterparts. Ongoing work to develop transgenic pig lines may help foster more overall interest in the model (Meier et al., 2018). Even in the context of decellularization for graft preparation, transgenic pigs may eventually facilitate a deeper understanding of both cell-type or molecule-specific mechanisms underlying how the aforementioned techniques alter the graft and ultimately impact the native, ocular environment.

2.5.2. Surgical methods and graft preparations in a cat model

Others have preferred the cat model for verifying graft preparation technique. Parameters of the cat eye may be biased towards certain surgical approaches and allow better access than other mammalian counterparts (Brunette et al., 2011). This may be especially true of full-thickness corneal transplantation as a result of the cat eye dimensions and relative size of the anterior chamber (Brunette et al., 2011). In pursuing surgical avenues for graft insertion, evidence suggests that a 5 mm biopsy (which represents ~8% of the cat eye’s endothelial surface) is required for full corneal endothelial reconstruction (Proulx and Brunette, 2012). In this model, injection of corneal endothelial cells into the eye appeared to minimally impact the functional endothelium of the grafted cornea (Proulx and Brunette, 2012).

Graft survival and long-term viability of grafts’ cellular components are also critical. Assessments of endothelial cell counts in the cat eye one month after allograft insertion demonstrated an average of 30% cellular loss over the entire donor area, compared to only 15% cell loss following homograft insertion (Cohen et al., 1990). Efforts to improve these outcomes have included removal of donor endothelium prior to graft insertion as well. Overall, results in cat eyes have been similar to those in humans, and the cornea of cats have a similarly limited regenerative capacity that mirrors human patients (Bahn et al., 1982). These factors support use of the cat model in work to achieve better outcomes for corneal graft preparation, insertion, and survival.

2.5.3. Pioneering transplantation in a non-human primate model

Ultimately, identifying technologies and surgical manipulations compatible in non-human primates will yield techniques and grafts with the greatest translational relevance in humans. Multiple groups have already reported some success engineering corneal endothelium for non-human primates; in non-human primate recipients, attempts to perform CEC transplantation of human and porcine xenografts, of monkey allografts, and of monkey homografts have been promising (Insler and Lopez, 1991, 1986; Koizumi, 2009; Koizumi et al., 2007; Okumura et al., 2011). Some of the previously explored model species include cynomolgus monkeys and rhesus monkeys. At this time, cost and availability remain limiting factors for use of such models in both North America and Europe (Isa et al., 2009). In Japan, historical availability of Japanese macaques due to the native species’ previous status as an agricultural pest has enabled slightly augmented access to research in non-human primate models (Isa et al., 2009). For scientific questions closest to impacting patient lives, furthering research in these models will provide a critical foundation.

Some of the early work culturing CECs onto type I collagen carriers in primates have led to successful graft transplantation through DSAEK (Koizumi, 2009; Koizumi et al., 2007). Further examination of each primate subject revealed graft detachment from the host within one week of graft insertion, but the corneas recovered transparency within six months of graft insertion. Several possible explanations exist for this, one of which includes CEC separation from the graft and subsequent repopulation of the cornea. Or, the graft transplantation itself may introduce cytokines or growth factors that promote native CEC division and repopulation of the cornea.

These advancements in non-human primates significantly propel CEC transplantation forward, while also raising new questions about the mechanisms of both the successful and unsuccessful efforts to date: Can ROCK inhibition be harnessed across corneal endotheliopathies to improve CEC migration and proliferation? Are there species-specific differences in CECs or CEC-graft attachment that limit the translation of findings (i.e., from a rabbit model to a non-human primate model)? What is the timescale of post-operative improvement in primate subjects, particularly for novel treatment options, and will that translate for human counterparts? Furthermore, exploring both successes and failures of the field will be necessary to provide a more complete understanding of the corneal microenvironment and to develop additional safe, minimally invasive treatment options.

3. Beyond grafts: Alternative transplantation Approaches

Building on the fundamentals underlying rabbit and non-human primate driven CEC transplantation in particular, several teams have generated tremendous momentum towards actualizing novel clinical solutions.

3.1. Combination therapeutics: injectable CECs and ROCK inhibition

In particular, Kinoshita group findings suggest that Rho-associated kinase (ROCK) signaling inhibition facilitates improved adhesion of cultured CEC grafts in both rabbit and non-human primate models (Okumura et al., 2012); this is especially salient in the context of known Rho-ROCK signaling that downregulates integrin-mediated adhesion of monocytes, while selective ROCK inhibition upregulates adhesion and cellular attachment (Okumura et al., 2009; Worthylake et al., 2001; Worthylake and Burridge, 2003). Maintaining corneal transparency following CEC transplantation is a major challenge to successful treatment, but the addition of a selective ROCK inhibitor appears to promote transplanted CEC function as supported by measures of appropriate lucency, cell density, morphology, and expression of function-related markers (Okumura et al., 2012). Efficiency of ROCK inhibitor treatment remains under consideration as well, with necessary attention to the proportion of a CEC graft that successfully and functionally integrates into the host. Furthermore, evidence has not yet clarified whether ROCK inhibition alone is sufficient to permit local CEC regeneration or whether contributions of the CEC graft itself are required for successful treatment.

Further studies indicated that supplementing CECs with the ROCK inhibitor Y-27832 helped to sustain the healthiest endothelium, even without CEC scraping (Brunette et al., 2011; Ichikawa et al., 2008). Evidence demonstrates that this ROCK inhibitor activates CEC migration and proliferation, in addition to increasing cloning efficiency of limbal stem/progenitor cells through improved cellular adherence and reactive oxygen species-scavenging capacity (Kinoshita et al., 2018; Zhou et al., 2013). Additional efforts to harness ROCK inhibition using ripasudil for therapeutic intervention appears to involve cell migration by enhancing cell-matrix adhesion, focal adhesion complex formation and breakdown, intercellular junction disassembly, and proteolytic microenvironment regulation (Schlötzer-Schrehardt et al., 2021). In the context of Fuchs endothelial corneal dystrophy (FECD) cell lines generated from FECD patients, these ripasudil-induced changes could be promoted by a single dose of ripsasudil (Schlötzer-Schrehardt et al., 2021). Evidence also suggests that ROCK inhibition promotes CEC proliferation and migration to improve outcomes (Guo et al., 2015; Okumura et al., 2016, 2012; Schlötzer-Schrehardt et al., 2021; Sun et al., 2015; Syed and Rapuano, 2021). Further investigation, through off-label use of netarsudil in a study of patients with FECD, indicates that ROCK inhibition reduces corneal edema (Price and Price, 2021). To examine this theory further, the ROCK inhibitor netarsudil was tested amongst a cohort of symptomatic FECD patients, with results associating netarsudil-mediated ROCK inhibition with reduction of corneal edema alongside improvement in scotopic corrected distance visual acuity (Price and Price, 2021). While additional research is necessary to fully assess visual acuity improvement under a wider range of lighting conditions, and the potential benefits across corneal endotheliopathies, this work highlights the successful therapeutic application of ROCK inhibition to date. Further review will be needed to formally approve this strategy for clinical use.

3.2. Injectable CEC Trx’ansplantation: Magnetic solutions

Sidestepping full thickness or even partial thickness transplantation, several groups approach CEC dysfunction by exploring cellular therapeutics consisting of injection of single cells or cellular clusters into the anterior chamber (Mimura et al., 2004b, 2005c; Oh et al., 2010; Yokoo et al., 2005). Multiple methods for pre-injection preparation have been studied as well. While cryoablation appears to irreversibly damage keratocytes, it may compromise the interior corneal surface thereby compromising CEC transplantation effort. Descemet membrane stripping enables injected cells to settle directly onto the host stroma (Mimura et al., 2004a; Oh et al., 2010). More specifically, Oh et al. found that CECs and Descemet membrane were particularly sensitive to transcorneal cryo-injury, while the corneal epithelium appears much more resilient to cryo-damage (Oh et al., 2010). With regards to cellular clusters, CEC spheroids have been cultivated in vitro with verified expression of key corneal endothelial markers, including nestin, beta3-tublin, glial fibrillary acidic protein, and alpha-smooth muscle actin (Mimura et al., 2005b; Yokoo et al., 2005). These spheroids have been successfully injected into the anterior chamber of rabbits, with postprocedural confirmation of hexagonal CE-like cells and some initial suppression of corneal edema (Mimura et al., 2005b, 2007). However, significant barriers remain, including resolution of post-procedural corneal edema and the difficulty of redirecting injected CECs to the desired aspect of the corneal surface.

To address the latter issue, some groups have relied on post-procedural prone positioning to harness gravity-assisted settling and adherence of injected CECs to the posterior corneal surface (Mimura et al., 2007). This positioning also minimizes cellular adhesion to the trabecular meshwork, iris, and lens and migration of transplanted cells into the posterior chamber.

Tackling adhesion obstacles from another angle, the Goldberg team demonstrated advantages of magnetic CEC guidance using U.S. Food and Drug Administration approved-50 nm biocompatible super-paramagnetic nanoparticles in rabbits (Xia et al., 2019). Notably, their results indicate no change in the light transmittance, viability, identity, or function of cultured human CECs with cytoplasmic loading with 50 nm magnetic nanoparticles. Magnetic nanoparticle loading enabled manipulation by an external magnetic field (Bartakova et al., 2018; Moysidis et al., 2015). Magnetic CEC delivery outperformed CECs delivered by gravity according to measures of delivery efficiency and cellular integration within this model (Xia et al., 2019). Utilizing GFP-expressing, donor human CECs, it was possible to immunostain and distinguish between donor and host CECs following gravity versus magnetic delivery (Xia et al., 2019). As these efforts translate from the lab into clinical trials, the field remains hopeful that these results indicate real possibilities for patients, and Fig. 1 highlights the major, currently available, treatment options.

Fig. 1. Treatment options for corneal endotheliopathies.

1–3, illustrating (in order) Descemet stripping endothelial keratoplasty/Descemet stripping automated endothelial keratoplasty, Descemet membrane endothelial keratoplasty, and Descemet stripping only. 4, illustrating magnetic human corneal endothelial cell transplantation by injection into the anterior chamber with endothelial cells containing endocytosed magnetic nanoparticles (pictured as solid red circles). 5, illustrating corneal endothelial cell transplantation by injection into the anterior chamber with addition of ROCK inhibitor (pictured as yellow molecules).

3.2.1. The future of investigation and treatments for corneal endotheliopathies

While the aforementioned approaches are distinct, they ultimately reached clinical trials through evidence of successful preliminary investigations in animal models. The advantages and disadvantages of multiple popular animal models for corneal endotheliopathies are summarized in Fig. 2. Thoughtful selection of each model accounts for specific technical advantages alongside parallels to the human corneal environment. Notably, these recent efforts repurpose the rabbit model, previously deemed most suitable for graft design and carrier studies, for studying new possibilities within corneal transplantation. Work to date undoubtedly serves as a foundation, but we should not assume that historical approaches to aligning models with scientific questions prescribe the singular way forward. Altogether, these considerations of various animal models provide a necessary platform for researchers and clinicians alike in the creative development of novel therapeutic solutions for patients impacted by corneal endotheliopathies.

Fig. 2. Comparison of research models across species.

Analysis of the advantages and disadvantages of multiple popular animal models for different aspects of corneal endotheliopathy research. Species shown: monkey, pig, cat, sheep, rabbit, rat, and mouse. Anatomic size of each animal’s eye significantly impacts the CEC transplantation methods most suitable for further investigation within that model.

Data availability

No data was used for the research described in the article.