Animal Models for Limbal Stem Cell Deficiency: A Critical Narrative Literature Review

This model has mainly been employed in mice and involves the use of a blunt metal spatula with a 0.3-mm tip for scraping the entire corneal epithelium from limbus to limbus [37‐40]. It is advantageous in that it limits the injury to the epithelium, with less inflammation, neovascularisation, and scarring, making it an ideal model for ascertaining the mechanisms through which a specific treatment influences the outcome [37]. However, this method is limited by its ability only to induce partial LSCD [37, 38]. Furthermore, achieving complete corneal epithelial removal with a spatula can present challenges when attempting to standardise and execute the procedure in a reproducible manner, due to potential variations in the surgical handling of the tissue among different operators [38]. It is important to note that when employing a dulled blade for debridement, the basement membrane and the basal surface of basal cells remain largely undisturbed [39, 40]. This preservation of structural integrity may impose constraints on experiments intended to evaluate myofibroblast formation [12].

The conventional procedure for limbectomy typically involves dissecting the limbus to a depth of approximately 30% and then surgically removing a 2-mm segment from both sides of the limbus. Following this, the peripheral cornea and the scleral and conjunctival tissues within the dissection boundary are excised [41‐43].

Several reports have demonstrated the retention of intact limbal epithelium when surgical limbectomy is performed as a standalone procedure [8, 44, 45]. In their seminal study, Huang and Tseng observed that only 33% of their rabbits developed corneal vascularisation 6 months after complete surgical limbectomy. However, when they combined limbal removal with corneal epithelial removal, this figure increased dramatically, to 96% [46]. Likewise, complete thermal obliteration of the limbus deliberately extending to the depth of the ciliary body proved inadequate in inducing corneal neovascularisation even 4 months after the initial insult in a murine model [15]. These observations provide compelling evidence for the recent discovery of corneal-committed cells resident within the cornea capable of dedifferentiating and repopulating the stem cell pool [46, 47]. Given the possibility that the corneal limbus might not be the sole niche housing stem cells, approaches targeting limbus removal should be supplemented with the removal of the corneal epithelium [15, 16, 48]. To address this issue, investigators have recently combined surgical limbectomy with the application of NaOH in addition to limbal epitheliectomy, aiming to cause more extensive injury to stem cells residing within the cornea and limbus [31, 49, 50].

There may be a learning curve associated with surgical limbectomy, which may result in inconsistencies across the created models over the course of the study within and between different operators and laboratories [44]. Furthermore, surgical limbectomy demands more vigilance during and after the procedure than other physical methods, such as physical scraping or the use of the AlgerBrush [28]. This heightened vigilance is crucial to alleviate the risk of unintended perforation and entry into the anterior chamber associated with surgical limbectomy. It is also noteworthy that surgical limbectomy in laboratory animals does not evoke the customary inflammatory response observed after chemical or thermal injury in humans [51]. This holds particular significance, as the extent of the inflammatory reaction plays a crucial role in dictating the fate of stem cells following such injury. Notably, a comparative study demonstrated that surgical excision leads to less limbal hyperaemia, corneal oedema, and neovascularisation than that with alkali exposure in New Zealand rabbit eyes [45].

The AlgerBrush is a motorised instrument used to remove corneal rust rings following removal of a metallic foreign body and can also be used during pterygium surgery for shaving fibrovascular tissue from the corneal surface. It has recently emerged as a valuable tool in inducing LSCD in rabbits, rats, and mice [44, 52]. Since it requires less force and yields a more uniform and reproducible injury, it has been found more advantageous for limbal and corneal epithelial removal when compared to a blunt spatula [38]. The procedure typically entails two passes, with the lateral side of the burr facing the limbal and corneal region, with each pass taking approximately 5–6 s, complemented by thorough rinsing of the ocular surface [38].

The device offers three distinct burr types, each featuring 0.5- and 1-mm-diameter pointed burr and 2.5-mm-diameter rounded burr (Table 1). The selection of burr size for AlgerBrush usage varies depending on the animal model under consideration. While the 0.5-mm burr was more frequently used in mice owing to the small size of their eyes [38, 53, 54], the 2.5-mm rounded burr was found more effective in removing corneolimbal epithelium in rabbits when compared to the pointed 1-mm burr [44]. It is imperative to note, however, that factors other than the size of the burr, such as the burr speed, pressure exerted, angle of the burr, and residence time, are also crucial in determining the degree and extent of the injury [12].

The injury inflicted by the AlgerBrush primarily encompasses the surface epithelium and the basement membrane, exerting minimal impact on the underlying stroma when compared to chemical methods, thanks to a pressure-sensitive clutch preventing stromal penetration [40, 57]. Consequently, this mitigates the risk of excessive inflammation and scarring, facilitating interventions targeting LSCD and making it easier to assess post-intervention outcomes [52]. However, it is worth noting that less inflammation and scarring may pose a limitation in replicating real-life scenarios, especially when attempting to simulate injury resulting from chemical trauma. The LSCD model induced by a rotating burr is sustainable, lasting at least 3 months after the induction [52]. Its application is easy even in the hands of less experienced operators; however, the learning curve can still hinder achieving consistent and reproducible outcomes across different operators and laboratories.

The use of the AlgerBrush was discredited by some authors due to its inefficiency in achieving complete removal of the limbal and corneal epithelium [38, 54]. To avoid incomplete epithelial removal, other authors have suggested monitoring the degree of epithelial debridement using fluorescein stain and thoroughly rinsing the ocular surface after the procedure [44]. This precautionary measure is essential due to the inherent tendency of the rotating burr to displace cells towards the periphery of the resulting wound, potentially resulting in the re-engraftment of dislodged cells back onto the ocular surface when the animal resumes blinking [12]. The ability of the AlgerBrush to effectively remove the basement membrane has also been a subject of debate, as highlighted in a study conducted by Stepp et al. [12]. Notably, while the rotating burr was found to completely eliminate the basement membrane at the central region of the mouse cornea, it yielded only a partial effect in the case of rat corneas [12, 40]. In contrast, Boote et al. observed that the basement membrane was left intact after a single application of the rotating burr in mouse eyes [58]. However, a second round of the AlgerBrush coupled with the exertion of pressure onto the cornea led to complete denudation of the basement membrane, along with minimal removal of anterior stromal tissue [58].

Akin to other physical methods that directly disrupt the microarchitecture of the limbal niche, it is plausible that using a rotating burr may limit the efficacy of cellular therapies by potentially impeding the engraftment of transplanted cells within their natural microenvironment [56]. Another drawback of the AlgerBrush is the potential for stromal injury or perforation stemming from excessive pressure or overtreatment and the risk of bleeding when procedures are conducted at a very slow pace [38, 52].

Models of NaOH exposure to induce LSCD generally involve the application of a pre-soaked circular filter paper of various sizes (Table 2) onto the ocular surface in direct contact with the central cornea, the limbus, or a combination of both. The width of the filter paper varies depending on the ocular dimensions of the experimental animal as well as the primary area of intended injury. An alternative method for limbal and corneal NaOH exposure was proposed with the use of custom-size filter papers (2 × 4 mm) that were laid onto the corneal-limbal regions for a predetermined duration, a rather less-controlled exposure when compared to the use of circular filter papers [79]. Another modification was the use of a sterile cotton swab pre-soaked with 1 N sodium hydroxide (NaOH), which was then applied to the rabbit’s eye for a duration of 30 s [80]. Notably, applying flat filter papers may be challenging with highly curved corneas of mice [72]. Furthermore, the use of filter papers may cause unintended injury to the fornix and adjacent conjunctiva [81]. To overcome these limitations, Shadmani et al. proposed the use of punch-trephine for a well-circumscribed and localised exposure [82]. Similarly, Schultz et al. utilised a well with an internal diameter of 12 mm which was firmly pressed onto the cornea and subsequently filled with 1 mL of 2 N NaOH and left intact for 60 s [83]. More recently, Żuk and associates topically administered a 0.1-mL volume of NaOH with a concentration of 10% onto the corneal surface for 10 s [84]. Holan et al. adopted a similar approach by instilling ten drops of 0.25 N NaOH on the corneal surface of rabbit eyes over the course of 60 s [85]. Likewise, two studies reported the use of topical NaOH (0.1 N and 0.15 mM) for inducing LSCD in mouse eyes [13, 86]. Despite the ease of application with the topical eye drop method, particularly in small animals, this method provides less control over the time of exposure, with an increased risk of unintended injury to adjacent conjunctiva, fornix, and eyelids [82]. Furthermore, it is presumably less likely to induce a full-blown picture of LSCD due to quick spread to other unrelated ocular surface areas, drainage from the nasolacrimal duct, and rapid dilution with the basal and reflex tear secretion.

In numerous studies, the conventional approach entails exposure to an extended region encompassing the whole circumference of the cornea in addition to the limbus using filter papers [65, 66, 71]. Notably, however, such an approach may result in severe consequences, including corneal stromal scarring, corneal ulcers, and in extreme cases, penetration of the chemical into the anterior chamber, giving rise to widespread, undesirable side effects such as damage to the ciliary body, contractable uveitis, and phthisis bulbi, which may jeopardise the efficacy of the treatment under investigation and/or animal dropout from the study [71]. On the other hand, exposure solely to the central cornea, described as central alkali burn (CAB) by Sprogyte et al., has been shown to heal without conjunctivalisation and, therefore, may not necessarily lead to LSCD, making it more suitable for corneal wound healing research [87‐89]. Hence, limiting the contact of chemicals exclusively to the limbal area (limbal alkali burn) may be a prudent strategy, as it not only ensures the effective depletion of stem cells in this region but also serves to mitigate the aforementioned complications [90]. To this end, researchers have adopted the use of doughnut-shaped filter papers, crafted using punch trephines of various sizes to achieve the desired dimensions, with the aim of preventing exposure to the central cornea (Fig. 2) [71‐73, 75, 77, 91]. In contrast to circular-shaped filter papers, the use of doughnut-shaped filter papers alleviates the issue with improper fit along the limbal circumference due to the convexity of the central cornea potentially causing fit issues along the limbal area. Nevertheless, it is essential to acknowledge the presence of committed corneal epithelial cells, which may migrate to the limbus and undergo dedifferentiation to become putative limbal stem cells, thereby repopulating the limbal stem cell pool [16, 47]. This may be a concern, particularly when limbal approaches are not supplemented with mechanical disruption of the limbal microarchitecture and corneal epithelial debridement. In view of this concern, Ma et al. advocated a standardised procedure involving the application of 1 N NaOH for 30 s to the albino rabbit, followed by corneal epithelial debridement [48].

The exposure time is among the other important determinants of the extent of injury in an LSCD animal model. Kethiri et al. posit that the duration of ocular surface exposure to NaOH should be restricted to a maximum of 30 s in animal models [13]. Prolonged exposures beyond this timeframe have been associated with adverse outcomes such as the development of corneal ulcers, increased susceptibility to infections, decreased palpebral fissure height, and the potential for subsequent perforations [51]. Conversely, shorter exposure durations will likely yield partial LSCD rather than total LSCD. It should be emphasised, however, that 45 s of exposure could be closer to what a patient would experience in a real-life situation [13]. To investigate the effect of varying NaOH exposure times in rabbit eyes, a group employed different time regimes to induce chemical burns of varying severity, specifically employing durations of 25 s to generate a mild burn and 45 s to generate a severe chemical burn, the latter of which resulted in more corneal stromal disruption and diminished transparency [92]. Nonetheless, it should be noted that exposure time is only one variable amongst a plethora of variables affecting the final outcome with exposure to NaOH. Other reported determinants governing the efficacy of an alkali-induced LSCD model are the concentration and volume of NaOH solution, the area of exposure (e.g., the diameter of the filter paper), mode of delivery, clearance method, and, notably, the ambient temperature and humidity [89, 93].

The concentration of NaOH can be titrated depending on the depth and extent of the desired iatrogenic injury. In their report, Villabona-Martinez et al. suggest that while a 15-s application of 0.5 N or 0.6 N NaOH may be optimal for fibrotic damage to the epithelium and the stroma, a concentration of 0.75 N of NaOH applied for 15 s would be more ideal for a full-thickness response involving Descemet’s membrane and the endothelium without causing corneal neovascularisation (NV) [93]. However, if the objective is to provoke a full-thickness reaction with persistent epithelial defects and corneal NV, 1 N NaOH may be the preferred concentration [93]. Nonetheless, one should be aware that these recommendations by Villabona-Martinez et al. are based upon a standard exposure time of 15 s with 5-mm circular filter paper followed by thorough rinsing with balanced salt solution [93].

There are numerous advantages and disadvantages associated with the use of NaOH in inducing LSCD in animal models. Alkalis cause chemical burns similar to those observed after domestic and industrial accidents, with widespread injury to the ocular surface [44]. They also effectively remove the basement membrane of the cornea [94]. Neovascularisation in alkali models of LSCD tends to be more pronounced, making it an ideal choice for studying and developing strategies to combat corneal neovascularisation. [45] However, the use of NaOH ultimately gives rise to variable phenotypes of LSCD, and the depth and extent of the elicited injury are difficult to control, which is a significant challenge, particularly when dealing with small animals such as mice [38, 94]. Complications with the use of NaOH tend to be more pronounced and occur with greater frequency, while the risk of damage beyond the limbus to the sclera and posterior ocular structure is heightened [28, 36]. Interestingly, significant disparities have also been observed in the responses of rabbit and mouse eyes to ocular surface NaOH exposure when compared with specimens obtained from human eyes afflicted with LSCD, with the latter showing more pronounced vascularisation, inflammation, and goblet cell formation [13]. Nonetheless, it should be emphasised that in the aforementioned study, total LSCD could only be reliably established in half of the experimental animals, likely due to topical single-drop application of NaOH on rabbit and mouse eyes, raising questions about the methodology employed for inducing LSCD.

n-Heptanol has been widely utilised for the purpose of removing corneal epithelial cells. However, various studies have demonstrated that the use of n-heptanol alone often leads to re-epithelialisation [66]. Additionally, its effectiveness in removing limbal epithelial cells is considerably lower [44]. However, when combined with surgical limbectomy, the success rate for LSCD development increased to 96% [41]. As a result, n-heptanol is rarely used in isolation and is commonly employed in conjunction with surgical limbectomy, debridement, or AlgerBrush application [44, 95‐98]. One advantage of n-heptanol is that it selectively removes the epithelium without affecting the stroma, thereby preserving the stromal microenvironment [97]. Various exposure regimens have been proposed, ranging from 1–2 min to as long as 5 min [95, 97, 98]. Although n-heptanol was reported to successfully remove the corneal epithelium when applied for 60 and 120 s, longer durations of up to 300 s are required to eradicate the limbal basal cells [41].

Sulphur mustard (SM) vapour is a reactive alkylating agent that is a potential irritant to the ocular surface. Its exposure is characterised by a biphasic tissue response, namely the acute and the delayed phases, documented in both humans and animal models [99]. The delayed response, occurring as early as 2 weeks after experimental exposure, leads to LSCD. Its effects are likely driven by protracted inflammation in the limbal stroma, impairment of corneal sensory innervation, and trophic effects on the limbal niche [100, 101]. Various exposure methods have been defined, ranging from topical application to whole-body exposure [25, 99, 102]. However, due to the hazardous nature of SM vapour and its significant environmental risk to laboratory staff, it is a less preferred method for inducing LSCD [36]. Additionally, the damage caused by SM vapour extends beyond the limbus and affects the corneal endothelial layer, which undermines its validity as an appropriate LSCD model [25]. A report also noted that limbal epithelial cells exhibiting slow cycle characteristics may be more resistant to SM toxicity than the more terminally differentiated epithelial cells found in the central cornea [103]. As a result, SM exposure leads to a spectrum of LSCD severity, ranging from complete healing to instances of partial and total LSCD in both humans and rabbits, raising doubts about its suitability for studying total LSCD [100, 103].

Thermal injury has only been documented in one study in conjunction with rotating burr shaving of the limbal and corneal epithelium. The authors observed a more severe LSCD phenotype with pronounced corneal scarring, advocating the utility of the model in studies where the objective is to investigate stromal responses after thermal injury to the ocular surface [38]. However, inherent limitations of the proposed thermal method are the risk of perforation, the difficulty in optimising the temperature at the tip of the cautery, and the lack a standard technique for delivering heat throughout the whole circumference of the limbus.

Psoralen AMT (Sigma-Aldrich, St. Louis, MO, USA), a planar tricyclic compound that forms cross-links between thymine bases upon exposure to light, has been employed as a topical agent to deplete corneal stem cells in frog eyes [28]. When applied topically, it interferes with DNA replication and cell division, invoking a state of stem cell deficiency in the cornea [104]. Although it stands out as a rapid, easily applicable, controllable, and single treatment option for inducing damage to limbal stem cells, it has not yet been explored in larger-scale animal models. It has also been shown to cause a delay in eye development in the treated eye, an effect that may confound the validity of the model [28].

An interesting LSCD model was proposed by Das et al. based on repeated exposure to a standardised UV-B light in mice to emulate the ocular surface exposure to solar UV-B radiation in outdoor workers [105]. Their UV-B dose in experimental mice was equivalent to 4.5–5.2 h of occupational or environmental sunlight exposure for outdoor workers during summertime in different parts of the world [106]. Although the proposed model may be representative of a real-life situation, it is somewhat impractical considering that controlled exposure for precisely 20 min/day during the same daytime hours for a duration of 10 weeks was necessary. Furthermore, their findings in UV-B-exposed mice indicated the development of partial LSCD with cellular and structural alterations in the limbal area rather than the total depletion of limbal epithelial stem cells.

Given its well-documented cytotoxicity, benzalkonium chloride (BAC) has also been utilised as a chemical agent to induce LSCD in mouse eyes. In a study conducted by Lin et al., various concentrations of BAC (0.1%, 0.25%, and 0.5%) were applied to the mouse eyes four times a day over a 28-day period [107]. On the 28th day, the continuous exposure to 0.5% BAC resulted in the development of LSCD, as evidenced by clinical manifestations such as corneal conjunctivalisation and neovascularisation, the presence of goblet cells on the cornea, and the loss of putative stem cell markers [107]. Nonetheless, the long-term sustainability of this model after cessation of BAC exposure, particularly in achieving total LSCD, remains unknown. Moreover, such effects of BAC could be specific to mouse eyes, necessitating further studies involving other animal models.

The majority of animal models employed in the study of LSCD have traditionally focused on acquired LSCD, while research into congenital LSCD has remained relatively limited. A well-established model for mimicking congenital LSCD involves the utilisation of the heterozygous Pax6^+/− mouse, which serves as a primary model for aniridia and the associated keratopathy; however, its concordance with human aniridia is debated [108, 109]. Notably, Wang et al. recently proposed a novel method to induce congenital LSCD through forced eyelid opening immediately after birth [110]. The authors posit that prematurely removing the eyelid ditch at birth would impair the protective function of the eyelids on the vulnerable and still immature ocular surface tissues, thereby leading to limbal stem cell loss [110]. Subsequent to this intervention, the animals were followed up for 1 month, at the conclusion of which they were euthanised to validate the LSCD model. The methodology was reportedly easy to perform, and the ensuing inflammation was identified as the primary driver of the long-term development of an LSCD-like phenotype akin to keratitis-induced LSCD observed in humans [110]. Nonetheless, to what extent the proposed model may resemble the primary cause of LSCD in practicality, namely ocular surface chemical and thermal injury, warrants further investigation.