Understanding Cornea Homeostasis and Wound Healing using a novel model of Stem Cell Deficiency in Xenopus

Abstract

Limbal Stem Cell Deficiency (LSCD) is a painful and debilitating disease that results from damage or loss of the Corneal Epithelial Stem Cells (CESCs). Therapies have been developed to treat LSCD by utilizing epithelial stem cell transplants. However, effective repair and recovery depends on many factors, such as the source and concentration of donor stem cells, and the proper conditions to support these transplanted cells. We do not yet fully understand how CESCs heal wounds or how transplanted CESCs are able to restore transparency in LSCD patients. A major hurdle has been the lack of vertebrate models to study CESCs. Here we utilized a short treatment with Psoralen AMT (a DNA cross-linker), immediately followed by UV treatment (PUV treatment), to establish a novel frog model that recapitulates the characteristics of cornea stem cell deficiency, such as pigment cell invasion from the periphery, corneal opacity, and neovascularization. These PUV treated whole corneas do not regain transparency. Moreover, PUV treatment leads to appearance of the Tcf7l2 labeled subset of apical skin cells in the cornea region. PUV treatment also results in increased cell death, immediately following treatment, with pyknosis as a primary mechanism. Furthermore, we show that PUV treatment causes depletion of p63 expressing basal epithelial cells, and can stimulate mitosis in the remaining cells in the cornea region. To study the response of CESCs, we created localized PUV damage by focusing the UV radiation on one half of the cornea. These cases initially develop localized stem cell deficiency characteristics on the treated side. The localized PUV treatment is also capable of stimulating some mitosis in the untreated (control) half of those corneas. Unlike the whole treated corneas, the treated half is ultimately able to recover and corneal transparency is restored. Our study provides insight into the response of cornea cells following stem cell depletion, and establishes Xenopus as a suitable model for studying CESCs, stem cell deficiency, and other cornea diseases. This model will also be valuable for understanding the nature of transplanted CESCs, which will lead to progress in the development of therapeutics for LSCD.

1. Introduction

The basal cornea epithelium includes a population of Corneal Epithelial Stem Cells (CESCs) that are responsible for corneal homeostasis, wound healing, and for maintaining corneal transparency, which is essential for normal vision (Cotsarelis et al., 1989). During homeostasis, CESCs undergo asymmetric divisions to generate Transit Amplifying Cells (TACs), which undergo more rapid and numerous cell divisions, and ultimately form terminally differentiated cells of the more superficial layers. Superficial cells are eventually sloughed off from the surface as they die (Beebe and Masters, 1996; Thoft and Friend, 1983). It is generally accepted that CESCs reside in the basal layer of the peripheral cornea or limbus, in crypt-like structures called Palisades of Vogt (Amitai-Lange et al., 2015; Davanger and Evensen, 1971; Di Girolamo, 2011; Dua et al., 2005; Pellegrini et al., 1999; Schermer et al., 1986; Shortt et al., 2007a; Townsend, 1991, Zhao et al., 2009). However, some studies suggest they may reside throughout the cornea in certain species (Chang et al., 2008; Dua et al., 2009; Majo et al., 2008).

Limbal Stem Cell Deficiency (LSCD) is a debilitating corneal disease resulting from functional or anatomical loss of the cornea epithelial stem cell population. In LSCD, the transparent cornea epithelium cannot be renewed and is ultimately replaced by conjunctival epithelial cells. Clinical features for diagnosis of LSCD include epithelial defects, corneal opacity and vascularization, with consequent visual impairment or blindness (Sejpal et al., 2013). Centripetal migration of pigment cells into the central cornea has also been observed during this process in rabbits, guinea pigs, and even humans (Cafaro et al., 2009; Cowan, 1963; Davanger and Evensen, 1971; Henkind, 1965; Mann, 1944; Michaelson, 1952; Wolosin et al., 2000). Clinical manifestations of LSCD vary based on the severity and extent of involvement of the cornea tissue (Dua and Azuara-Blanco, 2000). There can be partial LSCD, in which only a segment of the limbus is involved, or more severe cases, involving total LSCD.

Remarkable therapies have been developed to treat LSCD by utilizing epithelial stem cell transplants (Atallah et al., 2016; Galindo et al., 2017; Grueterich et al., 2002; Haagdorens et al., 2016; Holland 2015; Koizumi et al., 2001; Le and Deng, 2019; Ljubimov and Saghizadeh, 2015; Rama et al., 2010; Saghizadeh et al., 2017; Sasamoto et al., 2018; Shortt et al., 2007b; Utheim et al., 2018; Yazdani et al., 2019). However, effective repair and recovery depends on many factors, such as the source and concentration of donor stem cells, and the proper conditions to support these transplanted cells (Atallah et al., 2016; Ti et al., 2002). Our understanding of LSCD pathology, and the specific behavior of CESCs, is still rather limited. We do not fully understand mechanisms underlying how CESCs heal wounds or how the transplanted CESCs restore transparency in LSCD patients (Castro-Munozledo, 2013). A major hurdle to studying CESCs and LSCD has been a lack of convenient vertebrate models.

One excellent vertebrate model is the frog Xenopus. In the early larval stages (Nieuwkoop and Faber, 1994), the Xenopus cornea epithelium is a two-cell layer thick structure, comprised of an outer apical layer and a basal layer. A few scattered keratocytes lie below the basal layer. The cornea endothelium is not initially fused to the basal layer; the two layers being connected only at the center of the cornea by a structure called the corneal stalk or stroma attracting center (Hu et al., 2013). As the cornea develops, more keratocytes appear, the stroma develops, and additional structures are added (Bowman’s layer, and Descemet’s membrane) (Hu et al., 2013). In Xenopus, both cornea morphology and development are largely conserved with that of humans (DelMonte and Kim, 2011; Hu et al., 2013), which makes it a suitable model system to study the cornea epithelium, its stem cells, and potentially the pathology of corneal diseases.

One of the methods used to identify stem cells is via retention of incorporated thymidine analogs (BrdU, EdU). Unlike the highly proliferative TACs, the more quiescent stem cells retain the incorporated thymidine analogs for longer durations and are referred to as Label Retaining Cells (LRCs) (Yoon et al., 2014). Previous studies conducted in our lab show that while these LRCs are restricted to the peripheral limbal region in the adult frog cornea, (Hamilton and Henry, 2016) the LRCs appear to be distributed throughout the basal layer in the larval frog cornea (Perry et al., 2013). Interestingly, in both the larval and adult frog cornea, p63 (conventionally accepted as one of the CESC/TAC markers (Pellegrini et al., 2001)) labels all nuclei throughout the basal layer of larval and adult corneas (Perry et al., 2013; Sonam et al., 2019).

Here, we used a photoactivatable DNA cross-linker, Psoralen 4-amino-methyl trioxalen (AMT) hydrochloride, in combination with UVA exposure (PUV), to establish a novel frog model of stem cell deficiency (SCD) that mimics many of the changes seen in human LSCD. This combination of Psoralen and UVA is routinely used in labs to generate feeder cells (McGarry et al., 2009) for tissue culture, in clinical treatments of psoriasis (Stern, 2007), and to inactivate leukocytes for preventing transfusion-associated graft-versus-host disease (Grass et al., 1998). We show that treatments done on whole corneas result in cases of SCD, and these corneas do not regain transparency. Furthermore, we have characterized cellular/molecular events, including: mitosis, cell death, cell density, appearance of the Tcf7l2 labeled subset of apical skin cells, and changes in the percentage of p63 positive basal cells (as a potential indicator of basal proliferative cells, e.g., CESCs and TACs), over a period of up to 32 days following these treatments. We showed that PUV treatment results in cell death and a conversion to the skin phenotype. On the other hand, corneas subject to treatments on only one half of the cornea are eventually able to heal and restore their transparency.

2. Materials and methods

2.1. Animals

Adult Xenopus laevis frogs were obtained from Nasco (Fort Atkinson, WI). Tadpole larvae were reared following established protocols (Henry and Grainger, 1987; Henry and Mittleman, 1995). Tadpole stages were assessed using those defined by Nieuwkoop and Faber (1994).

2.2. Creating LSCD via photoactivatable (UV), chemical treatments (Psoralen AMT)

To deplete cornea stem cells we used Psoralen AMT (Sigma-Aldrich, St. Louis, MO). Psoralen is a planar tricyclic compound, which passes freely through cell and nuclear membranes, where it intercalates into the DNA strands (Cimino et al., 1985). It is a light-sensitive drug that absorbs ultraviolet light (long wave-UVA) to form monoadducts and diadducts (crosslinks) between thymine bases (Cimino et al., 1985; Deans and West, 2011; McGarry et al., 2009). Therefore, treatment with UVA forms covalent bonds and irreversible DNA inter-strand crosslinks (Cimino et al., 1985). Inter-strand crosslinking with Psoralen interferes with DNA replication, arrests cell division, and can cause cell death, effectively preventing proliferation of treated cells. This novel application of Psoralen/UV (PUV) treatment, when applied to the cornea, creates a state of stem cell deficiency within the cornea.

2.3. Topical application of Psoralen (Construction of the reservoir pipette tip)

A standard 1ml pipette tip (Catalog No. 1111–2021, Tip One, USA Scientific Inc.) was used to construct a modified reservoir tip for localized application of Psoralen to the cornea. The tip was cut at approximately 5mm from the dispensing end at a 45° angle (Figure 1A, 1B). This results in an oval opening with inner diameters of approximately 1.7mm by 1.4mm at the end of the pipette (Figure 1C). A notch of approximately 9mm in length was also cut into one side of the pipette, about 7mm above this opening (Figure 1B, 1C). This creates an open reservoir with a holding volume of about 40μl. The reservoir opening is sufficiently large to permit addition of Psoralen solution, without disturbing or moving the pipette when it’s placed tightly against the cornea. Lastly, the back end of the pipette was cut in half, about 20mm from its opening (Figure 1B), and half of the pipette was removed so that it can be attached to a metal rod held by a Narishige joystick hydraulic micromanipulator, using a small binder clip (Figure 1D, 1E).

Figure 1: Apparatus and procedure for the PUV treatment.

(A) Standard 1ml pipette tip, which is modified as (B) a special reservoir applicator pipette using a razor blade, according to the dimensions shown. The broad end of the pipette (mmc) is cut in half so that it can be attached to the metal rod held by the micromanipulator using a small binder clip. (C) Top view of the reservoir pipette showing the internal diameters (d1 and d2) of the pipette opening and side port or reservoir opening. (D–F) PUV-treatment procedure. (D) Anesthetized tadpole is placed in a petri dish on Kimwipes soaked in anesthetic solution. Reservoir pipette tip is affixed to a metal rod held by the micromanipulator using a small binder clip. The tip is gently placed around the cornea to form a sealed chamber. Psoralen solution is then added into the reservoir and allowed to contact the cornea. (E) Higher magnification view of area shown in panel D. (F) UV exposure using a 10X objective. UV is localized to the cornea, (bright blue spot), using the microscope iris diaphragm. (G) Schematic diagram of larval Xenopus eye. Structures are as labeled, cer, corneal end of reservoir; mmc, micromanipulator clamp end; ro, reservoir opening.

2.4. Psoralen-UV (PUV) treatment procedure

Xenopus tadpoles were treated at developmental stages 51–53 (Nieuwkoop and Faber, 1994). The tadpoles were first anaesthetized (~1–2 minutes) in MS-222 (ethyl 3-aminobenzoate methanesulfonate, Sigma-Aldrich, St. Louis, MO) diluted 1:2000 in 1/20× Normal Amphibian Media (NAM) (Slack, 1984). A 60mm petri dish was layered with anesthesia-moistened Kimwipes to form a trough to restrain the tadpole during treatment (Figure 1E). The tadpole was gently maneuvered into the trough with a small fire polished glass rod to position it on one side, such that the left lateral side was facing upwards toward the microscope objective. The reservoir pipette tip was maneuvered using the Narishige hydraulic micromanipulator, to lower it to the surface of the cornea, where the modified tip establishes a sealed, leak-free pocket with the small reservoir directly over the cornea (Figure 1E).

A fine gel loading pipette tip (Catalog No. 02-707-181, Fisher Scientific, Hampton, NH) was used to deliver 20μl of the Psoralen solution into the reservoir. After the treatment time had passed, the Psoralen was removed from the reservoir and the pipette tip was retracted from the eye. The tadpole, was immediately illuminated by UVA light (wavelength=346nm) for the required amount of time, using an upright fluorescence microscope (Zeiss Axioplan) with a 10X objective (Figure 1F). The area of UV exposure was constrained to the approximate diameter of the eye and the thin ring of skin immediately surrounding the eye by using the fluorescence microscope’s adjustable iris. In some cases only the posterior half of the cornea was irradiated. To accomplish this, a mask was placed over the anterior half of the eye that consisted of a rectangular sheet of aluminum foil, approximately 24mm × 15mm. One edge was folded back along the long axis to prepare a final size of 24mm × 11 mm. This was done since the trimmed edges might be sharp, and therefore only the smooth, folded edge was allowed to contact the cornea.

After illumination, the tadpoles were transferred to recovery bowls containing 1/20× NAM for about 30 minutes, and the bowls were placed on a rocker set at 20 rocks per minute to permit quick recovery of the tadpole from the anesthesia (Hamilton and Henry, 2014). The tadpoles were next transferred into a bowl with dechlorinated tap water. Corneas from the untreated right sides of these tadpoles were used as internal controls.

2.5. Phenotypic assessment of LSCD

In both treated and control cases, morphological changes were observed, including the invasion of pigment cells from the peripheral limbus towards the center of the cornea, changes in opacity, and vascularization of the cornea. Images for phenotypic assessment were taken under a compound microscope (Zeiss Axioplan) using external epi-illumination from a gooseneck fiber-optic light source. A SPOT camera (Spot Imaging, Sterling Heights, Ml) was used to capture images at multiple focal planes for each specimen. These images were then merged (flattened) using Helicon Focus software (Helicon Soft Ltd., Kharkov, Ukraine). FIJI (National Institutes of Health, Bethesda, MD, USA) was used for image analyses. The area of the cornea was approximated in these images by drawing a circle around the peripheral region overlying the diameter of the eye, and calculating the area inside this circle (boundary demarcated in Figure 1G).

The number of pigment cells were then counted in this area, and normalized to a standard area (1mm² for whole cornea treatments and 0.5mm² for each half following half cornea treatments). The normalized number of pigment cells were then plotted and compared to controls. Opacity and vascularization were scored on a scale of 0, 1, and 2, based on the degree of opacity or vascularization (see Table 1 for the scoring criteria), which are based on similar criteria reported in the literature (Le et al., 2018; Shortt et al., 2014).

Table 1: Scoring system used to grade corneal opacity and vascularization.

This table shows scoring system based on a scale of 0 (no signs), 1 (moderate), and 2 (severe). The criteria used to assign the score for corneal opacity and vascularization are described, and some examples of these assigned scores shown in various figures are also listed.

Score	Corneal Opacity	Vascularization
0 (No signs)	Cornea is transparent. Outlines of xanthophores, iridiophores and melanophores in the iris are clearly visible. Pupillary edge (free edge of the iris) is well defined and distinctly visible. Example: Figure 2A, 2A’	No blood vessels present. Example: Figure 2A, 2A’
1 (Moderate)	Cornea appears hazy, with a grayish translucency. Outlines of xanthophores, iridiophores and melanophores in the iris are blurry. Pupillary edge is discernable but blurry. Example: Figure 2B, 2B’	Blood vessels seen at the cornea boundary, but stay restricted to the peripheral cornea region. Example: Figure 2B
2 (Severe)	Cornea appears opaque. Outlines of xanthophores, iridiophores and melanophores in the iris are not visible. Pupillary edge is indistinguishable. Example: Figure 2D, 2D’, Figure 3B, 3B’	Multiple blood vessels are present. Blood vessels extend into the central cornea. Example: Figure 2D’, Figure 3C, 3C

Animals were reared for a period of up to 32 days. Challenges were associated with rearing and imaging the live animals, which are subjected to repeated anesthetization and observation. Some animals do not tolerate repeated, frequent anesthesia to permit daily observations. Therefore, the time intervals between repeated observations were lengthened and varied for different specimens.

2.6. Determination of Psoralen treatment concentration and duration

A stock solution of 5mM Psoralen AMT was first prepared in DMSO and diluted into 1/20×NAM (Slack, 1984), to make the required working solutions. A wide variety of conditions, including different Psoralen concentrations (between 5 and 25μM), treatment times (1 to 5 minutes), and UVA exposure times (between 60 and 75 seconds), were tested in preliminary experiments to determine a set of conditions that produced consistent results (preliminary data not shown). In some cases, the treatments were even repeated a second time on the same animals. It was found that a single application of Psoralen, at a concentration of 20μM, for 2 minutes, 45 seconds, with UVA exposure of 65 seconds, represented the lowest concentration and time interval with the most consistent effects, where 100% of the specimens exhibited some opacity at 32 days post treatment (dpt). 88% of these cases exhibited signs of neo vascularization by 32dpt. Furthermore, of these cases, 92% of the specimens exhibited some signs of opacity and neovascularization by 15/16dpt (Figure 4B, 4C, Table S1A, S1B). With lower concentrations and shorter exposure times the percentage of cases with these effects was smaller. Higher concentrations and longer treatment times were found to cause the eyeball to undergo a slight rotation away from the surface in some cases. Interestingly, when these same conditions were tested for the half-cornea treatments, the effects were found to be somewhat less severe. For example, only 50% of the cases showed some moderate opacity and only 50% of the cases showed some moderate neovascularization at 15dpt on the treated side. The other cases appeared to be normal. Therefore, a slightly higher concentration of Psoralen (25μM) with a longer application time (3 minutes, 30 seconds), and longer UVA exposure (75 seconds) were ultimately chosen for those half cornea treatments. These conditions resulted in some opacity in 100% of the specimens, and neo-vascularization in 86% of the specimens at 15dpt (Figure 4E, 4F, Table S1C, S1D). Even though this treatment was greater than that used on whole corneas, most of these half-treated corneas were eventually able to recover by 28dpt (unlike whole treated corneas, see Results below).

Figure 4: Quantification of pigmentation, opacity, and vascularization following PUV treatment on whole corneas and half corneas.

(A) Control corneas show very minimal levels of pigment cells (black line), whereas corneas that received PUV treatment show increased pigmentation at the corresponding time points (red line). (B) Control corneas show no opacity at all time points examined (silver columns), whereas opacity increased in PUV treated corneas as the time points progressed (orange and red columns). (C) A majority of control corneas showed no vascularization at all time points examined (silver columns), however, some corneas showed slight vascularization very close to the peripheral boundary of the cornea (gray columns). Vascularization scores in the PUV treated corneas increased as the time points progressed (orange and red columns). (D) In the cases where only half of the cornea received PUV treatment, the untreated half (blue line) shows minimal pigmentation, whereas the treated half (red line) shows an increase in pigmentation, through 15dpt, after which it starts declining as the pigment cells are lost from the cornea. (E) No corneal opacity was detected in the untreated half of the cornea (silver columns), whereas corneal opacity in the treated half of the cornea increased until 15dpt (orange and red columns), and then declined by 28dpt (blue column). (F) A majority of untreated halves of the corneas were scored 0 for vascularization at all time points examined (silver columns), however, some corneas showed slight vascularization very close to the peripheral boundary of the cornea (gray columns). Vascularization scores in the PUV treated corneas increased until 15dpt (orange and red columns), and then declined by 28dpt (blue and orange columns). Error bars represent standard error of the mean. Statistical analysis has been done for untreated control corneas vs. PUV treated corneas in (A), and for untreated control halves vs. the PUV treated halves of the corneas in (D). N = 7–12 for time points in (A–C), except for control (whole cornea) at 1dpt (N = 4). N = 5–9 for time points in (D–F). *** = p-value < 0.0002. ** = p-value ≤ 0.008; * = p value ≤ 0.05.

2.7. TUNEL assay

The TUNEL staining protocol was adapted from Hensey and Gautier (1998) with the following changes. Tadpoles were fixed in 3.7% formaldehyde for 2.5 to 3 hours at room temperature (RT). Eyeballs were removed and rinsed with 1X Phosphate Buffered Saline (1XPBS), 3 times for 15 minutes at RT. Permeabilization was done by incubating eyeballs in 1XPBS and 0.2% Triton X-100 (PBT) for 2 washes of 10 minutes each. The samples were then washed in 1XPBS 2 times for 15 minutes each, before incubating in 1X Terminal deoxynucleotidyl Transferase (TdT) buffer (Promega, Madison, WI) for 30 minutes at RT. Recombinant TdT (Promega, Madison, WI) was added at a concentration of 150 U/ml with 0.5μM digoxigenin-11-dUTP (Roche), and samples were incubated overnight at RT. Reactions were terminated with 1XPBS and 1mM EDTA at 65°C (two washes of one hour each). EDTA was removed with four one-hour washes in 1XPBS at RT. The chromogenic detection reactions have been adapted from Harland (1991) with the following changes. After the 1XPBS washes from the previous step, the eyes were washed in PBT for 15 minutes at RT, blocked in PBT with 20% goat serum, for one hour at RT, and incubated overnight at 4°C in Fab fragments of anti-digoxigenin-alkaline-phosphatase (Roche, Basel, Switzerland) diluted 1:2000 in PBT with 20% goat serum. Reactions were rinsed with PBT at RT for 6 washes of one hour each, and finally at 4°C overnight. The samples were briefly washed 2 times for 10 minutes each with alkaline phosphatase (AP) buffer (0.1 M Tris, pH 9.5 and 0.05M MgCl₂), before the staining was developed using nitro blue tetrazolium and 5-bromo-4-chloro-3-indolyl phosphate substrates, each diluted 1:300 in AP buffer. The chromogenic reaction developed at RT for approximately 25 minutes and was terminated by washing in 1XPBS once at RT for 5 minutes, followed by 2 washes of 1XPBS at RT for 30 minutes each to remove any leftover chromogenic reagents.

2.8. Immunofluorescence

The processed samples from TUNEL assays were washed in PBT, 2 times for 15 minutes each at RT. Blocking was done for one hour at RT with the blocking mixture prepared as follows: 1% Bovine Serum Albumin and 10% Goat Serum, diluted in PBT. The samples were then transferred into primary antibodies, which are all diluted in the immunofluorescence blocking solution, and incubated overnight at 4°C. The p63 antibody (1:125, ab735, Abcam) was used to detect basal cells (CESCs/TACs), and the phospho-Histone H3S10P antibody (1:400, sc-8656-R, Santa Cruz) was used to detect the subset of cells undergoing mitosis. Samples were rinsed in PBT 5 times for 10 minutes each before incubation in secondary antibodies (1:300, diluted in immunofluorescence blocking solution, Alexa Fluor goat anti-mouse 488 or Alexa Fluor goat anti-rabbit 546, Life Technologies) for 1.5 hours at RT. Next, samples were washed 2 times in PBT for 10 minutes each at RT. Nuclei were visualized by staining with Hoechst 33342 (Thermo Fisher Scientific, Waltham, MA) (2μg/ml in 1XPBS) for 30 minutes at RT. Lastly, samples were rinsed 2 times in PBT for 10 minutes each at RT, and once in 1XPBS for 10 minutes at RT.

For Tcf7l2 immunostaining, animals were fixed in Dent’s fixative (20% DMSO, 80% methanol) overnight at 4°C, followed by 2 washes with 100% methanol for 10 minutes each at RT. Eyeballs were removed and washed with 1X PBS and 0.5% Tween 20 once for 5 minutes at RT, followed by incubation in Dent’s blocking solution (1X PBS, 0.5% Tween 20, and 10% Goat Serum) for 2 hours at RT. Samples were incubated in the Tcf7l2 primary antibody (anti-TCF4, D-4: sc-166699, Santa Cruz Biotechnology, Dallas, TX) at 1:200 dilution in Blocking solution, overnight at 4°C. Afterwards, samples were washed with 1X PBS and 0.5% Tween 20, 6 times for 10 minutes each at RT, before incubating in secondary antibody (1:300, Alexa Fluor Goat anti-mouse 488, Life Technologies) for 1.5 hours at RT. The samples were then washed with 1X PBS and 0.5% Tween 20 four times for ten minutes each. Nuclei were visualized by counterstaining with Hoechst 33342 (1 μg/ml in 1XPBS) (Thermo Fisher Scientific) for ten minutes at RT. Finally, samples were rinsed 2 times in PBT for ten minutes each at RT, before mounting the corneas on microscope slides, as described above.

Following the staining, the corneas were gently detached from the eyes by lifting their edges using fine forceps, and cutting the cornea free from the corneal stalk (Figure 1G). The detached corneas were placed in a 60mm petri dish filled with 1XPBS. Ultra-fine iridectomy scissors were used to make 4–5 small cuts in each cornea from the periphery towards the center, which helps to flatten their spherical structure. The corneas were then placed into a drop of SlowFade Gold Antifade mountant (Thermo Fisher Scientific, Waltham, MA) on a glass microscope slide. Coverslips were placed on the samples and gently pressed to flatten the specimens.

2.9. Hematoxylin and Eosin Staining

PUV treated specimens were euthanized and fixed in 3.7% formaldehyde or Dent’s fixative as described in section 2.7 and 2.8 above. The samples were then dehydrated in ethanol for 10 minutes each, cleared in Xylene, and embedded in Paraplast Plus (McCormick Scientific LLC, St Louis, MO). The samples were sectioned at a thickness of 9μm, and stained with Harris Hematoxylin (Fisher Scientific, Hampton, NH), and Eosin, based on the established protocols (Humason, 1972; Wolfe and Henry, 2006).

2.10. Microscopy and image analysis

Samples were observed under a Zeiss Axioplan microscope, and imaged using a Axiocam 503-mono camera controlled by ZEN software (Carl Zeiss, Munich, Germany). The number of labeled nuclei were determined using Image J (U.S. National Institutes of Health, Bethesda, MD). To quantify the total number of nuclei (Hoechst labeled nuclei), p63 labeled nuclei, and cell death (TUNEL positive nuclei, and more highly condensed, fragmented Hoechst labeled pyknotic nuclei) three random areas (minimum area = 70μm × 70μm and maximum area = 345μm × 260μm (area of the entire 40X image)) were selected on PUV treated whole corneas, untreated controls, Psoralen only (without UV) controls, and UV only (without Psoralen) controls. For the half cornea PUV treated samples, and half cornea UV only treated controls, two random areas (as above) were selected on both the treated and untreated halves of these corneas. The number of nuclei were counted in these areas, and expressed as nuclei per unit area, to calculate the nuclear density for Hoechst, p63, TUNEL assays and pyknotic nuclei. Typically, the TUNEL positive nuclei were sporadically distributed across the cornea, so for all these cases the value was obtained by counting all the TUNEL positive nuclei within the cornea samples. In all cases, the total area of the cornea was calculated and the TUNEL positive nuclear density was then expressed as total TUNEL positive nuclei divided by total cornea area. To quantify the cells undergoing mitosis (H3S10 positive nuclei), all the labeled nuclei present in the cornea were counted.

To quantify pyknotic nuclei, Hoechst labeled speckles (HLS) smaller than approximately 3μm (about one third to one half of the usual size of a healthy nucleus) were considered as pyknotic nuclei. Closely spaced clusters of HLS were counted as one pyknotic nucleus. A distinct pyknotic nucleus was also scored when the distance between HLS was greater than approximately 3μm. Whenever the appearance of pyknotic nuclei was not well-defined (due to a greater degree of nuclea fragmentation) confirmation was obtained using the appearance of nuclear blisters in the corresponding DIC images, where multiple closely situated HLS, appearing in one nuclear blister, were counted as a single pyknotic nucleus. The Hoechst nuclear density was normalized to a standard area of 70μm × 70μm. Furthermore, the number of p63, H3S10, TUNEL positive, and pyknotic nuclei were divided by the total number of nuclei, to determine the number of respective positive nuclei per total number of nuclei present within the cornea or standard area. Expressing these numbers as a percentage of the total nuclei in the cornea allows us to control for the variation in cell density of the cornea and variations among individual specimens. Since there are very few mitotic nuclei, as compared to Hoechst labeled nuclei, mitosis was expressed as the number of mitotic figures per 1000 Hoechst nuclei. Pericorneal tissue, where pigmentation is present (Figure 1G), was excluded from analysis for all control samples. Likewise, in PUV treated samples where pericorneal tissue (skin cells/pigment cells) ultimately appeared over the edge of the eye to replace the cornea, only the area located directly over the eyeball was included in these analyses. For statistical analysis and error calculation each “N” is defined as the number of individual corneas analyzed for that particular set of conditions. Statistical analysis has been done by comparing standard error of the means in respective cases, and evaluated using the unpaired t-test.

3. Results

3.1. Characterization of corneal phenotypes after PUV treatment

We examined the effects of the PUV treatment on corneas to evaluate whether the phenotype recapitulated characteristics of LSCD seen in mammals, e.g., changes in corneal opacity, corneal neo-vascularization, and even the presence of pigment cells (melanocytes) in the cornea. Untreated control corneas never showed signs of SCD, and appeared completely transparent, and maintained a regular, intact, smooth, curved surface (Figure 2A, 2A’). Corneas that received PUV treatment started showing moderate effects, and appeared translucent by 3–7dpt (Figure 4B, Table S1 A). Dorsal views also show that the PUV treated corneas developed a bumpy appearance, with an irregular epithelial surface (Figure 2B, 2B’). A few blood vessels were also seen in the cornea, but were mostly restricted to the peripheral regions (Figure 4B, Table S1 A). Pigment cells started appearing in the cornea region by 7dpt, and by 15dpt the pigment cells were scattered over the entire cornea (Figure 2C, 2C’). At even later stages (25 days post treatment), pigment cells were still present throughout the cornea and blood vessels extended further inward into the central cornea region (Figure 2D’), both signs of a more severe SCD phenotype. None of the treated corneas returned to normal appearance and continued to show these phenotypes over the course of 32 days post treatment (described in more detail below).

Figure 2: Comparison of corneal phenotypes.

Phenotypes in untreated control corneas (A, A’), PUV treated whole corneas (B-D, B’-D’), PUV treated half corneas (E, F, E’, F’), Psoralen only controls (whole cornea) (G, G’), and UV only controls (whole cornea) (H, H’). Images were taken at various time points (dpt), as indicated. (A, A’) Control cornea is clear and pigment cells are restricted to the edges of the peripheral cornea. (B, B’, C, C’) Cornea after PUV treatment starts showing increased opacity and vascularization (white arrows). Pigment cells (white arrowheads) are also entering the cornea. (D, D’) Cornea is still opaque 25dpt after whole cornea PUV treatment. The normal corneal phenotype is lost, and pigment cells completely cover the cornea (white arrowheads). (E, F, E’, F’) Dotted blue lines indicate the boundary between the PUV treated half and the untreated (control) half. (E, E’) At 15dpt after PUV treatment, the treated half shows opacity and increased pigmentation (white arrowheads), whereas the untreated half is devoid of pigment cells and appears transparent. (F, F’) At 28dpt after PUV treatment on half of the cornea, the treated half has become clear of pigment cells and the normal corneal phenotype and transparency are restored. (G, G’, H, H’) Psoralen only treated corneas and UV only treated corneas, respectively, are clear and pigment cells are restricted to the edges of peripheral cornea. Green arrowheads in (A’, E’ to H’) show transparent cornea. Blue arrowheads point to wrinkled irregular epithelium, an, anterior side of the tadpole; dr, dorsal side of the tadpole; ps, posterior side of the tadpole; vr, ventral side of the tadpole. Scale bar in H’ equals 500μm.

A feature which is unique to aquatic vertebrates is a group of sense organs within the lateral line organ system (LLOS) that helps these animals navigate through water by detecting movements, vibrations, and pressure changes. These lateral lines are located at various positions along the body. Two of these, the supraorbital and infraorbital canals, extend around the eye (white arrows, Figure 3A, 3A’). In an untreated control cornea, these parts of the LLOS were restricted to the epidermis surrounding the cornea (Figure 3A, 3A’). In PUV treated corneas, the supraorbital and infraorbital canals became displaced towards the center of the cornea region that overlies the eye, together with pigment cells (Figure 3B, 3B’).

Figure 3: Severe phenotypes in whole cornea PUV treated cases.

PUV treated cases showing the more severe Stem Cell Deficiency (SCD) phenotypes. (A, A’) Control corneas depicting normal transparent cornea morphology. These corneas are free of vasculature, pigment cells, and the lateral line organs (white arrows) are restricted to the skin. (B, B’) PUV treated cases on the opposite sides of the control corneas shown in (A, A’) at 32 and 29 days post treatment, respectively, depicting the loss of the normal transparent corneal phenotype. The cornea is completely opaque and covered with epidermal tissue similar to that surrounding the eye. Skin cells, including pigment cells and the lateral line organs (white arrows), have invaded the corneal region. (C, C’) PUV treated cases at 15dpt, depicting vascularization with multiple blood vessels passing into the cornea (white arrowheads), an, anterior side of the tadpole; dr, dorsal side of the tadpole; ps, posterior side of the tadpole; vr, ventral side of the tadpole. Scale bar in C’ equals 200μm for (C, C’), and 665μm for (A, A’, B, B’).

Interestingly, over time, we also observed that PUV treated eyes would appear somewhat smaller in size compared to the untreated control eyes (Figure 2A, 2A’ vs. Figure 2C, 2C’). Measurements taken over time showed that PUV treated eyes showed continued growth that was delayed between 1–11dpt, compared to the untreated control eyes, before resuming similar growth rates to controls (Figure S1). We examined these eyes more closely in sections to determine if the smaller size might be caused by damage to the retina or the cells residing in the ciliary marginal zone. Histological examinations revealed no morphological differences between the untreated and the treated eyes at 15dpt and 30dpt, and these eye tissues appeared to be normal (Figure S2).

In order to examine how untreated cornea tissue may repair damage following PUV treatment, we restricted the treatments to the posterior half of the cornea. Superficially, the untreated halves showed a phenotype similar to untreated control (whole corneas); and changes in opacity or increases in pigmentation were not observed from 1–28dpt (Figure 2E, 2E’). On the other hand, the PUV treated half initially developed a phenotype similar to the PUV treated whole corneas, as described above (Figure 2E, 2E’). However, when the half-cornea treatment animals were followed through 28dpt, most of the treated posterior halves of the corneas appeared to restore their normal appearance and looked similar to the anterior untreated halves (Figure 2F, 2F’). Interestingly, some changes were initially observed in both the treated and untreated halves of these corneas in response to PUV treatments, which are described in greater detail, below.

3.2. The effect of PUV treatment on pigmentation

The cornea is typically devoid of pigment cells. To quantify the effects of PUV treatment on pigmentation in the cornea, changes in pigment cell counts were quantified. The natural pigment in these melanocytes allows for ready visualization of these cells and for tracing their movements, and provides one sign for the influx of cells from the epidermis surrounding the cornea.

The untreated control corneas showed a minimal number of pigment cells, with only 1–4 pigment cells in the standard 1mm² area for 1 to 22dpt time points, and an average of 6 pigment cells at the 32dpt time point, which may be related to the larger size of these eyes (Figure 4A). These few melanocytes were always observed close to the peripheral boundary of the cornea, and never extended into the central cornea region. The pigment cell counts in the PUV treated corneas were similar to the untreated control corneas at 1 to 5dpt, but started increasing at 7dpt. In contrast to the untreated control corneas, the pigment cell counts kept increasing over time (average of 53 pigment cells in the standard area) at 15dpt (Figure 4A). The highest number of pigment cell counts was observed at 32dpt (average of 141 pigment cells in the standard area), when the eyeball appears to be completely covered by epidermis (Figure 3B, 3B’). This sharp increase may also be influenced by the process of metamorphosis, which is accompanied by normal remodeling of the skin, head/skull, and retraction of the eyeball within the orbit (Figure 3B, 3B’).

In cases where PUV treatment was restricted to the posterior halves, the untreated control halves of these corneas showed minimal numbers of pigment cells between 1–28dpt (average of 1–4 pigment cells in a standard area). In the treated half of these corneas, the pigment cell counts were higher at 5dpt and continued to increase until 15dpt (average of 29 pigment cells in a standard area) (Figure 2E, 2E’, Figure 4B). After 15dpt, the number of pigment cells decreased at 22 and 28dpt (Figure 4B). The corneas didn’t become completely free of pigment cells over this time interval, but the number of pigment cells was greatly reduced (average of 5 pigment cells in a standard area) (Figure 2F, 2F’).

3.3. The effect of PUV treatment on corneal opacity

The untreated control corneas showed no signs of corneal opacity for 1 to 32dpt (Figure 4B, Table S1A). The PUV treated corneas did not show opacity at 1dpt, but started showing moderate opacity starting at 3dpt (Figure 4B, Table S1 A). As the PUV damage progressed, an increased number of specimens showed moderate opacity at 7dpt (Figure 4B, Table S1 A). The severity of opacity, as well as the percentages of specimens showing severe opacity, increased as the time points progressed, and 100% of the specimens exhibited some opacity at 32dpt (Figure 4B, Table S1A) (Figure 3B, 3B’). A few specimens examined at 11, 15, and 22dpt did not show signs of corneal opacity (Figure 4B, Table S1 A), which suggests that the accumulation of the PUV damage was delayed in those cases.

In cases that received PUV treatment only on the posterior halves, the untreated control halves of the corneas were clear and transparent, and showed no signs of opacity at any of the time points examined (Figure 4E, Table S1C). The treated posterior halves of the cornea initially did not show opacity at 1 and 3dpt, but started showing moderate corneal opacity starting at 5dpt (22% cases) (Figure 4E, Table S1C). The number of specimens that showed moderate to severe opacity increased from 5 to 11dpt, and peaked at 15dpt (57% moderate opacity, 43% severe opacity) (Figure 4E, Table S1C) (Figure 2E, 2E’). After 15dpt, the corneal opacity decreased at 22 and 28dpt. Some samples (40%) showed no signs of opacity at 22dpt (Figure 4E, Table S1C), and most cases were transparent by 28dpt (83%, Figure 4E, Table S1C) (Figure 2F, 2F’), indicating that most specimens underwent significant repair.

3.4. The effect of PUV treatment on vascularization

The untreated control corneas did not show signs of vascularization in most cases. In some specimens, slight vascularization was observed very close to the peripheral boundary of the cornea, but these vessels never extended into the central cornea region (Figure 4C, Table S1B). In PUV treated corneas, the number of specimens showing moderate vascularization started increasing at 3dpt, where either the supraorbital and/or the infraorbital artery would extend branches into the peripheral cornea (Figure 4C, Table S1B). The severity of vascularization also increased from 5 to 11dpt. Starting at 15dpt the ratio of specimens graded severe vs. moderate was higher, indicating increased severity of neo-vascularization at these time points (Figure 4C, Table S1B). In severe cases, multiple blood vessels branching from either the supraorbital and/or the infraorbital arteries could be seen extending into the cornea region (Figure 3C, 3C’). In a few cases, other arteries like the supratrochlear artery, that runs further away from the cornea was also seen to extend branches into the tissue over the eye. Only a few specimens of those examined at 15, 22, and 32dpt did not show signs of vascularization (Figure 4C, Table S1B).

In cases that received PUV treatment only on the posterior halves, some of the specimens, interestingly, showed some moderate vascularization in the untreated anterior halves of the corneas at 3 to 28dpt (Figure 4F, Table S1D). However, vascularization was never found to be severe in the untreated half, and blood vessels were restricted close to the peripheral boundary of those corneas. The treated posterior halves did not show any vascularization at 1dpt, but some specimens started showing moderate vascularization at 3dpt (Figure 4F, Table S1D). The number of specimens that showed vascularization increased as the time progressed from 3 to 15dpt, with 57% showing moderate and 29% showing more severe vascularization at 15dpt (Figure 4F, Table S1D) (Figure 2E’). The number of specimens exhibiting vascularization decreased at 22dpt. Finally, at 28dpt, 67% of the cases examined showed no signs of vascularization and 33% cases showed moderate vascularization (Figure 4F, Table S1D), (Figure 2F’).

3.5. The effect of PUV treatment on expression of Tcf7l2 (whole cornea treatments)

Tcf7l2 is a HMG box transcription factor, known to be a downstream effector in the canonical Wnt signaling pathway. Tcf7l2 has been implicated in maintaining human corneal epithelial stem cells and is localized to the cytoplasm during normal homeostasis, being translocated to the nucleus in response to wounding (Lu et al., 2012). Tcf7l2 has also been shown to be essential in long term maintenance and wound repair of skin epithelia (Nguyen et al., 2009). A recent study from our lab (Sonam et al., 2019) discovered that Tcf7l2 expression serves as a good marker for a subset of apical epithelial cells within the frog skin (Figure 5A–5F), which are not normally present in the cornea (Figure 5E’, 5F’).

Figure 5: Tcf7l2 immunostaining showing appearance of Tcf7l2 expressing skin cells in the cornea region following whole cornea PUV treatments.

(A–D) Normal skin tissue obtained from the flank region of a representative tadpole. (A) DIC image showing pigment cells, which are a characteristic of skin tissue. (B) Skin tissue shows positive immunostaining for Tcf7l2, which labels a subset of apical epithelial skin cells that are not normally present in the cornea. (C, D) are the corresponding Hoechst and merged images for (A, B), respectively. (E–H’) Control and PUV treated corneas obtained from the left and right sides of the same tadpole, respectively. (E, F) shows the peripheral epidermis region (skin that surrounds the cornea), where Tcf7l2 staining is normally observed. (E’, F’) shows the cornea region of the control (untreated) cornea, where no Tcf7l2 staining is normally observed. (G, H) is the peripheral skin region, and (G’, H’) is the cornea region overlying the eye and of a PUV treated cornea collected at 12dpt. The presence of Tcf7l2 expressing cells in the cornea region shows the presence of these skin cells. Note that these Tcf7l2 positive cells appear smaller and more compacted than those found in the surrounding skin. (F, F’, H, H’) are the corresponding merged (Tcf7l2+Hoechst) images of (E, E’, G, G’), respectively. Scale bar in D equals 100μm and applies to A-D and equals 40μm in H’ and applies to E–H’.

Expression of Tcf7l2 in the peripheral epidermis and the cornea were compared in both untreated controls and PUV treated corneas at 12 days post treatment (Figure 5E–5H, and 5E’–5H’, respectively). In contrast to the Tcf7l2 expression that is normally restricted to the peripheral epidermis (Figure 5E, 5F, 5E’, 5F’), Tcf7l2 expression in the PUV treated cornea is observed both in the peripheral epidermis and in cells extending into the central cornea region (Figure 5G, 5H, 5G’, 5H’). Interestingly, these Tcf7l2 positive cells in the cornea region and peripheral epidermis of PUV treated cases (Figure 5G, 5H, 5G’, 5H’) appear to be smaller and more compactly arranged when compared to the Tcf7l2 positive cells found in the peripheral epidermis of untreated control cases (Figure 5E, 5F).

3.6. The effect of PUV treatment on cell death via pyknosis

To examine cell death, the percentage of pyknotic nuclei were evaluated in the PUV treated corneas, and compared to the percentage of pyknotic nuclei in untreated control corneas. The control corneas exhibited a minimal level of cell death via pyknosis at all time points from 1 to 15dpt (Figure 6A, Figure S3A–S3C). On the other hand, the corneas that received PUV treatment showed significant increases in cell death at 1dpt (p<0.0001) and at 3dpt (p=0.0018) (Figure 6A, Figure S3A’–S3B’). At 5dpt, cell death levels declined to control levels and no significant increases in the percentages of pyknotic nuclei were observed for the 7, 11, and 15dpt time points, the last time points recorded, when compared to controls (Figure 6A, Figure S3C’).

Figure 6: Quantification of pyknotic nuclei and apoptotic cell death following PUV treatment on whole corneas and half corneas.

(A, B) Counts of pyknotic nuclei and TUNEL positive nuclei, respectively, in the untreated controls (whole cornea) (black line) and the PUV treated whole corneas (red line). (C, D) Counts of pyknotic nuclei and TUNEL positive nuclei, respectively, in the untreated (control) halves (blue line) vs. the PUV treated halves of the corneas (red line) of the same corneas. Control (whole cornea) from (A) has been included in (C) for convenience of comparison (black line). The observations have been taken at the time points indicated on the X-axis (dpt). The pyknotic nuclei and TUNEL positive nuclei are represented as a percentage of the total Hoechst nuclei. (A, C) Error bars represent standard error of the mean. (B, D) Boxes represent lower and upper quartiles; whiskers and outliers represent Tukey range. Statistical analysis has been done for untreated control corneas (black) vs. PUV treated corneas (red) in (A, B), and for untreated control halves (blue) vs. the PUV treated halves (red) of the corneas in (C, D). Outliers have been included in calculating p-values in (B, D). N = 9 for all time points in (A). N = 9–13 for time points in (B). N = 10 for all time points in (C). N = 10–15 for time points in (D). *** = p ≤ 0.001; ** = p ≤ 0.01; * = p ≤ 0.05.

In the corneas that received localized PUV treatment, the treated halves of the corneas showed a significant increase in the percentage of pyknotic nuclei at 1 and 3dpt (9.5%, p=0.0039 at 1dpt; 3.1%, p=0.0336 at 3dpt) (Figure 6C, Figure S6A’, S6B’), as compared to the untreated halves of the corneas, which showed a minimal percentage of pyknotic nuclei (Figure 6C, Figure S6A, S6B). The percentage of pyknotic nuclei in the treated halves of the cornea decreased to levels seen in the untreated halves starting at 5dpt (Figure 6C, Figure S6C, S6C’).

3.7. The effect of PUV treatment on cell death (TUNEL assays)

The effects of PUV treatment on cell death were also examined by TUNEL assays. In the control corneas, a baseline percentage of TUNEL positive cells was detected (average of 3% at 1dpt and 3.5% at 3dpt) (Figure 6B, Figure S4A column i, S4B column i). However, at 5dpt, an increase in the percentage of TUNEL positive cells was observed in some of the samples, and the individual samples showed variability in the percentage of cell death (minimum = ~0.1% to maximum = ~45.6%) (Figure 6B, Figure S4C column i). The percentage of cell death appeared to be similar at 7 and 11dpt, with a slightly lower variability among individual samples (Figure 6B). The percentage of cell death and the variability among individual specimens declined at 15 and 20dpt, before an increase was observed at 25 and 30dpt time points (Figure 6B). Increases in cell death at later stages are typically observed as animals approach metamorphosis (Ishizuya-Oka et al., 2010; Nakajima et al., 2005; Nieuwkoop and Faber, 1994).

When compared to control corneas, the PUV treated corneas exhibited a significant increase in the percentage of cell death at 1dpt (mean cell death = 7.6%, p=0.0198) (Figure 6B, Figure S4A’ column i). The percentage of cell death in PUV treated corneas appeared to be similar to control corneas at 3 and 5dpt (Figure 6B, Figure S4B’ column i, S4C column i). At 7dpt, an increase in the median percentage of cell death was observed in PUV treated corneas; however, the mean was not statistically different than the controls. The percentage of cell death decreased at 11–20dpt, and showed an increase at 25 and 30dpt, similar to control corneas (Figure 6B).

Corneas in which PUV treatment was restricted only to the posterior half also showed a significant increase in cell death in the treated half (7.9%, p=0.036) as compared to the untreated control half at 1dpt (Figure 6D, Figure S7A column i, S7A’ column i). The levels of cell death declined at 3dpt in both the treated vs. untreated sides (Figure 6D, Figure S7B column i, S7B’ column i). At 5dpt, a marked increase in the levels of cell death was observed in both the halves (Figure 6D, Figure S7C column i, S7C column i). From 7 to 15dpt, the levels of cell death declined in both the halves, before showing an increase at the 20 to 30dpt time points (Figure 6D, Figure S7D column i, S7D’ column i). Noticeable differences in cell death levels between the treated and the untreated halves of the corneas were only seen at 1dpt, while later time points (3 to 30dpt) showed no significant differences.

3.8. The effect of PUV treatment on the expression of the basal epithelial cell marker, p63

The transcription factor p63 is widely used as a marker for epithelial stem cells and TACs (Arpitha et al., 2008; Di lorio et al., 2005; Liu et al., 2018; Pellegrini et al., 2001). Furthermore, p63 has been shown to localize to the basal epithelium of the Xenopus cornea, which is where proliferative cells are localized (Perry et al., 2013). The p63 antibody detects the delta form of p63 (Perry et al., 2013; Tomimori et al., 2004). Here, the effects of PUV treatment on the expression of this marker were examined in the cornea (Di lorio et al., 2005; Perry et al., 2013; Sonam et al., 2019).

In the control corneas, the percentage of p63 labeled nuclei, (which is localized only to nuclei in the basal cells) remained fairly constant, ranging between 46.9% to 57.7% of all cells present between 1 to 30dpt time points (Figure 7A, Figure S5A column i – S5E column i). In PUV treated corneas, a diminished percentage of p63 positive nuclei was observed at 1dpt (~10% p63 positive nuclei, p<0.0001) (Figure 7A, Figure S5A’ column i). The percentage of p63 nuclei appeared to increase at 3dpt, and returned to control levels at 7dpt (Figure 7A, Figure S5B’ column i, S5C column i). The lower percentage of p63 labeled nuclei may be attributed to a greater initial effect on the proliferating population of cornea cells as expected from the PUV treatment. Compared to the controls, the percentage of p63 nuclei were still significantly lower at 3dpt (p<0.0001) and 5dpt (p=0.0295). From 11 to 30dpt, a slight decline was observed in the percentage of p63 positive cells in the PUV treated corneas. The percentage of p63 positive cells in PUV treated corneas was significantly lower than control corneas at 15 and 20dpt (Figure 7A, Figure S5D’ column i), but not at 25 and 30dpt, reflecting the higher variation among individual samples at these later time points (Figure 7A, Figure S5E’ column i).

Figure 7: Quantification of p63 positive nuclei, mitosis, and total cell density following PUV treatment on whole corneas and half corneas.

(A-C) Percentage of p63 positive nuclei, mitotic nuclei counts (H3S10), and total cell density, respectively, in the untreated controls (whole cornea) (black lines) vs. the PUV treated whole corneas (red lines). (A’–C’) Percentage of p63 positive nuclei, mitotic nuclei counts (H3S10), and total cell density in the untreated control half (blue lines) vs. the PUV treated half (red lines) of the cornea. The controls (whole cornea) from (A–C) have been included in (A’–C’), respectively, for convenience of comparison (black lines). The observations have been taken at the time points indicated on the X-axis, as days post treatment (dpt). Error bars indicate standard error of the mean. Statistical analysis has been done for untreated control corneas (black lines) vs. PUV treated corneas (red lines) in (A–C), and for untreated control halves (blue lines) vs. the PUV treated halves (red lines) of the corneas in (A’–C’). N = 7–10 for time points in (A). N = 10–15 for time points in (A’). N = 7–13 for time points in (B). N = 9–15 for time points in (B’). N = 9–13 for time points in (C). N = 10–15 for time points in (C’). *** = p ≤ 0.001; ** = p ≤ 0.01; * = p ≤ 0.05.

In cases where only the posterior half of the cornea received the PUV treatment, the percentage of p63 positive nuclei in the treated half was slightly lower when compared to the untreated half of the cornea at 1dpt (p=0.2754) (Figure 7 A’, Figure S8A column i, S8A’ column i). The percentage of p63 positive cells appeared to increase at 3dpt in both the treated halves and the untreated halves of the corneas (Figure 7 A’, Figure S8B column i, S8B’ column i). A decline in the percentage of p63 nuclei was observed at 5dpt in both the halves of the corneas, followed by an increase at 7dpt (Figure 7 A’, Figure S8C column i, S8C’ column i). As compared to the untreated controls (whole corneas) (Figure 7A), both the untreated and treated halves showed a lower percentage of p63 from 1 to 7dpt (Figure 7A’). At 11 and 15dpt, the percentage of p63 nuclei increased in both the untreated and treated halves (Figure 7 A’, Figure S8D column i, S8D’ column i). From 5 to 15dpt, the slight differences in the percentage of p63 positive cells between the treated and untreated halves were not statistically significant. Interestingly, at 20dpt, the percentage of p63 positive cells declined in both the halves, and the percentage of p63 positive cells in the treated half was significantly lower than the percentage of p63 positive cells in the untreated half (p=0.0085) (Figure 7A’). The percentage of p63 positive cells in the untreated half of the corneas remained consistent for the 25 and 30dpt time points, and was similar to the treated half of the corneas (Figure 7 A’, Figure S8E column i, S8E’ column i).

3.9. The effect of PUV treatment on mitosis in the cornea

Anti-phosphorylated-histone-H3S10 antibody, was used as a reliable marker to detect mitotic cells (Hans and Dimitrov, 2001; Perry et al., 2010; Thomas and Henry, 2014; Walter et al., 2008). The untreated control corneas showed an increase in mitosis from 1 to 5dpt (Figure 7B, Figure S5A column ii – S5C column ii). The mitotic nuclei counts then decreased at 7 and 11dpt, before increasing at 15dpt (Figure 7B, Figure S5D column ii). These represent the basal mitosis levels of the CESCs and TACs in the untreated control cornea, over this period of larval development. Additionally, an increase in mitosis was observed at 25 and 30dpt in the untreated control corneas (Figure 7B, Figure S5E column ii). These fluctuations are an indicator of natural variations associated with differences in growth rates and ages/stages of larval development.

In contrast to the untreated control corneas, the mitotic nuclei counts were significantly decreased at 1dpt in PUV treated corneas (Figure 7B, Figure S5A’ column ii). This decrease may be correlated with changes in cell death and the likelihood that cells become arrested in response to the DNA crosslinking from the PUV treatment (Cuddihy and O’Connell, 2003). Subsequently, a significant increase was observed in mitosis at 3dpt in the PUV treated corneas as compared to the control corneas (Figure 7B, Figure S5B’ column ii), suggesting that the surviving CESCs/TACs are stimulated to undergo mitosis and may be responding to mitigate the damage caused by PUV treatment. In contrast to the untreated control corneas, the mitotic counts continued to remain higher at 5, 7, and 11dpt in the PUV treated corneas (Figure 7B). At 15dpt, the mitotic counts declined to levels observed in untreated controls (Figure 7B, Figure S5D’ column ii). The mitotic counts subsequently dropped further at 20dpt in the PUV treated corneas and remained somewhat lower than the mitotic counts in untreated control corneas at 25 and 30dpt, respectively (Figure 7B, Figure S5E’ column ii).

In cases where the PUV treatment was restricted to the posterior half of the cornea (Figure 7B’), the treated half of the cornea initially showed a statistically significant decrease in mitosis compared to the untreated (control) half at 1dpt (Figure 7B’, Figure S8A column ii, S8A’ column ii). Both the treated and untreated halves of the cornea showed a sharp increase in mitosis at 3 and 5dpt (Figure 7B’, Figure S8B column ii, S8C column ii, S8B’ column ii, S8C’ column ii). The mitotic count in both the treated half and the untreated half stayed elevated at 7 and 11dpt, before declining at 15 and 20dpt (Figure 7B’, Figure S8D column ii, S8D’ column ii). Furthermore, both the treated and untreated halves showed increased mitosis at 25 and 30dpt. (Figure 7B’, Figure S8E column ii, S8E’ column ii). While changes in mitosis were observed, the differences between the treated and untreated halves for time points from 3 to 30dpt were not found to be statistically significant.

3.10. The effect of PUV treatment on total cell density in the cornea

In control corneas, the total cell density remained fairly consistent for the time points from 1 to 15dpt, except for a slight decrease seen at 3dpt (Figure 7C, Figure S5A column iii – S5D column iii). During the time points from 20 to 30dpt, an increase was observed in the total cell density (Figure 7C, Figure S5E column iii). At these time points the animals are at stage 58–66, which corresponds to the developmental stages approaching metamorphic climax (Nieuwkoop and Faber, 1994), where the corneal epithelial cells appear more condensed, and the cornea epithelium is known to thicken (Hu et al., 2013) and finally mature into a multilayered structure.

A significant decrease in the total cell density was observed in the PUV treated cases at 1dpt (p=0.0094) (Figure 7C, Figure S5A’ column iii), which is related to the increased levels of cell death at this time point. The cell density in PUV treated corneas showed an increase at 3dpt. From 5 to 20dpt, the cell density in PUV treated corneas continued to increase and became significantly higher when compared to the control corneas (Figure 7C, Figure S5C’ column iii, S5D’ column iii). This is likely due to a combination of increased mitosis of any surviving cornea cells, and primarily the invasion of skin cells from the peripheral epidermis. As mentioned earlier, we noted a difference in the morphology of Tcf7l2 positive cells within the PUV treated cornea (Figure 5G’, 5H’) vs. that normally seen in the peripheral epidermis (Figure 5E, 5F). This appears to be due to differences in the thickness of these tissues, because the tissue that replaces the cornea after the PUV treatment is thicker and these cells become more columnar in shape. The observed increase in total cell density in PUV treated corneas supports this conclusion. Although the total cell density in PUV treated corneas at 25 and 30dpt was higher than the control corneas, this difference was not statistically significant, which reflects the higher variation found among individual samples at these later time points (Figure 7C).

For those cases in which half of the cornea was treated, no statistically significant differences were observed in the cell density of the untreated halves vs. the treated halves (Figure 7C’, Figure S8A’ column iii – S8E’ column iii). The cell density in both the untreated and the treated halves showed a slight decrease at 3dpt, followed by an increase from 5 to 15dpt (Figure 7C’, Figure S8A column iii – S8D column iii). Cell density in both the halves of the cornea declined at 20 and 25dpt before an increase in cell density was observed at 30dpt (Figure 7C’, Figure S8E column iii). Interestingly, the cell density in both the untreated and the treated halves of these corneas appeared to follow a similar trend as compared to the cases that received PUV treatment on the whole corneas (Figure 7C).

3.11. The effects of Psoralen only and UV only treated controls

The effects of treatment with Psoralen without UV exposure (Psoralen only controls), and UV without Psoralen (UV only controls) were also examined. Both the whole cornea Psoralen only controls (Figure 2G, 2G’) and the UV only controls (Figure 2H, 2H’) showed no signs of SCD.

A minimal number of pigment cells were observed in the contralateral right (untreated) corneas, with an average of ~1–3 pigment cells in standard area (Figure 8A). Only slightly elevated levels of pigmentation were observed in the Psoralen only treated controls (average of ~2–6 pigment cells in standard area) from 1–15dpt. A significant increase in the Psoralen only treated controls was only observed at 7dpt time point. UV only controls showed a significant increase in pigment cell counts (average of ~3–7 pigment cells), as compared to their contralateral right (untreated) corneas (average of ~1–4 pigment cells) at 5 to 15dpt time points (Figure 8B). However, the pigmentation in the UV only treated corneas was still minimal and was restricted to the peripheral edge of the cornea. Pigment cells were never seen in the central cornea region.

Figure 8: Quantification of pigmentation in Psoralen only (whole cornea), UV only (whole cornea), and UV only (half cornea) controls.

(A) The contralateral control corneas (black line) show a minimal level of pigmentation. Only slightly elevated levels of pigmentation were observed in Psoralen only whole cornea treatment controls (blue line). A significant increase was only observed at 7dpt. (B) When compared to the contralateral untreated corneas (black line), UV only treatment on whole corneas (red line) appears to cause a significant increase in the levels of pigmentation at 5, 7, and 15dpt. However, pigmentation was still minimal in the UV only treated controls at these time points. (C) In the cases where only half of the cornea received UV treatment, the treated halves (red line) show minimal increases in pigmentation at 1–7dpt. Error bars indicate standard error of the mean. Statistical analysis has been done for control cornea vs. Psoralen only treated (A), control cornea vs. UV only treated (B), and untreated halves vs. the UV only treated halves of the corneas (C). N = 4 for all time points in (A-C). *** = p ≤ 0.001; ** = p ≤ 0.01; * = p ≤ 0.05.

Both the Psoralen only controls and the UV only controls did not show any signs of opacity (Figure 9A, 9B, Table S2A, S2C). In Psoralen only controls and UV only controls slight vascularization was observed very close to the peripheral boundary of the cornea in some specimens, but these vessels never extended into the central cornea region (Figure 9D, 9E, Table S2B, S2D).

Figure 9: Opacity and vascularization in Psoralen only (whole cornea), UV only (whole cornea), and UV only (half cornea) controls.

Stacked column graphs of corneal opacity and vascularization are shown. For each time point examined, the analysis has been done for the Psoralen only treated control corneas vs. their contralateral corneas (A, D), respectively, the UV only treated control corneas vs. their contralateral corneas (B, E), respectively, and the untreated half vs. the treated half of the UV only half treated control corneas (C, F), respectively. None of the controls showed any signs of opacity at any of the time points examined (silver and blue columns in A, B, E). Whereas a majority of control specimens showed no signs of vascularization (silver and blue columns in D, E, F), some control specimens showed slight vascularization very close to the peripheral boundary of the cornea (gray and orange columns in D, E, F), however, these vessels never extended into the central cornea region.

Both the Psoralen only controls and the UV only controls showed minimal levels of pyknotic cell death that were similar to the untreated control corneas (Figure 10A). This verifies that Psoralen or comparable doses of UV alone do not trigger a significant increase in cell death.

Figure 10: Quantification of pyknotic nuclei and apoptotic cell death in Psoralen only (whole cornea), UV only (whole cornea), and UV only (half cornea) controls.

(A) Similar to the untreated control corneas (black line), both the Psoralen only controls (blue line) and the UV only controls (red line) showed minimal levels of pyknotic cell death at all time points examined. (B) Compared to the untreated control corneas (black line), the average percentage of TUNEL positive nuclei was lower in both the Psoralen only (blue line) and UV only (red line) controls at all time points. (C, D) At each time point, both the untreated halves (blue lines) and the UV only treated halves (red lines) of the corneas showed similar minimal levels of pyknosis (C), and apoptosis (D). The untreated controls (whole cornea) from (A, B) have been included in (C, D), respectively, for convenience of comparison. Error bars represent standard error of the mean. (B, D) Boxes represent lower and upper quartiles; whiskers and outliers represent Tukey range. Statistical analysis has been done for control (whole cornea) vs. Psoralen only treated, and control (whole cornea) vs. UV only treated in (A, B), and untreated halves vs. the UV only treated halves of the corneas in (C, D). Outliers have been included in calculating p-values in (B, D). N = 6–7 for Psoralen only, UV only, and UV only half treated corneas at all time points in (A-D). N = 9 for control (whole cornea) for all time points in (A–D). * = p ≤ 0.05 (displayed for UV only).

TUNEL assays showed variations among individual specimens. However, many time points showed no significant differences between the Psoralen only controls vs. untreated controls, and UV only controls vs. untreated controls. Significantly lower levels of cell death, as compared to untreated controls, were detected for: Psoralen only controls at 5dpt (p = 0.0486) and 15dpt (p = 0.0415); and for UV only controls at 15dpt (p =0.0281) (Figure 10B).

No significant differences were observed in the levels of mitosis and the percentage of p63 positive cells in both the Psoralen only controls and the UV only controls, which were similar to those seen in the untreated control corneas (Figure 11 A, 11B). Changes in the total cell density in both the Psoralen only controls and the UV only controls followed similar trends, as compared to the untreated controls (whole corneas) (Figure 11C). However, both the Psoralen only controls and the UV only controls showed a higher total cell density as compared to the untreated controls, which was significant at 7 to 15dpt for the Psoralen only controls, and at 1 to 15dpt for the UV only controls.

Figure 11: Quantification of p63 positive nuclei, mitosis, and total cell density in Psoralen only (whole cornea), UV only (whole cornea), and UV only (half cornea) controls.

At each time point observed, the percentage of p63 positive nuclei (A), and mitotic nuclei counts (B) were similar among the untreated controls (black line), Psoralen only controls (blue line), and UV only controls (red line), respectively. (C) Compared to the untreated controls (black line), the total cell density is significantly higher at 7 and 15dpt in Psoralen only controls (blue line), and significantly higher at all time points in UV only controls (red line). Percentage of p63 positive nuclei (A’), mitotic nuclei counts (B’), and total cell density (C’) in both the untreated (blue line) and treated (red line) halves of the UV only treated half cornea controls show similar levels, respectively. The controls (whole cornea) from (A-C) have been included in (A’–C’), respectively, for convenience of comparison. Error bars indicate standard error of the mean. Statistical analysis has been done for controls (whole cornea) vs. Psoralen only treated, and controls (whole cornea) vs. UV only treated in (A–C), and untreated halves vs. the UV only treated halves of the corneas in (A’–C’). N = 6–7 for Psoralen only treated, UV only treated, and UV only half treated at all time points in (A–C, A’–C’). N = 9 for controls (whole cornea) for all time points in (A, C, A’, C’). N = 7–9 for controls (whole cornea) for time points in (B, B’). *** = p ≤ 0.001; ** = p ≤ 0.01; * = p ≤ 0.05 (for UV only vs. controls (whole cornea)). # = p ≤ 0.05 (for Psoralen only vs. controls (whole cornea)).

3.12. The effects of aluminum foil mask and localized UV only control treatments

To test whether localized UV irradiation alone, and the presence of the aluminum foil mask could be affecting the cornea, localized UV treatments were prefomred without Psoralen on half of the cornea (UV only half controls). The aluminum foil mask was placed on the anterior half of the corneas, and the posterior half of the corneas was exposed to UVA, for the same duration of 75 seconds as in the PUV half cornea treated cases.

Both the UV only treated and the untreated halves of the corneas had a minimal number of pigment cells, with an average of ~1–6 pigment cells in the standard area for all time points from 1 to 15dpt (Figure 8C). The slight increase in the number of pigment cells in the treated halves when compared to the untreated halves was only found to be significant at 3dpt (p = 0.037). Changes in corneal opacity were not observed in any of the specimens for both the treated and the untreated halves from 1 to 15dpt (Figure 9C, Table S2E). Both halves also did not show vascularization in most cases. However, in some specimens among both the halves, slight vascularization was observed very close to the peripheral edge of the cornea, but these vessels never extended into the central cornea region (Figure 9F, Table S2F).

Both the treated and the untreated halves of the corneas showed very minimal pyknotic cell death throughout the time points examined (Figure 10C), and no significant differences were observed between the UV treated half vs. the aluminum foil covered untreated half of the cornea. TUNEL assay results were varied among individual specimens, but the mean levels of cell death in the untreated half of the corneas did not show any significant differences when compared to the treated half of the corneas (Figure 10D). Additionally, the mitosis levels, percentage of p63 positive cells, and the total cell density showed similar levels in the treated vs. untreated halves of the corneas and no significant differences were observed between them (Figure 11A’–11C’). These findings reveal that the placement of the aluminum foil mask does not trigger significant changes in the cornea.

4. Discussion

Though remarkable therapies have been developed to treat LSCD using epithelial stem cell transplants (Atallah et al., 2016; Galindo et al., 2017; Grueterich et al., 2002; Haagdorens et al., 2016; Holland 2015; Koizumi et al., 2001; Le and Deng, 2019; Ljubimov and Saghizadeh, 2015; Rama et al., 2010; Saghizadeh et al., 2017; Sasamoto et al., 2018; Shortt et al., 2007b; Utheim et al., 2018; Yazdani et al., 2019), our understanding of LSCD pathology, and the specific behavior of CESCs, is still rather limited. Convenient vertebrate models to study stem cell deficiency in the cornea will help us answer these questions. In this study, we have established a new frog model to study CESCs. This model recapitulates the hallmarks of LSCD in humans and other vertebrates. Combined with significant experimental advantages afforded by the Xenopus model system, PUV treatments provide a robust, versatile, and tractable system to readily prepare animals to study SCD, with a rapid, single treatment. Furthermore, the combination of highly localized application of Psoralen and restricted UV irradiation using the microscope pinhole iris and aluminum foil masks, allows one to target specific regions of the cornea to create localized damage. This allows us to further study the response of the cells in the undamaged region of the cornea. Additionally, we have characterized various cellular changes associated with these PUV treatments, such as cell death, mitosis, total cell density, and presence of Tcf7l2 expressing subset of apical skin cells and p63 positive basal epithelial cells.

Changes in Cornea Morphology/Histology

The cornea is normally avascular owing to its anti-angiogenic properties (Azar, 2006) and a corneal epithelial barrier is thought to separate the epithelial cells of the cornea and conjunctiva (Kubilus et al., 2017). When the CESCs/TACs are depleted in the case of LSCD, structural mechanisms and cell signals that maintain this boundary are disrupted (Batlle and Wilkinson, 2012; Dahmann et al., 2011). As a result, blood vessels and cells of the conjunctiva begin to appear in the cornea region (Fini and Stramer, 2005) (Dua, 1998). Related events are observed in the case of Pterygium (Liu et al., 2013).

Likewise, following PUV treatments in the frog, we observed branches of the supraorbital and infraorbital arteries had entered the cornea region, (Figure 4B, 4C, Table S1A, S1B) (Amitai-Lange et al., 2015; Kadar et al., 2013; Le et al., 2018) (Figure 2B, 2C, 2D’, 2E’, 3C, 3C’). The larval frog eye does not possess an obvious conjunctival epithelium, rather the surrounding skin directly abuts the outer edge of the cornea. Following PUV treatments we found that skin cells ultimately invade the cornea region. This includes the pigmented melanophores. In fact, studies conducted in rabbits (Mann, 1944; Michaelson, 1952; Wolosin et al., 2000), guinea pigs (Cafaro et al., 2009; Davanger & Evensen 1971; Henkind, 1969), and also humans (Cowan, 1963; Davanger & Evensen, 1971; Mann, 1944; Wolosin et al., 2000) reported that pigment cell migration can accompany LSCD. In the present study, pigment cells started appearing in the cornea region seven days following PUV treatments, and the number of these cells was found to increase over time (Figure 4A). Likewise, we observed that cells of the adjacent lateral line organ system enter the cornea region following PUV treatments (Figure 3B, 3B’). Finally, the appearance of Tcf7l2 positive cells in the PUV treated corneas confirmed that apical skin cells had also invaded the cornea in response to changes caused by PUV treatments (Figure 5G’, 5H’). Together, these observations suggest that the surrounding epidermis eventually replaces the cornea tissue.

Changes in the Growth of the Eye

Although the eye continues to grow larger, its growth is initially delayed following PUV treatment compared to the contralateral, untreated control eyes (Figure S1). On the other hand, we did not observe any abnormal histological changes in the retina or the ciliary marginal zones (Figure S2). One possibility is that the cornea normally sends signals to deeper eye tissues to regulate their growth, and these signals may have been disrupted as a result of the PUV treatments slowing eye growth. In fact, it is known that pre-lens, placodal ectoderm (from which both the lens and cornea develop) sends signals to the optic vesicle to control its normal development (Hyer et al., 2003).

Changes in Cell Death

A significant amount of cell death is seen very early, within 1–3dpt. Most of this appears to result from pyknosis, as a primary mechanism of cell death, rather than apoptosis (Figure 6A). The stimuli that causes apoptosis can also induce necroptosis (a programmed form of necrosis) and other forms of cell death can also occur independently of apoptosis (Tait et al., 2014). While TUNEL detects both apoptosis and necrosis, apoptosis is detected more generally (Gold et al., 1994), and necrotic cells can go undetected by TUNEL assays. Interestingly a wound healing study conducted in chicken corneas detected necrotic cell death in the absence of delayed cell death or apoptosis (Ritchey et al., 2011).

Beyond the first day, we did not observe any significant differences in percentage of TUNEL positive cells in the PUV treated corneas vs. the untreated control corneas (3dpt to 30dpt), and the results of TUNEL assays showed variations in the rate of cell death among the samples (Figure 6B). The short window of detection of DNA fragmentation by TUNEL (typically 1–3 hours from initiation to termination) (Gavrieli et al., 1992), and natural variations among specimens developing at different rates, may have contributed to these variations.

TUNEL positive cells in the PUV treated corneas were most pronounced around 5dpt-7dpt, but were not significantly higher than the untreated control corneas (Figure 6B). This was surprising because the inter-strand crosslinks caused by the PUV treatment might be expected to cause a delayed onset of apoptotic cell death in moderately damaged surviving cells, as they attempt to undergo proliferation. Such delayed responses to wounding, and delayed loss of CESCs have been previously reported in rabbits (Kadar et al., 2011).

When the tadpoles are undergoing metamorphosis around 25dpt-30dpt, another increase in apoptosis was observed (Figure 6B). Apoptosis is commonly seen during metamorphosis when tissue remodeling occurs (Ishizuya-Oka et al., 2010; Nakajima et al., 2005). Consistent with our results, increases in epithelial cell death have previously been reported in Xenopus corneas around metamorphosis (Hu et al., 2013). The results suggest that apoptotic cell death does not appear to be the most significant driver of cell death following PUV treatment.

Changes in p63 Expression

The transcription factor p63, has been widely used as a marker for proliferative epithelial cells (stem cells and TACs) (Arpitha et al., 2008; Di lorio et al., 2005; Liu et al., 2018; Pellegrini et al., 2001). It has also been used to distinguish stem cells for transplants in patients with LSCD (Rama et al., 2010). Similar to humans and mice (Collinson et al., 2002; Dua et al., 2003), p63 antibody labeling is detected throughout the nuclei of all basal cells of the cornea in both larval and adult Xenopus (Perry et al., 2013; Sonam et al., 2019).

PUV treatment resulted in a marked decrease in the fraction of p63 labeled cells in the corneal epithelium at 1dpt (Figure 7A). This agrees with the results from other studies that reported a decreased basal cell density in the cornea and limbus in patients diagnosed with LSCD (Chan et al., 2015), and in the rabbit model of LSCD (Galindo et al., 2017). A combination of enhanced mitosis in any surviving cells and the influx of cells from the peripheral epidermis likely contributes to the observed subsequent increase in the percentage of p63 labeled cells in the PUV treated corneas at 3dpt-7dpt. However, since p63 labels all basal cells in both the epidermis and the cornea (Li et al., 2015; Mills et al., 1999; Perry et al., 2013; Yang et al., 1999), the contribution of the p63 cells from epidermal stem cells/TACs in the skin, vs. those remaining in the cornea, cannot be distinguished. So far, there is no marker that can differentiate the stem cells/TACs of the cornea from those of the epidermis in Xenopus (Sonam et al., 2019).

Changes in Mitosis

Phospho-histone H3S10 is a widely used marker for mitosis (Thomas and Henry, 2014; Walter et al., 2008). The PUV treatment significantly diminished mitosis by 1dpt, suggesting that PUV treatment was capable of affecting the proliferation within the cornea epithelium. However, the increased mitosis seen at 3dpt to 11dpt (Figure 7B), suggests that a subset of cells survived PUV treatments. This population of cells ultimately appears to respond to the PUV damage by undergoing increased levels of mitosis, possibly compensate for the damage. Our results appear similar to those showing that CESCs/TACs can be stimulated to undergo increased mitosis in response to other forms of wounding (Richardson et al., 2018; Sagga et al., 2018). The increased mitoses, however, did not appear to mitigate the PUV damage to restore cornea transparency. Note that mitosis of invading epidermal stem cells or epidermal TACs that repopulate the cornea region likely also contribute to the observed increases in mitosis. Mitotic counts were lower in the PUV treated corneas at later time points (15dpt to 30dpt) (Figure 7B), presumably because the tissue had been converted to skin.

Repair Occurs Following Localized PUV Treatments

We showed that localized PUV treatments restricted to half of the cornea, resulted in a transient response on the treated side, and subsequently this damage is restored, most likely from undamaged cells on the untreated side. Compared to whole cornea treatments, the severity of PUV damage was somewhat lower in half-treated corneas. The appearance of pigment cells in the treated half of the cornea region was also delayed. While the number of pigment cells continued to increase following whole cornea treatments, this number eventually returned to levels closer to those seen in the untreated (control) halves of the corneas (Figure 4B). In half cornea treatments, fewer samples showed severe opacity, as compared to the whole cornea treatments. Changes in opacity also appeared at later time points, as compared to respective time points when PUV treatment was performed on the whole cornea. Likewise, neovascularization was observed in fewer specimens at both earlier and later time points, as compared to the whole cornea treatments. Ultimately, many of these half-treated corneas regained a normal appearance. Following repair, although the anterior and posterior halves of the cornea looked similar in the lateral views (Figure 2F), closer observation from the dorsal side (Figure 2F’) revealed that the treated half did not regain the same curved profile of control corneas in some cases. Although this could represent a permanent change in cornea morphology, it may mean that full restoration of the normal structural phenotype may take longer than 28 days.

We have shown that the damage caused by PUV treatment and the overt phenotypic changes observed in whole treated corneas were mainly restricted to the treated half of the cornea. Interestingly, however, the untreated half was found to exhibit some cellular changes. This includes initial signs of some moderate neovascularization, increases in the levels of mitosis, increases in cell density, and changes in the percentage of p63 positive cells (Figure 7A’–C’). These latter changes are suggestive of a compensatory response that may be aiding repair of the treated side (such as the increased levels of mitosis). Ultimately, these numbers return back to levels seen in control corneas. The initial changes observed in the untreated half could also be the result of a more generalized inflammatory response within the eye that resulted from PUV treatments and the damage created on the treated side. This could be related to the release of interleukins and/or cytokines associated with immune responses accompanied by the arrival of various immune cells such as macrophages, etc., that causes the untreated half to undergo these changes. The initial signs of some moderate neo-vascularization on the untreated side seem to support these arguments.

Variations in the Response to PUV treatments

While the end results of PUV treatments appear to be similar, the rate of advancement of LSCD and appearance of the skin phenotype differs among the treated cases. The rate of cornea restoration was also different in half-treated specimens. These variations are likely due to the fact that these animals vary somewhat in overall size and developmental stage, and do not develop or grow at the same rate, which is subject to natural variation and competition for food when animals are cultured in groups (for example there is always a great deal of variation in overall animal size in lab reared specimens). Variability could also arise from the specimens being derived from different clutches/parents. Human cases of CESD also do not follow the same trajectories (Haagdorens et al., 2016; Shortt et al., 2014).

5. Conclusion and future applications

We have established a model of SCD that appears to recapitulate many of the hallmarks of human LSCD. Here we have also utilized this model to study the response of corneal cells to damage and repair. This system provides us with an excellent platform to address future questions related to the behavior of CESCs/TACs and their mechanism of repair. How do the proliferating cells migrate to heal SCD and where do they arise? How many CESCs/TACs are needed to participate in an effective healing process? The percentage of CESCs/TACs required to be transplanted into the LSCD patients for a successful outcome has been a crucial question for therapeutic treatments (Pellegrini et al., 2013). Future experiments can be done to determine the minimum percentage of untreated/undamaged tissue that is required to restore the cornea. Furthermore, experiments involving lineage tracing of the CESCs/TACs in our SCD model will provide us with a unique opportunity to answer many questions (Amitai-Lange et al., 2015; Majo et al., 2008; Richardson et al., 2018), and to tease apart the exact mechanisms by which cornea damage may be restored by CESCs/TACs. Studies using this PUV model can be combined with pharmacological or genetic tools to investigate the roles of specific signaling pathways during repair. One can also use this system to examine factors that may accelerate and enhance the contribution of available CESCs/TACs to restore the damaged tissue. Hence, we anticipate that this clinically relevant model will facilitate studies of stem cell deficiency and recovery, and will be useful in discovering therapeutics approaches to treat disease and injury of the cornea.

Supplementary Material

1. Supplementary Figure 1 (Figure S1): Quantification of cornea area following PUV treatment on whole corneas.

(supplements data in Figure 2A–2D, 2A’–2D’)

Control corneas that received no PUV treatment show a steady increase in cornea area (black line). Corneas that received PUV treatment (red line) showed continued growth that was initially delayed compared to the control cornea. Error bars represent standard error of the mean. Statistical analysis has been done for untreated control corneas vs. PUV treated corneas. N = 7–12 for the time points, except for control cornea at 1dpt (N = 4). *** = p-value ≤ 0.0003. ** = p-value ≤ 0.005; * = p value ≤ 0.05.

8. Supplementary Figure 8 (Figure S8): Percentage of p63 positive nuclei, mitotic nuclei, and total cell density following PUV treatment on half corneas.

(supplements the data in Figure 7A’–7C’)

Representative images of p63 positive nuclei are shown in the green fluorescence channel (Column i), mitotic nuclei (H3S10) are shown in the red fluorescence channel (Column ii), alongside corresponding Hoechst counterstain in the blue (UV) fluorescence channel (Column iii), for the time points indicated as days post treatment (dpt). (A–E) and (A’–E’) correspond to 1, 3, 5, 15, 25dpt for untreated control half of the corneas and the PUV treated half of the corneas, as indicated. PUV treatment reduced the percentage of p63 nuclei at 1dpt in both the halves, which recovered at 15 and 25dpt. The treated half of the cornea initially showed a decrease in mitosis compared to the untreated half at 1dpt. Similar levels of mitosis were observed between the treated and untreated halves for 3, 5, 15, 25dpt, where both the halves showed an increase in mitosis at 3, 5, 25dpt. The total cell density in the treated half of the cornea, appeared similar to the untreated halves of these corneas for the respective time points, and showed an increase starting at 5dpt. Scale bar in E’ column iii is 50μm.

2. Supplementary Figure 2 (Figure S2): Histological examination of eye tissue morphology.

(supplements data in Figure 2A, 2C, 2D, 2A’, 2C’, 2D’)

Representative images of Hematoxylin and Eosin stained eye cross sections are shown (Column i), along with the higher magnification images of their retina (Column iii) and ciliary marginal zones (Column ii and Column iv). (A, B) The untreated control and the PUV treated corneas, respectively, at 15 days post treatment (dpt). (C, D) The untreated control and the PUV treated cornea at 30dpt. No differences were observed in overall eye morphology between the untreated control and the PUV treated corneas, which includes the tissues of the retina and ciliary marginal zones. Scale bar in D column iv equals 50μm for images in Column ii - Column iv, and 200μm for images in Column i.

3. Supplementary Figure 3 (Figure S3): Cell death analysis via pyknotic nuclei assessment following PUV treatment on whole corneas.

(supplements the data in Figure 6A)

Representative images showing Hoechst staining for 1 to 5 days post treatment (dpt) as labeled. (A–C) and (A’–C’) correspond to 1–5dpt for untreated control corneas and PUV treated corneas, as indicated. (A–C) Control corneas show only a minimal number of pyknotic nuclei. (A’–C’) An increase in the number of pyknotic nuclei is observed at 1 and 3dpt in PUV treated corneas, after which the number of pyknotic nuclei is similar to control corneas at 5dpt. Red arrowheads indicate representative pyknotic nuclei. White dotted circles enclose single fragmented pyknotic nuclei. Scale bar in C is 50μm.

4. Supplementary Figure 4 (Figure S4): TUNEL assay for cell death analyses following PUV treatment on whole corneas.

(supplements the data in Figure 6B)

Representative brightfield images of TUNEL positive nuclei are shown (Column i), alongside corresponding Hoechst images (Column ii), for the time points indicated as days post treatment (dpt). (A–D) and (A’–D’) correspond to 1, 3, 5, 15dpt for untreated control corneas and PUV treated corneas, as indicated. A significant increase in apoptosis levels in PUV treated corneas, as compared to control corneas was only observed at 1dpt. No significant differences were observed at other time points. Note presence of invading pigment cells in the PUV treated case shown in (D’). Red arrowheads indicate representative TUNEL positive nuclei and black arrows indicate pigment cells, which are larger and have a distinct color and irregular shape. Scale bar in D’ column ii is 50μm.

5. Supplementary Figure 5 (Figure S5): Percentage of p63 positive nuclei, mitotic nuclei, and total cell density following PUV treatment on whole corneas.

(supplements the data in Figure 7A–7C)

Representative images of p63 positive nuclei are shown in the green fluorescence channel (Column i), mitotic nuclei (H3S10) are shown in the red fluorescence channel (Column ii), alongside corresponding Hoechst counterstain in the blue (UV) fluorescence channel (Column iii), for the time points indicated as days post treatment (dpt). (A–E) and (A’–E’) correspond to 1, 3, 5, 15, 25dpt for untreated control corneas and PUV treated corneas, as indicated. Percentage of p63 remains fairly consistent in untreated control corneas. PUV treatment reduced the percentage of p63 nuclei at 1dpt, and the nuclei appear damaged. The mitotic nuclei counts were decreased at 1dpt in PUV treated corneas, while an increase was observed in mitosis at 3 and 5dpt in the PUV treated corneas. After the initial increase, the mitotic counts declined at 15dpt to control levels, and remained lower than the mitotic counts in untreated control corneas at 25dpt. Total cell density remains fairly consistent from 1 to 15dpt in untreated control corneas, and an increase in the total cell density was observed at 25dpt. Total cell density decreased at 1dpt in PUV treated corneas, while increased total cell density was observed at 3, 5, 15, 25dpt. Scale bar in E’ column iii is 50μm.

6. Supplementary Figure 6 (Figure S6): Cell death analyses of pyknotic nuclei following PUV treatment on half corneas.

(supplements the data in Figure 6C)

Representative images showing Hoechst staining for 1 to 5 days post treatment (dpt). (A–C) and (A’–C’) correspond to 1–5dpt for untreated control halves and PUV treated halves of the corneas, as labeled. (A–C) Untreated control halves shows a minimal number of pyknotic nuclei. (A’–C’) An increase in the number of pyknotic nuclei was observed at 1 and 3dpt in the PUV treated halves of the corneas, after which the number of pyknotic nuclei is similar to control halves of the corneas at 5dpt. Red arrowheads indicate representative pyknotic nuclei. White dotted circles enclose single fragmented pyknotic nuclei. Scale bar in C’ is 50μm.

7. Supplementary Figure 7 (Figure S7): TUNEL assay for cell death analyses following PUV treatment on half corneas.

(supplements the data in Figure 6D)

Representative brightfield images of TUNEL positive nuclei are shown (Column i), alongside corresponding Hoechst images (Column ii), for the time points indicated as days post treatment (dpt). (A–D) and (A’–D’) correspond to 1, 3, 5, 15dpt for untreated control halves and PUV treated halves of the corneas, as indicated. An increase was observed in apoptosis at 1dpt in the treated half of the cornea vs. the untreated control half of the cornea, whereas no significant differences were observed at other time points. Red arrowheads indicate representative TUNEL positive nuclei. Scale bar in D’ column ii is 50μm.

Highlights:

• Psoralen followed by UV exposure creates stem cell deficiency in Xenopus.

• This frog model recapitulates the hallmarks of human cornea stem cell deficiency.

• Treatment causes cell death, mitosis, and conversion to skin epithelium.

• These effects can be easily localized to any part of the cornea.

• Undamaged cells will restore transparency after localized Psoralen-UV treatment.

Abbreviations:

AMT

4′-aminomethyltrioxsalen

AP

alkaline phosphatase

CESCs

cornea epithelial stem cells

LSCD

limbal stem cell deficiency

dpt

days post treatment

DMSO

dimethyl sulfoxide

EDTA

ethylenediaminetetraacetic acid

HLS

Hoechst labeled speckles

LLOS

lateral line organ system

LRCs

label retaining cells

NAM

normal amphibian media

PBS

phosphate-buffered saline

PBT

1XPBS and 0.2% Triton X-100

PUV

Psoralen and UVA

RT

room temperature

TACs

transit amplifying cells

Tcf7l2

Transcription factor 7 like 2

TdT

terminal deoxynucleotidyl transferase

TUNEL

terminal deoxynucleotidyl transferase dUTP nick end labeling