First pilot study of intrastromal rAAV-PAX6 gene therapy suggests improved corneal thickness and transcription correction in aniridic mouse Authors: Sif G. Kaad, Tess C. Lengyell, Seyedeh Zeinab Mirjalili Mohanna, Diana Djaksigulova, Andrea J. Korecki, Elizabeth M. Simpson Published: 2026 ABSTRACT Aniridia is a rare congenital vision-loss disorder caused primarily by loss-of-function variants of the paired box 6 (PAX6) gene. There is currently no cure. Augmentation gene therapy has emerged as a successful treatment for inherited ocular diseases. Here, we conducted the first pilot preclinical intrastromal augmentation gene therapy for aniridic cornea. It was undertaken in an aniridia mouse model, using adeno-associated virus 9 (AAV9), with a ubiquitous promoter driving PAX6, delivered by injection at 3 months, and harvest at 5 months post-injection (PI). The primary endpoint was histologically quantified epithelial thickness. The secondary endpoints were the slit lamp assessment of keratopathy and the RT-ddPCR measurement of four key signaling genes in the cornea. At 5 months PI, we demonstrated virally delivered PAX6 protein in the aniridic mouse corneal stroma. Critically, we found a statistically significant, albeit suggestive due to low n values, structural improvement in the corneal epithelial thickness. When the data were pooled, we also demonstrated a significant and complete transcription correction of four key genes—Notch1, Hes1, Wnt5a, and Mtor. We explore a variety of hypothesizes to explain these initial results, including that the virally delivered PAX6 is acting non-cell autonomously, moving from the non-dividing stroma into the epithelium, causing structural and molecular improvement. Introduction Aniridia is a rare congenital vision-loss disorder caused primarily by heterozygous loss-of-function variants of the paired box 6 (PAX6) gene (OMIM: 106210).1,2 PAX6 is a transcription factor with a tightly regulated dosage that is essential for normal eye development.3,4,5,6,7 Aniridia is primarily a panocular disorder but is named for the occurrence of iris hypoplasia.8 It typically results in low vision at birth due to retinal foveal hypoplasia. This is followed by progressive loss of vision primarily due to corneal aniridia-associated keratopathy, including pannas and neovascularization. There is currently no cure for aniridia, with treatments focusing on symptom relief, such as pharmacological approaches to glaucoma and dry eye and surgical interventions for iris hypoplasia and keratopathy.9 Augmentation gene therapy has emerged as a successful strategy for treating inherited ocular disease. The first FDA-approved ocular gene therapy was Luxturna in 2017, an rAAV (recombinant adeno-associated virus) treatment for RPE65-associated retinal dystrophy.10 Currently, clinical trials are ongoing of virus-based therapies for retinitis pigmentosa (ClinicalTrials.gov ID: NCT05203939, NCT06292650, and NCT06388200), Stargardt disease (ClinicalTrials.gov ID NCT07063251 and NCT06467344), and X-linked retinoschisis (ClinicalTrials.gov ID NCT06289452). Augmentation gene therapy may be particularly suitable for aniridia given that the functional gene renders patients immune tolerant to PAX6, and there are more than 600 PAX6 pathogenic variants that could be addressed with the same therapeutic.11,12 However, PAX6 augmentation therapy may be limited by a narrow therapeutic index due to PAX6 dosage sensitivity, and expression may need to be restricted to specific target cells.13,14 The cornea is made up of three cell layers—the middle multicellular stroma, underlain by the single-cell endothelial layer, and overlain by the multicellular epithelial layer.15 In mice, the basal corneal epithelial cells adjacent to the stroma become terminally differentiated and shed from the epithelium every 14 days.16 After mouse eye development is complete, the stroma and endothelium are virtually non-dividing.17 In healthy mouse eyes, slowly dividing limbal stem cells are present throughout life and continuously replenish the epithelium with new cells to maintain the transparency of the cornea.18 However, in the aniridic cornea, there is a slowly progressive loss of corneal transparency associated with inflammation, neovascularization, and an insufficiency of limbal stem cells.19,20,21 In human studies quantifying the thickness of the aniridic epithelium, results have varied, perhaps due to technical challenges.22,23 However, in mouse models, the results are more consistent, showing a thinner corneal epithelium.24,25,26,27 At the molecular level, key proteins have been identified as having altered expression in the human corneal epithelium and related tissues of normal versus aniridic corneas.28 The notch1 (Notch1) protein is reduced, while the hes family bHLH transcription factor 1 (Hes1), wingless-type MMTV integration site family, member 5A (Wnt5a), and mechanistic target of rapamycin kinase (Mtor) proteins are all increased.28 Overall, the time course of aniridia-associated keratopathy offers a potential therapeutic window. The work presented here, testing an rAAV PAX6 augmentation virus, builds directly on published control experiments using two capsids, five virus doses, three routes of administration, and two mouse strains.29 For that work, they delivered a “historical marker”—the improved Cre recombinase (iCre),30 which rearranges the genome and drives expression of the fluorescent marker tdTomato. This system was used to permanently mark transduced cells. Critically, after intrastromal injection. The results demonstrated the safety of the rAAV9 vector and extensive labeling of all three corneal layers and the limbal stem cells and their daughter cells. There was also escape from the cornea to the liver, heart, and blood vessels of the retina. To ensure these control experiments were applicable to the work presented here, we used the following: the same capsid, rAAV9; the same mouse strain and causal variant, B6129F1-Sey; intrastromal injection by the same person; injection at the same developmental time, 3 months; the same ubiquitous promoter, smCBA; same viral backbone; and a similar dose (7.6 × 1010 vg/eye and 8.6 × 1010 vg/eye). The AAV9 capsid has the ability to cross the blood-brain barrier,31,32,33 and it received FDA approval in 2019 as part of the augmentation gene therapy onasemnogene abeparvovec (Zolgensma) for spinal muscular atrophy.34 Such previous approval of a capsid can render future approvals easier. The B6129F1-Sey is a mouse model of aniridia, carrying a heterozygous loss-of-function variant of PAX6.35,36,37 This variant has been reported in a patient with aniridia38 and is a premature termination codon, the type of variant found in ∼70% of human cases of aniridia.39 For the cornea, on the 129S1/SvImJ or B6129F1 genetic backgrounds, this variant results in a thinner epithelium and slowly developing keratopathy, which mimic human aniridia.24,25,26,29 These mice have been used previously for preclinical studies.40,41 Typically, ocular gene therapy has been delivered by the subretinal route.42 Recently, a preclinical pilot study of PAX6 gene therapy used intravenous delivery,43 and a clinical trial (NCT06474416) is currently testing intrastromal delivery of lentivirus as a treatment for patients with herpes simplex virus 1 stromal keratitis. Here, we used intrastromal injection to achieve direct delivery to the target tissue and minimize the possible broader impacts of PAX6, a repetitively used potent transcription factor.3,44,45 In humans, intrastromal injection is routinely used to administer antibiotics and antifungals.46,47 Virus delivery at 3 months of age was as early as possible, determined by the size of the Sey/+ eye, which needed to be fully grown to reliably inject intrastromally. A ubiquitous promoter was chosen to have as broad transduction as possible because the therapeutic target cell was unknown. Once the target for a positive impact is known, one anticipates that a more specific promoter would be optimal for efficacy and safety. Finally, a maximal dose was delivered, given the maximal titer, thus favoring observing an impact. Once known, the lowest effective dose would be used. In contrast, for this first pilot intrastromal study, the iCre was replaced by 3xFLAG/PAX6 (an amino-terminal 3xFLAG fused to the PAX6 protein). The functionality of the amino-terminal FLAG-tagged PAX6 has already been demonstrated by intravenous injection and in the genome.43,48 In addition, the 16–68 days post-injection (PI) harvest time was replaced with a time point of 5 months PI. This long harvest time would preclude detection of direct transduction of the rapidly turning over epithelium16 but was chosen for the possibility of capturing a positive impact of virus-delivered PAX6 in limbal stem cells and their progeny repopulating the epithelium.29,49 The primary endpoint of this study was epithelial thickness, with slit lamp assessment of keratopathy and transcription levels of key genes being secondary. Results Experimental mouse cohort, treatment groups, outcome measures, and no sex effect The experimental cohort of 27 age-matched B6129F1 wild-type (WT) mice and the aniridia mouse model Sey/+ were intrastromally injected into the left eye only, at 3 months of age, with either vehicle (the virus delivery solution: phosphate-buffered saline and 0.001% pluronic acid [PBS + P]) or augmentation rAAV FLAG/PAX6 (Figure S1). Slit lamp imaging was conducted twice on the injected eye at 2–2.5 and 5 months PI. Subsequently, at 5 months PI, injected and uninjected eyes were harvested either for the histological detection of virally delivered PAX6 and quantification of the corneal layer thickness or for transcript quantification of key genes by droplet digital PCR (ddPCR) (Figure S1). No animals became ill or died during the course of the experiment. Elevated frequency of glaucoma is part of the aniridia phenotype,50 and one Sey/+ mouse, injected with the FLAG/PAX6 augmentation virus, developed a swollen eye (possibly glaucoma) in the uninjected eye and thus was harvested early (3 months PI). Because aniridia is thought to occur irrespective of sex,2 we used mice of both sexes (female, n = 14; male, n = 13). To reveal any sex differences, all numerical data in this study were analyzed by sex, and only one of 29 comparisons was significant at the p ≤ 0.05 level, with p = 0.0194. However, applying the Bonferroni multiple-comparisons correction rendered p ≤ 0.05/29 = 0.00172; thus, no significant sex differences were found. Slit lamp imaging revealed no impact of intrastromal injections Slit lamp imaging was performed at 2.5 and 5 months PI for vehicle-injected eyes and at 2 and 5 months PI for FLAG/PAX6-injected eyes (Figures 1, S2, and S3). Since eyes were injected at 3 months, these eyes were imaged at 5–5.5 and 8 months of age. Visual inspection of slit lamp images revealed that all WT corneas remained clear throughout the time course, showing no negative impact of either vehicle or FLAG/PAX6 injection, potentially due to healing of the injection site. All Sey/+ eyes showed the typical aniridia-associated keratopathy, including impingement of blood vessel, throughout the time course. These eyes also showed no negative impact of either vehicle or FLAG/PAX6 injection, potentially due to healing of the injection site. Visual inspection of the slit lamp images by a blinded observer did not reveal any difference between vehicle and injected Sey/+ eyes on the keratopathy phenotype, based on blood vessel incursion, blood vessel length, or degree of microphthalmia.Figure 1Slit lamp imaging did not reveal changes post vehicle or FLAG/PAX6 injection in WT or Sey/+ eyesImaging of the injected eye of the same mouse at 2–2.5 and 5 months PI after vehicle or FLAG/PAX6 injection in WT (upper row) or Sey/+ (lower row) eyes. Slit lamp images of all injected WT and Sey/+ eyes in this study at both time points are available in Figures S2 andS3, respectively. PI, post-injection. Slit lamp imaging did not reveal changes post vehicle or FLAG/PAX6 injection in WT or Sey/+ eyes Imaging of the injected eye of the same mouse at 2–2.5 and 5 months PI after vehicle or FLAG/PAX6 injection in WT (upper row) or Sey/+ (lower row) eyes. Slit lamp images of all injected WT and Sey/+ eyes in this study at both time points are available in Figures S2 andS3, respectively. PI, post-injection. Virally delivered PAX6 protein was present in the corneal stroma 5 months PI in Sey/+ eyes To ensure antibody detection of FLAG-tagged PAX6 protein was robust, we first stained corneas from WT (negative control) and Fax/+ (heterozygous) (positive control) mice (Figure S4). These mice were not injected with virus, but the Fax/+ mice express FLAG-tagged PAX6 protein from their genome (see material and methods).41,48,51 As expected, we detected no FLAG signal in WT eyes but observed robust FLAG and co-labelling of FLAG with PAX6 in the corneal epithelium of Fax/+ eyes. A subset of 14 injected eyes, vehicle (WT, n = 2; Sey/+, n = 2), or FLAG/PAX6 (WT, n = 4; Sey/+, n = 6), were histologically investigated for the presence of virally delivered PAX6 in the cornea, indicated by the co-labeling of FLAG and PAX6 antibody (Figure 2). Evaluation of the corneal images indicated that no eyes showed injection track marks, potentially due to healing by 5 months PI, nor other structural abnormalities attributable to the intrastromal injections. As expected, no FLAG signal was present in the cornea of vehicle-injected WT or Sey/+ eyes. In addition, FLAG signal was virtually absent from the corneal epithelium of FLAG/PAX6-injected WT or Sey/+ eyes. However, FLAG staining was detected in the corneal stroma of 1 of 4 FLAG/PAX6-injected WT eyes. In addition, FLAG staining was detected in the corneal stroma of 6 of 6 FLAG/PAX6-injected Sey/+ eyes. Although PAX6 is not normally expressed in the stroma, this result is attributable to the ubiquitous promoter used here to drive the viral FLAG/PAX6 protein.Figure 2Histological imaging showed virus-delivered PAX6 protein in the cornea of Sey/+ eyes and increased corneal thickness(A–D) Immunofluorescence staining was performed on paraffin-embedded sections 5 months PI after vehicle or FLAG/PAX6 injection in WT or Sey/+ eyes. Magnification is indicated on the first image and was the same for all. (A) No positive FLAG staining was found in the cornea of vehicle-injected WT eyes. (B) Positive FLAG staining was found in the cornea of one FLAG/PAX6-injected WT eye. (C) No positive FLAG staining was found in the cornea of vehicle-injected Sey/+ eyes. (D) Positive FLAG staining was found in the cornea of all FLAG/PAX6-injected Sey/+ eyes. Representative images in (D) show the thicker epithelial layer in the FLAG/PAX6-injected Sey/+ eyes compared with (C) vehicle-injected Sey/+ eyes but not to the level of thickness seen in (A) vehicle-injected WT eyes. Co-labeling with PAX6 indicates the presence of virally delivered PAX6 protein (yellow, white arrows). EPI, epithelium; STR, stroma; END, endothelium. Histological imaging showed virus-delivered PAX6 protein in the cornea of Sey/+ eyes and increased corneal thickness (A–D) Immunofluorescence staining was performed on paraffin-embedded sections 5 months PI after vehicle or FLAG/PAX6 injection in WT or Sey/+ eyes. Magnification is indicated on the first image and was the same for all. (A) No positive FLAG staining was found in the cornea of vehicle-injected WT eyes. (B) Positive FLAG staining was found in the cornea of one FLAG/PAX6-injected WT eye. (C) No positive FLAG staining was found in the cornea of vehicle-injected Sey/+ eyes. (D) Positive FLAG staining was found in the cornea of all FLAG/PAX6-injected Sey/+ eyes. Representative images in (D) show the thicker epithelial layer in the FLAG/PAX6-injected Sey/+ eyes compared with (C) vehicle-injected Sey/+ eyes but not to the level of thickness seen in (A) vehicle-injected WT eyes. Co-labeling with PAX6 indicates the presence of virally delivered PAX6 protein (yellow, white arrows). EPI, epithelium; STR, stroma; END, endothelium. Intrastromally injected rAAV9 escaped the cornea, resulting in virally delivered PAX6 present in the retina 5 months PI The work of Mirjalili Mohanna et al.29 demonstrated that after central intrastromal injection, as performed here, rAAV9 can escape the cornea to be found at 16–68 days PI in the liver, heart, and retinal blood vessels. Thus, we first histologically demonstrated robust antibody detection of FLAG-tagged PAX6 protein in Fax/+ (positive control) retinas (Figure S5). Then, we examined the same 14 injected eyes, vehicle (WT, n = 2; Sey/+, n = 2) or FLAG/PAX6 (WT, n = 4; Sey/+, n = 6), used for corneal histology, but now focusing on the retina (Figure 3). As expected, endogenous PAX6 was present in the ganglion cell layer (GCL), inner nuclear layer (INL), and outer nuclear layer (ONL). Again, as expected, no FLAG staining was present in the retina of vehicle-injected WT or Sey/+ eyes. FLAG staining was detected in the retina of 1 of 4 FLAG/PAX6-injected WT eyes. This was the same WT mouse that showed FLAG staining in the corneal stroma. Finally, FLAG staining was detected in the retina of 6 of 6 FLAG/PAX6-injected Sey/+ eyes. These eyes showed virally delivered PAX6 indicated by co-labelling of FLAG and PAX6 throughout the retina, with a concentrated expression centered around the optic nerve head. Overall, this broader retinal expression of escaped virus compared with that seen previously29 may be due to the longer (5 months PI) time to harvest.Figure 3Histological imaging showed virus “escaped” and delivered PAX6 protein to the retina of Sey/+ eyes(A–D) Immunofluorescence staining was performed on paraffin-embedded sections 5 months PI after vehicle or FLAG/PAX6 injection in WT or Sey/+ eyes. Magnification is indicated on the first image and was the same for all. (A) No positive FLAG staining was found in the retina of vehicle-injected WT eyes. (B) Positive FLAG staining was found in the retina of one FLAG/PAX6-injected WT eye. (C) No positive FLAG staining was found in the retina of vehicle-injected Sey/+eyes. (D) Positive FLAG staining was found in the retina of all FLAG/PAX6-injected Sey/+ eyes. Co-labeling with PAX6 indicates the presence of virally delivered PAX6 protein (yellow, white arrows). GCL, ganglion cell layer; INL, inner nuclear layer; ONL, outer nuclear layer. Histological imaging showed virus “escaped” and delivered PAX6 protein to the retina of Sey/+ eyes (A–D) Immunofluorescence staining was performed on paraffin-embedded sections 5 months PI after vehicle or FLAG/PAX6 injection in WT or Sey/+ eyes. Magnification is indicated on the first image and was the same for all. (A) No positive FLAG staining was found in the retina of vehicle-injected WT eyes. (B) Positive FLAG staining was found in the retina of one FLAG/PAX6-injected WT eye. (C) No positive FLAG staining was found in the retina of vehicle-injected Sey/+eyes. (D) Positive FLAG staining was found in the retina of all FLAG/PAX6-injected Sey/+ eyes. Co-labeling with PAX6 indicates the presence of virally delivered PAX6 protein (yellow, white arrows). GCL, ganglion cell layer; INL, inner nuclear layer; ONL, outer nuclear layer. Statistically significant data suggested increase in corneal epithelial thickness after FLAG/PAX6 injection in Sey/+ eyes The same 14 injected eyes, vehicle (WT, n = 2; Sey/+, n = 2) and FLAG/PAX6 (WT, n = 4; Sey/+, n = 6), which were histologically investigated for virally delivered PAX6, were examined visually (Figure 2) and then quantified for changes in the corneal thickness and cell count (Figures 4 and S6). Representative histological images are presented in Figure 2. For clarity, selected statistical differences are presented in Figures 4A, 4B, and 4D, and for comprehensiveness, all comparisons are presented in Figure S6.Figure 4Significant increase in corneal epithelial thickness after FLAG/PAX6 injection in Sey/+ eyes(A–D) Quantification of (A) epithelial thickness, (B) epithelial cell count, (C) stromal-endothelial thickness, and (D) stromal-endothelial cell count after vehicle or FLAG/PAX6 injection in WT or Sey/+ eyes. Statistics for the subset of most relevant post hoc comparisons are shown; for all comparisons, see Figure S6A. The expected reduction in epithelial thickness in Sey/+ compared with WT eyes was observed. FLAG/PAX6 injection did not impact the epithelial thickness in WT eyes. However, it significantly increased the expected reduced thickness in Sey/+ eyes, although not back to WT levels. (B) The expected reduction in epithelial cell count in Sey/+ relative to WT eyes was observed. FLAG/PAX6 injection did not impact the epithelial cell count in WT eyes. However, it increased the expected reduced thickness in Sey/+ eyes, although not significantly and not back to WT levels. (C) Stromal-endothelial thickness was not significantly impacted by vehicle or FLAG/PAX6 injection in WT or Sey/+ eyes. (D) Stromal-endothelial cell count, despite a significant ANOVA, showed no significant differences on post hoc analysis by vehicle or FLAG/PAX6 injection in WT or Sey/+ eyes. Data shown are the mean ± SEM, with each dot indicating a single cornea analyzed from each mouse. Statistical significance was determined using ANOVA, and if significant, then a post hoc Tukey’s multiple-comparison test. p ≤ 0.05 was considered as statistically significant. ∗, p ≤ 0.05; ∗∗∗, p ≤ 0.001; ns, not significant. Significant increase in corneal epithelial thickness after FLAG/PAX6 injection in Sey/+ eyes (A–D) Quantification of (A) epithelial thickness, (B) epithelial cell count, (C) stromal-endothelial thickness, and (D) stromal-endothelial cell count after vehicle or FLAG/PAX6 injection in WT or Sey/+ eyes. Statistics for the subset of most relevant post hoc comparisons are shown; for all comparisons, see Figure S6A. The expected reduction in epithelial thickness in Sey/+ compared with WT eyes was observed. FLAG/PAX6 injection did not impact the epithelial thickness in WT eyes. However, it significantly increased the expected reduced thickness in Sey/+ eyes, although not back to WT levels. (B) The expected reduction in epithelial cell count in Sey/+ relative to WT eyes was observed. FLAG/PAX6 injection did not impact the epithelial cell count in WT eyes. However, it increased the expected reduced thickness in Sey/+ eyes, although not significantly and not back to WT levels. (C) Stromal-endothelial thickness was not significantly impacted by vehicle or FLAG/PAX6 injection in WT or Sey/+ eyes. (D) Stromal-endothelial cell count, despite a significant ANOVA, showed no significant differences on post hoc analysis by vehicle or FLAG/PAX6 injection in WT or Sey/+ eyes. Data shown are the mean ± SEM, with each dot indicating a single cornea analyzed from each mouse. Statistical significance was determined using ANOVA, and if significant, then a post hoc Tukey’s multiple-comparison test. p ≤ 0.05 was considered as statistically significant. ∗, p ≤ 0.05; ∗∗∗, p ≤ 0.001; ns, not significant. The known aniridia phenotype in Sey/+ mice includes a thinner corneal epithelium with fewer cells.24,25,26,27 Thus, it was reassuring that this reduction was maintained after vehicle injection of Sey/+ eyes versus WT eyes. This is shown in representative images (Figure 2) and quantitatively with non-overlapping groups giving a significant difference in both thickness (Figure S6A) and cell count (Figure S6B). No significant difference was found between vehicle- and FLAG/PAX6-injected WT eyes in epithelial thickness (Figures 2 and 4A) or cell count (Figures 2 and 4B). This is presumably due to only 1 of 4 eyes showing virally delivered PAX6 staining. However, an improvement of epithelial thickness was seen after FLAG/PAX6 injection of Sey/+ eyes versus vehicle-injected Sey/+ eyes. This is presumably due to 6 of 6 eyes showing virally delivered PAX6 staining. This result is shown in representative images (Figure 2) and quantitatively with non-overlapping groups giving a statistically significant improvement (Figure 4A). However, the n = 2 vehicle control mice make this result suggestive. This was a partial improvement, attaining 63.1% of WT levels. This result was paralleled by epithelial cell count, which in Sey/+ eyes was increased after FLAG/PAX6 injection, compared with vehicle-injected eyes (Figures 2 and 4B). This also was a partial improvement, attaining 68.8% of WT levels. In contrast to the corneal epithelium, there is no known aniridia phenotype in Sey/+ mice for the stromal-endothelial layers either in thickness or number of cells when measured peripherally, avoiding central cornea-lens adhesions.26,27 Thus, as expected, there was no statistical impact of vehicle or FLAG/PAX6 injection in WT or Sey/+ eyes (Figure 4C). For stromal-endothelial cell count, the initial ANOVA was significant; however, subsequent post hoc analysis revealed no significant difference between the groups (Figures 4D and S6C). Uninjected contralateral eyes were not impacted by injection in partner eye Initially, a linear regression analysis was performed to investigate the correlation between corneal thickness and cell count by using the data from the same 14 injected eyes, vehicle (WT, n = 2; Sey/+, n = 2), and FLAG/PAX6 (WT, n = 4; Sey/+, n = 6) (Figure 5A). A statistically significant positive correlation was found for all eyes, showing that increased corneal thickness is accompanied by more cells, not larger cells. In addition, the graph further demonstrated that all the FLAG/PAX6-injected Sey/+ eyes (red) showed increased thickness and cell count without overlap compared with the vehicle-injected Sey/+ eyes (blue) and that for none did the improvement reach WT mouse levels (black and green).Figure 5Correlations show that uninjected contralateral eyes were not impacted by injection in partner eye(A and B) Correlation analysis shows that the uninjected contralateral eyes were not impacted by injection in the partner eye. A positive and significant correlation between corneal epithelial thickness and epithelial cell count was seen for both the (A) injected eyes (data from Figures 4A and 4B), and (B) contralateral eyes. WT corneas cluster as the thickest with the most cells (top right of the correlations), demonstrating no impact of the FLAG/PAX6 injection (green) compared with vehicle (black). Sey/+ corneas are thinner with less cells (bottom left of the correlations). (A) FLAG/PAX6-injected Sey/+ corneas (red) are shifted toward thicker with more cells (up and to the right), showing the positive impact of FLAG/PAX6 injection, compared with vehicle (blue). (B) Sey/+ corneas of the contralateral eyes cluster as the thinnest with the least cells (bottom left of the correlation), showing no impact of FLAG/PAX6 injection of the partner eye (red), compared with vehicle injection of the partner eye (blue). Data were analyzed using simple linear correlation. p ≤ 0.05 was considered as statistically significant. Correlations show that uninjected contralateral eyes were not impacted by injection in partner eye (A and B) Correlation analysis shows that the uninjected contralateral eyes were not impacted by injection in the partner eye. A positive and significant correlation between corneal epithelial thickness and epithelial cell count was seen for both the (A) injected eyes (data from Figures 4A and 4B), and (B) contralateral eyes. WT corneas cluster as the thickest with the most cells (top right of the correlations), demonstrating no impact of the FLAG/PAX6 injection (green) compared with vehicle (black). Sey/+ corneas are thinner with less cells (bottom left of the correlations). (A) FLAG/PAX6-injected Sey/+ corneas (red) are shifted toward thicker with more cells (up and to the right), showing the positive impact of FLAG/PAX6 injection, compared with vehicle (blue). (B) Sey/+ corneas of the contralateral eyes cluster as the thinnest with the least cells (bottom left of the correlation), showing no impact of FLAG/PAX6 injection of the partner eye (red), compared with vehicle injection of the partner eye (blue). Data were analyzed using simple linear correlation. p ≤ 0.05 was considered as statistically significant. For each mouse, we administered injection only in the left eye, and it has been shown by others that a unilateral gene therapy can affect both eyes.52,53,54 Thus, we asked the question: did the suggestive positive impact of FLAG/PAX6 in Sey/+ eyes extend to the contralateral uninjected eye (Figure 5B)? The contralateral eyes were histologically quantified for corneal thickness and cell count in the same manner as the injected eyes. Then, a linear regression analysis was performed, and a statistically significant positive correlation was found for all eyes. But, this time, the graph demonstrated that all the uninjected contralateral FLAG/PAX6 Sey/+ eyes (red) showed no increased thickness or cell count and clustered with the uninjected contralateral vehicle injected Sey/+ eyes (blue). Virally delivered FLAG-PAX6 transcripts in Sey/+ cornea were highest but not significantly different from those of controls Ldha, Ppia, Rpl5, and Tbp have previously been identified as ubiquitous genes for reverse-transcription droplet digital PCR (RT-ddPCR) normalization in the adult cornea.55,56 To validate these genes for this study, we quantified their mRNA transcript levels across controls, genotypes, and treatment groups. All four genes demonstrated no significant difference in transcript levels (Figure S7), and, thus, the geometric mean of these genes was used for normalization of the experimental genes. To validate the Endo-Pax6 RT-ddPCR assay, endogenous Pax6 transcript levels were quantified in WT and Sey corneas injected with vehicle only (Figure S8). As expected for B6129F1 adult cornea,24 Sey mice exhibited 55.2% of the Pax6 transcript levels observed in WT mice, consistent with the presence of a single functional Pax6 allele. To validate the FLAG-PAX6 RT-ddPCR assay, we analyzed corneas from Fax/+ and Fax/Fax (homozygous) mice (Figure S9). The assay detected a sequence of the transcript from this genomic FLAG/PAX6, which is the same as that in the FLAG/PAX6 virus. As expected, both genotypes exhibited similar levels of Endo-Pax6 transcription (Figure S9A), while Fax/+ mice exhibited 46.6% of the Pax6 transcript levels observed in Fax/Fax mice, consistent with the presence of a single functional Pax6 allele in the former (Figures S9B and S9C). Finally, as expected, FLAG-PAX6 transcript levels were not significantly different from Endo-Pax6 in Fax/Fax mice (Figure S9D). A new subset of 14 corneas from injected eyes, vehicle (WT, n = 3; Sey/+, n = 2) or FLAG/PAX6 (WT, n = 5; Sey/+, n = 4), were investigated by RT-ddPCR for mRNA transcript levels of virally delivered FLAG/PAX6 (FLAG-PAX6 assay) or endogenous Pax6 (Endo-Pax6 assay) (Figures 6 and S10). These eyes were injected and harvested at the same time as the histological subset, but they were from different mice. For clarity, selected statistical differences are presented in Figure 6, and for comprehensiveness, all comparisons are presented in Figure S10.Figure 6FLAG/PAX6-injected Sey/+ corneas had the most FLAG-PAX6 transcripts, but the difference was not significant(A–C) Transcript levels of virally delivered FLAG-tagged PAX6 (FLAG-PAX6 assay) or endogenous Pax6 (Endo-Pax6 assay) were quantified in corneas with RT-ddPCR. (A) FLAG-PAX6 detected mRNA transcripts in corneas after vehicle or FLAG/PAX6 injection in WT or Sey/+ eyes. FLAG/PAX6-injected Sey/+ corneas had the most FLAG-PAX6 transcripts, but the difference was not significant. (B) Reanalyzing the data in (A), conservatively removing the 1.768 FLAG/PAX6-injected Sey/+ data point, even though it was not a significant outlier. Statistics for the subset of most relevant post hoc comparisons are shown; for all comparisons, see Figure S10A. Overall significance was detected, and FLAG/PAX6-injected Sey/+ corneas still had the most FLAG-PAX6 transcripts, but the difference was not significant compared with vehicle-injected Sey/+ eyes. (C) Endo-Pax6 detected mRNA transcripts in corneas after vehicle or FLAG/PAX6 injection in WT or Sey/+ eyes. Statistics for the subset of most relevant post hoc comparisons are shown; for all comparisons, see Figure S10B. The expected reduction in Endo-Pax6 transcripts in Sey/+ compared with WT eyes was observed. However, there was no significant impact of vehicle or PAX6/FLAG injection. Data shown are the mean ± SEM, with each dot indicating a single cornea analyzed from each mouse. Statistical significance was determined using ANOVA, and if significant, then a post hoc Tukey’s multiple-comparison test. The robust regression and outlier removal (ROUT) test was used to detect outliers. p ≤ 0.05 was considered as statistically significant. ∗, p ≤ 0.05; ns, not significant, RT-ddPCR, reverse transcription droplet digital PCR. FLAG/PAX6-injected Sey/+ corneas had the most FLAG-PAX6 transcripts, but the difference was not significant (A–C) Transcript levels of virally delivered FLAG-tagged PAX6 (FLAG-PAX6 assay) or endogenous Pax6 (Endo-Pax6 assay) were quantified in corneas with RT-ddPCR. (A) FLAG-PAX6 detected mRNA transcripts in corneas after vehicle or FLAG/PAX6 injection in WT or Sey/+ eyes. FLAG/PAX6-injected Sey/+ corneas had the most FLAG-PAX6 transcripts, but the difference was not significant. (B) Reanalyzing the data in (A), conservatively removing the 1.768 FLAG/PAX6-injected Sey/+ data point, even though it was not a significant outlier. Statistics for the subset of most relevant post hoc comparisons are shown; for all comparisons, see Figure S10A. Overall significance was detected, and FLAG/PAX6-injected Sey/+ corneas still had the most FLAG-PAX6 transcripts, but the difference was not significant compared with vehicle-injected Sey/+ eyes. (C) Endo-Pax6 detected mRNA transcripts in corneas after vehicle or FLAG/PAX6 injection in WT or Sey/+ eyes. Statistics for the subset of most relevant post hoc comparisons are shown; for all comparisons, see Figure S10B. The expected reduction in Endo-Pax6 transcripts in Sey/+ compared with WT eyes was observed. However, there was no significant impact of vehicle or PAX6/FLAG injection. Data shown are the mean ± SEM, with each dot indicating a single cornea analyzed from each mouse. Statistical significance was determined using ANOVA, and if significant, then a post hoc Tukey’s multiple-comparison test. The robust regression and outlier removal (ROUT) test was used to detect outliers. p ≤ 0.05 was considered as statistically significant. ∗, p ≤ 0.05; ns, not significant, RT-ddPCR, reverse transcription droplet digital PCR. As expected, the FLAG-PAX6 assay detected no virally delivered transcripts in the cornea of vehicle-injected WT or Sey/+ eyes (Figure 6A). Consistent with the predominantly negative histological results, the assay also detected no virally delivered transcripts in the cornea of FLAG/PAX6-injected WT eyes. The highest level of virally delivered transcripts was detected in the cornea of FLAG/PAX6-injected Sey/+ eyes, although even this did not reach significance. The question arose if removing the 1.768 FLAG/PAX6 injected Sey/+ data point in would change the result in (Figure 6A). Using the robust regression and outlier removal (ROUT) test, we found no outliers. Nevertheless, the reanalysis was done without the data point (Figure 6B). Statistics for the subset of most relevant post hoc comparisons are shown; for all comparisons, see Figure S10A. Overall significance was detected, and FLAG/PAX6-injected Sey/+ corneas still had the most FLAG-PAX6 transcripts, but even with this analysis, the difference was still not significant when compared with vehicle-injected Sey/+ eyes. The known aniridia phenotype in Sey/+ cornea includes a reduction in the endogenous PAX6 transcripts in comparison to WT counterparts.24 The Endo-Pax6 assay detected a reduction of 44.8% after vehicle injection (Figure 6C), although this did not reach significance (Figure S10B). Consistent with the previous lack of significant changes in FLAG/PAX6 transcript (Figures 6A and 6B), the Endo-Pax6 assay detected no significant difference in endogenous Pax6 transcripts between vehicle- and FLAG/PAX6-injected WT corneas (Figure 6C). Similarly, the Endo-Pax6 assay detected no significant difference in endogenous Pax6 transcripts between vehicle- and FLAG/PAX6-injected Sey/+ corneas (Figure 6C). Four genes impacted by the aniridia phenotype show a significant pattern of completely corrected transcription Notch1, Hes1, Wnt5a, and Mtor have previously been identified as having altered protein expression in the human corneal epithelium and related tissues of normal versus aniridic corneas.28 To assess the impact of virally delivered PAX6 on these key signaling genes, we quantified their endogenous mRNA transcript levels in cornea by using RT-ddPCR (Figure 7). A subset of the cornea investigated for PAX6 transcription was used—vehicle (WT, n = 3; Sey/+, n = 2) or FLAG/PAX6 (Sey/+, n = 4). We excluded FLAG/PAX6-injected WT corneas because only 1 of 4 of them showed virally delivered protein.Figure 7Overall significant pattern of the correction of gene transcription back to WT levels(A–D) Endogenous transcript levels in corneas after vehicle or FLAG/PAX6 injection in WT or Sey/+ eyes were quantified with RT-ddPCR for (A) Notch1, (B) Hes1, (C) Wnt5a, and (D) Mtor. (A) In vehicle-injected WT or Sey/+ eyes, Notch1 endogenous transcripts were significantly elevated in Sey/+ eyes. After FLAG/PAX6 injection in Sey/+ eyes, the transcript levels were significantly reduced back to a level not significantly different from that in the vehicle-injected WT eyes. (B–D) In vehicle-injected WT or Sey/+ eyes, endogenous transcripts of Hes1, Wnt5a, and Mtor were elevated in Sey/+ eyes, although not significantly. After FLAG/PAX6 injection in Sey/+ eyes, the transcript levels were reduced significantly for Hes1 and Mtor, back to a level not significantly different from that of vehicle-injected WT eyes. In (A)–(D), boxplots show the maximum, mean, and minimum, with each dot indicating a single cornea analyzed from each mouse. (E) shows the mean difference and 95% confidence interval (CI), with all genes pooled for each pair of conditions. Overall, (1) there was a significant increase in the endogenous levels of these genes between vehicle-injected WT and vehicle-injected Sey/+ eyes; (2) there was a significant decrease in the endogenous levels of these genes between vehicle-injected Sey/+ and FLAG/PAX6-injected Sey/+ eyes; and (3) there was no significant difference in the endogenous levels of these genes between vehicle-injected WT and FLAG/PAX6-injected Sey/+ eyes. p ≤ 0.05 was considered as statistically significant. ∗, p ≤ 0.05; n, number of corneas; ns, not significant; RT-ddPCR, reverse transcription droplet digital PCR. Overall significant pattern of the correction of gene transcription back to WT levels (A–D) Endogenous transcript levels in corneas after vehicle or FLAG/PAX6 injection in WT or Sey/+ eyes were quantified with RT-ddPCR for (A) Notch1, (B) Hes1, (C) Wnt5a, and (D) Mtor. (A) In vehicle-injected WT or Sey/+ eyes, Notch1 endogenous transcripts were significantly elevated in Sey/+ eyes. After FLAG/PAX6 injection in Sey/+ eyes, the transcript levels were significantly reduced back to a level not significantly different from that in the vehicle-injected WT eyes. (B–D) In vehicle-injected WT or Sey/+ eyes, endogenous transcripts of Hes1, Wnt5a, and Mtor were elevated in Sey/+ eyes, although not significantly. After FLAG/PAX6 injection in Sey/+ eyes, the transcript levels were reduced significantly for Hes1 and Mtor, back to a level not significantly different from that of vehicle-injected WT eyes. In (A)–(D), boxplots show the maximum, mean, and minimum, with each dot indicating a single cornea analyzed from each mouse. (E) shows the mean difference and 95% confidence interval (CI), with all genes pooled for each pair of conditions. Overall, (1) there was a significant increase in the endogenous levels of these genes between vehicle-injected WT and vehicle-injected Sey/+ eyes; (2) there was a significant decrease in the endogenous levels of these genes between vehicle-injected Sey/+ and FLAG/PAX6-injected Sey/+ eyes; and (3) there was no significant difference in the endogenous levels of these genes between vehicle-injected WT and FLAG/PAX6-injected Sey/+ eyes. p ≤ 0.05 was considered as statistically significant. ∗, p ≤ 0.05; n, number of corneas; ns, not significant; RT-ddPCR, reverse transcription droplet digital PCR. For all four genes, vehicle-injected Sey/+ cornea had increased endogenous mRNA transcript levels compared with vehicle-injected WT. For Notch1, this was significant (Figure 7A), although it did not reach significance for Hes1 (Figure 7B), Wnt5a (Figure 7C), and Mtor (Figure 7D). Excitingly, for all four genes, FLAG/PAX6-injected Sey/+ cornea had decreased endogenous transcript levels compared with vehicle-injected Sey/+. This correction of transcription was significant for Notch1, Hes1, and Mtor but not for Wnt5a. Furthermore, for all four genes, the level was completely corrected and thus not significantly different compared with vehicle-injected WT. When the data for all four genes were pooled (corneas, n = 5, 6, and 7), vehicle-injected Sey/+ corneas were found to have significantly increased endogenous transcript levels compared with vehicle-injected WT (Figure 7E). Furthermore, FLAG/PAX6-injected Sey/+ cornea had a significant correction of endogenous transcript levels compared with vehicle-injected Sey/+. This corrected level was complete, not differing significantly from that of vehicle-injected WT. Discussion This is the first preclinical demonstration of intrastromal augmentation gene therapy for the aniridic cornea. We addressed the PAX6 haploinsufficiency of aniridia by delivering PAX6 in rAAV9 by intrastromal injection in 3-month-old Sey mice, a model of aniridia. At 5 months PI, no ongoing virally delivered PAX6 transcription was noted. However, we demonstrated the presence of virally delivered PAX6 protein in the corneal stroma of 6 of 6 mice. We saw a statistically significant, albeit suggestive due to low n values, structural improvement in the corneal epithelial thickness. In contrast, slit lamp imaging revealed no impact of intrastromal injections on keratopathy. When the data were pooled, we demonstrated a significant increase in the transcription of the four key signaling genes in untreated aniridic mouse cornea and a significant and complete correction of this molecular transcription change with gene therapy. All this occurred without affecting the transcription of endogenous Pax6, nor any negative impact structurally or molecularly on the injected eye, on the contralateral eye, or on the overall health of the animal. One of the limitations of this work is that it was a first-time intrastromal pilot study that needs to be repeated by an independent group. Toward this end, we have made available the mice from The Jackson Laboratory (JAX) (Bar Harbor, ME, USA) and the rAAV FLAG/PAX6 virus genome plasmid from Addgene (Cambridge, MA, USA) who have chosen to distribute it both as a plasmid and virus. We started this study with n = 27 mice, which we now know was suboptimal.26 These mice were divided by two genotypes 1:1 (WT and Sey/+), assigned to two possible injections 1:2 (vehicle control and experimental rAAV FLAG/PAX6) and two types of analysis 1:1 (histology and ddPCR). This resulted in the vehicle control groups containing n = 2–3 mice and the experimental rAAV FLAG/PAX6 groups with n = 4–5. This bias was introduced in case there was loss of experimental mice due to rAAV FLAG/PAX6 injection, which did not happen. Thus, for the quantification of corneal thickness, we had n = 2 vehicle control mice, and n = 6 rAAV FLAG/PAX6 experimental mice (Figure 4). When compared, the groups had no overlap and gave a statistically significant p = 0.0006. Nevertheless, the n = 2 value makes the result suggestive. Finally, our control was the liquid vehicle in which the rAAV was delivered. Thus, if the presence of the virus or FLAG was responsible for the therapeutic effect, we would not know. If the presence of the virus contributes to the therapeutic effect, that is not a concern because virus will be part of the final therapy. However, if the FLAG fused to PAX6 is contributing to the therapeutic effect, that is a concern because the final therapy would not have FLAG. In this first study, we found the FLAG tag very useful to identify the virus-delivered PAX6, but in the future, a virus without FLAG must also be included as the actual gene therapy. In this work, we assume, based on the experimental similarities, that viral transduction was the same as reported by Mirjalili Mohanna et al.29 With their “historical” system, they observed extensive transduction of all three corneal layers and the transduction of limbal stem cells. Thus, we assume in our experiment too that the virus initially transduced all three corneal layers and limbal stem cells. However, in our work, detection of the virally delivered FLAG/PAX6 protein is a “current” marker, showing where the virus is still present and is being transcribed and translated at 5 months PI. The corneal epithelium rapidly turns over via cell division every 14 days.16 Thus, it was anticipated that at 5 months PI, no virally delivered FLAG/PAX6 protein would be detected in the epithelium, and it virtually was not. It was less clear how quickly the transduced limbal stem cells, which reside in the corneal limbus and regenerate the corneal epithelium,57 would lose the virus presumably by cell division. Because we detected virtually no virally delivered FLAG/PAX6 protein in the epithelium, we conclude that limbal stem cells and their progeny had lost the virus by 5 months PI. Finally, the adult corneal stromal layer is non-dividing,17 and virally delivered FLAG/PAX6 protein was detected in 6 of 6 Sey/+ mice, even at 5 months PI. A future experiment would include a series of harvest times to re-demonstrate transduction of the epithelium but also discover if this is direct or secondary to basal cell or limbal cell transduction, and then discover the time course of virus loss. Virally delivered FLAG/PAX6 was seen in the corneal stroma of only 1 of 4 injected WT mice, compared with being seen in the corneal stroma of 6 of 6 Sey/+. Is this result random procedural failure, or is there a different mechanism acting in WT compared with Sey/+ cornea? Perhaps, there is faster elimination of virus or a more efficient regulatory mechanism that shuts down virus transcription or translation in WT cells. The transcription mechanism would need to function independently of the native PAX6 gene signals, which are absent in the virus. Another hypothesis is that the cells continuing to express virally delivered FLAG/PAX6 are not present in the healthy WT cornea. Regardless of the mechanism, it is operating in the therapist favor, because gene therapy was more prevalent in the aniridic mouse eye. Given the lack of PAX6 transcripts and the relatively modest level of virally delivered protein located in the corneal stroma at 5 months PI, we were surprised but encouraged to see a statistically significant, albeit suggestive due to low n values, structural improvement in the corneal epithelial thickness. In considering alternative mechanistic hypotheses to explain this result, it would have been interesting to know if the amount of viral PAX6 transcription in the cornea would correlate positively with increased thickness. But, because the injected eye of each mouse was taken either for histology or molecular studies, but not both, no such correlation could be generated. Nevertheless, assuming a corneal thickness improvement, two not mutually exclusive hypotheses regarding mechanism may be considered. In both hypotheses, the significant transcription correction of the four key signaling genes in the Sey/+ cornea would be secondary, not to ongoing virally delivered PAX6 transcription, but rather to the improvement in epithelial thickness, thus supporting that structural change. First, we hypothesize that the virus was present in the epithelium due to direct transduction, and more importantly, limbic stem cell transduction, as predicted by previous studies,29 long enough to have a positive impact on epithelial thickness. If this was the case, the impact may have been fading by 5 months, because the PAX6 protein was no longer in the epithelium, and we may not have captured the greatest extent of this suggestive improvement. Second, we hypothesize that the virally delivered PAX6 is not acting cell autonomously (intrinsically to a single cell), but rather acting non-cell autonomously (influencing other cells) moving from the stroma into the epithelium to cause the structural improvement. Although there are situations in which PAX6 acts cell autonomously,58,59 it is also known to act non autonomously.60,61,62 If non-cell autonomous actions were the case, the suggestive positive impact of the augmentation gene therapy may be long term, driven by viral expression in the non-dividing stromal cells.17 These initial suggestive positive results, in the presence of no harm, argue for the continued preclinical exploration of PAX6 augmentation therapy for aniridia. Key parameters to understand the mechanisms and potential of the treatment include time of delivery, and time of harvest. As with most therapies, we anticipate a benefit from earlier treatment. We administered intrastromal PAX6 virus injections in mouse adulthood (3 months) when aniridia-associated keratopathy was fully developed in the Sey/+ mouse, which may explain why no positive impact on the slit lamp images was seen. Thus, we conclude that late treatment was unable to reverse keratopathy. However, if delivered earlier, could it have prevented or halted the development of keratopathy? Unfortunately, in this aspect, the small size of the mouse eye is limiting. In our hands, intrastromal injection is reliable only in the Sey/+ eye at 3 months when it has fully grown. For earlier delivery experiments, the research would benefit from intravenous delivery or a larger animal model of aniridia. However, eye size will not be a limitation in humans where intrastromal injections have been done in childhood.63 A series of harvest times would allow a further understanding of the mechanisms involved in the suggested improvement of the corneal epithelial thickness. When, and by what mechanism, did escape to the retina occur in 6 of 6 mice, and could that be harnessed to treat aniridic phenotypes in both tissues simultaneously? Was there complete correction of corneal thickness earlier, which was waning at 5 months, and would it be gone later? Or, was the suggestive therapeutic effect persistent, and would it last throughout the animal’s lifetime? Use of cell type-specific promoters driving PAX6 in the augmentation gene therapy might not only be an effective tool in this regard but also add safety to the final clinical approach. For example, a stromal-specific promoter might demonstrate that a relatively stable tissue (the stroma) could be used to continuously deliver PAX6 non-cell autonomously to improve the epithelium, thus providing a long-term treatment for the aniridic cornea. Furthermore, this strategy may be generalizable for the treatment of other corneal diseases. Material and methods Breeding and genotyping of mice The mice were bred and housed in the pathogen-free mouse core facility of the Center for Molecular Medicine and Therapeutics at The University of British Columbia. All experimental procedures were done within this facility; so, no acclimatization period was needed. C57BL/6J (B6)-Pax6Sey-em1(3xFLAG)Ems mice (The Mutant Mouse Resource and Research Center [MMRRC], RRID: MMRRC_066963-MU) were backcrossed to the genetic background 129S1/SvImJ (JAX, RRID: IMSR_JAX:002448) to produce 129-Pax6Sey-em1(3xFLAG)Ems (RRID: MMRRC_071297-MU). Experimental mice were age-matched B6129F1 hybrids, produced by timed pregnancies mating B6 dams (RRID: IMSR_JAX:000664) with 129-Pax6Sey-em1(3xFLAG)Ems/+ N2 stud males, some of which also carried an X-linked lacZ/0 transgene (RRID: IMSR_JAX:038651) (Figure S1). Neither the amino 3xFLAG (DYKDDDDK-tag) peptide on the genomic Sey allele48 nor lacZ transgene on the X chromosome64 was used for the experiments undertaken here; so, for clarity and brevity, the mice are referred to as Sey/+ (heterozygote). An experimental cohort of 27 mice was generated to provide at least 3 animals per group, but given the power analysis data now available,26 an even larger cohort would have been planned. Negative control mice were B6 WT. Positive control mice were B6-Pax6em3Ems, with the amino 3xFLAG on the genomic WT Pax6 allele; for clarity and brevity, these mice are referred to as Fax/+ (heterozygote) or Fax/Fax (homozygote) (RRID: MMRRC_066967-MU).41,48,51 The mice were weaned and ear-notched on postnatal day (P) 18–21, and genotyping PCR was performed as described previously.24,48 The mice were weaned, genotyped, and assigned to four treatment groups for comparison, using constrained randomization (sex and genotype) by one individual. Intrastromal injections were administered on a separate day by a second individual. The injector was blinded to the genotype and did not re-examine the eyes, which would have suggested the genotype. Tissue harvest was done by a third person, and a fourth person imaged and scored the eyes. The scorer was blinded to the treatment group, but the genotype was obvious due to the Sey/+ phenotype. RT-ddPCR was undertaken by a fifth person. Prior to the study, the development of a swollen eye (possibly glaucoma) was established as an exclusion criterion. No other inclusion or exclusion criteria were set. No additional confounds were identified. The mice were kept on a 7 am–7 pm light cycle and had access to food and water ad libitum. All procedures involving mice were done in accordance with the Canadian Council on Animal Care and UBC Animal Care Committee (protocols A17-0204, A17-0205, A21-0140, and A21-0184) and adhered to the ARVO statement for the use of animals in ophthalmic and vision research. Cloning and virus production rAAV genome plasmid pEMS2244 smCBA-3xFLAG/PAX6 was assembled in a multistep cloning strategy from a synthesized smCBA plasmid (pEMS2075)65,66 and a synthesized 3xFLAG/PAX6 plasmid (pEMS1906). It was in our “plug and play” ssAAV2 backbone plasmid version 1, which included the following additional elements: multiple cloning site for promoter insertion, chimeric intron, NotI for protein sequence insertion, AsiSI-flanked woodchuck hepatitis virus posttranscriptional regulatory element (WPRE), and SV40 polyA.67,68 The rAAV genome plasmid was propagated in E. coli SURE cells (Agilent Technologies, Santa Clara, CA, USA). DNA was prepared using the Spin MiniPrep Kit (QIAGEN, Germantown, MD, USA), and the plasmid was confirmed free of rearrangements by AhdI digest; ITRs were verified by SmaI digest, and cloning sites were verified by sequencing. The confirmed plasmid was sent to the University of Pennsylvania Vector Core (Philadelphia, PA, USA) for large-scale DNA amplification using the EndoFree Plasmid Mega Kit (QIAGEN). Quality control on the plasmid preparation was done by SmaI, PvuII, and SnaBI digests, and confirmed plasmids were packaged into AAV9 capsid, with titer determined by ddPCR. The virus produced, rAAV9 smCBA-3xFLAG/PAX6-WPRE, is hereafter referred to for brevity as FLAG/PAX6. Intrastromal injections Adult mice (aged 3 months) were weighed, anesthetized with isoflurane, and injected with 5 mg/kg Metacam (DIN: 02240463, Boehringer Ingelheim, Ingelheim am Rhein, Germany) for pain management. Alcaine eye drops (DIN: 00035076, Alcon, Geneva, Switzerland) were applied to numb the eye receiving the injection, and OptixCare lubricating eye gel (VP 914304, Aventix, Burlington, Canada) was applied to the contralateral eye to prevent dryness. Custom-made 34-gauge removable small hub RN needles (length: 9.53 mm; point style: 4; bevel angle: 30°) (Hamilton Company, Reno, NV, USA) were used for intrastromal injections. Once the mice reached surgical plane of anesthesia, 2 μL of undiluted FLAG/PAX6 at a titer of 4.3 × 1013 vg/mL in PBS + P (Cat #24040032, Gibco Poloxamer 188, Thermo Fisher Scientific, Waltham, MA, USA) was loaded into a 2.5 μL syringe (Cat #7632-01, Hamilton Company). The final intrastromal injection dose was 8.6 × 1010 vg/eye. For vehicle injections, 2 μL of PBS + P was loaded into the syringe. Central intrastromal injections were administered following a previously described method.29,41,49 The mice were weighed, and their eyes were inspected daily for five days PI, followed by weekly monitoring until the end of the study. Split lamp imaging Slit lamp imaging was undertaken to monitor the impact of treatment on keratopathy. It was performed using the Phoenix Micron Anterior Segment Imaging System (Phoenix Research Labs, Pleasanton, CA, USA) and followed a previously described protocol.48 Imaging was performed at 2.5 and 5 months PI for the vehicle-injected eyes and at 2 and 5 months PI for the FLAG/PAX6-injected eyes. Imaging was not performed on the uninjected eyes. The time difference was due to an unavoidable delay. Tissue harvest After slit lamp imaging, the mice at 5 months PI were sacrificed by cervical dislocation, and the eyes were removed for either histological or molecular quantification (Figure S1). Eyes used for histological quantification were placed in 4% PFA for 2 h, followed by PBS + 10% sucrose at 4° for storage until use. Eyes used for molecular quantification were placed in PBS. Using a dissecting microscope, corneas were isolated by cutting along the corneal/scleral divide, and the lens and retina were discarded. In case of Sey/+ eyes with corneal-lens adhesions, best efforts were made to keep only the corneal tissue. The corneas were then snap frozen in liquid nitrogen and kept at −80°C for storage until use. Tissue histology Eyes were removed from PBS + 10% sucrose and washed 3× 5 min in PBS, then placed in histology cassettes and immersed in 70% EtOH before embedding. Paraffin embedding and 5-μm sectioning of eyes for histological quantification were performed at British Columbia Children’s Hospital Research Institute Core Technologies and Services, Vancouver, Canada. A light microscope (Olympus, Tokyo, Japan) was used for the selection of tissue sections. Corneal slides with clearly defined and intact epithelium and stromal-endothelial layers in the peripheral cornea were selected for nuclear staining. Retinal slides containing sections with the optic nerve head protruding into the retina and intact retinal sub-layers were selected for nuclear, FLAG, and PAX6 staining. Prior to staining, all sections were subjected to deparaffinization. In brief, the sections were immersed in the following order in 2× 100% xylene, 1× 100 xylene 1:1 100% EtOH, 1× 100% EtOH, 1× 95% EtOH, and 1× 70% EtOH and 50% EtOH; thereafter, the sections were kept in dH2O. Upon antigen retrieval, the sections were removed from dH2O and incubated in antigen retrieval buffer (20 mL 10× Dako Target Retrieval Solution [#S1699, Agilent Technologies, Santa Clara, CA, USA] with 180 mL dH2O) in a 95° water bath (70 min). The sections where then rinsed in 0.1 M PBS (5 min) and 0.1 M PBS-T (10 min) before blocking with blocking buffer for 60 min at room temperature (RT). Then, the sections were incubated with the following primary antibodies at 4° overnight: FLAG (1:250, #F1804, Millipore Sigma, Burlington, MA, USA) and PAX6 (1:500, #60433, Cell Signaling, Danvers, MA, USA). The following day, the sections were rinsed in 0.1 M PBS-T (3× 10 min) and incubated with the secondary antibodies FLAG (1:1000, #A11032, Invitrogen, Waltham, MA, USA), PAX6 (1:1000, #A32734, Invitrogen), and Hoechst (1:1000, Millipore Sigma), and then nuclear stained for 2 h in a humid chamber at RT. Staining media was then removed, and slides were washed in 0.1 M phosphate buffer (PB) (2× 10 min), with a final wash in 0.01 M PB (1× 10 min). Coverslips with ProLong Gold Antifade mounting solution (Thermo Fisher Scientific) were placed on slides, and the slides were stored at 4°C at least overnight. Tissue imaging and quantification Imaging and quantification were performed as previously described.26 In brief, histology slides were imaged using the Olympus BX61 Microscope at 10× magnification in the DAPI channel, using Z stacking. ImageJ (http://rsbweb.nih.gov/ij/, version 1.52) was used to measure the thickness of corneal layers, using the straight-line tool in 2 × 500 μm windows, each located peripherally left or right of center. Windows were chosen to avoid structural abnormalities such as keratolenticular adhesions to minimize variability in the measurements, and in the majority of cases, measurements were taken near the center of the windows. Thus, two thickness measurements were obtained and averaged for each eye, for the epithelium or stromal-endothelial layers. Cell counts were performed manually in 2 × 100 μm windows by using the cell counter tool in ImageJ within the same windows. Thus, two cell count measurements were obtained and averaged for each eye, for the epithelium or stromal-endothelial layers. RT-ddPCR Corneas were homogenized, and RNA was purified using PureLink RNA Micro Kit (#12183016, Thermo Fisher Scientific). cDNA was synthesized using SuperScript VILO (#11754050, Thermo Fisher Scientific) and then diluted to 0.5 ng/μm. RT-ddPCR was performed on the QX200 Droplet Digital System (Bio-Rad, Hercules, CA, USA) using ddPCR EvaGreen Supermix (#1864034, Bio-Rad). Four ubiquitous genes were chosen based on previous publications.55 The primers for three genes, lactate dehydrogenase A (Ldha), peptidylprolyl isomerase A (Ppia), and ribosomal protein L5 (Rpl5), were selected from the same paper,55 and the fourth, TATA box-binding protein (Tbp), from a different paper.56 Endogenous Pax6, Notch1, Hes1, Wnt5a, and Mtor transcripts were quantified using commercially available assays (Mm.PT.58.12286940, Mm.PT.58.5718719, Mm.PT.58.41697865, Mm.PT.58.16402801, and Mm.PT.58.42252266, respectively, Integrated DNA Technologies, Coralville, IO, USA). The virally delivered FLAG-PAX6 transcripts were quantified with the forward primer 5′-TGGACTACAAAGACCATGACGG-3′ and the reverse primer 5′-TACAATCTTCTGCCGGGTGG-3′. All transcript values were normalized to the geometric mean of the four ubiquitous genes. Statistical analysis Generation of graphs was performed using GraphPad Prism (version 10.1.2, Boston, MA). The experimental unit was the individual eye; data are expressed as the mean ± standard error of the mean. Given the sample size, normality was assessed by visual inspections of quantile-quantile plots, rather than normality testing. The data had no major deviation from the diagonal line and were, therefore, treated as being normally distributed. An unpaired parametric two-tailed t test was used to analyze the mean difference between two groups and was considered significant at p ≤ 0.05 (GraphPad Prism). Ordinary one-way ANOVA was used to analyze the difference between more than three groups, and when significant (p ≤ 0.05), a post hoc analysis with Tukey’s multiple-comparisons test was performed (GraphPad Prism). Simple linear regression was performed to investigate the correlation between epithelial thickness and cell count and between the stromal-endothelial thickness and cell count in the injected eyes and uninjected eyes, respectively (GraphPad Prism). For endogenous transcripts, linear regression models were fit separately for each gene to estimate mean differences in gene expression by treatment group, using SAS (version 9.4, SAS Institute Inc., Cary, NC, USA). Data for all genes were then pooled and analyzed with a mixed-effects linear regression model. A random effect for a mouse was included to adjust for repeated measures from the same mouse, and the model was adjusted for gene. The ROUT test was used to identify outliers. Data and code availability Plasmids pEMS2075 and pEMS2244 used in this study are available to the research community through the nonprofit distributor Addgene (www.addgene.org). All mouse strains used in this study are available through the nonprofit distributor JAX (www.jax.org). Acknowledgments The authors thank Melissa Braschel (Biostatistical Analyst, Clinical Research Support Unit, BC Children’s Hospital Research Institute) for the calculation of linear regression models. We also thank the British Columbia Children’s Hospital Research Institute (BCCHRI) Core Technologies and Services for paraffin embedding of eyes and tissue sectioning. The authors also thank the Gut4Health Microbiome Core Facility at BCCHRI (RRID: SCR_023673) for support in establishing the RT-ddPCR assays used in this study. Finally, this work was funded by a Canadian Institutes of Health Research (10.13039/501100000024CIHR) grant awarded to E.M.S. (PJT-159484, NLTR GR008316). Author contributions Conceptualization, funding acquisition, and supervision: E.M.S.; formal analysis, S.G.K. and T.C.L.; investigation and validation, S.G.K., T.L.C., S.Z.M.M., D.D., and A.J.K.; visualization, S.G.K.; writing – original draft, S.G.K., T.L.C., S.Z.M.M., D.D., and E.M.S.; writing – review and editing, S.G.K. and E.M.S. Declaration of interests The authors declare no competing interests.