TNFα-dependent modulation of WT1-MMP9 regulatory axis links developmental and inflammatory pathways in glaucoma

Summary

Glaucomas are heterogeneous optic neuropathies associated with extracellular matrix dysregulation, abnormal ocular morphogenesis, and inflammatory signaling. Targeted deep sequencing of 586 primary congenital glaucoma (PCG) cases and 1,757 controls identified rare pathogenic variants in multiple genes, including WT1 and MMP9. Notably, WT1 variants clustered within the nuclear export sequence. Further, functional analyses showed that combined wt1-pax6 suppression in zebrafish disrupted ocular morphogenesis, highlighting developmental interdependence. In human trabecular meshwork cells, WT1 acted as a transcriptional repressor of MMP9, while TNF-α signaling triggered nitric oxide-dependent nuclear export of WT1, resulting in delayed MMP9 upregulation. This effect was reversible by inhibiting nuclear export or nitric oxide synthase. A patient-derived mutation in the nuclear-export region of WT1, disrupted this regulatory switch, causing abnormal MMP9 expression. These findings position WT1 as an important regulator linking developmental and inflammatory mechanisms in glaucoma pathogenesis.

Graphical abstract

Highlights

• •Glaucoma-associated WT1 variants cluster within the nuclear export sequence
• •WT1 perturbation produces ocular developmental abnormalities in zebrafish
• •WT1 directly represses MMP9 expression in trabecular meshwork cells
• •TNFα-induced nuclear export of WT1 controls inflammatory induction of MMP9

Molecular biology; Molecular mechanism of gene regulation; Developmental biology

Introduction

Primary congenital glaucoma (PCG) and primary angle closure glaucoma (PACG) are at two ends of the broader glaucoma spectrum disorders. PCG is an extreme form of developmental glaucoma that occurs due to developmental defects in the trabecular meshwork (TM) and anterior chamber angle and manifests at birth to early infancy.¹ If untreated, these diseases lead to visual impairment and blindness. The other end of the spectrum harbors PACG associated with minor iris defects such as plateau and steep iris and pose a risk for narrow occludable angle, thereby resulting in raised IOP and subsequent optic nerve damage.² The spectrum of glaucoma, including PCG and PACG, are complex diseases with multifactorial etiologies. PCG occurs due to multiple events during ocular development that are regulated by precisely coordinated interactions of transcription factors and signaling molecules expressed in a spatiotemporal fashion.³ PACG, on the other hand, is an age-related condition with subtle changes in the anterior chamber angle and is progressive in nature.⁴ Both these diseases are attributed to multiple genes and their interactions lead to pathogenesis.

While direct genetic overlaps between PCG and PACG are limited, several common pathways have emerged across these conditions. Rather than being driven by single-gene defects, both disorders are increasingly recognized as the consequence of perturbations in interconnected developmental, extracellular matrix (ECM), and inflammatory pathways, making the investigation of shared regulatory networks an important complementary approach to genetic studies. Variants in genes regulating ECM remodeling (e.g., MMP9, LTBP2, and collagens^5,6,7), anterior segment morphogenesis (PAX6, FOXC1, PAWR, and WT1^8,9,10,11), and immune-ECM crosstalk (ST18, TNF^12,13) highlight the shared molecular mechanisms that converge on dysregulated ECM turnover and abnormal anterior chamber angle development. Rare pathogenic variants in the known candidate genes account for a small proportion of PCG and PACG phenotypes and do not fully explain their molecular basis of pathogenesis. While PCG is primarily attributable to variants in the CYP1B1, LTBP2, TEK, MYOC, and FOXC1 and their interactions,^14,15,16,17 PACG on the other hand is an age-related condition and 8 loci including genes like PLEKHA7, COL11A1, PCMTD1,⁷ and ABCC5¹⁸ have been associated through genome-wide association studies (GWAS). A haplotype-based GWAS on PACG suggested association of CNTNAP5 genic region in Indian PACG patients,¹⁹ which was further validated functionally using zebrafish model.²⁰ Additionally, a few candidate genes such as HGF,²¹ MMP9,²² HSP70,²³ and eNO²³ have also been implicated in PACG pathogenesis, although their functional characterizations are not fully elucidated. Collectively, these observations suggest that defining the molecular mechanisms linking developmental regulation, ECM remodeling, and inflammatory signaling may provide insights that extend beyond individual genetic associations.

To better understand the molecular mechanisms underlying these interconnected pathways, we screened a cohort of PCG patients negative for mutations in known candidate genes. The NGS panel sequencing revealed mutations in 10 different genes. Among the identified pathogenic genes, transcription factor WT1 and its reported target MMP9 have been reported with PCG¹⁷ and PACG,¹² respectively. WT1, being a transcription factor, is known for its promiscuity, acting as both activator and repressor of transcription in various tissue types.²⁴ Bioinformatic analysis of the MMP9 promoter identified putative binding sites for both WT1 and PAX6. PAX6 being one of the master regulators of the eye and central nervous system morphogenesis, linked with a debilitating eye anterior segment defect, aniridia, when mutated. Interestingly, WT1 is in close genomic proximity with PAX6.²⁵ In WAGR syndrome, where aniridia is one of the major pathophenotypes, a contiguous gene deletion encompassing both PAX6 and WT1 has been observed.²⁶ MMP9 is an ECM metalloproteinase, and its optimum expression and proper function is important for anterior segment development.²⁷ MMP9 expression is often closely regulated by transcription factors and cytokines like TNF-α.²⁸ In PACG, higher levels of MMP9 and TNF-α are reported in patients across multiple studies.²⁹ TNF-α, reportedly, plays a significant role both in neuroprotection and in neurodegeneration by regulating the activation and downregulation of pro-apoptotic signals.³⁰ A promoter variant of TNF-α, c.-308G>A (rs1800629) is associated with different subtypes of glaucoma across multiple ethnicities.^{31,32,33,34,35,36,37,38,39,40,41} Together, these observations suggest that defining the molecular mechanisms linking developmental regulation, ECM remodeling, and inflammatory signaling may provide insights that extend beyond individual genetic associations.

Here, we investigated the molecular relationship between WT1 and MMP9 and examined how inflammatory signaling regulates this interaction. The identification of rare WT1 and MMP9 variants in our PCG cohort provided the biological rationale for investigating the WT1-MMP9 regulatory axis, with the objective of defining its underlying molecular mechanism rather than establishing a direct genetic contribution of WT1 to glaucoma. We hypothesized that WT1 functions as a transcriptional regulator of MMP9 and that TNF-α dependent nucleocytoplasmic trafficking of WT1 modulates this regulatory interaction. To test this hypothesis, we investigated the regulation of WT1 localization and MMP9 expression using inflammatory and pharmacological perturbations in human trabecular meshwork (HTM) cells and complementary zebrafish studies, and further evaluated the functional consequences of a patient-derived WT1 variant within this regulatory framework.

Results

Identification of potentially pathogenic variants through targeted gene panel sequencing

Deep sequencing followed by further validations in 586 PCG cases revealed 29 potentially pathogenic heterozygous variants in ten genes associated with glaucoma (COCH, FBLN5, HSF4, LOXL1, MMP9, and WDR36), anterior segment dysgenesis (WT1), and optic atrophy (ACO2, NBAS, and OPA1). These variants were either absent or rarely present in the control subjects (n = 1757). Among these ten potentially pathogenic genes that exhibited rare pathogenic variants in this cohort, WT1 and MMP9 were prioritized for further investigation based on bioinformatic network analyses. Gene regulatory network analysis performed using the TF-gene interaction module in NetworkAnalyst, incorporating data from the TRRUST database, identified WT1 as a potential upstream transcriptional regulator of MMP9 (Figure S1).^42,43,44 This predicted regulatory relationship was further supported by protein-protein and functional association analyses using the STRING and GeneMANIA database (Figure S2).^45,46

Overall, we observed 62 rare variants in WT1 (Table S3), of which four were pathogenic across four unrelated PCG cases (Figure 1A). All the pathogenic WT1variants were restricted to the nuclear export site (NES) (Figure 1B). Wild-type amino acid residues of these four pathogenic variants are highly conserved across species (Figure 1C). The REVEL scores for the p.P264L and p.M250K variants and ACMG criterion (PM2 supporting) indicated evidence of pathogenicity. Likewise, we observed 45 rare variants in MMP9 in PCG cases (Table S4). Among them, two variants p.R98Q and p.W116C were pathogenic (PM2 supporting), while the p.L187F that co-occurred with a WT1 variant (p.M250K) was strongly pathogenic (ACMG criterion: PS3). The pathogenic variants in both these genes were rarely detected in our controls and other global databases. Segregation analysis of the families of the probands harboring pathogenic WT1 (n = 4) and MMP9 (n = 2) alleles were done. The data revealed that one of the parents of the probands in these families carried either a copy of the heterozygous WT1 or MMP9 allele and were asymptomatic. While some probands harbored only a heterozygous allele, we believe the other pathogenic allele is yet uncharacterized and may be unraveled through further assessments of the whole genome. Further, PCG is known to harbor co-occurring alleles across multiple genes based on their genetic and physical interactions.^17,47 Additionally, a complete list of common variants in WT1 (n = 17) and MMP9 (n = 11) are provided in Tables S3 and S4, respectively. However, there was no association with any of the common variants of WT1 and MMP9 in PCG.

Figure 1.

Pathogenic variants in WT1 and MMP9 identified in PCG patients

(A) Representative clinical data of patients harboring rare pathogenic alleles of WT1 and a combination of WT1 and MMP9. The table at the bottom indicates their pre-and post-operative corneal clarity and visual acuity.

(B) Pedigree information of PCG patients with WT1 pathogenic variants.

(C) Schematic showing different domains of WT1 along with the location of the pathogenic variants.

(D) All the amino acid residues of pathogenic variants in WT1 are conserved across mammalian species,.

(E) Pre and post-operative clinical data of patients. PCG = primary congenital glaucoma, IOP = intraocular pressure, CD = corneal diameter, CDR = cup to disc ratio, OD = right eye, OS = left eye, NPL = no perception of light, FFL = fixing and following light, VA = visual acuity, 1∗hazy cornea and buphthalmos with Haab’s striae, 2∗ Haab’s striae, 3∗ leucomatous grading corneal scar, 4∗ cornea clear, 5∗ megalocornea with mild haze, 6∗ clear cornea “- ” = data not available.

Genotype-phenotype correlation indicated relatively poor prognosis in terms of post-operative IOP control and visual acuity in patients harboring WT1 variants (Figure 1D). As evident from the figure, the patient (PCG159) harboring the p.P264L had poor corneal clarity and fixing and following of light (FFL). Interestingly, we observed the co-occurrence of heterozygous WT1 (p.M250K) and MMP9 (p.L187F) variants in a PCG case (PCG799). Importantly, the corneal clarity resolved, and visual acuity improved in this patient harboring heterozygous alleles of WT1 and MMP9 following post-operation (Figure 1E).

Knockdown of wt1 and pax6 genes induces distinct ocular and developmental defects in zebrafish

In case of morpholino-mediated wt1 knockdown in zebrafish, morphological abnormalities emerged predominantly between 6 and 7 days post fertilization (dpf) (Figures 2A and 2B). These included gross craniofacial deformities, curvature of the body, and prominently enlarged pericardial sacs indicative of heart edema (Figure 2B). wt1 morphant embryos exhibited a microphthalmic phenotype. Quantitative measurements of the major and minor ocular axes confirmed a significant reduction in eye size compared with control embryos (Figure S3). The gross phenotypes observed upon WT1 knockdown in zebrafish were further validated by generating F0 mosaic crispants. Cas9-only injected control embryos did not display ocular or developmental abnormalities. Importantly, wt1a crispants reproduced the ocular phenotypes previously observed following wt1 morpholino-mediated knockdown. The CRISPR-mediated gene disruption was also sequence validated by sanger sequencing (Figures 2A and 2B).

Figure 2.

Phenotypic characterization of wt1(a/b) morphants, wt1 knockout and pax6 morphants

(A) Schematic showing the wt1 splice site blocking morpholino binding site.

Exon 1 of wt1a was targeted for CRISPR-Cas9 mediated knockout. Sanger sequencing showed random indels in wt1a knockout zebrafish.

(B) Comparable morphological phenotypes between wt1 a/b morphant (MO) and wt1a knockout fish (crispant) with mismatch control (MM); scale bar, 0.5 mm.

(C) Scatterplot showing gross morphological phenotypes observed including cranial deformation, severe bending, smaller eye size and edema between morphants and mismatch controls. Data are presented as mean and error bars represent standard error of the mean (SEM). Each data point represents one biological replicate (N = 3). Statistical significance was assessed by unpaired t test; ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001 and ∗∗∗∗p < 0.0001.

(D) Schematic representation of pax6a and pax6b morpholino binding site. Translation blocking (TB) and splice site (SC) blocking morpholinos were used for pax6a and pax6b, respectively.

(E) Morphological phenotypes observed included mainly iris deformities, cranial deformation, tail bending and edema. Scale bar, 0.5 mm.

(F) Scatterplot shows comparison between morphants and mismatch controls. MO- pax6 morphant, MM- Mismatched control. Each data point represents one biological replicate (N = 3). Error bars represent standard error of the mean (SEM). Statistical significance was assessed by unpaired t test ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001 and ∗∗∗∗p < 0.0001.

In pax6 morphants, distinct ocular phenotypes were evident by 5 dpf, including significant reduction in eye diameter and features resembling aniridia (Figures 2D and 2E). Additionally, these embryos exhibited a curved body axis and bent tail, none of which were observed in control groups (Figure 2F). Quantitative measurements of eye diameter at both 5 and6 dpf confirmed a significant reduction in pax6 morphants compared to mismatched and uninjected controls (Figure S3).

Genetic interaction between wt1 and pax6 revealed by combined morpholino knockdown

To investigate a potential genetic interaction between wt1 and pax6 in zebrafish eye development, morpholino doses were serially diluted to determine sub-effective concentrations that produced no visible phenotypes when injected individually (Figure S4). These diluted doses for wt1 and pax6 were then combined and co-injected into zebrafish embryos at the single-cell stage.

By 5 dpf, embryos injected with the combined sub-effective dose morpholinos exhibited a range of developmental abnormalities, including reduced eye diameter, iris malformations, curved tail, bent body, and pericardial edema (Figures 3A, 3B, and S5). These phenotypes were not observed in embryos injected with the individual sub-effective doses.

Figure 3.

Combination of sub effective dose morpholino injection reverted ocular phenotype

(A) Combination sub-effective (SE) dose morpholino injected fish showed severe ocular as well as gross morphological phenotypes in comparison to combination mismatch control fish; scale bar, 0.5 mm.

(B) Scatterplot showing gross morphological phenotypes observed including cranial deformation, severe bending, smaller eye size and edema between morphants and mismatch controls. Each data point represents one biological replicate (N = 3); error bars represent mean ± SEM; unpaired t test is performed to assess statistical significance.

(C) H&E staining of effective dose of wt1(a/b) morpholino, effective dose of pax6 (a and b) morpholino and combination sub effective morpholino injected fish for deep phenotyping showed changes in the different neuronal layers and lens formation with respect to the mismatch controls; scale bar, 200 μm. Representative brightfield images are also shown. SE- sub-effective, scale bar, 0.5 mm. Unpaired t test is performed to assess statistical significance; ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001 and ∗∗∗∗p < 0.0001.

To further examine eye-specific defects, the affected embryos were fixed, embedded in OCT medium, and cryosectioned primarily through the head region to capture the eyes. Hematoxylin and eosin staining revealed disrupted ocular architecture, providing histological evidence that both wt1 and pax6 are in a genetic interaction which is instrumental in regulating zebrafish eye development (bottom panel, Figure 3C), whereas pax6 knock-down showed incomplete ocular development (middle panel, Figure 3C) and wt1 knock-down indicated ocular phenotype is only restricted to reduction in the eye size (upper panel, Figure 3C).

Combinatorial regulation of WT1 and PAX6 on human MMP9 promoter

Chromatin immunoprecipitation followed by PCR (ChIP-PCR) demonstrated that both WT1 and PAX6 bind to the MMP9 promoter region in HTM cells. Specific amplification of MMP9 promoter fragments was observed in the samples immunoprecipitated with WT1 and PAX6 antibodies, as indicated by distinct bands on the agarose gel (Figures 4A and 4B). No such bands were detected in the GFP antibody (negative control) immunoprecipitated samples, confirming the specificity of the assay. The input control yielded strong amplification, validating the presence of the target sequence in the chromatin sample, while the histone H3 acetylation antibody served as a positive ChIP control. These findings confirm that both WT1 and PAX6 have binding sites on MMP9 promoter in HTM cells.

Figure 4.

WT1 over expression represses the MMP9 promoter transactivation by binding to its promoter

(A) Schematic showing predicted WT1 and PAX6 binding sites on MMP9 promoter.

(B) ChIP PCR showing WT1 and PAX6 have binding affinity to the MMP9 promoter regions.

(C) WT1 overexpression represses MMP9 promoter trans-activity. Short, mid and large fragments of MMP9 promoter were cloned as depicted in the schematic diagram. Each data point represents average value of relative luciferase activity in fold change from one biological replicate; N = 3. Error bars represent mean ± SEM; unpaired t test is performed to assess statistical significance. Representative blot for WT1 overexpression (D) and its densitometric analysis (E).

(F) Enzymatic activity of MMP9 upon WT1 overexpression (OE) condition. Data points represent gelatinase activity in fold change; N = 4; error bars represent mean ± SEM. Unpaired t test is performed to assess statistical significance; ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001 and ∗∗∗∗p < 0.0001.

To further investigate the transactivation effect of WT1 and PAX6 on MMP9 promoter, three constructs have been created, and dual luciferase assay was performed to show how WT1 overexpression modulates MMP9 promoter activity differentially, depending on the presence of canonical and non-canonical PAX6 binding motifs. It is to be noted that the HTM secondary cell line has abundant PAX6 but low WT1 endogenous expression (Figure S6A). In the construct containing only the WT1 binding site (short), WT1 acted as a transcriptional repressor, significantly reducing luciferase activity upon WT1 overexpression compared to the control. The mid construct, which included both WT1 and a non-canonical PAX6 binding site, exhibited a moderate but statistically insignificant increase in luciferase activity upon WT1 overexpression. In contrast, the large construct, containing WT1 site, a non-canonical PAX6 site, and a canonical PAX6 paired domain (PD) site, showed a significant increase in luciferase activity (Figure 4C). We used MCF7 cell line as a negative control for our luciferase assay as it doesn’t show any detectable expression of both WT1 and PAX6 (Figure S6A). It was observed that there was no significant change in the luciferase activity upon WT1 overexpression in MCF7 cell line (Figure S6B). Moreover, it has also been observed that transiently overexpressing WT1 in HTM cell line causes a reduction in the MMP9 protein levels (Figure 5H) and gelatinase activity (Figure 4F).

Figure 5.

WT1 and TNF-α functionally interact to regulate MMP9

(A) Summary of epigenomic signatures of TNF-α promoter region showing the positions of rs1800629 and rs2516482.

(B) rs1800629 lies in WT1 binding motif in the TNF-α promoter. Allele-specific luciferase activity of TNF-α promoter under.

(C) No treatment and (D) TNF-α treated conditions. Each data point represents relative luciferase activity in fold change from one independent biological replicate; N = 3; error bars show mean ± SEM; ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001 and ∗∗∗∗p < 0.0001. Relative gene expression levels of (E) TNF-α and (F) MMP9 under wild type WT1 and P264L mutant WT1 overexpression and TNF-α treatment combination conditions. Individual data points represent relative expression in fold change from one biological replicate; N = 3.

(G) Gelatinase activity of MMP9 upon WT1 overexpressed (HTM WT1), TNF-α treated and combination (TNF HTM WT1) conditions. Data points represent average gelatinase activity in fold change from independent biological replicates; N = 3.

(H) Immunoblot of MMP9 with WT1 overexpressed (WT1 OE), TNF treated and combination (TNF WT1 OE) conditions.

(I) Densitometric analysis of MMP9 intensity; Individual data points represent relative expression in fold change from one biological replicate; N = 3; error bars represent mean ± SEM. Unpaired t test is performed to assess statistical significance; ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001 and ∗∗∗∗p < 0.0001.

TNF-α promoter SNP rs1800629, associated with late onset PACG, harbors WT1 binding site and shows allele specific regulatory activity

In a previous study by Chakraborty et al., rs2516482 showed a statistically significant association with PACG.¹⁹ Interestingly, rs1800629, located in the promoter region of TNF-α, showed complete LD (linkage disequilibrium) with rs2516482, whose association with glaucoma has already been reported in multiple studies. Independently, rs1800629 showed statistically significant association both in our discovery (p = 4.27 × 10⁻⁶) and replication (9.89 × 10⁻³) cohort of PACG GWAS (Figures S7A and S7B). All these reports including ours have been listed with ethnicities and odds ratio (OR) per A allele (Figure S8A). From the forest plot it is obvious that the A allele is bidirectional in terms of being “risk” or “protective” in glaucomatous neurodegeneration (Figure S8B). These observations encouraged us to look more closely at how rs1800629 is involved in the disease context. Further, we have plotted four epigenomic features—CTCF, DNase, H3K4Me3, and H3K27Ac of the region (Figure 5A). This 100 kb region harbors both rs2516482 and rs1800629, among which rs1800629 falls in the active chromatin region, potentially involved in regulation of TNF-α. Transcription factor binding site analysis in MEME suite revealed that rs1800629 falls in a region having WT1 binding motif (Figure 5B). For determining the role of regulatory activity of rs1800629 on TNF-α expression, we performed a dual luciferase reporter assay. Luciferase assay showed that the promoter activity was significantly increased for the plasmids containing allele A in comparison to the allele G in HTM cells (Figure 5C). The trend remains the same with overexpressed WT1. Upon TNF-α treatment, the relative luciferase activity was still higher in the A allele constructs, but the trend becomes non-significant when we compared G and A allele constructs with WT1 overexpression (Figure 5D).

TNF-α and WT1 functionally interacts to regulate MMP9

We have observed from previous experiments that WT1 represses MMP9 gelatinase activity (Figure 4F) and that there is an interaction between WT1 and TNF-α, evident from the promoter activity assays. It is well known that TNF-α induces MMP9 through the NF-κB pathway, so we wanted to check if WT1 exerts a transcriptional regulation on TNF-α and MMP9 and how it changes under exogenous TNF-α stimulated conditions. Thus, we exposed untransfected, wild-type WT1 overexpressed and mutant WT1 (P264L) overexpressed HTM cells to TNF-α (50 ng/mL) for 24 h. Further, qPCR assay confirmed that WT1 behaves as a transcriptional repressor for both TNF-α and MMP9 genes. Interestingly, exogenous TNF-α stimulation upregulated both MMP9 and TNF-α transcription by several folds in both untransfected and WT1 overexpressed conditions. It was also very intriguing to note that this exogenous TNF-α-mediated upregulation of MMP9 and TNF-α was severely suppressed if a mutant (P264L) WT1 clone was transfected (Figures 5E and 5F). This is one of the WT1 mutations that was reported after the PCG cohort patients, and it lies in the nuclear export sequence of the WT1 transcription factor. Transient WT1 overexpression causes a repression of the MMP9 protein levels and gelatinase activity, whereas TNF-α treatment reverses this repression resulting in a significant upregulation of MMP9 protein levels and activity even under WT1 overexpressed condition (Figures 5G and 5H).

TNF-α induces translocation of WT1 from the nucleus and upregulation of MMP9 in a time dependent manner

While exploring the relationship between WT1 and TNF-α, an earlier report suggested TNF-α-induced translocation of WT1 from nucleus to cytosol through nitric oxide-mediated pathway in lung epithelial cells.⁴⁸ To investigate if the reversal of repressed MMP9 expression by TNF-α is due to WT1 translocation from nucleus to cytosol, HTM cells were treated with TNF-α (50 ng/mL) and the subcellular localization was examined using immunofluorescence assay across different time points. Under no treatment condition, WT1 is mostly seen to be localized in the nucleus with some diffused cytoplasmic staining (Figure 6Ai). Interestingly, following 10-min treatment with TNF-α, a perinuclear punctate staining of WT1 is observed with reduced nuclear staining indicating translocation of WT1 from nucleus (Figure 6Aii). Further, upon treatment with TNF-α at 30 m, 1, 2, and 3 h, the presence of WT1 puncta is observed in the cytosol and nuclear intensity remains low (Figure 6Aiii–vi). As time progresses the size of cytosolic puncta reduces, and their distribution seems to diffuse. At 6 h, WT1 nuclear intensity is comparable with the untreated condition suggesting restoration of nuclear WT1 (Figure 6Avii). At 24 h, WT1 intensity appears to be slightly higher than the untreated condition (Figure 6Aviii). The nuclear intensity and puncta number are plotted for representation (Figures 6B and 6C).

Figure 6.

TNF-α induced translocation of WT1 and upregulation of MMP9

(A) WT1 staining in TNF-α treated conditions show its translocation from the nucleus to cytoplasm (perinuclear localization). (i) Untreated HTM cells (UT) show nuclear localization. At 10, 30 min, 1, 2, 3 h perinuclear puncta are observed, which diminish at time point 6 and 24 h (ii–viii). The white boxes highlight the localization of WT1 and shown as inset; scale bar, 20 μm.

(B) Scatterplot shows nuclear intensity of WT1 across time points. Each data point represents average nuclear intensity from one independent experiment; N = 3.

(C) WT1 puncta numbers in the cytoplasm across these time points are plotted. Each data point represents average of a field from three biological replicates; n = 6.

(D) Immunoblot for MMP9 across time points of TNF-α treatment. Upper band represents pro MMP9 and lower band represents active MMP9.

(E and F) Densitometric analysis of active and pro MMP9 levels. Individual data points represent relative expression in fold change from one biological replicate; N = 3; Error bars are presented as mean ± SEM. Statistical significance was assessed by unpaired t test.; ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001 and ∗∗∗∗p < 0.0001.

Next, we investigated whether WT1 localization has any impact upon MMP9 expression. For this, we treated HTM cells with TNF-α (50 ng/mL) for similar time points. In untreated conditions MMP9 has lower expression, which remains to be similar up to 4.5 h time point. At 6.5, 8, and 12 h time points, we observed a significant upregulation in both pro and active MMP9 levels (Figures 6D–6F).

Nuclear export inhibitor leptomycin B restricts WT1 in the nucleus and represses MMP9 expression

WT1 localization and corresponding MMP9 levels were quite intriguing in terms of their temporal regulation. WT1 translocation is more evident in the earlier time points of TNF-α treatment whereas MMP9 levels do not increase in those time points; rather, they increase later. Hence, we intended to block the nuclear export using nuclear export inhibitor leptomycin B (LMB) and observe its effect on WT1 localization and MMP9 expression. HTM cells were treated with 20 nM LMB prior to TNF-α treatment. Confocal imaging of the cells revealed LMB treatment resulted in inhibition of TNF-α-induced WT1 translocation. At 10 and 30 m time points reduced perinuclear puncta were observed suggesting nuclear arrest of WT1 (Figure 7Ai–v). The nuclear intensity and puncta number summarize the observation (Figures 7B and 7C). At the 3 h time point we can see no visible changes in the localization consistent with our previous finding.

Figure 7.

Leptomycin B inhibits the TNF-α induced nuclear export of WT1 and upregulation of MMP9

(A) Untreated cells show nuclear localization (i) and TNF-α treated conditions at 10 and 30 m show perinuclear localization (ii and iv). LMB pretreated cells show WT1 localization in the nucleus at both 10 and 30min time points (iii and v); scale bar, 20 μm. White boxes highlight the localization of WT1.

(B) Scatterplot shows nuclear intensity of WT1 across different treatment conditions. Data points represent average nuclear intensity from five fields of three independent experiments, N = 3.

(C) WT1 puncta numbers in the cytoplasm across these conditions are plotted. Each data point represents average of a field from three biological replicates; N = 3.

(D) Immunoblot for MMP9 across time points of TNF-α treatment and LMB treated cell lysates.

(E) Densitometric analysis of active MMP9 levels. Individual data points represent relative expression in fold change from one biological replicate; N = 3; Error bars are presented as mean ± SEM . Statistical significance was assessed by unpaired t test. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001 and ∗∗∗∗p < 0.0001.

While looking at MMP9 expression levels, at 10 m time point there is no evident change between LMB treated and untreated groups. At 10 h and 12 h time points, there is a drastic reduction of MMP9 levels in the LMB treated groups in comparison to the only TNF-α treated groups (Figures 7D and 7E). These results indicated WT1 translocation at earlier time points triggers MMP9 upregulation in later time points that can be reverted by arresting WT1 in the nucleus.

L-NAME inhibits NOS activity and restricts TNFα mediated MMP9 upregulation

Nuclear export inhibitor LMB treatment resulted in inhibition of WT1 translocation and reduced MMP9 levels, which might involve other factors along with WT1. Hence, we tried to look at these events upon using L-NAME (Nω-Nitro-L-arginine methyl ester hydrochloride), a specific inhibitor of nitric oxide synthase (NOS) which is reported to be crucial for WT1 translocation. Upon L-NAME treatment, at 10 and 30 min time points we observed localization of WT1 in the nucleus (Figures 8A–8C). Further, similar to LMB treatment, L-NAME treatment showed a reduction of MMP9 protein levels at 12 h time point in comparison to only TNF-α treated cells, suggesting the WT1 translocation might be NOS dependent, which is known to be triggered following TNF-α treatment. (Figures 8D and 8E).

Figure 8.

NOS inhibitor L-NAME inhibits WT1 translocation and restricts upregulation of MMP9

(A) Immunofluorescence of WT1 shows its localization in no treatment (i), TNF-α treated 10 and 30 min (ii and iv) and TNF-α and L-NAME co-treated cells (iii and v); scale bar, 20 μm.

(B) Scatterplot shows nuclear intensity of WT1 across different treatment conditions of TNF and LNAME. Data points represent average nuclear intensity from five fields of three independent experiments; N = 3.

(C) WT1 puncta numbers in the cytoplasm across these conditions are plotted. Each data point represents average of a field from three biological replicates; n = 8.

(D) Immunoblot for MMP9 across time points of TNF-α treatment and L-NAME treated cell lysates.

(E) Densitometric analysis of active MMP9 levels. Individual data points represent relative expression in fold change from one biological replicate; N = 3. Error bars are presented as mean ± SEM . Statistical significance was assessed by unpaired t test. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001 and ∗∗∗∗p < 0.0001.

L-NAME pretreatment rescues the LPS-mediated upregulation of MMP9 in zebrafish embryos

We wanted to replicate our cellular findings in zebrafish model, where we could mimic the inflammation in zebrafish and results in an increased production of MMP9. In line with this, a detailed literature review encouraged us to use the LPS immersion technique.⁴⁹ 6 h post LPS immersion, the 3-dpf embryos had distinct observable phenotypes namely, tail fin edema, edema around the heart, slower heartbeat (Figure S9; Videos S1 and S2) and slower locomotory movement (Video S3) in comparison to the untreated control embryos (Figure 9A). LPS-treated embryos exhibited pericardial enlargement and tail fin swelling. These morphological features are consistent with inflammatory phenotypes previously reported in zebrafish following LPS exposure, including edema associated with systemic inflammation.⁵⁰ The qPCR results confirmed significant upregulation of mmp9 mRNA in zebrafish larvae treated with LPS compared to untreated controls. In the LPS + L-NAME treated group, expression levels of mmp9 were notably reduced (Figure 9B). We have previously reported higher tubulin acetylation as a marker for retinal neurodegeneration.²⁰ Upon acetylated tubulin staining, we observed comparatively higher intensity in LPS treated embryos than the untreated ones, indicating neurodegeneration in the ganglionic cell layer (GCL) and the inner plexiform layer (IPL). However, in the LPS + L-NAME treated embryos, partial rescue has been observed depicted with significant reduction in acetylated tubulin staining, indicating potential neuroprotective effect of L-NAME (Figures 9C and 9D). NeuroTrace staining of zebrafish eye cryosections revealed significant changes in neuronal density and structure following LPS treatment. Sections from larvae exposed to LPS displayed reduced staining intensity of Nissl granules in retinal neurons, indicating neuronal degeneration (Figures S10A and S10B).

Figure 9.

L-NAME pretreatment rescues the LPS mediated upregulation of mmp9 in zebrafish embryos

(A) Gross morphological changes observed in LPS treated zebrafish embryos compared to untreated embryos. The embryos exhibited phenotypes like heart and tail fin edema. Tail fin edema is shown in the lower figure and the white arrows mark the area that shows the phenotype. Scale bar, 500 μm.

(B) qPCR shows L-NAME treatment reducing the LPS induced mmp9 upregulation. Each dot represents an independent biological replicate; N = 3; error bars show mean ± SEM.

(C) Acetylated tubulin staining showing higher intensity in LPS treated sections and relatively reduced intensity LPS+L-NAME treated embryos. The white boxes compare regions in the retinal layers in the three representative images. White arrowheads compare the acetylated tubulin staining in the GCL and IPL layers across the three groups.

(D) Acetylated tubulin ROI mean intensity analysis of the treatment conditions. Each dot represents an independent experiment; error bars show mean ± SEM. Unpaired t test is performed to assess statistical significance; ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001 and ∗∗∗∗p < 0.0001. GCL, ganglionic cell layer; IPL, inner plexiform layer; INL, inner nuclear layer; OPL, outer plexiform layer; ONL, outer nuclear layer; PR, photoreceptor layer; RPE, retinal pigmented epithelium.

Discussion

Our study identifies a TNF-α responsive WT1-MMP9 regulatory mechanism integrating developmental and inflammatory signaling pathways that are relevant to ocular homeostasis and glaucoma (as depicted in the graphical abstract). This connection is particularly significant as MMP9 is a vital ECM remodeling enzyme involved in regulating aqueous humor dynamics and IOP—two critical elements in glaucoma pathology.

Pathogenic variants in WT1 and MMP9 and their interactions in PCG

Genetic screening revealed that a proportion of PCG cases harbored rare variants in WT1 and MMP9 genes, thereby expanding the mutation spectrum of these genes (Tables S3 and S4). Notably the 4 mutations observed in the nuclear export sequence of WT1 in PCG patients signify the importance of WT1’s cellular localization in the disease context. We also observed the co-occurrence of WT1 and MMP9 variants in PCG suggesting a possible genetic interaction. Our earlier studies have revealed such co-occurrences in CYP1B1 and TEK indicating genetic and physical interactions.⁴⁷ Genotype-phenotype correlation indicated relatively poor prognosis in cases harboring WT1 variants. Interestingly, we have observed the patient harboring both WT1 and MMP9 variants demonstrated relatively better prognosis compared to those with only WT1 variants. The co-occurrence of these heterozygous alleles provided insights on improved post-operative outcomes in terms of corneal clarity and visual acuity. Notably, all WT1 and MMP9 variants identified in our cohort were present in the heterozygous state, and no homozygous variants were detected. This observation is consistent with the study design, which focused on PCG cases lacking mutations in established major causative genes such as CYP1B1. The absence of homozygous variants suggests that WT1 and MMP9 may not act as classical recessive disease genes in these patients, but instead contribute as genetic modifiers within a broader pathogenic framework. The absence of homozygous variants suggests that WT1 and MMP9 are unlikely to represent classical recessive disease genes in this cohort. Rather, the identification of these rare heterozygous variants provided the biological rationale for investigating the WT1-MMP9 regulatory mechanism explored in this study. WT1 is a dosage-sensitive transcription factor whose partial functional perturbation may have substantial downstream consequences. Consistent with this notion, heterozygous deletions encompassing WT1 in WAGR syndrome are sufficient to produce developmental abnormalities, supporting the concept that WT1 haploinsufficiency can contribute to disease phenotypes.⁵¹ Furthermore, the co-occurrence of WT1 and MMP9 variants in at least one patient raises the possibility of oligogenic contributions to glaucoma susceptibility. Further bioinformatic analyses predicted WT1 to exert a transcriptional control over MMP9. These findings highlight the importance of considering multi-allelic inheritance to functionally decode gene-gene interactions in the clinical assessment of PCG.

WT1 has well-established roles in kidney development and tumor suppression, and pathogenic variants in WT1 have been implicated in disorders such as Wilms tumor, Denys-Drash syndrome, Frasier syndrome, and WAGR syndrome.⁵² Notably, many disease-associated WT1 mutations affect the zinc-finger DNA-binding domain or result in large genomic deletions that disrupt WT1 function. In contrast, the variants identified in the present study are heterozygous and cluster within the nuclear export sequence (NES), a region whose contribution to renal disease susceptibility or tumor predisposition remains poorly understood. Although our findings suggest that NES variants can alter WT1 localization and downstream regulatory functions, further clinical and functional studies will be required to determine whether these variants have broader implications beyond ocular pathology.

wt1 and pax6 genetic interaction is necessary for anterior segment morphogenesis in zebrafish

Our earlier study described a complex regulatory network of transcription factors (PITX2, FOXC1, and FOXC2) and WT1 regulator (PAWR) is critical for ocular development and associated with anterior segment defects when mutated.^11,53,54 Here, we have shown wt1 knockdown in zebrafish embryos resulted in ocular developmental abnormalities, underscoring its role in anterior segment morphogenesis. In contrast pax6 knockdown resulted in severe iris deformities like aniridia, ARS, and corectopia consistent with its established function in iris development. The specificity of the gross phenotypes observed in the wt1 morphants were validated by mosaic wt1a knockouts, which showed comparable phenotypes with respect to wt1 morphants. Previous studies have demonstrated functional divergence between the zebrafish wt1 paralogs, with wt1a exhibiting a broader developmental role than wt1b.^55,56 Consistent with this, CRISPR-mediated disruption of wt1a results in severe developmental defects, whereas wt1b mutants are largely phenotypically normal, supporting a predominant role for wt1a during development.⁵⁷ The concordance between wt1 morpholino-induced phenotypes and those observed in wt1a crispants supports the specificity of wt1a disruption in mediating the observed ocular defects. pax6 knockdowns in zebrafish resulted in well-defined iris phenotypes in the zebrafish embryos. Notably, co-injection of sub-effective doses of wt1 and pax6 morpholinos led to a reappearance of ocular and systemic defects not observed with either sub-effective morpholino alone, suggesting genetic interaction between the two transcription factors. This combined effect suggests a genetic interaction between wt1 and pax6 in anterior segment development, reminiscent of WAGR syndrome in humans.⁵¹ These findings suggest that WT1 and PAX6 operate within a shared regulatory network, important for maintaining anterior eye structure and function.

WT1-MMP9 regulatory axis is modulated by TNF-α in HTM cells

In HTM cell line, transient overexpression of WT1 exhibited differential regulation of MMP9 promoter activity depending on the presence of binding sites. This indicated that WT1 functions as a transcriptional repressor in the presence of WT1 binding site only (short) and as an activator in the presence of both WT1 and PAX6 binding motifs (mid and large). These results suggest a context-dependent regulatory role of WT1 on the MMP9 promoter, with potential combinatorial regulation when PAX6 binding sites are present. Moreover, we could not observe any significant change in the promoter trans-activity after WT1 upregulation in any of the three constructs in MCF7 cell line. This consolidates that a stable expression of PAX6 is necessary for the WT1-mediated differential promoter transactivation. ChIP-PCR confirmed direct binding of WT1 and PAX6 to the MMP9 promoter, further consolidating the transcriptional control of MMP9 by these transcription factors. The functional relevance of WT1’s transcriptional control was also highlighted by gelatinase activity assays, which demonstrated that WT1 overexpression significantly reduced MMP9 enzymatic activity, aligning with its repressive role in HTM cells. Together, these results propose an intricate model in which MMP9 expression is finely tuned by both WT1 and PAX6, with promoter architecture dictating their combined regulatory output.

The TNF-α promoter SNP rs1800629 showed a differential luciferase activity indicating a relatively higher level of TNF-α in the individuals harboring A allele. We were able to uncover that WT1 overexpression leads to reduced TNF-α expression influencing its biological activity. Though the complexity cannot be completely explained, WT1 poses to be an important regulator of TNF-α. Supporting this, a previously reported study done on lung epithelial cells, showed that TNF-α induces MMP9 via a NOS pathway and that WT1 is transported out of the nucleus upon exogenous TNF-α treatment.⁴⁸ We validated the phenomenon in HTM cells, demonstrating that TNF-α induces nuclear export of WT1 through a nitric oxide-dependent mechanism. It is evident from the WT1 puncta formation in TNF-α-treated HTM cells across different time points. The perinuclear puncta are prominent in the initial time points of treatment and gradually diffuse with time. This translocation correlated with a delayed yet pronounced upregulation of MMP9, suggesting a regulatory mechanism where WT1 repression is attenuated following cytoplasmic export. Pharmacological (small molecules) inhibition of WT1 export using leptomycin B or NOS inhibition with L-NAME prevented both WT1 translocation and subsequent MMP9 induction at later timepoints, reinforcing the role of WT1 nuclear localization in transcriptional repression. Although LMB is a well-established inhibitor of CRM1/exportin-1-mediated nuclear export,⁴⁴ its effects are not specific to WT1 and may influence the nuclear-cytoplasmic trafficking of multiple proteins. To address this limitation, we employed the mechanistically distinct NOS inhibitor L-NAME, which targets a pathway previously implicated in TNFα-mediated WT1 translocation.⁴⁸ The concordant effects of LMB and L-NAME on WT1 localization and MMP9 expression strengthen the conclusion that TNFα-induced WT1 nuclear export contributes to MMP9 regulation in this context.

Patient-derived WT1 mutation perturbs its nuclear export affecting MMP9 expression

Our qPCR analyses confirmed that presence of WT1 in the nucleus keeps both TNF-α and MMP9 expression levels at a repressed state. Exogenous TNF-α stimulation overcomes this repression by translocating WT1 out of the nucleus, significantly upregulating both TNF-α and MMP9. Interestingly, this induction was attenuated upon expression of a patient-derived WT1 mutant (P264L). This suggests the relevance of nuclear export of WT1 on MMP9 expression. A proline to leucine change at 264^th amino acid in the nuclear export sequence disrupted the nuclear export of WT1 upon exogenous TNF-α treatment, linking patient mutation to functional impairment in inflammatory regulation.

In vivo rescue of MMP9 upregulation mirrors in vitro mechanisms

To investigate whether the TNF-α/WT1/MMP9 regulatory axis operates in vivo, we used an LPS-induced inflammation model in zebrafish embryos. LPS treatment via immersion of embryos led to a robust upregulation of mmp9, mimicking inflammatory stress conditions. This suggests activation of the NF-κB pathway and its downstream targets, confirming the induction of inflammatory response. Interestingly, pre-treatment with the NOS inhibitor L-NAME significantly rescued this upregulation, confirming the involvement of nitric oxide in the inflammatory modulation of MMP9. The pericardial enlargement and tail fin swelling observed following LPS treatment are consistent with inflammatory phenotypes previously reported in zebrafish following endotoxin exposure, including edema associated with systemic inflammation.⁵⁰ Furthermore, the accompanying molecular changes observed in our study, including altered expression of mmp9, provide additional support for the inflammatory nature of these phenotypes. It has been reported in earlier studies that increased MMP9 levels are correlated with neurological disorders like Alzheimer’s disease,⁵⁸ cerebral amyloid angiopathy,⁵⁹ and ischemia.⁶⁰ In our study also we observe an increased acetylated tubulin and reduced NeuroTrace staining in LPS treated group. Higher tubulin acetylation levels are reported to be a marker for neurodegeneration and disrupted cyto-architecture.⁶¹ It was intriguing to observe that L-NAME pretreatment of zebrafish embryos showed reduction in the LPS mediated increased acetylated tubulin staining suggesting potential neuroprotective effect. These results corroborate our findings in HTM cell line and support a conserved role for the TNF-α-NO-WT1 pathway in regulating MMP9 expression. Importantly, these zebrafish experiments demonstrate the importance of inflammatory pathways in modulating anterior segment development and remodeling in vivo, relevant to mechanisms implicated in glaucoma.

MMP9 dysregulation in developmental and late-onset glaucoma

Matrix metalloproteinase-9 (MMP9), a key ECM remodeling enzyme, plays a critical role in maintaining anterior segment integrity and TM function.⁶² Its expression and activity are precisely regulated to preserve aqueous humor outflow and IOP homeostasis. Moreover, impaired eNOS activity has been implicated with elevated IOP.⁶³ Dysregulation (both downregulation and upregulation) of MMP9 is implicated in glaucomatous pathogenesis, through distinct mechanisms across different disease stages.^27,64,65

In PCG, reduced MMP9 expression or function can impair ECM turnover during early eye morphogenesis, leading to abnormal anterior segment development and impaired outflow of aqueous humor. Mouse models deficient in MMP9 exhibit anterior segment malformations, phenocopying features of congenital glaucoma.⁶²

Conversely, in late-onset PACG and other adult glaucoma, elevated MMP9 expression, often driven by inflammatory cytokines such as TNF-α and TGF-β2, results in excessive ECM degradation.^66,67 This dysregulation weakens the structural integrity of the TM and disrupts aqueous outflow resistance, contributing to chronic IOP elevation.^68,69,70 Overactivation of MMP9 can also compromise the mechanical resilience of TM tissue, ultimately disturbing the fine-tuned ECM balance necessary for normal ocular physiology.

Limitations of the study

While this study identifies a previously unrecognized TNFα-responsive WT1-MMP9 regulatory mechanism linking developmental and inflammatory signaling pathways, certain limitations should be acknowledged. The genetic findings should not be interpreted as establishing disease causality but rather as providing the biological rationale for the mechanistic investigations presented in this study. Here, we studied the functional consequences of WT1-MMP9 regulatory axis rather than establish genetic association or penetrance. In addition, the zebrafish experiments were intended to examine developmental and inflammatory mechanisms and do not fully recapitulate the complex pathophysiology of glaucoma. These findings provide a mechanistic framework for future investigations into the contribution of WT1-dependent signaling to ocular physiology and disease.

In conclusion, our findings identify WT1 as a transcriptional regulator whose activity is dynamically modulated by inflammatory signaling through TNFα-dependent nitric oxide-mediated nucleocytoplasmic trafficking. By establishing WT1 as a direct regulator of MMP9, our study identifies a previously unrecognized regulatory mechanism linking developmental and inflammatory signaling pathways relevant to ocular homeostasis. These findings provide mechanistic insight into WT1-dependent regulation of ECM remodeling and establish a framework for future studies exploring the physiological and pathological significance of the TNFα-responsive WT1-MMP9 regulatory axis in glaucoma.

Resource availability

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Dr. Moulinath Acharya (Email: ma1@nibmg.ac.in).

Materials availability

All unique reagents generated during this study, including expression constructs and other experimental materials, are available from the lead contact upon reasonable request, subject to institutional regulations and material transfer agreements where applicable.

Data and code availability

• •All data supporting the conclusions of this study are included within the article and its supplemental information.
• •This study did not generate any custom computer code or software.
• •Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

Acknowledgments

The authors would like to thank all the patients and their family members for their participation in this study. We appreciate the efforts of Drs. Ravi Thomas, Rajul Parikh, Harsha L Rao, and Sirisha Senthil in the clinical diagnosis of the PCG patients. The authors would like to thank Prof. Michael Walter, University of Alberta and Dr. Hemant Khanna, University of Massachusetts, for their critical comments on structuring the manuscript. The authors would also like to thank Dr. Sagar Sengupta, Director, BRIC-NIBMG for his insightful suggestions and critical comments throughout the study. The study was supported in parts by extramural research grants from the Indian Council of Medical Research (2021-14979/Genomics/Adhoc-BMS) to Dr. Subhabrata Chakrabarti, and Science and Engineering Research Board, Government of India (SERB; EMR/2016/006979) to Dr. Subhabrata Chakrabarti and Dr. Moulinath Acharya, Intramural fundings from the BRIC-National Institute of Biomedical Genomics (BRIC-NIBMG). Kalyani and the Hyderabad Eye Research Foundation (HERF), Hyderabad, India are duly acknowledged. T.A. is currently supported by a research fellowship from the Department of Biotechnology (DBT), Government of India. J.S. and A.K.M. are supported currently by research fellowships from the University Grants Commission (UGC), India. S.B.(45/04/2022/HUM/BMS) and A.M. (3/1/3/JRF-2020/HRD/134416/33) are currently supported by fellowships from the Indian Council of Medical Research (ICMR), Government of India and G.P. was supported by the fellowship from (09/815(0023)/2017-EMR-I) of the Council of Scientific and Industrial Research (India) (CSIR), Government of India.

Author contributions

T.A. and J.S. performed all zebrafish, biochemistry, molecular, and cell biology immunofluorescence and confocal imaging experiments and analyses, and M.K., S. Bera, G.P., and A.M. generated the human genetics data by deep sequencing and performed further validations. A.K. Mishra and R.C.K. provided the patients and controls samples and were involved in the clinical diagnosis and were involved in genotype-phenotype correlation. T.A., J.S., S. Bera, and A.M. drafted and edited the original manuscript. S.S.R. performed zebrafish maintenance and tissue processing. A.K. Mandal generated all WT1 mutant clones through site-directed mutagenesis. D.R. assisted in generating the LPS-induced model of inflammation in zebrafish embryos. S. Bagchi helped in tissue section preparation and immunofluorescence in zebrafish embryos. S. Chakraborty performed GWAS analyses and identified the associated TNF promoter SNP. M.M. supervised all the functional analyses, helped in conceiving and designing the study, reviewing and editing of the manuscript draft. S. Chakrabarti and M.A. conceived the study, secured fundings, supervised all the analyses and interpretation of results and drafted the manuscript. All authors read and approved the final version of the manuscript for publication.

Declaration of interests

The authors declare no conflict of interest.

STAR★Methods

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Anti-WT1	Sigma-Aldrich	Cat#SAB2102716; RRID: AB_1061142
Anti-MMP9	Abcam	Cat#ab283575; RRID: AB_2928971
Anti-PAX6	Abcam	Cat #ab5790; RRID: AB_305110
Anti-β-Actin	Santa Cruz Biotechnology	Cat#sc-47778; RRID: AB_626632
Anti-GAPDH	Cell Signaling Technology	Cat#2118; RRID: AB_561053
Goat anti-Mouse IgG-HRP	Abcam	Cat#ab205719; RRID: AB_2755049
Goat anti-Rabbit IgG-HRP	Abcam	Cat#ab97051; RRID: AB_10679369
Mouse anti-Acetylated Tubulin	Sigma-Aldrich	Cat#T7451; RRID: AB_609894
Goat anti-Rabbit Alexa Fluor 488	Abcam	Cat#ab150077; RRID: AB_2630356
Goat anti-Mouse Alexa Fluor 568	Abcam	Cat#ab175473; RRID: AB_2895153
Chemicals,peptides andrecombinantproteins
Recombinant Human TNF-α	Abclonal	Cat#RP00001
Leptomycin B	Sigma-Aldrich	Cat#L2913
L-NAME	Sigma-Aldrich	Cat#483125-M
Lipopolysaccharide (Escherichia coli O111:B4)	Sigma-Aldrich	Cat#L2630
OCT Compound	Tissue Tek	Cat #4583
DPX Mountant	SIgma-Aldrich	Cat#06522
DAPI	Invitrogen	Cat#D1306
ProLong Gold Antifade Mountant	Invitrogen	Cat#P36930
NeuroTrace 435/455 Blue Fluorescent Nissl Stain	Thermo Fisher Scientific	Cat#N21479
Criticalcommercialassays
RNeasy Plus Mini Kit	Qiagen	74134
SuperScript IV VILO Master Mix	Invitrogen	11756050
TranscriptAid T7 High Yield Transcription Kit	Thermo Fisher Scientific	K0441
Q5 Site-Directed Mutagenesis Kit	New England Biolabs	Cat#E0554
Dual-Luciferase Reporter Assay System	Promega	E1910
Gelatin Degradation Assay Kit	Abcam	Cat#ab234057
SuperSignal West Pico PLUS	Thermo Fisher Scientific	Cat# 34580
Experimentalmodels:Celllines
HEK293	American Type Culture Collection	Cat# CRL-1573
Immortalized Human Trabecular Meshwork (HTM) cells	Gift from Prof. Michael Walter, University of Alberta	N/A
Experimental models:Organisms/strains
Wild-type Tubingen zebrafish (Danio rerio)	NIBMG zebrafish facility	Tubingen strain
RecombinantDNA
pCMV6-WT1 expression plasmid	OriGene	Cat #RC220079
pGL3-Basic vector	Promega	212936
pRL-TK Renilla plasmid	Promega	E2241
Oligonucleotides
wt1 morpholino	Gene Tools	Sequence in Table S2
pax6a morpholino	Gene Tools	Sequence in Table S2
pax6b morpholino	Gene Tools	Sequence in Table S2
WT1 sgRNA primer	This study	aggcactcaacccacagtga
PCR primers	This study	Table S5
Software andalgorithms
CRISPOR	CRISPOR	https://crispor.gi.ucsc.edu/
STRING	STRING Consortium	https://string-db.org/
ImageJ	NIH	NA
Leica Application Suite (LAS)	Leica Microsystems	NA
NIS-Elements	Nikon	NA
GraphPad Prism	GraphPad	Version 10
QuantStudio Design & Analysis	Thermo Fisher Scientific	Used with QuantStudio 5
Instrument
ChemiDoc XRS+	Bio-Rad	NA
Tecan Infinite 200 PRO	Tecan	M Plex
Nikon A1 Confocal Microscope	Nikon	NA
Leica Stereomicroscope	Leica Microsystems	NA
Leica CM1860 UV Cryostat	Leica Microsystems	NA

Experimental model and study participant details

Enrollment of patients and controls

The study protocols adhered to the tenets of the Declaration of Helsinki and were approved by the Institutional Review Boards of the L V Prasad Eye Institute, Hyderabad, Telangana (LEC 08-15-097).

Our study cohort consisted of unrelated patients with PCG (n = 586), who were devoid of homozygous mutations in the known candidate genes (CYP1B1, LTBP2, TIE2, MYOC, FOXC1), along with their unaffected parents and relatives (n = 687) and ethnically matched normal volunteers from the same geographical region of habitat of the patients (n = 1757). Detailed clinical inclusion and exclusion criteria for the enrollment of PCG and controls, have been described earlier.^47,71,72,73 Each subject was independently diagnosed by at least two glaucoma specialists and there was a good inter-observer agreement based on the kappa statistics (κ = 0.94). Cases with discordant diagnosis were excluded. The normal volunteers (controls) belonged to a longitudinal cohort (The Andhra Pradesh Eye Disease Study [APEDS]) who are being followed up for the last 20 years⁷⁴

Animal experiments

Zebrafish husbandry and maintenance

Zebrafish (Danio rerio) were maintained in recirculating benchtop rack systems (Tecniplast. Inc, Italy) under standard laboratory conditions on a 14-h light/10-h dark cycle. Adult zebrafish were paired for natural spawning to collect synchronized batches of embryos. Typically, the Tubingen strain initiates spawning at the onset of light (around 7 AM). After feeding at 6:30 PM on the preceding day, pairwise breeding was established on a sloped breeding tank filled with zebrafish water. To maintain a 2:1 female to male ratio, four females and two males were housed in the breeding tank. When the light comes on at 7AM in the morning, they often breed the next day. Around 500 embryos were collected from the bottom of the breeding tank. After utilizing a strainer to gather the eggs, the strainer was rinsed with 1X E3 media (5 mM NaCl, 0.17 mM KCl, 0.33 mM CaCl2, 0.33 mM MgSO4, and 10−5% methylene blue) before the eggs were put into the Petri dish (with an average of 25 embryos/Petri dish) and kept in the incubator at 28.4 °C. All experiments in this study were conducted in compliance with institutional animal protocols under Committee for Control and Supervision of Experiments on Animal (CCSEA) approval (No. NIBMGZF/001/2022-23).

Cell culture experiments

Cell culture, transfections and chemical treatments

Immortalized HTM cells were obtained as a kind gift from Prof. Michael Walter in the Department of Medical Genetics at the University of Alberta. HEK293 cells were obtained from ATCC (Cat.no- CRL1573). For both these cell lines, we have not performed any authentication test, since we obtained these from reliable sources. Both the cell lines were tested against mycoplasma contamination using PCR based methods and found negative for mycoplasma. Immortalized HTM and HEK293 cells were cultured in DMEM (Gibco) supplemented with 10% fetal bovine serum (HI FBS, Gibco) and 1% anti-anti (Gibco) at 37°C in 5% CO₂. WT1 overexpression was done using pCMV6-WT1 clone (Origene) with FuGENE HD transfection reagent (Promega) as per manufacturer’s protocol. For TNFα treatment experiments, Recombinant Human TNF-alpha (Abclonal, RP00001) was used at 50ng/mL concentration unless mentioned otherwise. The treatments were given for different time points as necessary for the experiment and mentioned accordingly. Leptomycin B (Sigma Aldrich, L2913) was used to block nuclear export at a concentration of 20nM in HTM secondary cells 3 h prior to TNF-α treatment. L-NAME (Nω-Nitro-L-arginine methyl ester hydrochloride) (Sigma Aldrich, 483125-M) was co-treated with TNF-α as an inhibitor of NOS at a concentration of 500μM in HTM secondary cells.

Method details

Genetic screening

Genetic screening was accomplished by deep sequencing of our cohort of patients and controls with a targeted gene panel, which was designed using the Ion Ampliseq Designer (Life Tech, Foster City, CA). Further details are provided elsewhere.⁴⁷ Briefly, sequencing was undertaken on an Ion Torrent platform using the Ion Ampliseq chemistry following the manufacturer’s guidelines (Both from Life Tech). Bar coded targets were amplified, ligated and sequenced at an average depth of 500X (range from 120X-1100X coverage). The analysis pipeline included standard quality control measures followed by alignment with the hg19 sequence. Each sample exhibited 800-900 variants with >98% being on target. Samples that failed in the quality control measures described above were repeated. All potential variants were stringently filtered and validated by Sanger sequencing on an automated 3130 XL DNA sequencer (Applied Biosystems, CA) using the BigDye chemistry with pre-designed primers (Table S1).

Data analysis

Standard analysis packages of GATK and Varseq were used to identify the potential variants. Initially, the involvements of the observed variants to PCG were ascertained based on their segregation with the disease phenotype, frequencies in the controls, conservation of the wildtype residues, pathogenicity based on REVEL (rare exome variant ensemble learner) scores along with the revised ACMG guidelines.^75,76 The segregations of multiple pathogenic rare variants across all the genes were considered. PCG cases harboring multiple heterozygous rare pathogenic variants were additionally assessed for co-occurrences of such variants in their parents and relatives as well as normal controls. Genes exhibiting co-occurring variants were then chosen for network analysis. STRING analysis was undertaken to assess the genetic interactions of these co-occurring genes and were considered for functional analysis (as detailed below). Along with the rare variants, common variants across the WT1 and MMP9 genes were also screened for associations to PCG. Allele and genotype frequencies of each variant was calculated along with their odds ratios and 95% confidence intervals. The frequencies of the rare and associated common alleles were compared to the global databases comprising ExAC (Exome Aggregation Consortium Database), 1000 genomes and gnomAD.^77,78,79 Genotype-phenotype correlation was undertaken on cases harboring specific rare alleles based on their clinical parameters comprising IOP, corneal diameter, cup to disc ratios, visual field changes at presentation and final follow up. The severity of the disease was assessed based on prognosis at their last visit.

Microinjection of morpholino

For wt1, a splice-site blocking morpholino (MO6-wt1a), capable of targeting both wt1a and wt1b transcripts, was selected based on ZFIN database and previous literature.⁸⁰ For pax6, MO2-pax6a (a translation-blocking morpholino targeting pax6a) and pax6b-MO (targeting pax6b) were used.^81,82 Morpholinos along with the 5 bp mismatched control oligos were sysnthesized at Gene Tools, LLC. (sequences are listed in Table S2). Morpholino oligos were diluted to a concentration of 2.5 μg/μL, with 10 mM KCl and 25% Phenol Red and 1 nL (2.5 ng) was injected into wild-type zebrafish embryos at the single-cell stage.

RT-PCR validation of wt1a/b morpholino-mediated splice disruption

Total RNA was isolated was done using RNeasy plus mini kit (Qiagen) as per the kit protocol. cDNA synthesis was done using SuperScript™ IV VILO™ Master Mix (Invitrogen). To assess morpholino-mediated splice disruption, RT-PCR was performed using a forward primer located within exon 8 and a reverse primer located within the adjacent intronic sequence. Primers used in the assay are listed with their sequences in Table S5. Amplification products were resolved on an agarose gel and visualized by ethidium bromide staining. Detection of an intron-containing transcript in morpholino-injected embryos, but not in mismatch control embryos, was interpreted as evidence of aberrant splicing resulting from effective splice-blocking activity of the morpholino (Figure S11).

Quantification of ocular phenotype

To quantify the microphthalmic phenotype, embryos were imaged using a Leica stereomicroscope and ocular dimensions were measured using Leica Application Suite (LAS) software. The major and minor axes of the eye were measured from lateral images, and measurements from control and experimental groups were subjected to statistical analysis. Data are presented as mean ± SEM.

CRISPR-Cas9 mediated generation of F0 mosaic knockout zebrafish embryos

Single-guide RNAs (sgRNAs) were designed using the CRISPOR online tool (http://crispor.tefor.net/)⁸³ to target the exon 1 of zebrafish WT1a gene minimal off-target effects (wt1_sgRNA exon1-aggcactcaacccacagtga). sgRNA templates were generated and transcribed in vitro with the TranscriptAid T7 High Yield Transcription Kit (Thermo Fisher Scientific), according to the manufacturer’s protocol.

For in vivo gene editing, 2 nL of ribonucleoprotein (RNP) complex—comprising 400 pg sgRNA and 800 pg Cas9 protein—were microinjected into the yolk of one-cell stage zebrafish embryos. Embryos injected with Cas9 protein alone served as procedural controls. Injected embryos were incubated at 28.5°C and monitored for developmental and morphological phenotypes. Genomic DNA was extracted from individual embryos, and the target region was amplified and sequenced via Sanger sequencing to confirm mosaic knockout efficiency and mutational spectrum. Primers used are listed with their sequences in Table S5.

Zebrafish embryo fixation and cryo-sectioning

Five days post-fertilization (dpf) zebrafish embryos were collected into a 12-well plate. The embryos were fixed in 4% paraformaldehyde (PFA) and incubated overnight at 4°C on a rocking platform. Following fixation, embryos were washed with 1X PBST (PBS +0.1% Tween 20) for 5 minutes at room temperature. Embryos were then gradually dehydrated through a methanol series: 30%, 50%, and 70% methanol in 1X PBST, each for 10 minutes. Complete dehydration was achieved by transferring embryos to 100% methanol and storing them at −20°C for 14–24 h. For rehydration, embryos were passed through a descending methanol series (70%, 50%, and 30% methanol in 1X PBST), each for 10 minutes, followed by a final wash in 1X PBST for 5 minutes at room temperature. After PBST removal, embryos were incubated in 30% sucrose (prepared in deionized water) until they sank to the bottom of the wells, indicating adequate infiltration. Once fully sunk, embryos were embedded in OCT compound using molds and stored at −20°C until further processing. Cryosections were cut at 15 μm thickness using a cryostat (make) (LEICA CM 1860 UV) and collected directly on charged slides (Autofrost, Cancer Diagnostics). Sections were adhered to the slides by drying overnight at room temperature and stored at -80°C until use. Before immunostaining, slides were thawed at room temperature for 30–60 minutes and allowed to dry fully before initiating the staining process.

Eosin and hematoxylin staining

Cryosections were thawed individually onto plain, untreated, charged glass slides to minimize tissue degradation. The sections were fixed in 70% ethanol for 10 seconds and rinsed with deionized water. Slides were briefly dipped in fresh Mayer’s hematoxylin and then washed under a gentle stream of running water for 1 minute. After rinsing with 70% ethanol, the sections were immersed in eosin for 60 seconds. Dehydration was performed through a graded ethanol series—70%, 90%, and 100% ethanol, for 10 seconds each. The slides were then cleared in xylene for 30 seconds, air-dried at room temperature, and mounted using DPX mountant.

Site directed mutagenesis to generate mutant WT1

Mutation was incorporated in the SDM primers by using NEBaseChanger™ tool and SDM was carried out in the wild type pCMV6-WT1 clone, using Q5® Site-Directed Mutagenesis Kit (E0554) according to the manufacturer’s protocol. (P264L_FWD– GTACTCGGTGctgCCCCCGGTCT; P264L_REV- TGCTGCTCACCCAGCGAGC).

Immunoblotting and antibodies

Cell lysates were prepared using appropriate cell lysis buffer and protein concentrations were estimated using Bradford reagent (Bio-Rad). Equal amounts of protein (10-50ug) were resolved by 8-12% SDS-PAGE gel and transferred to the nitrocellulose membrane. The membranes were blocked with 5% non-fat milk in TBST (Tris-buffered saline-Tween 20) for 1 h and incubated with primary antibodies overnight at 4°C. The primary and secondary antibodies along with their dilutions are: anti-WT1, anti-MMP9, anti-Pax6, anti-Beta actin (1:5000), anti-GAPDH (1:1000, 2118, CST), Goat anti-Mouse IgG H&L (HRP) (1:10,000, ab205719, Abcam), Goat anti-Rabbit IgG H&L (HRP) (1:10,000, ab97051, Abcam). The blots were visualized using SuperSignal™ West Pico PLUS Chemiluminescent Substrate (Thermofisher) in ChemiDoc XRS+ System (Bio-Rad). Densitometric analysis was performed on non-saturated images using ImageJ software (NIH, Bethesda, MD, USA).

Dual luciferase reporter assay

Three different regions of the human MMP9 promoter—short (WT1 binding site only), mid (WT1 and PAX6 non-canonical binding site), and large (WT1, PAX6 non-canonical, and PAX6 paired domain binding sites) were PCR-amplified and cloned into the pGL3-Basic vector upstream of the firefly luciferase gene. To determine the allele specific regulatory effect of TNF-α promoter SNP rs1800629 and putative binding of WT1 in that region, we carried out a dual luciferase reporter assay. A 400bp region flanking the rs1800629 (G allele) variant was PCR amplified from genomic DNA and cloned into the multiple cloning sites upstream of firefly luciferase cassette using KpnI and XhoI enzymes in the pGL3-basic vector. Site directed mutagenesis was performed to get the A allele construct. All the primers used in the assay are listed with their sequences in Table S5. HTM cells were cultured in 24-well plate and were co-transfected with 50ng of G or A allele constructs along with 20ng pRL-TK plasmid (Renilla) per well using FuGENE HD transfection reagent (Promega). 100ng of WT1 construct and empty vector were also transfected as and when necessary. After 48 h of transfection, cell lysates were harvested, and Firefly and Renilla luciferase activities were measured using the Dual-Luciferase Reporter Assay System (Promega) following the manufacturer’s instructions on a Tecan Infinite 200 PRO (Tecan). Promoter activity was normalized by calculating the Firefly to Renilla luminescence ratio. Differences in normalized luciferase ratios between comparison sets were plotted as fold change values across three biological replicates and were statistically compared by Student’s t-test.

Immunofluorescence staining and confocal microscopy

Briefly, cells grown in coverslips were fixed with 4% paraformaldehyde at room temperature for 15 minutes. Permeabilization was done with 0.25% triton-X 100 in PBS for 20 minutes and afterward blocking was done using 3% BSA in PBST for 1 h. WT1 primary antibody was used (1:1000 dilution) for overnight incubation at 4-degree celcius. Cells were washed with PBS and incubated with secondary antibody Goat anti rabbit Alexa Fluor 488 (1:1000, Abcam). Cells were rinsed once more, counterstained with DAPI (Invitrogen), and mounted on slides using ProLong Gold antifade reagent (Invitrogen).

Immunofluorescence staining of zebrafish cryosections was carried out using established protocols.²⁰ Briefly, cryosections were thawed and air-dried at room temperature for 30 minutes before staining. Sections were rehydrated in PBS and blocked for 1 h at room temperature with 5% BSA (Gibco) in PBS supplemented with 0.3% Triton X-100. The sections were incubated overnight at 4 °C using mouse anti-acetylated tubulin (1:1000, Sigma Aldrich). Following three PBS washes, sections were incubated with Alexa Fluor 568–conjugated anti-mouse secondary antibody (1:1000, Abcam) for 2 h at room temperature in the dark. Nuclei staining was done using DAPI (1 μg/mL, Invitrogen) for 5 minutes at room temperature, protected from light. After final washes in PBS, sections were mounted with ProLong Gold Antifade Mountant (Invitrogen) and imaged using a Nikon A1 confocal microscope. Analyses were done using NIS elements and ImageJ software as necessary.

Negative control samples were processed identically except that the primary antibody was omitted and only the secondary antibody was applied. Representative negative control images are shown in Figures S12 and S13.

Lipopolysaccharide (LPS) and L-NAME treatment in zebrafish

To induce systemic inflammation, zebrafish larvae at 3 days post-fertilization (dpf) were exposed to ultrapure lipopolysaccharide (LPS) derived from Escherichia coli (Sigma al, USA) at a final, working concentration of 200 μg/mL.⁸⁴ LPS was administered via immersion by adding it to 2 mL of E3 medium containing approximately 30 embryos per well in a 12-well plate. For combinatorial treatment, L-NG-Nitro arginine methyl ester (L-NAME), a non-specific nitric oxide synthase (NOS) inhibitor, was used at a concentration of 250 μM. L-NAME was dissolved in embryo medium and pre-administered 1 h prior to LPS exposure to inhibit nitric oxide production. Zebrafish were maintained under standard laboratory conditions throughout the experiment. The inflammatory response was assessed through various approaches depending on experimental goals, including phenotypic observation of morphological or ocular abnormalities, fluorescence microscopy, and quantitative PCR (qPCR) analysis of inflammation-associated gene expression, including mmp9.

RNA isolation and RT-qPCR analysis

RNA isolation was done using RNeasy plus mini kit (Qiagen) as per the kit protocol. cDNA synthesis was done using SuperScript™ IV VILO™ Master Mix (Invitrogen). The RNA expression levels were measured using SYBR reagent in Quantstudio 3. The relative mRNA expression of gene of interest was calculated by using comparative threshold (ΔΔCt) method normalizing with ACTB levels. Fold change values were represented in graphs and statistical tests were done accordingly. Primers used in the assay are listed with their sequences in Table S5.

Gelatinase activity measurement

Gelatinase activity was measured in the cell lysates using a gelatinase activity assay kit (ab234057 Gelatin Degradation Assay Kit; Alternative to Zymography), following the manufacturer’s instructions. Fluorescence was measured in kinetic mode at 37 °C for 2 h on a Tecan Infinite 200 PRO M Plex Multimode Microplate Reader (30050303).

Neurotrace staining

Charged slides were used to mount cryosections of 12–15 μm thickness. Zebrafish eye sections were stained with NeuroTrace 435/455 Blue Fluorescent Nissl Stain (1:50, N21479, Thermo Fisher) for 30 minutes at room temperature, following the manufacturer’s protocol. After staining, sections were washed three times with PBS and mounted using ProLong Gold Antifade (Invitrogen). Confocal images were acquired using a Nikon Ti2 Eclipse microscope. Image intensity was analyzed with Nikon NIS-Elements software by selecting the entire eye as the region of interest (ROI), and fluorescence intensity within ROIs was quantified. ROI intensities were statistically compared using a two-sample Student’s t-test in GraphPad Prism (v9, Dotmatics).

Quantification and statistical analysis

Data are presented as mean ± SEM as mentioned in the corresponding figure legends. Prism software version 10 (GraphPad) was used to analyze the data. An unpaired two-tailed Student’s t-test (p < 0.05 was considered statistically significant) was performed for statistical analysis unless mentioned otherwise. For all statistical analyses, significant differences are denoted by ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001 and ns-not significant. All experiments had at least three biological replicates (N = 3) unless specified otherwise.

Supplemental information

Table S3. Distributions of the rare and common variants in WT1 gene

(A) contains the rare variants; (B) has the common variants and (C) provides a funnel plot of the odds ratios for the common variants.

Table S4. Distributions of the rare and common variants in MMP9 gene

(A) contains the rare variants; (B) has the common variants and (C) provides a funnel plot of the odds ratios for the common variants.