Congenital microcoria deletion in mouse links Sox21 dysregulation to disease and suggests a role for TGFB2 in glaucoma and myopia

Abstract

Congenital microcoria (MCOR) is a rare hereditary developmental defect of the iris dilator muscle frequently associated with high axial myopia and high intraocular pressure (IOP) glaucoma. The condition is caused by submicroscopic rearrangements of chromosome 13q32.1. However, the mechanisms underlying the failure of iris development and the origin of associated features remain elusive. Here, we present a 3D architecture model of the 13q32.1 region, demonstrating that MCOR-related deletions consistently disrupt the boundary between two topologically associating domains (TADs). Deleting the critical MCOR-causing region in mice reveals ectopic Sox21 expression precisely aligning with Dct, each located in one of the two neighbor TADs. This observation is consistent with the TADs' boundary alteration and adoption of Dct regulatory elements by the Sox21 promoter. Additionally, we identify Tgfb2 as a target gene of SOX21 and show TGFΒ2 accumulation in the aqueous humor of an MCOR-affected subject. Accumulation of TGFB2 is recognized for its role in glaucoma and potential impact on axial myopia. Our results highlight the importance of SOX21-TGFB2 signaling in iris development and control of eye growth and IOP. Insights from MCOR studies may provide therapeutic avenues for this condition but also for glaucoma and high myopia conditions, affecting millions of people.

Keywords: DCT enhancer adoption; SOX21-TGFB2 signalling glaucoma and myopia; Sox21 ectopic expression; congenital microcoria; developmental genetics; genetic eye disorders; genome architecture; mouse model; topologically-associated domain; translational medicine.

Graphical abstract

Figure 1

Chromatin organization of the human and mouse genomes in wild-type and MCOR conditions (A) Chromatin interaction heatmaps on human chromosome 13q32.1 encompassing the MCOR locus with positions referencing the human reference sequence hg19. Hi-C sequencing data in human neural progenitor cells (hNPCs) reveal that the MCOR locus is located within a 250 kb TAD (T2) flanked by larger TADs of 1.2 Mb (T1) on the left and 500 kb (T3) on the right. Insulation scores and CTCF ChIP-seq tracks below the heatmap delineate TAD boundaries. The MoDLE Hi-C map predicted using CTCF ChIP-seq data (WT Predict) demonstrates a similar chromatin organization. Removal of CTCF peaks within the MCOR deletion (MCOR Predict) results in the expansion of the T2 TAD into T3, encompassing SOX21. (B) Magnified view of the predicted WT and MCOR heatmaps, gene track, and insulation scores calculated using hicFindTADs tool on MoDLE highlighting chromatin interaction changes due to MCOR-causal deletions. The projection of the most distant deletion boundaries (MAX) and the critical MCOR deletion (MIN) shows all deletions impact the boundary between T2 and T3 in the wild-type condition. A single asterisk (^∗) denotes the critical MCOR region from the smallest MCOR-causing deletion in Family FR1 (Fares-Taie et al.⁴). A double asterisk (^∗∗) indicates the deletion in the patient with elevated TGFB2 levels in their AH. The dashed blue line shows the SOX21 position across all panels. (C) Virtual circular chromosome conformation capture (4C) sequencing plot anchored on the SOX21 promoter shows SOX21 interacting within the T3 domain in hNPCs (red arrows). The critical MCOR deletion shifts and enhances SOX21 promoter interactions to the T2 domain, particularly with DCT (blue arrows). (D) Chromatin interaction heatmap and CTCF ChIP-seq track from mouse neural progenitor cells (mNPC) at chromosome 14qE4 show a 3D organization similar to chromosome 13q32.1 in humans, with the MCOR region located within a 200 kb TAD (t2), flanked by larger TADs (t1 and t3), each spanning 800 kb; positions refer to the mouse reference sequence mm10. Below the mNPC heatmap, MoDLE predictions with normal CTCF ChIP-seq data (mWT Predict) suggest t2 is partitioned into two subTADs (t2.1 and t2.2). This slight difference does not affect the pattern of chromatin reorganization in the MCOR condition compared to the human counterpart. When the three CTCF peaks in the MCOR deletion are removed (mMCOR Predict), the t2.2 subTAD expands into t3 to include Sox21, whose promoter gains interaction with the region comprising Dct, as depicted in (F). (E) Magnified view of the predicted heatmaps form the WT (mWT) and MCOR (mMCOR) mice, gene track insulation scores calculated using hicFindTADs tool on MoDLE highlighting chromatin interaction changes caused by deleting the mouse region homologous to the critical MCOR-causal deletion in humans. This region crosses the boundary between t2.2 and t3 in the wild-type condition. The dashed blue line marks Sox21 position across the panels. (F) Virtual 4C plot anchored on the mouse Sox21 promoter shows interaction within t3 in the wild-type condition (red arrow) and a gain of interaction within t2, particularly with Dct (blue arrows).

Figure 2

Pupillary and molecular phenotype of the B6.cΔMCOR/+ mouse (A) Assessment of pupil diameter in B6.cΔMCOR/+ mice using a pupillometer reveals a moderate but statistically significant reduction in baseline pupil size compared to B6.WT animals (^∗∗p < 0.01, n = 8 and 7 two-month-old animals per group, respectively). The graph plots individual data points with mean values and SEM error bars. The p value was calculated using one-way/two-way ANOVA. (B) Analysis of the impact of the critical MCOR deletion on the expression of Dct, Tgds, Gpr180, Sox21, and Abcc4 mRNAs using RNA-seq datasets from IrCB of B6.cΔMCOR/+ and B6.WT mice shows that the deletion induces ectopic expression of Sox21; Note that the abundance of Tgds and Gpr180 is halved in B6.cΔMCOR/+ mice carrying the heterozygous deletion of the MCOR region. Erros bars depict SEM. The p values were calculated using an unpaired t test (∗∗p < 0.01, ∗∗∗p < 0.001; ns, not significant). (C) Real-time qPCR analysis of Dct and Sox21 abundance in IrCB RNA extracts from newborn B6.cΔMCOR/+ and B6.WT mice (n = 5 and 3, respectively) are consistent with RNA-seq data analysis. The graph plots individual data points with mean values and SEM error bars. Statistical significance was determined using Fisher's PLSD test following ANOVA (∗∗∗p < 0.01; ns, not significant). (D) Western blot analysis of IrCB protein extracts shows detection of SOX21 in B6.cΔMCOR/+ but not in B6.WT counterparts. β-actin (BACT) serves as the loading control, and C+ denotes glial cells which express endogenously SOX21.

Figure 3

SOX21 and DCT expression in the IrCB of newborn albino SW.cΔMCOR/+ and SW.WT mice Immunohistochemistry analysis of iris and CB in newborn SW.cΔMCOR/+ and SW.WT mice illustrate the detection of SOX21 and DCT in the PEL of the iris and the AEL of the CB. AEL, anterior epithelium layer; CB, ciliary body; CP, ciliary process; DM, dilator muscle; PEL, posterior epithelium layer; S, iris stroma. Scale bars: 20 μm.

Figure 4

Spatiotemporal pattern of expression of Sox21 and DCT in the developing eye of albino SW.cΔMCOR/+ and SW.WT mice (A) In situ hybridization (ISH) combined with immunohistochemistry analysis of Sox21 mRNA and DCT during eye development, spanning from the invagination of the optic vesicle (OV) into the optic cup (OC) to the formation of the IrCB. Sox21 mRNA is stained in red, and DCT is stained in green. Autofluorescence mainly arising from blood vessels appears in cyan. In SW.cΔMCOR/+ eyes, Sox21 expression begins at E10.5, covering the entire outer layer of the OC. This layer develops into the retinal pigment epithelium (RPE) and the anterior epithelium of both the ciliary body (CB) and iris (AEL) at its edge (outlined by white dotted lines) by E15.5. Sox21 aberrant expression precedes the appearance of DCT in the developing eye. As the RPE differentiates, DCT expression decreases, mirroring the decline in Sox21 expression during the late stages of development. In the anterior epithelium of the ciliary body (aCB), both genes maintain stable expression from development to maturation, unlike the iris anterior epithelium (AEL) and iris posterior epithelium (PEL). DCT and Sox21 are expressed during the prenatal period in the developing AEL but not postnatally. In contrast, in the PEL, expression of both genes begins prenatally and continues into the postnatal period. In SW.WT eyes, Sox21 expression initiates in the proximal region of the developing OV at embryonic day 10 (E10) but is absent in the forming OC) (depicted by the orange bar in the schematic summary); note that Sox21 is detectable by E15.5 in the eyelid and hair follicles (white arrows). AEL, anterior epithelium layer; CB, ciliary body; CP, ciliary process; DM, dilator muscle; L, lens; PEL, posterior epithelium layer; R, retina; S, iris stroma; SM, sphincter muscle. Scale bars: 100 μm. (B) Schematic summary of Sox21 and DCT expression dynamics during iris development showing that under normal conditions, Sox21 expression is confined to the OV stage (orange bar), whereas in mutant eyes, it persists throughout eye development (red bars) and mirrors DCT expression (green bars); note that at the earliest stage (E10), Sox21 mRNA precedes detection of DCT, possibly due to enhancer priming before gene transcription initiation. Interestingly, the expression patterns of Sox21 and DCT in mutant iris epithelia correspond to reported changes in iris pigmentation in pigmented mice: initially heavily pigmented AEL loses melanin postnatally, while the PEL gains melanin centrifugally, reaching full pigmentation in adulthood. These observations suggest that DCT regulates Sox21 expression in mutant animals. The dilator muscle (depicted in dark blue) begins forming shortly after the sphincter (light blue), around E17.5 and E18.5, respectively. Thus, during the initial stages of dilator muscle development, Sox21 expression is present in the AEL, which contributes to the dilator muscle (DM). The cessation of Sox21 expression in the AEL by P0 in mice indicates a likely early developmental defect underlying the disease. The shaded box denotes the period from birth to 2 months, where information on iris AEL pigmentation and the stages of DM development are currently limited. aCB, ciliary body anterior epithelium; AEL, iris anterior epithelium; C, cornea; DM, dilator muscle; EL, eyelid; IL, optic cup inner layer; L, lens; LP, lens placode; LV, lens vesicle; NE, neuroectoderm; OC, optic cup; OL, optic cup outer layer; OV, optic vesicle; pCB, ciliary body posterior epithelium; PEL, iris posterior epithelium layer; R, retina; RPE, retinal pigment epithelium; S, iris stroma; SE, surface ectoderm; SM, sphincter muscle. Scale bars: 100 μm.

Figure 5

SOX21 binding sequences in mouse and human TGFB2 and in vitro and in vivo analysis of TGFB2 expression relative to SOX21 expression (A) The DNA sequence targeted by SOX21 in the first intron of Tgfb2, identified by ChIP-seq analysis in the mouse B6.cΔMCOR/+, is shown on the left. The corresponding sequence from the human genome assembly GRCh38, located in the first intron of TGFB2 is presented on the right panel. The mouse and human SOX21-consensus motifs, according to the JASPAR database, are displayed at the bottom of each panel. The mouse binding sequence (JASPAR: PB0069.1) of SOX21 is highlighted in blue, found in both the human and mouse sequences in the intron. The human motif (JASPAR: MA0866.1), shown in green, is also present in the TGFB2 intron 1, suggesting a potential binding of SOX21 in this location. (B) RNA-seq heatmap focusing on genes in the TGFB pathway reveals consistent overexpression of Tgfb2 in all four B6.cΔMCOR/+ irises, despite some variations observed among individuals within the same group. (C) RT-PCR analysis in human glioma cells shows a significant decrease in TGFB2 abundance when SOX21 expression is knocked down using siRNA, compared to treatment with non-targeting siRNA. This supports the role of SOX21 in regulating TGFB2 expression. The graph plots individual data points with mean values and SEM error bars. The statistical significance was assessed using an unpaired t test (∗∗p < 0.01; ∗∗∗∗p < 0.0001). (D) Analysis of TGFB2 concentration in the AH of an individual affected with MCOR from Family FR1 initially reported by Fares-Taie et al. whose deletion is shown in Figure 1, and 11 controls using ELISA show an accumulation of the protein in the MCOR subject compared to controls. The graph plots individual data points (each representing TGFB2 levels from a single eye) with mean values and SEM error bars. Statistical significance was not assessed due to the availability of only one MCOR individual.