Distribution of TGFBI variants in patients with early onset glaucoma

• Introduction
• Methods
• Results
• Discussion

Abstract

Purpose: To describe a novel association of TGFBI variants with congenital glaucoma in a family with GAPO (growth retardation, alopecia, pseudoanodontia, and progressive optic atrophy) syndrome, as well as among other unrelated cases of juvenile onset open-angle glaucoma (JOAG) and primary congenital glaucoma (PCG).

Methods: This study of one family of GAPO with congenital glaucoma and three unrelated patients with JOAG analyzed a common link to glaucoma pathogenesis. Three girls with GAPO syndrome born to consanguineous parents in a multi-generation consanguineous family were identified. Two of the girls had congenital glaucoma in both eyes, while the elder sibling (a 10-year-old female) had features of GAPO syndrome without glaucoma.

Results: A genetic evaluation using whole exome sequencing revealed a novel homozygous ANTXR1 mutation in all three affected siblings with GAPO. No other mutations were detected in the genes associated with glaucoma. A rare missense variant in the TGFBI gene was shared in the two siblings with congenital glaucoma and GAPO syndrome. We found three other unrelated patients with JOAG and one patient with primary congenital glaucoma with no known glaucoma causing gene mutations, but having four different missense variants in the TGFBI gene. One of these patients with JOAG had familial granular corneal dystrophy. Molecular dynamic simulations of TGFBI and 3-D structural models of three of its variants showed significant alterations that could influence TGFBI protein function.

Conclusions: The possibility that variations in the TGFBI gene could have a possible role in the pathogenesis of congenital and juvenile onset open-angle glaucomas needs further evaluation.

Primary congenital glaucoma (PCG; OMIM 231300) and juvenile onset open-angle glaucomas (JOAG) are generally monogenic, with genes having a strong biologic effect that plays a role in their pathogenesis. Along with MYOC, CYP1B1 and LTBP2 [1-7], newer genes associated with either form of glaucoma include the CPAMD8, GPATCH, and EFEMP1 [8-10]. However, a large number of PCG and JOAG patients do not have a known mutations in any of these genes [11], indicating that chromosomal loci harboring putative genes associated with PCG and JOAG remain to be discovered [12]. The advent of next-generation sequencing now offers the opportunity for new gene discoveries, especially for diseases where genetic variations have a significant biologic effect on disease causation.

One of these genetic diseases is growth retardation, alopecia, pseudoanodontia, and progressive optic atrophy (GAPO; OMIM 230740) syndrome, an extremely rare genetic disorder, with only about sixty cases reported to date. The ocular features of GAPO syndrome include puffy eyelids, sparse eyelash hair, poliosis, hypertelorism, ptosis, strabismus, nystagmus, megalocornea, keratoconus, keratopathy, optic atrophy, myelinated retinal nerve fiber layer, and congenital glaucoma (buphthalmos) [13]. The diagnosis of GAPO is usually made early due to the characteristic premature “geriatric” appearance of the patients and the presence of cardinal features like alopecia and pseudoanodontia, as well as the associated features like dwarfism, prominent supraorbital ridges, frontal bossing depressed nasal bridge, long philtrum, and a wide open anterior fontanelle. Late onset alopecia has also been frequently documented [14,15]. Patients with GAPO have a reduced life span and usually die in their third or fourth decades of life due to generalized interstitial fibrosis of the lungs and atherosclerosis [16,17].

The invariable consanguinity in families with GAPO points toward an autosomal recessive mode of inheritance caused by mutations in ANTXR1 at chromosomal position 2p14; this gene codes a protein essential for intracellular actin assembly [18,19]. Defective production of this protein alters cell adhesion properties, leading to excessive deposition of extracellular matrix (ECM) secondary to reduced turnover and resulting in the various phenotypic features characteristic of this syndrome [19,20]. While the occurrence of glaucoma is reported among patients with GAPO syndrome [17,21-25], whether glaucoma develops due to mutations in the ANTXR1 gene itself or in some other (glaucoma-associated) genes is unknown.

This report analyses the genetic background of two of three sibling children with GAPO syndrome who also had congenital glaucoma. Whole exome sequencing (WES) analysis pointed toward a Transforming Growth Factor Beta Induced (TGFBI) gene variant common to both siblings. We found other TGFBI variants in three unrelated patients with JOAG, one of whom had familial granular corneal dystrophy, and in one patient with PCG.

Four girls from a multigenerational consanguineous family presented to us. Three of the girls were affected with GAPO, and two also had congenital glaucoma in addition to GAPO features (Figure 1). The family resided in North India, where a strong tradition of consanguinity persists. The three cases in the family described here are the three girls with features of GAPO syndrome.

We found a missense homozygous ANTXR1 gene mutation; (NM_032208.2:c.572T>C, p.Ile191Thr) in all three children affected with GAPO. This change was deemed novel and disease causing by mutation taster (0.99), PolyPhen-2 (1.0), and PROVEAN (−4.0). It is classified as a variant of uncertain significance as per ACMG guidelines. Another change at this position, c.572T>G, p.Ile191Ser is reported in the COSMIC database (COSV100362660).

Homozygosity mapping applied to the WES data of the consanguineous family detected one large (11Mb) homozygous region involving the TEK gene (i.e., chromosome 9) in only one of the two sisters with congenital glaucoma, among the genes known for glaucoma. However, no CNVs or intronic splice region variants were detected in the TEK gene in either of the sisters.

No missense variants (as per ACMG criteria) or CNVs were detected that co-segregated with glaucoma in the two girls with congenital glaucoma (Cases 1 and 2) at any of the known loci for Mendelian forms of glaucoma (GLC1A, GLC1B, GLC1D-Q) or PCG (GLC3A-D), including the known causative genes, MYOC (GLC1A), CYP1B1(GLC3A), WDR36 (GLC1G), ASB10 (GLC1F), OPTN (GLC1E), NTF4 (GLC1O), TBK1(GLC1P), LTBP2 (GLC3C), FOXC1, PITX2, TEK, COL4A1, and PAX6.

Three missense variants in PTPN4, PCDHB4 and GGT2, were observed exclusively in the two sisters with GAPO and congenital glaucoma (case 1 and case 2; Appendix 1). The variants of PTPN4 and PCDHB4 were rejected as causal because of the absence of any association in glaucoma pathogenesis pathways. The GGT2 variant was also not considered as it was earlier reported that hGGT2 does not encode a functional enzyme and therefore is unlikely to play a role in glaucoma pathogenesis [32]. By contrast, the TGFBI is known to have a strong role in ocular anterior segment development [33], and a missense heterozygous variant in TGFBI, 5–135388650-C-T NM_000358.3:c.968C>T NP_000349.1:p.Ala323Val (Figure 1A) was found. While this variant is likely benign according to the ACMG criteria, the LoF score for this variant (0.132) is suggestive of it being possibly damaging. The variant is highly conserved, with a genomic evolutionary rate profiling (GERP) value of 5.65. The Clinvar (RCV000279769.3) star rating score extracted for the variant was 1/4 with two submissions, benign, and uncertain significance for corneal dystrophy.

We then looked for the presence of TGFBI gene variants in a cohort of 40 patients with JOAG (diagnosed before 25 years of age) in whom WES had previously been done. We found two TGFBI variants present in two unrelated patients with JOAG. Both patients had nonfamilial glaucoma diagnosed before the age of 40 years. Neither had any other mutations (either in exonic or intronic regions) of significance in the known glaucoma genes. In one patient, a TGFBI heterozygous missense variant was found in the same codon as observed in the described cases of GAPO, but with a different amino acid residue; 5–135388650-C-A; NM_000358.3:c.968C>A, NP_000349.1:p.Ala323Glu. The second patient with JOAG had a novel null variant (frameshift mutation) 5–135397226-C-; NM_000358.3:c.1944del, NP_000349.1:p.Ser649LeufsTer22. Both patients had a normal anterior segment with a clear cornea. None of their family members were reported to have any ocular disorders.

A third patient with JOAG had concomitant granular corneal dystrophy (GCD) and had a heterozygous TGFBI mutation NM_000358.3:c.1663C>T, NP_000349.1:p.Arg555Trp, which is reported as pathogenic in Clinvar (VCV000007866.7) with 3 submissions associated with GCD (Table 1). We looked for other mutations or CNVs in the known genes for glaucoma, but we did not find any. The patient had an untreated IOP of 30 mmHg in BE with advanced glaucomatous optic atrophy. This patient had 2 other first degree relatives (a brother and his daughter) also affected with GCD but without having glaucoma or raised IOP (Figure 3). The clinical features of these three patients with JOAG are given in Table 2. However, the clinical features did not differ from those of other JOAG patients.

We also found a TGFBI variant (likely pathogenic according to ACMG) in another PCG patient. Of 15 unrelated patients with PCG who underwent WES, we found that this patient with congenital glaucoma had a rare TGFBI variant and no other associated known glaucoma gene mutation (Table 1).

A comparative modeling of the TGFBI variants in this study (p.Ala323Val, p.Ala323Glu and p.Ser649LeufsTer22) was performed since the available crystal structures lack the C-terminus (Figure 4). The model was subjected to a 200 ns MD simulations for refinement. During this simulation, the C-terminal region of the protein was gradually stabilized through interactions with the FAS1–4 and FAS1–3 domains. The last frame of the simulation was used as a template to generate TGFBI variants A323E, A323V, and 649 frameshift, which were then subjected to MD simulations to understand the effect of the mutations on the local and global structure of TGFBI and its variants. Analysis of the MD trajectory yielded various parameters, such as the root mean square deviation (RMSD), root mean square fluctuation (RMSF) and radius of gyration (RG), which allow assessment of structural deviation, residue fluctuations, and compactness of the proteins, respectively (Appendix 2). The MD simulations were followed by Principal Component Analysis (PCA) to study the overall motions of the TGFBI and these 3 variants. PCA extracts correlated motions of the protein from the MD trajectory, thereby allowing us to study the impact of mutations on distant regions of the protein and subsequent changes in protein movement. Projection of the trajectory along PC1 and PC2 depicted correlated motions of various domains of the protein, as shown in Appendix 3. We found that TGFBI fluctuated between an elongated and a banana-like shape due to bending between the FAS1–2 and FAS1–3 domains. The bending movement was more prominent in all variants compared to wild-type (WT) TGFBI, and occurred majorly in the FAS1–2 and FAS1–3 domains. The free energy landscape (FEL) was also generated to visualize the potential energy of TGFBI and mutants. The impact of the three variants was also evident on the FEL, since the free energy wells were shallow and distorted compared to the deeper well of the WT TGFBI (Figure 5).

This family of three children affected with GAPO, two of whom presented with bilateral congenital glaucoma, allowed us to study the genetic association of glaucoma with GAPO syndrome. All three carried a novel missense homozygous ANTXR1 gene mutation.

Glaucoma with GAPO syndrome has been reported in ten cases to date [14,15,17,20-25]. Glaucoma in five of these patients was early onset open angle glaucoma [14,15,24,25]. Primary congenital glaucoma (PCG) is described in five patients [17,21-25], and asymmetric glaucoma, as seen in Case 1 of the GAPO family that presented to us, has also been documented [17,21], while others have described a late presentation with end stage glaucoma [21,22,25]. The pathogenesis of congenital glaucoma in children with GAPO syndrome is not clearly understood. If mutations in the ANTRX1 gene are postulated as a cause of glaucoma in GAPO, then this does not explain why so few patients with GAPO develop glaucoma and why only 2 of the 3 siblings with the ANTRX1 mutation in the family described by us developed glaucoma. A variable expression of ANTRX1 mutation could be one reason why the elder sister (case 3) with ANTRX1 mutation and GAPO syndrome did not have glaucoma. However, even if we consider a variable expressivity in ANTRX1 as the reason for one sibling with GAPO syndrome not developing congenital glaucoma in this family, we believe the presence of the TGFBI variant may have a role in modification of this expression. The clinical significance of the TGFBI p.Ala323Val variant, although conflicting in Clinvar, cannot not be ruled out and its association with disease requires further investigation.

The abnormal expression of TGFBI is related to the occurrence and development of some types of cancers, as well as different types of corneal dystrophies, including lattice corneal dystrophies (LCD) and granular corneal dystrophies (GCD) [34]. The TGFBI gene was one of the first discovered to cause corneal dystrophies [35] and since then, numerous reports on genotype–phenotype correlations have been published. The human protein transforming growth factor-beta-induced protein ig-h3 (Beta ig-h3), encoded by TGFBI, has four FASI domains, and the mutated amino acid observed in our two patients corresponds to the FAS1–2 domain of TGFBIp. The TGFBI mutations associated with various corneal dystrophies are located in the FASI-1 and FASI-4 domains [36]. No association with developmental glaucoma has been noted.

The TGFBI gene lies within the linkage area mapped as a glaucoma locus 5q22.1-q32 [37]. A gene expression profile of human trabecular meshwork (TM) tissue by SAGE (serial analysis of gene expression) has shown a higher expression of TGFBI, along with other glaucoma causing genes and genes involved in typical TM maintenance functions [38]. A study by Kim et al., using Alb-hßigh3 transgenic mice, showed that overexpressed human βigh3/ (hßigh3)/ TGFBIp in the blood might be involved in anterior segment morphogenesis and eye development in mice. The phenotype observed in transgenic mice is similar to that seen in human eye disorders, such as anterior segment dysgenesis and Peters' anomaly [39]. These observations support a role for TGFBI in congenital glaucoma, as observed in the two sisters with GAPO. A string analysis also showed that TGFBI and LTBP2 (associated with congenital glaucoma) are co-expressed, possibly interacting with MYOC and CYP1B1. Glaucoma has also been observed in a patient with a TGFBI, R124H mutation, though the phenotypic details of this patient are not known [40]. These observations suggest a role for TGFBI in glaucoma pathogenesis, in conjunction with other genes.

We also found 3 unrelated (non-dominant) patients with JOAG and one patient with PCG who had with TGFBI variants in a separate cohort; none of these patients had any known glaucoma gene mutations. One patients had GCD; however, two other members of his family with GCD and the TGFBI mutation did not have glaucoma. The differences in phenotypic expression among family members in this family with GCD and with the same TGFBI mutation may also reflect differences in genetic background/modifier alleles that might affect the expression of TGFBI co-regulators. No report in the literature has described an association between JOAG and GCD, and the finding of the TGFBI mutation leads us to postulate that variations in this gene could be contributing to glaucoma pathogenesis.

These 6 patients with early onset glaucoma and having variants in the same TGFBI gene support a role for TGFBI in glaucoma pathogenesis. A high degree of phenotypic variability and incomplete penetrance is known for TGFBI mutations; however, the fundamental reason why different gene variations cause morphologically distinct phenotypes remains elusive.

The modeling studies of the TGFBI wild-type, along with the three variants A323E, A323V, and 649-frameshift, have provided insights into the effect of these variations on the full-length structure and its dynamics. While specific aggregation mechanisms cannot be deduced at this stage, the alteration of the protein dynamics was clearly evident in the simulation studies. In this study, we built models of mutant forms of TGFBI protein and subjected them to molecular dynamics simulations (MDS), which attempt to mimic the protein's physiologic aqueous environment in silico. Subsequent principal component analysis (PCA) was performed on the MDS trajectory to understand the motions of the protein and its mutant forms. The altered dynamics due to the mutations can greatly change the molecular interactions, thereby affecting the normal physiologic function of the protein. The A323 is located at the interface of the CROPT and FAS1–1 domains, and perturbation at this position can alter inter-domain interactions and even global protein structure. For A323V, the presence of a hydrophobic residue of valine on the surface can destabilize the protein, and the same was observed in molecular dynamics studies. In the case of A323E, the glutamic acid was observed to form strong intra-domain interactions with K72 and N207 from the CROPT domain and FAS1–1 domain, respectively. In the case of the frameshift variant of TGFBI, the sequence at the C-terminal is greatly altered compared to the WT.

The comparative modeling studies revealed that the C-terminal region forms two alpha-helices packed against each other with hydrophobic residues, further interacting with the FAS1–4 and FAS1–3 domains. Due to the frameshift, packing of the helices was disturbed, leading to increased solvent exposure of the C-terminal tail. The structural differences observed during the simulations for TGFBI and its mutant forms were depicted using RMSD plots which show the global structural changes in protein, RMSF plots highlighting the residue-wise fluctuations, and RG plots, which measure of the compactness of the protein. The structural changes observed in these plots for the frameshift mutant were significantly distinct compared to the wild-type.

Further understanding of the motions of the proteins was obtained by performing principal component analysis (PCA). PCA extracts the correlated motions of the protein from the MD trajectory, thereby allowing us to study the impact of mutations on distant regions of the protein and subsequent changes in protein movement. The porcupine plots recapitulate the major movements of the proteins as deduced from PCA. The observed bending movement of the protein was more pronounced in the frameshift variant, while minor variations in the other mutants were observed compared to the wild-type. The free energy landscape (FEL) is deeper than the shallow FELs of the mutant forms due to decreased protein stability. This in silico analysis conclusively showed that the mutations destabilized the TGFBI protein and could therefore affect its function.

The limitations of our study include its small number of patients and our inability to genotype the parents and siblings of the patients with JOAG and the patient with PCG who harbored the TGFBI variants. Gene expression and functional studies using cell-based assays and transgenic animal models of the mutant are also needed to examine the effect of these variants on the protein and the downstream targets to establish a cause-and-effect relationship. Moreover, potential interactions between TGFBI and other genes in the causation of glaucoma also need to be explored. Nevertheless, based on the findings of this preliminary report, TGFBI may be an interesting candidate for glaucoma studies, although we cannot identify it as a cause of glaucoma based only on our present findings. More studies in a larger patient population, as well as functional studies, are needed to further elucidate its role in glaucoma pathogenesis.