Genetic Association of TCF4 Intronic Polymorphisms, CTG18.1 and rs17089887, With Fuchs' Endothelial Corneal Dystrophy in an Indian Population

Abstract

Purpose.: To assess the genetic association of transcription factor 4 (TCF4) intronic polymorphisms and CTG18.1 allele in individuals with Fuchs' endothelial corneal dystrophy (FECD) individuals from a sample Indian population.

Methods.: Forty-four FECD patients and 108 unrelated age-matched controls were recruited with informed consent for this study. Three, single nucleotide polymorphisms (SNPs) spanning the third intronic region of TCF4 (rs613872, rs17089887, and rs17089925) and an unstable trinucleotide repeat CTG18.1 allele were genotyped by direct sequencing using Sanger's method. The association of polymorphisms was analyzed using χ² test and logistic regression.

Results.: SNP rs17089887 (P = 0.013) and CTG18.1 (P = 2 × 10⁻⁴) alleles were found to be significantly associated with FECD in the sample Indian population. However, the other two SNPs, rs613872 and rs17089925, were not likewise associated. Thirty-four percent of FECD subjects and 5% of control individuals harbor more than 50 trinucleotide repeats, which was considered as the disease threshold.

Conclusions.: TCF4 poses a major contributor to FECD manifestation globally, with a significant association of rs17089887 and CTG18.1 allele in the Indian population.

Introduction

Progressive deterioration of corneal endothelial cells and formation of corneal guttae characterizes Fuchs' endothelial corneal dystrophy (FECD; MIM# 136800). In 1910, Ernst Fuchs¹ reported about this bilateral corneal dystrophy that predominantly affects women. As an autosomal dominant and heterogeneous disease, FECD displays a peculiar trait of corneal guttata formation and excrescences from the Descemet membrane (DM) occurring in the fourth to fifth decade of life, which are clinically detected using specular microscopy.^2,3 On further progression, the affected person acquires a decreased visual acuity due to endothelial cell loss and corneal edema that finally lead to painful epithelial bullae formation.^4,5 So far, many genes have been attributed to cause this disease, thereby revealing its genetic complexity. Transcription factor 4 (TCF4), that encodes for E2-2 protein, a group of E protein transcription factors known for cellular growth and differentiation, is one such gene that has been associated with this disease. 

Baratz et al.⁶ showed a significant (P = 2.3 × 10⁻²⁶) association of TCF4 single nucleotide polymorphism (SNP) rs613872 with FECD by performing a genome-wide association study among subjects of Caucasian descent. This association was reconfirmed by Li et al.⁷ by conducting genome-wide linkage scan and association with 64 multiplex Caucasian families where they identified rs613872 to be coinciding with the FECD2 locus previously found by Sundin and colleagues.⁸ However, Thalamuthu et al.⁹ indicated that two other polymorphisms (rs17089887 and rs17089925) and not the much evident rs613872 of TCF4 gene were strongly associated with FECD among the Chinese cohort. Apart from these SNP associations, TCF4 has also been ascribed with trinucleotide repeat expansion among FECD-affected individuals.¹⁰ Mootha et al.¹¹ has recently reported the association of both the polymorphisms, that is, rs613872 and CTG18.1, trinucleotide intronic expansion with the disease in Caucasian cohorts. These studies reflect that the region surrounding rs613872 harbor potential disease-causing elements for FECD. 

Statistical data comparing the prevalence of Fuchs' dystrophy in India remain elusive. Although a study group from an eminent tertiary eye care center in South India conducted a frequency distribution and outcome analysis of patients requiring penetrating keratoplasty (PK) as a surgical intervention for relief from ocular dystrophies; and FECD was accounted as the third most frequent (16.6%) reason for PK.¹² Studies comprising Indian late-onset FECD subjects are very scant and, hence, the genetic contributors for the disease specific to this region of the globe are veiled.¹³ The aim of this study was to assess the genetic association of various TCF4 polymorphisms (i.e., rs613872, rs17089887, rs17089925, and CTG18.1 alleles) among Indian FECD patients. In our population, we observed rs17089887 and CTG18.1 alleles to be significantly associated with FECD. 

Materials and Methods

Participants

A total of 108 (63 males and 45 females) control and 44 patients (18 males and 26 females) of Indian origin within the age range of 38 to 81 years (average age: 61.5 years including control and patients), were recruited at L V Prasad Eye Institute (a tertiary eye care institute) in Bhubaneswar (Odisha, India); after acquiring their written consent for enrollment in the study. We obtained detailed reports concerning the medical and family history of all the participants after clinically screening them via slit-lamp examination and categorizing on modified Krachmer FECD grade scale of zero through six (grade 0 individuals exhibiting no central guttae; grade 1: up to 12 scattered central guttae; grade 2: ≥12 scattered central guttae; grade 3: 1- to 2-mm central guttae; grade 4: 2–5 mm of clustered central guttae; grade 5: 5-mm confluent central guttae without edema; grade 6: ≥5-mm confluent central guttae with edema).¹⁴ The inclusion criteria for FECD patients involved individuals above 35 years of age exhibiting moderate to severe disease condition (grade 2 and above). Among the control population were the subjects with neither guttata nor compromised endothelium, but meeting the age criteria of the study. In a few cases slit-lamp examination was not possible in both the eyes/one eye due to advanced disease, for these cases we ascertained the inclusion criteria by histopathological tests reporting thickened DM with nodular excrescence and sparse endothelium. Subjects not matching the age criteria and/or with inconclusive specular/histopathology reports were excluded from the study. All the procedures for the study are in accordance with the Declaration of Helsinki. The ethics committee of National Institute of Science Education and Research and L V Prasad Eye Institute, reviewed and approved the study prior to the recruitment of subjects. 

Genotyping

The genomic DNA was extracted from the peripheral blood leucocytes (by DNA salting–out method) of the participants (controls and patients) and was maintained at 100 ng/μl.¹⁵ Genotyping of TCF4 SNPs (rs613872, rs17089887, and rs17089925) was done by designing primers flanking the polymorphisms (Table 1) for PCR using Eppendorf thermocycler system (Mastercycler pro; Eppendorf AG, Hamburg, Germany). The PCR mix comprising 100 ng template DNA, 0.5 μM of the flanking primers, 100 mM dNTP mixture (GeNei, Bangalore, India), 2.5 μL of DMSO, Taq Buffer A (GeNei, Bangalore, India), 1.5 mM MgCl₂, and 1.0 U Taq polymerase was constituted into 25-μL reaction volume. These reactions were incubated at 94°C for 5 minutes followed by 35 cycles of 45 seconds at 94°C, 45 seconds at annealing temperatures (Table 1), and 45 seconds at 72°C and finally incubated at 72°C for 10 minutes, and stored at 4°C until further use. These amplified products were electrophoresed on 1% agarose gel and subsequently eluted using QIAquick Gel Extraction Kit (QIAGEN, Hilden, Germany). These eluents served as template for Sanger sequencing reaction performed by employing Big Dye Terminator v.3.1 cycle sequencing kit (TX78744; Applied Biosystems, Austin, TX, USA) on an automated sequencing platform (3130xl genetic analyzer; Applied Biosystems). The sequences were analyzed using BioEdit v7.1 (Tom Hall; Ibis Biosciences, Carlsbad, CA, USA). 

Table 1

 

List of Primers Employed in This Study

Primer Name	Sequence (5′→3′)	Annealing Temperature	Source
TCF4rs17.87 F	TGGGCATAGAAGGCAAGAGAGA	60.0°C	This study
TCF4rs17.87 R*	CAGGATTTCGTCTTTATCCACAGGC		This study
TCF4rs17.25 F	TGGGTTGTGATGCTGTCTCAGTGT	60.3°C	This study
TCF4rs17.25 R*	TTCCTGCTTCTGACCCTCCAATGT		This study
TCF4rs61 F	CCCAGTAGGGTTGTGATGATGATG	60.9°C	This study
TCF4rs61 R*	CAGTTGGGAACACCCATTTGTCTG		This study
CTG18.1 F	CAGATGAGTTTGGTGTAAGATG	61.1°C	Wieben et al.10
CTG18.1 R*	ACAAGCAGAAAGGGGGCTGCAA		Wieben et al.10

 * Reverse primers.

Trinucleotide-Repeat Assessment

The region of interest encompassing the CTG18.1 allele (annotated as rs193922902, dbSNP) was amplified by PCR using Invitrogen Platinum PCR Super Mix High Fidelity and a primer pair; employing previously reported reaction conditions.¹⁰ The first cue on hetero/homozygosity was obtained by checking the band intensity of the PCR products on 2% agarose gels. To confirm heterozygosity of the individuals for CTG18.1 allele, the PCR products were gel eluted from 1% agarose gels and sequenced bidirectionally using ABI 3130xl Genetic Analyzer by Sanger's method for both the alleles. To confirm homozygosity, the PCR products were purified using QIAGEN PCR purification kit (QIAGEN) followed by ExoSAP (USB Products Affymetrix, Inc., Cleveland, OH, USA) treatment and directly used as sequencing template according to the manufacturer's protocol. Such two-tier verification confirms the number of individuals having at least more than 50 repeat units. 

Statistics

Comparison of age and sex between the case and control populations was done by Student t-test and Fischer's exact test, respectively. Deviance from Hardy Weinberg equilibrium for the genotypic frequencies of all the considered SNPs was calculated between cases and controls using Pearson's χ² test. Genetic power for the study was calculated using G*Power v3.1.9.2 statistical power analysis software (University of Düsseldorf, Düsseldorf, Germany). Association of polymorphisms with FECD were analyzed by χ² test and binary logistic regression taking age and sex as covariates using SPSS 20.0 statistical software for Windows (IBM SPSS, Inc., Chicago, IL, USA). We tested various models of genetic inheritance for each polymorphism using multiple logistic regression in SPSS: dominant (DOM; major allele homozygotes versus heterozygotes + minor allele homozygotes), recessive (REC; major allele homozygotes + heterozygotes versus minor allele homozygotes), and additive (ADD; major allele homozygotes versus minor allele homozygotes). Genotypic variables used for coding were 0, 1, and 2 with homozygous risk alleles for all the SNPs coded as 2; allele coding for CTG18.1 was adapted from Mootha and colleagues.¹¹ Linkage disequilibrium (LD) computation, LD plot generation, and haplotype frequency calculation was done using Haploview 4.2 (Broad Institute, Cambridge, MA, USA). Genetic association at a threshold of 5% was considered significant. 

Results

Demographics

Diagnosed with FECD, 44 unrelated patients enrolled for this study, were recruited from two different provinces of India (Odisha and West Bengal). As is seen in earlier demographic studies, FECD affected population in India replicates the fact that females are the most susceptible group for the disease (males, 40%; females, 59%).^9,16,17 All the cases were within the age range of 38 to 80 years, with the mean age being 60 years (Table 2). Not significantly distinct (P = 0.168) were the control group participants falling in the age range of 47 to 81 with a mean of 63 years. However, males (58%) slightly dominated the control population. Despite the sex bias, the study groups were not significantly distinct (P = 0.084). 

Table 2

 

Demographics of the Study Participants

Variables	Cases	Control	P Value
Number of Individuals	44	108	–
Number of females (%)	26 (59.09)	45 (44.11)	0.084
Mean Age (SD) in y	60 (9.24)	63 (8.52)	0.168

SNP rs17089887 is Significantly Associated With FECD in an Indian Population

Genetic association analysis was done for TCF4 intronic polymorphisms and FECD. All the study participants have their genotype frequencies in Hardy-Weinberg equilibrium. The allele distribution and association results for all the SNPs are described in Table 3. After age and sex correction, we found that only one of the intronic TCF4 SNPs, rs17089887, is significantly associated (Genotype P = 0.008) under dominant model of inheritance (P = 0.019) in our population with “T” allele (Allelic P = 0.013) imparting 2.073 (95% confidence interval [CI] = 1.18–3.62) with increased chances of acquiring FECD (Table 3). We observed that individuals with TT genotype for rs17089887 are at 65.9% increased risk. Its significance remains even after carrying out 10,000 permutations (P = 0.048). Keeping type I error at 5% and degree of freedom as 1, the current study is at 75% power with effect size of 0.4. The SNP rs17089925, which was previously outlined to be associated in the Chinese population, but fails to be so in our population.⁹ Whereas, the much cited polymorphism, rs613872, show a marginal association with respect to genotype (Df2, P = 0.03), but shows no association with respect to alleles (Df1, P = 0.246). 

Table 3

 

Association Results of Genotyped SNPs

SNP	Type	Control		FECD		Model	OR (95% CI)	P Value†	Tests	P Value‡
		n	%	n	%
rs17089887
Genotype	TT	48	44	29	65	DOM	2.37 (1.14–4.93)	0.02	χ2 (Df2)	0.008
	TC	47	43	13	29	REC	0.37 (0.81–1.75)	0.19
	CC	12	11	2	4	ADD	0.37 (0.10 −1.41)	0.12
Allele	T*	141	66	70	80		2.07 (0.81–1.93)	0.01	χ2 (Df1)	0.016
	C	71	33	17	19				10,000 permutation	0.048
rs613872
Genotype	TT	86	79	29	65	DOM	0.49 (0.22–1.07)	0.07	χ2 (Df2)	0.03
	TG	18	16	14	31	REC	1.65 (0.17–15.22)	0.65
	GG	4	3	1	2	ADD	1.26 (0.21–7.48)	0.98
Allele	T	188	87	72	82		1.50 (0.50–1.88)	0.79	χ2 (Df1)	0.246
	G	26	12	15	17				10,000 permutation	0.552
rs17089925
Genotype	CC	57	53	26	59	DOM	1.26 (0.62–2.57)	0.51	χ2 (Df2)	0.35
	CT	42	39	16	36	REC	0.589 (0.12–2.89)	0.51
	TT	8	7	2	4	ADD	0.63 (0.17–2.29)	0.71
Allele	C	154	72	68	77		1.37 (0.50–1.71)	0.24	χ2 (Df1)	0.405
	T	60	28	19	22				10,000 permutation	0.755

 Associated SNP is written in bold.

*  Risk allele.

†  P Value = P values obtained from logistic regression under DOM, REC, ADD models after correcting for age and sex; excluding four controls for which genotype data for any of the three SNPs were missing.

‡  P Value = P values obtained after data were subjected to various tests such as, χ² (Df2) = χ² test for difference between case and control genotypes; χ² (Df1) = χ² test for difference between case and control alleles and 10,000 permutation tests for allele distribution.

TCF4 Intronic CTG Trinucleotide Repeat Shows Significant Association With FECD

The threshold limit of CTG trinucleotide repeats (>50) in TCF4 gene assumed by Weiben and colleagues¹⁰ as a disease susceptibility marker for FECD was used in this study. Agarose gel electrophoresis followed by direct sequencing was employed to efficiently detect at least 50 repeats of CTG18.1 allele. Representative chromatograms depicting normal and extended repeats are as shown in Figure 1. The frequency distribution of both the alleles of control and FECD populations illustrates that 15/44 (34%) FECD cases and 5/97 (5%) control individuals harbored these extended trinucleotide repeats (≥50) (Fig. 2, Table 4). Out of these 15 patients, 14 individuals had at least one “T” allele of rs17089887 and the combined haplotype (T-X) shows significant association with the disease (P = 2 × 10⁻⁴). 

Figure 1

 

Sequence chromatogram of a control individual (A) with 12 repeats and an FECD patient (B) with 61 repeats.

Figure 2

 

Frequency distribution of both the alleles of CTG18.1 in FECD and control population.

Table 4

 

Result of Genotype and Haplotype Analysis of TCF4 SNP rs17089887 and CTG18.1 Variants in Indian FECD Cohort.

	Cases, n = 44	Controls, n = 97	P Value
CTG18.1
XX	1	0	2 × 10−4
XS	14	5
SS	29	92
rs17089887	n = 44	n = 107
TT	29	48	0.013
TC	13	47
CC	2	12
Haplotype
T-S	0.676	0.644	0.5977
C-S	0.138	0.319	1.2 × 10−3
T-X	0.131	0.024	2 × 10−4
C-X	0.056	0.013	0.0303

 P values of alleles and haplotype significantly associated to the disease are denoted in bold.

Discussion

Employing keratoplasty procedures alone, 14,153 corneal transplantations were done solely for correcting endothelial cell failure associated with Fuchs' Dystrophy in 2013 in the United States.¹⁸ According to an earlier report conducted at a tertiary eye care center in South India in 2004, 24 (16.6%) of the 144 corneal buttons of corneal dystrophy patients who underwent PK, tested positive for FECD.¹² Therefore, an early-stage FECD diagnosis and comprehension of the disease pathomechanism has become an absolute necessity for the rapidly rising number of patients undergoing keratoplasty globally every year. The present study attempts to elucidate the Indian scenario of FECD. However, the study participants gathered for this purpose exhibit a slight male dominance among the control population. As reported by Lewallan and colleagues,¹⁹ women in developing countries face sociocultural influences and financial barrier due to which they have lesser access to treatments for visual impairment in comparison with men. Being a developing country, India faces similar hindrances that are reflected in the marginally skewed sex ratio of the control population. Despite the sex bias, we ensured that the study groups are not significantly distinct. 

With increasing evidence of TCF4 being a prominent genetic contributor for FECD and scarce studies to elaborate the Indian scenario, our study aimed to investigate the association of various polymorphisms of TCF4 gene with patients suffering from FECD in the Indian cohort. We observed that rs17089887 and CTG18.1 show significant association with the disease, even after 10,000 permutation tests (Table 3). Unlike the Chinese cohort, the allele “T” of rs17089887 was found to be the risk allele for our population. These flip-flop associations can be reasoned as a consequence of difference in the LD structure of the gene in ethnically distinct groups.²⁰ When the LD structure of the region comprising these SNPs was compared between Chinese Han population residing in Beijing (CHB) and Gujarati Indians from Houston (GIH), we found it to be distinct in each group (Fig. 3). All the three SNPs are in complete LD in GIH cohort, coinciding with the LD plot for this study; whereas in CHB, rs17089925 and rs17089887 are the only pair to be in LD with each other. Indian FECD subjects do not show distinct association with rs613872 at the current power of the study; despite the fact that this population exhibits variation in polymorphism (T: 0.87, G: 0.12) unlike the Chinese population that are not polymorphic for this genomic position.⁹ However, the genotypic frequency of this SNP shows marginal association with FECD (P = 0.03), which explains that patient and control cohorts exhibit distinct distribution of genotypes such that individuals with genotype TT are at higher risk (65.9%) for contracting the disease. For rare diseases like FECD, the critical sample size sufficing the required genetic power (80%) is often difficult. Similar studies in future with a larger sample size and adequate power are required to improve such marginal associations. 

Figure 3

 

Linkage disequilibrium plot comparisons of three TCF4 intronic SNPs in different ethnic groups. Situated at the third intron of TCF4, the LD plot of rs17089887, rs613872, and rs17089925 depicts high-linkage disequilibrium among these SNPs in our study population and the GIH cohort; whereas only rs17089887 and rs17089925 are in LD in CHB population. The plot was constructed using Haploview 4.2 and the numbers in the diamonds represents the D' values.

A genetic marker, be it a microsatellite with 2, 3, or 4 bp of a short, tandem, repeat pattern or a variation at a single DNA base pair position (SNP), is polymorphic if the variation is seen in more than 1% of a population. Such polymorphic markers are said to be associated with a disease if there is a difference between the allele frequencies of unrelated diseased subjects (cases) and those without disease (controls). The association between the genotyped marker and the disease phenotype can be due to various reasons. The associated marker can be a causal genetic variant where the change in the DNA sequence is itself contributing toward the disease trait or that the association reflects a nearby DNA element that is in LD with the polymorphic marker and is functionally responsible for the disease.^21,22 The association can also be a consequence of population stratification. The genetic markers associated in this study are rs17089887 and CTG18.1 allele. In context with the previously associated polymorphisms of TCF4 gene in various populations, the CTG18.1 allele remains consistently associated throughout. This suggests that CTG18.1 can be a putative causal variant for FECD; although it is subject to rigorous evaluation in order to avoid false assignment of causality. 

The threshold of minimum repeat length beyond which it is rendered as a marker for FECD has not yet been demarcated. Weiben and colleagues,¹⁰ with the help of direct sequencing, STR assay and genomic Southern blots, were the first to report that repeat lengths arbitrarily assumed to be more than 50 units was associated with FECD in the Caucasian cohorts. Using a combination of STR assay and triplet repeat primed PCR (TP-PCR), Mootha et al.¹¹ confirmed its strong association with the disease. Such aberration in the repeat region, although present in approximately 5% of heathy population, is highly prevalent among FECD affected subjects. The techniques used in our study are PCR-based amplification of the microsatellite region, which consequence into a certain degree of differential amplification of size-variant alleles. The competitive nature of PCR favors amplification of shorter alleles over the extended ones.²³ Therefore, this technique has a slight possibility of underestimating the repeat lengths in a minority of individuals unlike the methodologies used in previous reports.^10,11 Despite this underestimation, the prevalence percentage (34%) obtained is still significantly associated (P = 2 × 10⁻⁴) with FECD in our population. However, the role of this unstable trinucleotide repeats in contributing toward the pathogenesis of the disease remains uncertain. It was revealed through recent studies in neurodegenerative diseases, like myotonic dystrophy and Huntington's disease, that such unstable trinucleotide repeats can lead to selective cell death through a toxic gain-of-function mechanism by sequestering these RNA in ribonuclear foci.²⁴ Taking into account, the apoptotic nature of degenerative endothelial cells in FECD, some researchers have also hypothesized that these unstable intronic expansions might be responsible for incurring RNA stress in these cells, and thereby causing apoptosis.^24,25 

The haplotype (G-X) of rs613872 risk allele (G) and expanded CTG allele (represented as X), respectively, depicted significant (P = 5.9 × 10⁻¹⁹) association with the disease trait in Caucasians.¹¹ In this study, we replicate similar association with an Indian cohort; the expanded CTG repeats accompanying the disease trait along with the “T” allele of rs17089887; with the haplotype T-X being significantly associated (P = 2 × 10⁻⁴). These SNPs, rs613872 and rs17089887, are situated very close to each other (2 kb), and thereby share an appreciable LD between them (Fig. 3). Combined signal of rs17089887 and CTG or rs613872 and CTG suggest the presence of significant disease-causing changes in the nearby regions of these alleles that are physically linked. Further studies catering to the quest of mRNA profile of the various repertoire of TCF4 RNA may solve its role in the involvement of disease progression. 

With recent studies adding on more genes as contributors toward FECD, it is becoming even more difficult to understand its pathomechanism. Our study reports, for the first time, the genetic scenario of a sample Indian population showing significant association between TCF4 polymorphisms and FECD; thereby strengthening its role as a major contributor globally. 

Acknowledgments

The authors thank the study participants for their contribution and consent to this study. They also thank Satabdi Sundaray for sample collection, Vinay K. Malloji for his contribution in genomic DNA extraction, and Sreedevi Yadavalli for assisting in professional English editing of the manuscript. 

Supported by National Institute of Science Education and Research (NISER; Bhubaneswar, Odisha, India), Department of Atomic Energy (Mumbai, Maharashtra, India). Senior research fellowships from the University Grants Commission (GGN) and NISER (BP). 

Disclosure: G.G. Nanda, None; B. Padhy, None; S. Samal, None; S. Das, None; D.P. Alone, None