A Functional Polymorphism at the FGF10 Gene Is Associated With Extreme Myopia

Abstract

Purpose: Fibroblast growth factor-10 (FGF10) can modulate extracellular matrix associated genes and, therefore, it could be a myopia susceptibility gene. This study used an animal model, single nucleotide polymorphisms (SNPs) association, and genetic functional assay to evaluate FGF10 gene for myopia.

Methods: The expression levels of FGF10 gene were compared among the form deprivation myopic (FDM) eyes, the fellow eyes of the FDM group, and the healthy eyes of experimental mice. In the present study 1020 cases (≤−6.0 diopters [D]) and 960 controls (≥−1.5 D) were enrolled from a Chinese population. Eight tagging SNPs were genotyped to test for an association between genotypes and myopia. The luciferase reporter assay was conducted for the particular SNP to assess the allelic effect on gene expression.

Results: The sclera of FDM eyes had a 2.57-fold higher level of FGF10 mRNA (P = 0.018) than the fellow eyes. Although no SNP was associated with high myopia, SNP rs339501 was significantly associated with extreme myopia (≤−10 D, P = 0.008) and the odds ratio (OR) was 1.58 for G allele carriers. The luciferase assay showed that the risk G allele significantly caused a higher expression level than the A allele (P = 0.011).

Conclusions: The evidence suggested FGF10 to be a risk factor for myopia. The sclera of myopic eyes had higher FGF10 levels. The risk G allele of SNP rs339501 was associated with extreme myopia in human and caused a higher gene expression in the luciferase assay. It is concluded that the FGF10 could have been involved in the development of myopia.

SIK2 negatively regulates Müller cell survival under normal conditions and chronic hyperglycemia.

Introduction

Myopia is a common eye condition worldwide, and its prevalence varies widely among populations and ages and between the sexes.^1–3 When the definition of less than −6 diopters (D) was used, the prevalence of high myopia was found to be 18% among young Taiwanese men and 24% among young Taiwanese women¹; both of which are higher than the 13.1% reported among young men in Singapore.² Furthermore, an increased frequency of high myopia (<−6.0 D) was found in young Taiwanese people: 10.9% in 1983 and 21% in 2000.⁴ Myopia progresses quickly in childhood, especially around the teenage years. By early adulthood, the rate of change in ocular refraction tends to decline and the prevalence of myopia stabilizes.⁵ Several studies have also shown the family history of myopia to be a significant risk factor.^6,7 Recently, genetic association studies including genome-wide association studies (GWAS) have reported several susceptibility genes to nonsyndromic myopia.^8–17 

Scleral remodeling is one of the important mechanisms for the development of myopia,¹⁸ Several genes are associated with the scleral remodeling including sulfated glycosaminoglycans (GAGs),¹⁹ matrix metalloproteinases (MMPs),²⁰ tissue inhibitors of metalloproteinases (TIMPs),²⁰ and TGF-β.²¹ The expression of fibroblast growth factor 10 (FGF10) has been shown to be abundant in the murine retina and sclera.^22,23 FGF10 plays a pivotal role in controlling elongation of lacrimal gland bud by modulating GAGs, MMPs, and TIMPs^24,25 during ocular gland development. Furthermore, defect in FGF10 leads to the development and differentiation of several ocular tissues.^26–28 In our previous study of microRNA-328, retinoic acid, and TGF-β were shown to be involved in the same network regulation leading to the development of myopia.²⁹ A recent study indicated that retinoic acid could regulate the TGF-β pathway in the control of FGF10 expression.³⁰ Therefore, the present study focuses on the involvement of FGF10 in the development of myopia. 

The present study aims to discover the role of FGF10 in myopia by the following methods: (1) measuring the FGF10 expression during the development of myopia in mice, (2) testing for the association between single nucleotide polymorphisms (SNPs) of FGF10 and high myopia in a Chinese cohort residing in Taiwan, and (3) evaluating the possible biologic consequence of the associated SNP. 

Materials and Methods

Animal Model

The C57BL/6J mice were purchased from the National Laboratory Animal Center, Taipei, Taiwan. The mice (23 days old) were randomly assigned to the form deprivation myopia (FDM) and healthy control eye (free of form deprivation). Animals in the FDM group wore diffuser goggles³¹ that covered the right eye for four weeks. After four weeks, the goggles were removed from the animals. After sacrifice, total RNAs were extracted from the retinal and scleral tissues separately. The gene expression levels were determined by quantitative PCR. 

The animal care guidelines are comparable with those published by the Institute for Laboratory Animal Research. The animal research in this study was approved by the Animal Care and Ethics Committee at Kaohsiung Medical University, Kaohsiung, Taiwan. The treatment and care of animals were conducted according to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. 

Gene Expression

The mRNA levels of FGF10 from 10 FDM eyes, 10 fellow eyes, and 4 healthy eyes were respectively measured by real-time PCR. The PCR reactions were performed by Applied Biosystems 7500 Real-Time PCR System using ×2 SYBR Green PCR Master Mix (Applied Biosystems, Foster City, CA). Total RNA was extracted from the retina and sclera separately with Trizol reagent (Invitrogen, Grand Island, NY), and quantity of RNA was confirmed by spectrophotometry and formaldehyde/agarose gel electrophoresis. In order to remove contaminating genomic DNA, 1 μg of total RNA was treated with 1 U RNase-free DNase I (Promega, Madison, WI) at 37°C for 30 minutes and then heated with 1 μL stop solution (Promega) at 65°C for 10 minutes. Subsequently, 0.5 μg of total RNA in each sample was reversely transcribed (M-MLV reverse transcriptase; Promega) using 0.04 μg random primers (Promega) in a total volume of 20 μL according to manufacturer's instructions. To confirm the unique PCR product for each sample, the melting curve after each reaction was checked. The expression level of FGF10 mRNA was normalized to that of an internal control Glyceraldehyde 3-phosphate dehydrogenase (GAPDH) by using the equation of log₁₀ (2^−ΔCt), where ΔCt = (Ct_FGF10 − Ct_GAPDH). The median and mean of log₁₀ (2^−ΔCt) and its SD were calculated. The relative expression level was used to indicate the fold change between different types of eyes by using the equation of 2^−ΔΔCt. The paired t-test was used to compare the difference of FGF10 expression between FDM and fellow eyes, and unpaired t-test was used for the data from different groups of animals. 

Genetic Association Study

The subjects for the current study were selected from existing data. The volunteer study subjects were recruited from the communities through poster advertisements or oral solicitation in southern Taiwan. The recruitment was not based on refraction errors, any particular ocular traits, or family history. All the participants were of Chinese descents. The subjects who met the following criteria were selected for the current study: (1) age between 18 and 40 years, (2) myopia in both eyes and a spherical refraction less than or equal or to −6.0 D in at least one eye for a case, (3) a spherical refraction greater than or equal to −1.5 D in the more myopic eye for a control, and (4) none of the controls had received any previous refractive surgery. The refractive error was measured without cycloplegia in the subjects. The refractive error was measured using an autorefractometer (Topcon KR-8100 or RM-8800; Topcon, Tokyo, Japan) for all eyes. A written informed consent was obtained from each subject. The study was approved by the institutional review board at the Kaohsiung Medical University Hospital. The research followed the tenets of the Declaration of Helsinki. 

Tag SNP Selection and Genotyping

The tag single nucleotide polymorphisms (tSNPs) at the FGF10 gene were selected from the release 3.0 Phase II data of the HapMap Project using the Tagger Pairwise method.³² tSNPs were chosen according to the following criteria: r² greater than or equal 0.8 and the minor allele frequency (MAF) greater than or equal to 10% in the Han Chinese population. A total of eight tSNPs met the selection criteria, which included three promoter SNPs: rs1482668, rs723166, rs2929855; and five intronic SNPs: rs1384449, rs2128433, rs339501, rs1011814, and rs1374962. Genotyping was performed by using the TaqMan technology (assay IDs are listed in Supplementary Table S1). Briefly, PCR primers and TaqMan minor groove binder (MGB) probes were designed and reactions were performed in 96-well microplates with ABI 9700 thermal cyclers (Applied Biosystems). Genotype data were analyzed with its System SDS software version 1.2.3 (Applied Biosystems). 

Statistical Analysis for Polymorphism Association Study

The allele frequency was obtained by direct gene counting. Hardy–Weinberg equilibrium (HWE) was tested in controls by using the χ² test for each SNP. According to the myopic status and three genotypes of each SNP, logistic regression was performed to calculate the significance with adjustment for age, sex, and college education. 

For multiple testing, family-wise permutations were performed to obtain an empiric P value for each SNP. In the present study, the minP permutation procedure was used to calculate the empiric P values. The minP is a distribution-free statistics that uses random label swapping to generate the null distribution in a population. The null distribution was obtained with the following steps: (1) swapping refractive errors randomly among all study subjects, (2) performing logistic regression on the permuted data to obtain the minimum P value of each run, and (3) repeating steps 1 and 2 to obtain 10,000 permutations. The empiric P values were calculated by the following formula: (R + 1)/(N + 1), where R is the number of runs in which the minimum P value is less than that observed, and N is the number of permutations. 

Construction of rs339501 Luciferase Vectors

The PCR reactions were carried out with the sequence-specific primer pairs: rs339501-A-allele 5′-GCAGGTACCTATTTGTGGTTCTCC-3′ and 5′-GGGACGCGTCTCTTTAATCAGTC-3′. For cloning the region containing the significant SNP rs339501, the primers were designed to contain a KpnI site and MluI cutting site. Desired DNA fragments were amplified by PCR and then inserted into luciferase reporter vector pGL3-Basic (Promega). The inserts were positioned in sense orientation relative to the luciferase coding sequence. The construct containing A allele of rs339501 was confirmed by DNA sequencing. To establish the construct carrying the rs339501 G allele, the site-directed mutagenesis experiments were conducted by overlap extension PCR. The mutant primer pairs were rs339501-G-allele 5′-CGTGTTAGTCATGGCTCAGACT-3′ and 5′-AGTCGAGCCATGACTAACACG-3′. The underline bases in the mutant primers indicated the position for mutating A allele to G allele. The sequence of rs339501 G allele construct was also confirmed by DNA sequencing. 

Transient Transfection and Luciferase Assays

The HRPE-19 cell line (American Type Culture Collection, Manassas, VA) was cultured in Dulbecco's modified Eagle's medium (DMEM; Gibco-BRL, Gaithersburg, MD) supplemented with 10% bovine fetal serum (Gibco-BRL), 2 mM l-glutamine (Sigma-Aldrich, St. Louis, MO), and 1 mM pyruvate (Sigma-Aldrich). The cells were maintained at 37°C in an atmosphere of 5% CO₂. The constructs (400 ng) were transfected into the RPE-19 cells (Lipofectamine 2000; Invitrogen, Carlsbad, CA). The luciferase assay was then performed according to the manufacturer's protocol. The cells were lysed in the passive lysis buffer (Promega) for 24 hours after transfection, then luciferase activity was measured (Luciferase Assay System; Promega). The enhanced green fluorescent protein expressing vector (pEGFP-C1) was also cotransfected into the RPE-19 cells, and the fluorescence level was used as the internal control. Each experiment was independently repeated three times, and each sample was studied in triplicate. A two-sided P value less than 0.05 was considered significant, according to the results of the Mann-Whitney U test. 

Results

Altered FGF10 Gene Expression in Form Deprivation Myopic Eye

The mRNA levels of FGF10 were altered in scleral cells in FDM mice, but not in retinal cells among three types of eyes (Fig. 1). Using the data obtained from the sclera of healthy eyes as the reference, mRNA levels were increased by 1.62-fold in the sclera of FDM eyes while the levels were decreased by 37% in the sclera of fellow eyes of the FDM group. The difference in the levels of FGF10 mRNA between the sclera of FDM and fellow eyes of the same animal was found to be significant (by 2.57-fold, P = 0.018). Although the FDM eyes had a 1.6-fold increase of FGF10 expression than the healthy eyes, there was not statistical significance for the difference between fellow eyes and healthy eyes, or FDM and healthy eyes. 

Figure 1

 

FGF10 mRNA levels in FDM, fellow, and healthy eyes (a) in the sclera and (b) in the retina. The expression level in the sclera tissues of FDM eyes was significantly higher than fellow eyes (P = 0.018). *P < 0.05.

Genetic Association Study for FGF10 Gene

A total of 1020 highly myopic cases and 960 controls were included in the present study. The mean age was 21.4 years for cases and 20.6 years for controls. The spherical refractions, which were measured in diopters (D), of the more myopic eyes had the mean ± SD of −8.0 ± 1.7 D for cases and −0.5 ± 0.6 D for the controls. The distributions of age, sex, myopia severity, and college education levels are shown in Table 1. 

Table 1

 

Characteristics of the Study Subjects

The call rates for the eight SNPs ranged from 91% to 99%. Three SNPs (rs2128433, rs2929855, and rs339501) had the call rates lower than 95%. To be sure of the genotyping quality, Sanger sequencing was performed for these three SNPs in 30 subjects. The sequencing data were completely consistent with the previous genotypes determined by the TaqMan assay. All the SNPs were in HWE in the controls, and the linkage disequilibrium plot is shown in Figure 2. The frequencies of the genotypes and the associations between the status of high myopia (≤−6.0 D) and the genotypes are shown in Table 2. Absence of a significant difference was observed both on the allele frequencies (data not shown) and genotype frequencies (Table 2) between the cases and the controls. Further an exploratory analysis was carried out to compare 125 subjects with extreme myopia (≤−10 D) versus controls. Two SNPs, rs339501 and rs1384449, had significant associations with extreme myopia with P values of 0.008 and 0.011, respectively (Table 3). In addition, the other two SNPs, rs2128433 and rs1011814, had borderline significance with P values of 0.063 and 0.086, respectively. However, only rs339501 remained significant after multiple testing correction (permutation P value = 0.049). The subjects carrying the minor G allele (i.e., either AG or GG subjects) were more likely to have extreme myopia with the odds ratio (OR) of 1.58. The genotype distributions of AA, AG, and GG were 70.5%, 24.6%, and 4.9%, respectively, in the extreme cases, and 76.7%, 21.5%, and 1.7%, respectively, in the controls. We also performed sex-specific analysis, and the results indicated that the G allele carriers in extreme myopia were always over-represented in the extreme myopia in both men and women (Supplementary Table S2). 

Figure 2

 

The linkage disequilibrium plots for the eight tag SNPs of FGF10. The darker gray cells indicate stronger linkage (higher D'). The numbers indicate r² (%) of pairwise tag SNPs.

Table 2

 

The Association Between tSNP Genotypes and High Myopia

Table 3

 

The Association Between Eight tSNPs and Extreme Myopia

Functional Assessment of rs339501 by Luciferase Assays

Since SNP rs339501 is an intronic SNP, the University of California, Santa Cruz (UCSC), Santa Cruz, CA genome browser was checked (in the public domain, http://genome.ucsc.edu/) to explore potential functional consequence of SNP rs339501. This SNP is located in the overlap region of three binding sites by three transcription factors (Fig. 3) based on the ENCODE TF ChIP-seq experiment.³³ The three transcription factors were BAF155, JUND, and CEBPB. Further, to validate that the A/G allele of SNP rs339501 could cause differential FGF10 expression levels, two luciferase reporters, each carrying one allele, were constructed (Figs. 4a, 4b). As shown in Figure 4c, the reporter carrying the G allele had a significantly higher expression level of luciferase activity than the reporter carrying the A allele (P = 0.011). These results suggested that the risk G allele causes a higher expression level than the protective A allele. 

Figure 3

 

Genomic location of rs339501. rs339501 is at the overlap region of three transcription factor binding sites.

Figure 4

 

Functional assessment by luciferase reporter assay for SNP rs339501. (a) Double-stranded oligonucleotides (containing rs339501) were cloned into the pGL3-basic luciferase vector to construct two different reporter constructs carrying the protective A allele or risk G allele. (b) Two sets of primers for the vector construction. (c) Luciferase activity of the two constructs in the RPE-19 cell line. Significant difference of luciferase activities between two reporter constructs (P = 0.011), using the data (mean ± SEM) from three independent transfection experiments. *P < 0.05.

Discussion

The present study provides evidence to support FGF10 as a risk factor for myopia. The first evidence was from the FDM animal study where higher scleral FGF10 level was found in the FDM eyes than the fellow eyes and healthy control eyes. Subsequent genetic association studies using 1980 Chinese subjects demonstrated that the risk G allele of SNP rs339501 was associated with extreme myopia (≤10 D). This significant SNP rs339501 was located at the overlap region of three transcriptional factor binding sites. Using the luciferase reporter assay, it was determined that the vectors containing the risk G allele of rs339501 had an increased luciferase activity. All these results consistently indicate that an over-expression of FGF10 could increase the risk for myopia. 

Sclera remodeling involves alterations in both the degradation and the synthesis of extracellular matrix (ECM) components. There are 18 FGF members that could be grouped into six subfamilies based on differences in sequence homology and phylogeny.³⁴ Several FGF members have been shown to regulate ECM.³⁵ FGF10 has been reported to increase MMP2 and MMP9 expression. All FGF members, when bound to the FGF receptors, would subsequently inhibit the TGF-β signaling.³⁶ Our recent study also indicated that TGF-β signaling could play a role in the development of myopia.²⁹ Hence, FGF10 is likely to have involved in the development of myopia by mediating TGF-β signaling to further remodel ECM. 

Previously, our group reported that the FGF2 gene could have been involved in the development of myopia, although polymorphisms of the FGF2 gene were not statistically associated with high myopia.³⁷ However, FGF2 and FGF10 do not belong to the same subfamily. So far, FGF10 has not been reported to be a regulatory factor for myopia. Since both FGF2 and FGF10 could bind with FGF receptors to regulate TGF-β signaling, it was interesting to further investigate their interaction during myopia formation. 

Based on the ENCODE ChIP-seq data,³³ three transcription factors (BAF155, JUND, and CEBPB) can bind at the region where SNP rs339501 is located. In murine retina and sclera, JUND has the most abundant expression, followed by BAF155, and then CEBPB.^22,23 However, the exact effect of these transcription factors on FGF10 expression has not been experimentally conducted in the scleral cells. Two clones containing A and G of rs339501 and flanking transcription factors binding site were constructed. The luciferase activity of the clones was higher in the G allele than the A allele. Taken together, the overall data implied that the risk G allele of rs339501 might increase FGF10 expression level by enhancing transcription factor binding. 

FGF10 has not been identified as a myopia susceptibility gene by previous genome-wide experiments, including cDNA microarray, proteomic analysis in sclera tissue of murine FDM eyes,^22,38 and GWAS.^10–17 Several possible reasons could explain their failure to identify FGF10 as a susceptibility gene. A well-recognized fact is that microarray experiments for global gene expression are subject to both type I and type II errors. Moreover, the current microarray platforms cannot provide good quantification to differentiate gene expression. Therefore, a subtle change in gene expression could be overlooked. In GWAS myopia studies, no significant SNPs were found in 5q12, where the FGF10 gene was located. A recent GWAS consortium analyzed more than 37,000 individuals from 32 studies, including 5 Asian cohorts, and identified more than 24 myopia susceptibility loci.³⁹ However, our significant SNP rs339501 and its four tagging SNPs (rs10512851, rs10512848, rs1037181, and rs16901825) have not been included in the Illumina 610 platform that was used to perform genotyping in the five Asian cohorts. Furthermore, only two GWAS studies focused on high myopia (refraction error ≤ −6 D),^11,13–15 while the others used the axial length,¹⁰ refraction error,^16,17 or high myopia with retinal degeneration as the outcomes.¹² Given that the FGF10 polymorphisms are only significant for extreme myopia (refraction error ≤ −10 D), discovering this gene effect may not be easy by a genome-wide approach. 

In FDM mouse, the mRNA level of FGF10 was significantly higher in treated eyes than fellow eyes. It is not clear why the fellow eyes had a lower level of FGF10 than the healthy eyes, although the difference was not statistically different. It has been reported that monocular visual treatment could produce binocular changes in scleral mRNA levels.^20,40 However, it is unclear whether a lower level of FGF10 level in the fellow eyes than the healthy eyes was due to random variation or a compensation effect in the fellow eyes. Further studies are warranted to clarify this point. 

The study design has strengths and limitations. The animal studies, functional genetic assay, and a large number of study subjects consistently implied the role of FGF10 in myopia development. The study population was relatively homogenous in terms of ethnicity, geographic location, and age. Previous studies have also indicated no concern of population stratification in the Taiwanese population.^41,42 However, the main finding of a significant association was based on 122 genotyped subjects with extreme myopia (≤−10 D) and, thus, future studies to replicate our finding are warranted. Although the significant SNP rs339501 might directly regulate FGF10 expression, fine mapping is still needed to be sure of the causal SNP in the region. 

In conclusion, firstly an increased expression level of FGF10 gene in murine FDM eyes was found. Using a large population, the minor G allele of SNP rs339501 was identified as a risk genetic marker for extreme myopia. Three transcriptional factors could bind at the region where the SNP is located and the reporter assay supported the fact that the risk G allele could cause a higher FGF10 gene expression. The present study suggested that higher expression of FGF10 could be a risk for myopia. 

Acknowledgments

Supported by grants from the Taiwan National Science Council (NSC95-3112B037-003 and NSC 99-2628-B-037-037-MY3). 

Disclosure: E. Hsi, None; K.-C. Chen, None; W.-S. Chang, None; M.-L. Yu, None; C.-L. Liang, None; S.-H.H. Juo, None