|Home | About | Journals | Submit | Contact Us | Français|
Primary open-angle glaucoma (POAG) is a genetically complex common disease characterized by progressive optic nerve degeneration that results in irreversible blindness. Recently, a genome-wide association study (GWAS) for POAG in an Icelandic population identified significant associations with single nucleotide polymorphisms (SNPs) between the CAV1 and CAV2 genes on chromosome 7q31. In this study, we confirm that the identified SNPs are associated with POAG in our Caucasian US population and that specific haplotypes located in the CAV1/CAV2 intergenic region are associated with the disease. We also present data suggesting that associations with several CAV1/CAV2 SNPs are significant mostly in women.
Primary open-angle glaucoma (POAG) is a phenotypically complex disease that causes progressive optic nerve degeneration resulting in irreversible blindness. Elevated intraocular pressure (IOP), caused by abnormal regulation of fluid dynamics within the eye, is a major risk factor for POAG-related optic nerve disease, although optic nerve disease can occur in the setting of normal IOP (normal-tension POAG; abbreviated NTG) (1,2).
Family and twin studies suggest that POAG has a significant heritability (3); however, the discovery of POAG susceptibility genes has been challenging. Genetic and genomic approaches have not yielded POAG genes with significant population attributable risks, suggesting that the underlying genetic architecture includes many genes of modest effect (4).
Recently, a genome-wide association study for POAG identified significant associations in an Icelandic case–control sample for two single-nucleotide polymorphisms (SNPs) located in an intergenic region between the CAV1 and CAV2 genes on chromosome 7q31, which code for two members of the caveolin family of proteins (5). Caveolins are implicated in a wide range of processes including modulation of the endothelial cell membrane (6), a process that participates in drainage of fluid (aqueous humor) from the eye (7). Caveolin-mediated endothelial cell function is also influenced by eNOS (8), and in previous studies, we identified interactions between NOS3 SNPs and postmenopausal hormone use in women in relation to high-tension POAG (HTOAG), suggesting that eNOS (NOS3 gene) under the influence of exogenous sex hormones could contribute to POAG (9). Loss of caveolin-1 function can cause increased expression of NOS3 (10), suggesting that caveolin-1 could modulate IOP through a mechanism that includes eNOS. Moreover, recent data suggest that caveolin-1 is required for the activation of endothelial nitric oxide synthase in response to 17-beta-estradiol (11), providing additional evidence that these interactions could be sex related. As a sex effect has not been considered for the CAV1/CAV2 POAG association, the purpose of this study was to first replicate an association between CAV1/CAV2 SNPs and POAG in our study population and then investigate the association in men and women independently.
Table 1 shows the baseline characteristics of the 1000 POAG cases and 1183 controls from the Primary Open-Angle Glaucoma Genes and Environment (GLAUGEN) study collected as part of the Gene Environment Association Studies (GENEVA) (12) consortium. POAG cases and controls were not significantly different with regard to age and sex or environmental exposures (Table 1). All patients were self-described Caucasians.
Using a model controlling for age (years), sex, method of extraction (QIAGEN, GENTRA, DNAzol), DNA specimen (from blood or buccal sample), race (three Eigen vectors) and study site [Genetic Etiologies of Primary Open-Angle Glaucoma (GEP), Nurses’ Health Study (NHS), and Health Professionals Follow-up Study (HPFS)], we initially tested for the effect of the minor allele at the two SNPs reaching genome-wide significance in the Icelandic study: rs4236601, P= 5.00E–10, odds ratio (OR)= 1.36 (95% confidence interval (CI): 1.23–1.50); rs1052990, P= 1.10E–09, OR= 1.32 (95% CI not reported) (5). We found that the minor alleles at these two SNPs were significantly associated with POAG in our study sample: rs4236601, P= 0.0014, OR= 1.28 (95% CI: 1.10–1.48); rs1052990, P= 0.0003, OR= 1.30 (95% CI: 1.13–1.50) (Table 2). Next, we examined the association between an additional 61 SNPs located in a 50 kB region flanking either side of the CAV1/CAV2 genomic region and POAG. Fourteen additional SNPs in the region were also positively associated with POAG risk overall (Fig. 1, Supplementary Material, Table S1) with ORs ranging from 1.16 to 1.30, and one SNP, rs6466579 showed a protective association (OR= 0.80). None of the additional 61 SNPs was independently associated with POAG after adjusting for rs4236601 and rs1052990 (Supplementary Material, Table S2). The SNPs significantly associated with POAG were found in the intergenic region 3′ to CAV2 and 5′ to CAV1 (Fig. 1).
The CAV1/CAV2 genes are located in a 300 kb region on chromosome 7q31. The SNPs associated with POAG in this study are clustered in a region between the 3′ end of the CAV2 gene and the 5′ end of the CAV1 gene (Fig. 1). Haplotype analyses using sliding windows across the region identified two haplotypes that remained significant after permutation testing: rs10256914 (G), rs3807986 (A), rs959173 (A), rs10270569, (A) P= 0.027; and rs6975771 (G), rs6976316 (G), rs12530912 (C), rs17138756 (A), P= 0.029 (Supplementary Material, Fig. S1). Interestingly, the two SNPs identified in the Icelandic study (rs1052990 and rs4236601) were not part of the two significant haplotypes, although a conditional analysis did not provide evidence that these two haplotypes were independent of rs1052990 and rs4236601, potentially due to linkage disequilibrium. Both associated haplotypes occur less frequently than the minor alleles of rs1052990 and rs4236601, although the effect sizes are comparable (OR 1.27 for rs6975771–rs17138756 and 1.28 for rs10256914–rs10270569). It is possible that the haplotypes are capturing a rarer SNP not present on the Illumina 660W platform that could influence disease development.
Our previous studies identifying an interaction between NOS3, postmenopausal hormone use and POAG in women suggested that sex may significantly influence specific pathways responsible for POAG development. We analyzed the CAV1/CAV2 SNPs in women and men independently and found that both rs1052990 and rs4236601 were significantly associated with POAG in women but not in men (Table 3). To assess the possibility that sex and genotype have a synergistic effect on POAG risk, we included interaction terms in our main regression models. The P-for-interaction based on sex was not significant for rs1052990 or rs4236601, possibly due to inadequate overall sample size. The P-for-interaction based on sex was <0.05 for 2 of the 63 SNPs in the greater CAV1/CAV2 region; however, the main effects for these SNPs were not significant (data not shown).
In our previous studies, NOS3 SNPs were preferentially associated with women with the high-tension variant of POAG. To determine whether the CAV1/CAV2 SNPs were preferentially associated with high-tension or normal-tension forms of POAG, we investigated the association between CAV1/CAV2 SNPs with normal-tension POAG versus controls and high-tension POAG versus controls adjusting for sex and found that rs1052990 was nominally significantly associated with normal-tension glaucoma (Table 3), while rs4236601 was not. Neither SNP was associated with high-tension POAG.
Although POAG has significant heritability, linkage studies, candidate gene studies and genome-wide association studies have not identified genes with a major population effect indicating that the underlying genetic architecture of the condition is complex. In this study, we confirmed an association between SNPs located in the CAV1/CAV2 genomic region in an American Caucasian population. Additionally, haplotype analysis suggests that the conserved genomic region 5′ to the CAV1 gene may be biologically relevant. This region contains a number of transcription factor-binding sites that could influence CAV1 gene expression (http:www.genome.ucsc.edu). Finally, our exploratory secondary analyses examining the role of sex and IOP-related endophenotypes suggest that this association may be enhanced in women. Interestingly, a recent report of 545 Caucasian POAG cases and 297 controls from Iowa and one of the CAV1/CAV2 SNPs (rs4236601) identified in the original study (5) described a lack of association with POAG (13); however, this study did not stratify by IOP endophenotypes or sex and had limited power to detect these modest effects.
Sex effects in POAG have been suggested by other studies. Estrogen receptors are expressed on the optic nerve retinal ganglion cells (14), and estrogen may have a neuroprotective effect in animal models of glaucoma (15,16). Postmenopausal hormone use results in modest IOP reductions in several studies (17–22). Cohort analyses have revealed relations between age at menarche, oral contraceptive use, age at menopause, postmenopausal hormone use and the various subtypes of POAG (23–26). Two groups have shown an inverse relation between body mass index and POAG in women, but not men (27,28), indicating that this association might be hormonally driven. Further studies examining sex effects in POAG, and sex interactions with genes contributing to POAG will be of interest.
Overall, the CAV1/CAV2 SNPs conferred modest risk to glaucoma susceptibility accounting for a small percentage of POAG heritability, suggesting that other, as yet unknown, genetic and/or environmental factors also contribute to this condition. As with other common complex disorders, it is likely that multiple molecular events and biological pathways influence disease development (29). Some POAG-associated pathways may be influenced by hormonal effects and would be expected to exhibit a significant sex effect. Specific pathways may also influence primarily IOP, while others may have a more substantial effect on retinal ganglion cell disease. This potential for multiple predisposing pathways suggests that stratification of the aggregate data by sex, IOP and other phenotypic variables may be an important strategy for POAG gene discovery.
Caveolin-1, the product of the CAV1 gene, is the scaffolding protein of caveolae and serves an important regulatory signaling function in endothelial cells (30). Endothelial cell function could broadly influence glaucoma pathogenesis; however, a major pathway for ocular fluid flow related to IOP involves the endothelium of Schlemm's canal, which connects the trabecular meshwork outflow pathways to the episcleral venous system (31). In response to increased fluid flow and increased IOP, Schlemm's canal inner wall endothelial cells form giant vacuoles (32). Vacuole formation involves modulation of endothelial cell signaling and vascular tone (33), processes that could be influenced by both caveolin-1 and eNOS (34) (Fig. 2). Interestingly, increased caveolin-1 expression has been observed soon after elevation of IOP in an in vitro glaucoma model (35). In our study, the CAV1/CAV2 SNPs were significantly associated with POAG overall; however, in our population, these associations were mostly significant in women, a finding that is supported by possible molecular interactions between caveolin, eNOS and 17beta-estradiol (11). These results, taken together with our previous studies on the relation of NOS3, POAG and sex (9) suggest that caveolin proteins, potentially through an interaction with eNOS and estrogen, may regulate IOP through a mechanism involving endothelial cell function. Our study also suggests that phenotype stratification and data analysis using molecular pathway approaches will be important strategies for successful POAG gene discovery.
After extensive quality control, we analyzed 1000 cases and 1183 controls drawn from three different studies: the GEP, the NHS and the HPFS. The GEP is a clinic-based case–control set, and the NHS and HPFS are case–control sets nested within population-based studies. The institutional review boards of the Massachusetts Eye and Ear Infirmary, the Harvard School of Public Health and the Brigham and Women's Hospital approved this study. Cases were defined as individuals with characteristic visual field defects consistent with glaucomatous optic neuropathy on reliable tests. Visual field tests were reproduced on a subsequent visual field or the cup disc ratio was 0.7 or more in at least one eye. Anterior segment exam did not show signs of secondary elevated IOP, such as exfoliation syndrome or pigment dispersion syndrome, and the filtration structures were deemed to be open based on clinical measures. Elevation of IOP was not a criterion; however, 67% of cases did have elevated IOP (>21 mmHg; HTOAG). Cases with IOP < 22 mmHg at diagnosis were classified as NTG. Controls had normal optic nerves (cup-to-disc ratios < 0.6), and normal IOP (<22 mmHg). Among controls, 9.65% had a positive family history of glaucoma. The majority of DNA samples were prepared using Qiagen extraction kits (Invitrogen). More than half of the samples were derived from buccal cells, and we previously demonstrated the feasibility of genotyping buccal cell DNA on the Illumina 660W Quad platform (36).
SNPs located within the CAV1/CAV2 region were genotyped using the Illumina 660WQuad platform at the Broad Institute, Cambridge, MA, USA.
Samples were plated to allow equal representation of cases and controls from each study site in order to minimize batch effects. The Illumina BeadStudio and Autocall software along with genotype cluster definitions based on study samples were used to generate genotyping calls. SNPs with GenTrain score <0.6, cluster separation score <0.4 and call rate <97% were considered technical failures at the genotyping center and were automatically zeroed before release for further quality control. Subsequent data quality-control measures consisted of identifying and removing samples with sex misidentification, unexpected duplicates and unexpected relatedness. Analysis of connectivity removed samples that appeared to be related to other samples and suggestive of contamination. Analysis of relative intensity data and allelic imbalance revealed no gross chromosomal anomalies on chromosome 7 in any sample. Any SNP with missing call rate >2% or with Hardy–Weinberg P-value < 10–4 was excluded.
Study site (GEP, NHS or HPFS), DNA source (blood or cheek cell) and DNA extraction method (DNAzol, QIAGEN or GENTRA) were independent predictors of genotyping call rate; hence, these variables along with age and sex were adjusted for in logistic regression models to assess the association between individual CAV1/CAV2 SNPs and POAG using PLINK v1.07. Since environmental exposures might be different between men and women, we performed secondary analyses in men and women separately, with further controlling for race (three eigenvectors), type 2 diabetes mellitus (yes/no), hypertension (yes/no), smoking history (ever/never), body mass index (kilograms per meter squared), alcohol intake (grams per day) and family history of glaucoma. For NHS and HPFS participants, this information was collected from biennial questionnaires and updated to the date of diagnosis for cases and matched controls. For cases and controls in the GEP, covariate data were collected at the time of DNA collection. Data on body mass index and alcohol consumption were not available for participants in GEP. The conditional analyses evaluating the SNP associations independent of the main Icelandic SNPs rs4236601 and rs1052990 were performed similarly as the single SNP analyses with the exception of the inclusion of SNPs rs1052990 and rs4236601 as covariates in the logistic regression models.
Haplotype analysis (c2) was performed with PLINK v1.07 using sliding windows of two to five SNPs across the region. Permutation testing (100 000 permutations) of haplotype blocks was performed using Haploview v4.2 for Windows. ORs for the two significant haplotypes (rs6975771–rs17138756 and rs10256914–rs10270569) were calculated using the same logistic regression model used for the single SNP associations.
Conflict of Interest statement. None declared.
This project was supported by grants to Pasquale (NHGRI U01HG004728; NEI R01EY015473; NCI CA87969, CA49449, CA055075; Research to Prevent Blindness Physician Scientist Award; the Margolis Fund (shared with J.L.W.) and J.L.W. (NEI R01EY015872). Genotyping was performed at the Broad Institute of MIT and Harvard, with funding support from the NIH GEI (U01HG04424). The GENEVA Coordinating Center (U01HG004446) provided assistance with genotype cleaning.