Evaluation of LOXL1 polymorphisms in primary open-angle glaucoma in southern and northern Chinese

Navigate by section:

• Introduction
• Methods
• Results
• Discussion

Abstract

Purpose: The lysyl oxidase-like protein 1 (LOXL1) gene is strongly associated with exfoliation glaucoma, which is very rare in the Chinese population. The implicated LOXL1 polymorphisms have not been associated with primary open-angle glaucoma (POAG). In this study, we investigated three of the LOXL1 polymorphisms in POAG in a southern Chinese population of Hong Kong and northern Chinese from Beijing.

Methods: The Hong Kong group included 293 POAG patients and 250 controls, and the Beijing group included 169 POAG patients and 197 controls. LOXL1 single nucleotide polymorphisms (SNPs), rs1048661, rs3825942, and rs2165241, were genotyped by direct DNA sequencing. Individual association was analyzed using the χ² test, and haplotype-based association analysis was performed in WHAP.

Results: Each of the candidate SNPs was not statistically associated with POAG in either group (p>0.017, Bonferroni correction). Haplotype-based association analysis had identified a significant omnibus association (Omnibus χ²=18.16, p=0.00115) between these SNPs and POAG in the Hong Kong group. A minor haplotype (T-G-T) showed significant statistical association with POAG. It presented in 2.1% of cases and 0.4% of controls, conferring a 5.24 fold of increased risk to the disease (95% CI: 1.17–23.54, P_perm=0.00108). However, this haplotype was absent in the Beijing group.

Conclusions: Individual LOXL1 SNPs, rs1048661, rs3825942, and rs2165241, were not associated with POAG in the Chinese population. However, a minor haplotype T-G-T was found to be associated with the disorder in the southern Chinese. The low frequencies of the at-risk alleles at rs1048661 and rs2165241 may be one of the factors that led to the low prevalence of exfoliation syndrome in the general populations of the Chinese.

Introduction

Glaucoma is a group of heterogeneous disorders that can lead to progressive optic neuropathy and loss of vision with or without the association of elevated intraocular pressure (IOP). It is a leading cause of irreversible blindness worldwide [1]. Primary open-angle glaucoma (POAG) is a major form of glaucoma. POAG is a complex disease with multiple genetic risk factors. So far, there are at least 24 candidate loci that have been linked to POAG [2-16]. Three genes have been identified for POAG from the reported loci, myocilin (MYOC, OMIM 601652) [17,18], optineurin (OPTN, OMIM 602432) [6,19], and WD repeat-domain 36 (WDR36, OMIM 609669) [12,20]. However, these genes altogether account for less than 10% of POAG cases [12,18,19].

Recently, in the populations of Iceland and Sweden, three single nucleotide polymorphisms (SNPs; rs2165241, rs3825942, and rs1048661) in the lysyl oxidase-like protein 1 gene (LOXL1, OMIM 153456) have been found to be strongly associated with exfoliation glaucoma (XFG, OMIM 177650), which is the most common form of secondary open-angle glaucoma associated with exfoliation syndrome (XFS; OMIM 177650) [21]. Subsequent studies have replicated the association of LOXL1 SNPs with XFG and XFS among populations in different regions including the United States [22-24], Central Europe [25,26], India [27], and Japan [28-31].

In contrast, rs2165241 in LOXL1 was only marginally (p=0.04) associated with POAG in the Icelandic population while the other two SNPs, rs3825942 and rs1048661, did not show any significant association [21]. Follow-up studies also demonstrated the lack of association between the LOXL1 SNPs and POAG in Swedish [21], Australian Caucasian [32], American Caucasian [23], African American [33], Indian [34], and Japanese populations [30,35]. These findings altogether suggested that the LOXL1 polymorphisms are risk factors for XFS/XFG but not for POAG. However, the role of these LOXL1 polymorphisms on POAG in Chinese populations is not known, and it is important to look for these SNPs in the general populations of China. The occurrence of XFS/XFG is very infrequent among the Chinese populations compared with the Japanese and Caucasian populations [36,37]. Investigation of the major LOXL1 SNPs in the general Chinese populations may provide further insight into the discrepancy in the disease prevalence of XFG/XFS between the Chinese and other ethnic groups.

In the Chinese population, we had previously mapped a POAG locus to 15q22-q24 (GLC1N) within a genetic distance of 16.6 Mb flanked by D15S1036 and rs922693 [14]. LOXL1 is located in this genetic region. Therefore, it is desirable to evaluate the possibility of LOXL1 being a POAG candidate gene in the Chinese. We have previously found different MYOC, OPTN, and WDR36 mutation patterns between Chinese and Caucasian POAG patients [38-40]. Whether the lack of association between LOXL1 and POAG in other populations also occurs in the Chinese populations requires further investigation. Furthermore, we have recently found that the distributions of variants in MYOC, OPTN, and WDR36 are different between Hong Kong Chinese (southern Chinese) and a northern Chinese population from Beijing (unpublished data). In this present study, we investigated the association between the three LOXL1 SNPs (rs2165241, rs3825942, and rs1048661) and POAG in two groups of Han Chinese, one from southern China and one from northern China, to explore their distributions.

Methods

Results

Discussion

The present study investigated the involvement of three LOXL1 polymorphisms in POAG among Han Chinese populations recruited from Hong Kong (southern China) and Beijing (northern China). Univariate analyses for each of the LOXL1 SNPs (rs1048661, rs3825942, and rs2165241) revealed that none of the three SNPs was significantly associated with POAG individually, regardless of the localization. The non-significant association between individual LOXL1 SNPs and POAG was consistent with that reported in other populations including Caucasians [21,23,32], Africans [33], Indians [34], and Japanese [35].

In the present study, all the southern Chinese subjects were localized in Hong Kong, and relative genetic homogeneity among the subjects could be assumed. As for the northern Chinese recruited in Tongren Hospital, they came from Beijing or regions around the city, a relatively large and dispersed region north of the Yellow River. To reduce the influence of historical migrations and genetic admixture between northern and southern Chinese, a northern subject was defined and included only if they and their families assured their nativities to be in this region for at least three generations. By using the Hardy–Weinberg equilibrium test, we observed that all three LOXL1 SNPs followed HWE in each group. Therefore, concerning the LOXL1 SNPs, there was no significant genetic admixture within each of the two study populations. When comparing the genotypic distributions of these variants between the two groups, no significant difference was detected, indicating that the distributions of the genotypes of individual LOXL1 SNPs did not vary across the southern and northern Chinese.

However, LD analysis and haplotype-based association analyses revealed that the combinatorial effect of the LOXL1 SNPs on POAG was different between the southern and northern Chinese populations. In the southern group, pairwise LD analysis showed that the SNP rs1048661 was in strong LD with rs3825942 (D’=0.957, 95% confidence bounds: 0.86–0.99) but in weaker LD with the intronic SNP rs2165241 (D’=0.721, 95% CB: 0.55–0.84). This provided strong evidence for historical recombination between the exonic SNP rs1048661 and intronic SNP rs2165241 as the upper confidence bound on D’ is less than 0.9 according to the criteria of Gabriel et al. [43]. In the Beijing group, all three SNPs were in strong LD, indicating less chance of recombination among the variants.

Using haplotype-based association analysis, we detected significant omnibus association between the three LOXL1 SNPs and POAG in the Hong Kong group. Subsequently, the haplotype-specific test and sole-variant test revealed that haplotype T-G-T (H5) was the sole haplotype, which can explain the omnibus association (P_HS<0.005, P_SV>0.05; Table 5). Other haplotypes were not associated with POAG. Carriers of haplotype H5 have a 5.24 fold increased susceptibility to POAG compared to non-carriers (95% CI: 1.17–23.54). However, such a high odds ratio might be due to the low frequencies of the haplotype so that the effect of H5 might have been overestimated. Due to the low frequencies of the H5 haplotype with 2.1% in cases and 0.4% in controls, Fisher’s exact test was used to confirm the significance. The results (Fisher’s exact test: p=0.027) suggested that the significant association between the T-G-T haplotype and POAG was less likely found purely by chance. Therefore, the H5 haplotype may be a genetic risk factor of POAG for a small fraction of southern Chinese. In contrast, in the Beijing group, no significant omnibus haplotype-based association was detected, and the T-G-T haplotype was not identified. The association between H5 and POAG could not be replicated in the northern Chinese. Absence of the T-G-T haplotype might be due to the strong LD between the three SNPs so that less haplotypes had been formed. Therefore, the association between the T-G-T haplotype and POAG may occur only in the southern Chinese. In recent literature, this association had not been identified in other populations. Therefore, a larger sample size of southern Chinese is required to replicate the finding and to better evaluate the effect of LOXL1 H5 haplotype on POAG genetics. A mutational screening of LOXL1 in the southern Chinese pedigree in which we had mapped the novel glaucoma locus on 15q22–24 [14] should help to evaluate its contributions to the development of glaucoma. POAG is a multigenic disease. The hitherto identified causative or modifier genes, MYOC, OPTN, and WDR36, account for less than 10% of POAG cases [12,18,19], suggesting the existence of other genetic risk factors underlying this disease. Therefore, if this association between the LOXL1 H5 haplotype is proven to be true, LOXL1 may be involved in the genetic architecture of POAG in a small fraction of southern Chinese, implying that a subgroup of POAG may share a common etiological pathway to exfoliation glaucoma. So far, the etiology of XFS/XFG is not fully known. Identification of LOXL1 may provide a breakthrough, but the functional roles of the LOXL1 variants in XFS/XFG are unclear. Therefore, to what extent POAG and XFG share a common etiology and the effect of the LOXL1 H5 haplotype on POAG still need to be investigated. It is possible that some patients who have XFS in a subclinical state manifest open-angle glaucoma as the initial sign. In these patients, the fibrillar extracellular materials may be too fine to be detected clinically, but they may accumulate in the trabecular meshwork, thus obstructing the outflow of aqueous humor and leading to increased IOP and subsequently glaucoma. Therefore, these patients might have been misdiagnosed as POAG rather than XFG.

The prevalence of XFS is very low in Chinese populations with 0.4% in Hong Kong Chinese [36] and 0.7% in Singaporean Chinese [37] aged 60 or above. In contrast, the prevalence rate of XFS varied between 0.8% and 3.2% in the Japanese [44,45], about 3.4% in the Australian population [32], and as high as 29% in the Icelandic population [21]. Genetic factors underlying such discrepancy in disease occurrence are unclear. LOXL1 is so far the most strongly associated gene with XFS/XFG having been reported in various populations [21-32]. We summarized the allele frequencies of the LOXL1 SNPs in normal subjects and POAG cases among different populations (Table 7) and observed variation in allelic frequencies of these SNPs among different ethnic groups. In the control subjects, the allele frequencies for rs3825942 G were similar among Chinese, Japanese, Icelandic, and Swedish populations (frequency: 0.847–0.879) [21,30] but were slightly lower in the American Caucasians (0.795) [23] and Indian (0.750) populations [34]. The lowest frequency was in the African American population (0.599) [33]. Concerning the relatively low frequency of rs3825942 G in the American Caucasians (0.795) reported by Fan et al. [23], it should be noted that the subjects in their study were with broad ethnic diversity, therefore the allele frequency might be biased. In another study by Challa et al, [46] the allele frequency of rs3825942 G in American Caucasians was 0.844, which is similar to other Caucasian populations. In contrast, allele frequencies of rs1048661 G and rs2165241 T were obviously lower in the Chinese and Japanese populations than other populations. Compared with the Icelandic population [21], the allele frequency of rs1048661 G was about 1.3 fold lower, and the frequency of rs2165241 T was fivefold to eightfold lower in the Japanese [30] and Chinese populations. Therefore, the low frequencies of the two at-risk alleles in the Japanese and Chinese may be a factor that leads to a lower prevalence of XFS in these two populations. However, when comparing the disease prevalence, it is 40−70 fold lower in the Chinese [36,37], and 9−36 fold lower in the Japanese [44,45] than in the Icelandic population [21]. Such distinction in disease prevalence cannot be fully explained by the discrepancies in the at-risk allele frequencies. Therefore, there may be lower penetrance of the LOXL1 variants in the Japanese and Chinese, leading to lower disease occurrence than other populations. Low penetrance of the LOXL1 variants was also found in the Australian population [32]. Moreover, the prevalence of XFS is lower in the Chinese than Japanese, although the allele frequencies were similar among the two populations. This indicates that other unidentified environmental or genetic factors may be involved in the etiology of XFS among different populations.

In summary, each of the three candidate SNPs in LOXL1 was not associated with POAG in Chinese populations. A minor haplotype (T-G-T) defined by these SNPs was associated with an increased susceptibility of POAG in the southern Chinese population, although further replication study is required. Moreover, in the general populations of the Chinese, the low frequencies of the at-risk alleles at rs1048661 and rs2165241 may be one of the factors that lead to a lower prevalence of XFS/XFG in the Chinese.

Acknowledgment

Professors Pang and Wang contributed equally to the research project and can be considered co-corresponding authors. This study was supported in part by a block grant of the Chinese University of Hong Kong and a direct grant, 2041222, from the Medical Panel of the Chinese University of Hong Kong. We thank all the participants in this study.