A Study on the Candidate Gene Association and Interaction with Measures of UV Exposure in Pseudoexfoliation Patients from India

Abstract Purpose Environmental and genetic factors are associated with development of Pseudoexfoliation syndrome (XFS). Here we intended to elucidate the association of candidate genes in relevance to UV exposure in these patients. Methods This is a case-control study of 309 subjects (N = 219 controls and 90 XFS cases) from India. PCR based direct sequencing was performed for candidate genes (LOXL1, POMP and TMEM136) followed by genotype and haplotype analysis. The promoter methylation status was assessed by Methylation specific PCR based direct sequencing of genomic DNA for all samples. The methylation status was compared with that of primary fibroblasts cultures established from patient’s Tenon’s tissue samples in subset of these patients. Results SNPs rs3825942, rs41435250, rs8818 (LOXL1) and rs3737528 (POMP) showed significant association with XFS. LOXL1 gene haplotype GAGC (rs1048661- rs3825942- rs41435250–rs8818) was associated with lower risk for XFS with a p value 4.1961 × 10−6 (OR =0; 95%CI, 0.000–0.003). POMP gene haplotypes for intronic SNPs (rs1340815- rs3737528- rs913797) TCC and TTC were associated with increased risk for the disease (OR > 1.0). Significant correlation for SNPs rs3825942 of LOXL1 (ρ= −0.132) and rs3737528 of POMP (ρ = 0.12) was observed with measure of lifetime UV exposure (CUVAF value). Reduced LOXL1 gene expression was observed in cultured tenon fibroblasts from the patients that correlated with differential methylation of the Sp-1 binding sites at -253, -243bp upstream to the transcription start site of LOXL1 promoter region. Conclusion Our results suggest a possible interaction for LOXL1 gene haplotype (GAGC) with the measure of ocular UV exposure in pseudoexfoliation syndrome.


Introduction
Genetic aetiology in XFS is evident, based on the varied incidence worldwide, with a possible interplay between genetic and/or environmental factors. 1 The role of other risk factors such as, latitude (probably related to UV exposure), 2 reactive oxygen species, 3 homocysteinaemia, 4 folate intake, 5 dysregulated retinoic acid signalling, 6 impaired autophagy 7 and elevation of several signal molecules such as transforming growth factor beta 1 (TGF-b1) in the aqueous humour (AH) 8 are reported in the pathophysiology of XFS/XFG in addition to the genetic aetiology. 91][12][13] Two common nonsynonymous protein-coding SNPs in exon 1, rs1048661 (p.R141L) and rs3825942 (p.G153D), has been consistently replicated in various population.A significant association with other risk factors were also reported for these genes from population and invitro studies. 14,15A reduced expression of POMP and TMEM136 genes was observed in iris and ciliary body tissue from patients. 16An increased POMP gene expression in lens capsular epithelium of XFS patients when compared to age-matched controls was also reported. 9,17In addition, many environmental factors like increased latitude was earlier correlated between POMP and XFS. 16Upregulated LOXL1 gene expression was also observed in cultured human Tenon's capsule fibroblasts stimulated by UV radiation and other factors like TGF-b1, hypoxia, oxidative stress. 18The effect of geographic and climatic factors on XFS development was also demonstrated in various studies. 19Our group has earlier shown effective lifetime UV exposure as a strong risk for XFS similar to other studies.Epigenetic mechanisms on LOXL1 gene expression which is shown to be altered in cases of XFS/XFG. 20,21Here we have assessed the possible association status of coding variants in LOXL1, POMP and TEMEM136 genes and their interaction with measures of environmental risk factors in a subset of the cohort reported earlier. 22The LOXL1 gene expression status has also been assessed in the human Tenon fibroblasts (HTFs) of these patients.

Patient recruitment, inclusion and exclusion criteria
The study (study code: 729-2018-P) was approved by the Institutional Ethics Committee of Vision Research Foundation and adhered to the tenets of the Declaration of Helsinki for research involving human subjects.Participants aged !18 years, with unilateral or bilateral cataracts without XFS as controls and with the presence of XFS material on the pupillary border, lens capsule, or anterior chamber angles were included in the study.Participants with any history of ocular trauma/ocular surgeries (other than cataracts) or glaucoma were excluded.The association of lifetime ocular UV exposure and its impact on the risk of development of XFS was assessed and documented earlier in these patients. 22The patient details including age, gender, socio-economic background and any other health-related issues were recorded for analysis purposes in the clinical proforma.Briefly, the protocol involves the measurement of lifetime UV exposure using a standardized questionnaire.It involves details of their residence, daily tasks, exposure to outdoor activities, use of protective eyewear, etc. Location details and the UV Index values were fetched from Tropospheric Emission Monitoring Internet Service (TEMIS), a web-based service.Lifetime UV exposure was then calculated.Conjunctival ultraviolet autofluorescence (CUVAF) photography (as a measure of preclinical UV-induced conjunctival damage) was taken to detect the conjunctival changes with exposure to UV radiation.

Genotype and haplotype analysis of LOXL1, POMP, TMEM136 genes
Genomic DNA was extracted from N ¼ 219 controls and 90 XFS cases using NucleoSpinV R Blood XL kit (Macherey-Nagel, GmbH, Germany) as per the manufacturer's instructions and assessed for its quality and quantity by standard agarose gel electrophoresis and Thermo Scientific NanoDrop TM 1000 Spectrophotometer respectively.PCRbased direct sequencing was performed in ABI Prism 3500 Genetic analyzer (Applied Biosystems, Foster City, USA) and analyzed in ABI PRISM DNA Sequencing Analysis Software v.5.1.1.The primers flanking the coding regions of LOXL1 were as given by Yilmaz, et al. 23 and that for POMP and TMEM136 genes were designed in Primer 3 input.The primers and the optimized conditions for PCR are given in Supplementary table 1. Haplotype analysis was performed and pairwise linkage disequilibrium (LD) was computed in Haploview program, version 4.1. 24The parameters for LD computation include the correlation coefficient (r2), haplotype estimation using accelerated EM algorithm similar to the partition/ligation method described in Qin, et al. 25 Interaction between haplotypes and the CUVAF (measure of UV damage), Comet assay (measure of DNA damage) were analyzed in THESIAS v3.1 (Testing Haplotype Effects In Association Studies) software. 26The significance is calculated based on the maximum likelihood model and linked to the SEM algorithm which can perform the simultaneous estimation of haplotype frequencies and their associated effects on the phenotypes.
Methylation status and its correlation with transcript levels (i) Methylation analysis: The promoter region of LOXL1, POMP and TMEM136 genes were analyzed in Methyl Primer express software (v1.0) (Supplementary figure 1) to identify the CpG islands Methylation and unmethylation specific primers (5'AGAGTGGGAAAGCGTTAGTC3'(MFP)&5'ACGACTC-CTCCGATAACAA3'(MRP) and 5'AGAGTGGGAAAGTG-TTAGTT3'(UFP) &5'ACAACTCCTCCAATAACAA3' (URP)) were designed using Methyl Primer Express software (v1.0) to amplify 170 bp length region (-316 to -147bp upstream to transcription start site of LOXL1 promoter region).Methylation analysis was performed for all the genomic DNA.Comparison of the methylation levels between blood and primary cultures of fibroblasts established from human Tenon's tissue samples was performed in a subset of these samples (N ¼ 21) and DN-A was extracted using PureLink Genomic DNA Kit.Bisulfite conversion was done using "Epitect bisulfite kit" as per the manufacturer's protocol.
(ii) Estimation of Transcript levels: The LOXL1 gene transcripts were correlated with the methylation status in HTFs from 21 XFS cases and controls.The TF culture was established, followed by the estimation of LOXL1 transcript levels by RT PCR.Human tenon fibroblast's excised during cataract surgery from 21 patients categorized as high and low-risk groups was used for assessing the LOXL1 gene methylation status and its expression.The excised tenon's tissue (1x1 mm) was placed in phosphate buffer saline (PBS) with 1X antibiotics and immediately transported on ice to the lab for further processing.The tissues samples were washed twice with 1X PBS and explant cultures were initiated in CO2 incubator at 37 C using DMEM-F12 containing 20% FBS.The media was replenished every other day for a period of 15-20 days until a confluent monolayer of cells was established.The Tfs were further sub-cultured in DMEM-F12 media supplemented with 10% FBS and antibiotics and verified for specific markers such as fibroblast surface protein-1.Cells from passages P1 to P5 were used further for methylation analysis.Total RNA was extracted from the Tf's cells using the conventional Trizol method according to manufacturer's protocol, and converted to cDNA using the iSCRIPT cDNA conversion kit (Bio-rad) and Pre-designed primers (FP 5'TGGCTGAACTCGTCCAT-GCTGTG3' and RP 5'ACTACGATGTGCGGGTGCTACT-G3').All reactions were performed in duplicates.

IV. Statistical analysis
Data were analyzed using SPSS software (version 20a for Windows; SPSS, Inc., Chicago, IL, USA) and p < 0.05 was considered statistically significant.The genotype frequencies between the cases and controls were compared using the Independent t-test for continuous variables and categorical variables using v2 test.The comet tail length and ascorbic acid levels were measured and reported earlier. 18Binary logistic regression was performed to analyze the risk of development of XFS with various factors like age, gender, UV exposure, CUVAF damage, comet tail length and ascorbic acid.Variants which showed statistical significance with p-value <0.05 were further analyzed for correlation with clinical parameters by Spearman correlation test.Functional association of the methylation status and expression levels was performed by Mann Whitney U test 27,28 and Student T-test 29 wherever indicated.

I. Patient recruitment and clinical examination
Complete demographic characteristics of the study population are shown in Table 1.The cases were significantly older 65.1 ± 7.5 than controls 57.3 ± 10.7 with p ¼ <0.001.We didn't find any significant association with any of the systemic disorders, personal habits, gender, Intra-ocular pressure, and lens status.

Distribution of POMP and TMEM136 genotypes in cases and controls
Three intronic SNPs (rs1340815, rs3737528 and rs913797) were identified in POMP gene (Supplementary Table 2).The TC and CC genotypes of rs3737528 were seen at comparatively higher frequency in cases (42% and 30%) when compared to the controls (28% and 9%) (Supplementary Table 2) and were statistically significant (p < 0.001) in additive model with OR ¼4.09; 95% CI (2.400 -7.001) (Supplementary table 2).Two coding variants (rs1893261, rs543156640) were observed in TMEM136 gene that were however not statistically significant (Supplementary Table 2).The electrophoretogram of the identified variants in LOXL1, POMP and TMEM136 genes are as depicted in Supplementary Figure 2.
SNPs rs3825942, rs41435250, rs8818 (LOXL1) and rs3737528 (POMP) were significantly associated with XFS in additive/autosomal dominant or recessive models (Supplementary Table 2) and were further analysed for correlation with measures of UV exposure (UV damage and CUVAF values), DNA damage (Comet tail length) and ascorbic acid levels by Spearman correlation method.The GA genotype of rs3825942 showed a negative correlation (q¼ À0.132; p < 0.05) with CUVAF value which is a measure of UV damage.Intronic variant, rs3737528 showed positive correlation (q ¼ 0.122 and (q ¼ 0.12; p < 0.05) with UV damage and CUVAF values, respectively (Table 2).Other SNPs did not show any significant correlation with any of the parameters.Binary logistic regression analysis was performed for SNPs rs3825942 and rs41435250 of LOXL1 that showed statistical significance with XFS and adjusted for age and gender with other covariates [UV exposure (UV damage and CUVAF values), Comet tail, and Ascorbic acid].The GA genotype of rs3825942 found statistically significant even after adjusting for other risk factors (p value >0.05; OR >1.0) (Table 3).

III (i) Methylation analysis of LOXL1 gene
Methylation status was scored for the 12 CpG sites as described by Debret et al. 30 (supplementary table 3) and we did not find any difference in the methylation pattern between cases and controls (Figure 2).

III (ii) correlation of LOXL1 levels with methylation status in HTFs
The mean expression level (Cq) of LOXL1 were 21.33 ± 1.71 in cases and 20.76 ± 1.71 in controls with Mann-Whitney U ¼ 46, Z score À0.59855 and p value 0.5485.We did not find any difference in the overall methylation pattern.However, decreased LOXL1 levels were observed for those samples that were partially methylated at CpG 9 and CpG 10 (mean Cq ¼20.91 ± 1.79) when compared to the unmethylated samples (mean Cq ¼ 21.73 ± 0.74) respectively (p ¼ 0.451538).The electrophoretogram of all samples representing the methylation status at CpG sites 9 and 10 are as depicted in Supplementary Figure 3.

Discussion
In the current study we intend to identify the potential association of these genes with lifetime UV exposure in XFS patients.The participants included here were evaluated for the effect of lifetime ocular UV exposure based on questionnaires, CUVAF and biochemical analyses.The results showed that ultraviolet radiation and geographic factors play a critical roles in the development of XFS patients from India. 22LOXL1, POMP and TMEM136 genes are reported with a significant role in UV related disease pathology.We have replicated the significant association of SNPs rs3825942, r41435250 and rs8818 with XFS similar to our earlier findings in 2 independent cohorts 31,32 from our center.2][33] except for rs1048661. 34The frequency distribution of SNP rs41435250 was similar to that reported by Dubey et al.; 34 however, the SNP was never reported in other similar studies form India.
The associations at both the markers (rs1048661, rs3825942) were widely replicated at levels surpassing genome-wide significance even in studies with modest sample sizes.Although the effect sizes exerted by both variants were uniformly large in the replication studies, the SNP exhibits allele reversal in certain populations 16,[35][36][37] (East Asians and Africans); however, a comprehensive analysis demonstrated a significant association of specific haplotypes harbouring these SNPs with XFS rather than the specific alleles. 38aplotypes GAGC, GAGG harbouring the low-risk allele A of rs3825942 were more represented in the controls and conferred low risk for the disease.Similarly, haplotypes, TGGG and GGTC with the risk-associated G allele of rs3825942 were relatively more frequent among XFS cases (Table 4).
][41][42] Specific LOXL1 haplotypes and fibulin-5 (FBLN5) gene variants were correlated with coexisting connective tissue comorbidities suggesting a possible syntropic effect. 43It will be interesting to further study the possible role of syntrophy in XFS and associated comorbidities.
Spearman Correlation analysis suggests association of rs3825942 with a measure of UV exposure (Table 2).Similarly, the GAGC haplotype flanking the A allele of SNP rs3825942 was indicative of a positive interaction with CUVAF value (Table 5).The GAGC haplotype with the indexed GA haplotypes which were correlated with higher mean enzyme activity 32 is suggestive of a protective role for the genotype against lifetime UV exposure.5][46] The haplotype correlation as observed in our study needs to be substantiated by further studies.Increased LOXL1 expression upon UV radiation in cultured human Tenon's capsule fibroblasts was studied in controlled experimental models. 18he expression levels and promoter methylation status of LOXL1 gene did not show any significant differences in human tenon fibroblasts (HTFs) from cases and controls.We observed that the LOXL1 methylation status was different between the genomic DNA and HTFs.The promoter activity of LOXL1 gene was reported to be regulated by the binding of Sp-1 transcription factor to proximal GC-rich regions in its promoter and hypermethylation of Sp-1 binding region was correlated with reduced expression in the fibroblasts from cutis laxa patients. 47,48Hence, we compared the methylation status of such Sp-1 binding sites corresponding to CpG 9 and 10 (Supplementary table 3) as described by Debret et al. 30 Samples that were partially methylated at these 2 sites showed decreased expression when compared to those unmethylated samples but without statistical significance (U ¼ 42, Z score 0.88022, p ¼ 0.451538).The current study however, lacks the methylation data for the other consensus binding sites (Sp-1A and Sp-1B) in the promoter regions that also regulate the LOXL1 gene expression.Hypermethylation of CpG islands in the LOXL1 promoter region has been observed in the anterior lens capsules of XFS patients, 49 Tfs and blood of XFG patients.Silencing of LOXL1 gene due to hypermethylation was reversed by inhibition using the DNA methyltransferase inhibitor 5-azacytidine in cultured skin fibroblasts from autosomal recessive Cutis Laxa patient, 30 in skin fibroblasts 50 and human bladder cancer cell lines, 51 in cultured human TFs when treated with 5-azacytidine in XFG patients.This suggests that targeting LOXL1 methylation may provide a therapeutic target in the management of XFS 20 .
LOXL1 promoter activity has also been shown to be regulated by variants located in the promoter region of LOXL1-AS1 transcript associated with XFS/XFG risk. 52Sequencing of the LOXL1-AS1 transcript, analysis for epigenetic modification in this region, and correlation with the expression levels in these samples are required to understand the regulatory effect of methylation on LOXL1 gene expression.Altogether, available data from our study suggests that epigenetic modification and sequence variation at LOXL1, increasing the risk of XFS, are associated with reduced LOXL1 expression which can, however, be regulated by other environmental factors.
In this study, we identified putative association of rs3737528 in POMP gene with XFS and also a positive correlation with CUVAF value (q ¼ 0.122; p value ¼ 0.044) and UV damage (q ¼ 0.12; p value¼ 0.036) in autosomal dominant and recessive model respectively (Table 2).However, the functional implication of this intronic SNP is not known.
In conclusion, our study results are suggestive of possible functional interplay between LOXL1 (rs3825942), POMP (rs3737528) and ocular UV exposure.However, further studies on characterizing the POMP gene promoter, its functional implications, methylation status of these region in patients are required for a comprehensive understanding on their potential role in disease pathology.Deep-sequencing the entire LOXL1 gene may identify additional rare variants with potential functional roles in the pathogenesis of XFS.

Figure 1 .
Figure 1.Linkage Disequilibrium Plot of LOXL1 (A) and POMP (B).(Pairwise r2 values are shown in diamonds that represent the pairwise LD between the 2 SNPs.Colour Scheme: White, shades of pink for D' < 1).

Figure 2 .
Figure 2. (A) Illustration of LOXL1 promoter sequence analysed in the current study (Red box) for methylation status with reference to the sequence containing GCrich motifs and transcription binding sites published by Debret et al. (28).CpG methylation analysis in tenons (B) and genomic DNA (C) using QUantification tool for Methylation Analysis (QUMA-web-based bisulfite sequencing analysis tool).Open circle, unmethylated CpG site; closed circle, partially methylated CpG site.

Table 2 .
Spearman correlation analysis for LOXL1 and POMP genotypes.

Table 4 .
Haplotype Analysis of SNPs in XFS Cases and Controls.

Table 5 .
Interaction between haplotypes LOXL1 and POMP gene variants with covariates as generated by THESIAS software v3.1.Bold indicates statistically significant p value.