Next Article in Journal
Confirmation of a Phenotypic Entity for TSPEAR Variants in Egyptian Ectodermal Dysplasia Patients and Role of Ethnicity
Previous Article in Journal
Old and New Systemic Immune-Inflammation Indexes Are Associated with Overall Survival of Glioblastoma Patients Treated with Radio-Chemotherapy
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Genes
Volume 13
Issue 6
10.3390/genes13061055
genes-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite question_answer Discuss in SciProfiles thumb_up
...
Endorse textsms
...
Comment
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Bioinformatic Prioritization and Functional Annotation of GWAS-Based Candidate Genes for Primary Open-Angle Glaucoma

by
Nigus G. Asefa
1,*,†,
Zoha Kamali
1,2,†,
Satyajit Pereira
3,
Ahmad Vaez
1,2,
Nomdo Jansonius
4,
Arthur A. Bergen
5 and
Harold Snieder
1
1
Department of Epidemiology, Unit of Genetic Epidemiology and Bioinformatics, University of Groningen, UMCG, Hanzeplein 1, 9713 GZ Groningen, The Netherlands
2
Department of Bioinformatics, Isfahan University of Medical Sciences, Isfahan P.O. Box 81746-7346, Iran
3
Department of Cell and Molecular Biology, Manipal School of Life Sciences, Manipal Academy of Higher Education, Manipal 576104, India
4
Department of Ophthalmology, University of Groningen, UMCG, Hanzeplein 1, 9713 GZ Groningen, The Netherlands
5
Department of Clinical Genetics, Amsterdam University Medical Center, Meibergdreef 9, 1105 AZ Amsterdam, The Netherlands
*
Author to whom correspondence should be addressed.
†
These authors contributed equally to this work.
Genes 2022, 13(6), 1055; https://doi.org/10.3390/genes13061055
Submission received: 17 April 2022 / Revised: 29 May 2022 / Accepted: 30 May 2022 / Published: 13 June 2022
(This article belongs to the Topic Big Data in Healthcare, Bioinformatics and Precision Medicine)
Browse Figures
Review Reports Versions Notes

Abstract

:
Background: Primary open-angle glaucoma (POAG) is the most prevalent glaucoma subtype, but its exact etiology is still unknown. In this study, we aimed to prioritize the most likely ‘causal’ genes and identify functional characteristics and underlying biological pathways of POAG candidate genes. Methods: We used the results of a large POAG genome-wide association analysis study from GERA and UK Biobank cohorts. First, we performed systematic gene-prioritization analyses based on: (i) nearest genes; (ii) nonsynonymous single-nucleotide polymorphisms; (iii) co-regulation analysis; (iv) transcriptome-wide association studies; and (v) epigenomic data. Next, we performed functional enrichment analyses to find overrepresented functional pathways and tissues. Results: We identified 142 prioritized genes, of which 64 were novel for POAG. BICC1, AFAP1, and ABCA1 were the most highly prioritized genes based on four or more lines of evidence. The most significant pathways were related to extracellular matrix turnover, transforming growth factor-β, blood vessel development, and retinoic acid receptor signaling. Ocular tissues such as sclera and trabecular meshwork showed enrichment in prioritized gene expression (>1.5 fold). We found pleiotropy of POAG with intraocular pressure and optic-disc parameters, as well as genetic correlation with hypertension and diabetes-related eye disease. Conclusions: Our findings contribute to a better understanding of the molecular mechanisms underlying glaucoma pathogenesis and have prioritized many novel candidate genes for functional follow-up studies.

1. Introduction

The term ‘glaucoma’ refers to a group of ocular disorders characterized by the loss of retinal ganglion cells and the degeneration of their axons [1]. Primary open-angle glaucoma (POAG) is the most common form of glaucoma. While the exact cause of POAG is still unknown, there is clear evidence that age, sex, and intraocular pressure (IOP) are important risk factors. However, genetic factors also play a significant role [1]. Indeed, early evidence from twin and family studies revealed substantial glaucoma heritability [2]. Later on, linkage studies enabled researchers to map the chromosomal locations of a number of rare pathological variants in genes (MYOC, OPTN, and WDR36) that co-segregated with the disease in families [3,4]. More recently, genome-wide association studies (GWASs) have identified a large number of common genomic variants associated with POAG in unrelated individuals. For example, a recent meta-GWAS in 12,315 POAG cases and 227,987 controls, from the Genetic Epidemiology Research on Adult Health and Aging (GERA) cohort and the UK Biobank, identified or replicated more than 70 genomic regions associated with the disease, of which 14 were novel genetic loci [5]. However, a major drawback of GWASs is that the identified variants merely tag genomic regions without providing definitive information on the likely causal genes and functional mechanisms underlying the statistical associations with a particular disease or phenotype [6].
Often, GWASs simply report the nearest genes for each locus. However, in many cases, there are several co-inherited variants in strong linkage disequilibrium (LD) with one another, making it difficult to distinguish the causal variants underlying the statistical association [7]. Moreover, several examples have clearly shown that functionally relevant genes are sometimes located at large distances from the significant GWAS loci [8,9]. This calls for further post-GWAS bioinformatics follow-up studies mapping the identified GWAS associations to the likely causal genes using more robust evidence rather than physical distance to the GWAS loci. Only a limited number of previous studies have attempted to determine the functional characteristics of GWAS-based POAG candidate genes [10,11]. However, a systematic post-GWAS approach that ranks the candidate genes in order of their relevance/causality based on different sources of biological evidence has not been performed yet. Therefore, deep investigations of disease-related genes’ expressions across different tissue types are required to better understand disease mechanisms and to avoid off-target reactions in subsequent therapeutic approaches.
The relationship between a GWAS association signal, the so-called single-nucleotide polymorphism (SNP), and the related causal gene (variant) is often not clear. An associated SNP may: (i) alter the amino acid coding (i.e., a nonsynonymous SNP (nsSNP)), changing the protein structure and potentially the function of a gene directly, or (ii) may indirectly exert its phenotypic effect through influencing regulatory sequences and, hence, the expression of a gene. Therefore, bioinformatic post-GWAS pipelines will typically check whether GWAS signals will be in high linkage disequilibrium with nonsynonymous SNPs within nearby genes and use publicly available expression quantitative trait loci (eQTL) resources from relevant tissues to check whether associated SNPs in the identified loci overlap with genomic loci related to altered gene expression [12]. Despite being helpful to some extent, finding such an overlap can also be misleading, as it does not provide conclusive evidence that the target gene expression is also associated with the phenotype. There are a number of ways to provide additional information, including transcriptome-wide association studies (TWAS).
TWAS is a popular approach to test gene expression–phenotype associations. However, this approach does not infer causality, as it is highly vulnerable to environmental confounder effects. Moreover, it requires individual-level phenotype and expression data from the same population, which is hardly ever available on a large scale, limiting the power of detecting true associations. A recently developed new TWAS approach uses previously trained transcriptome models to predict the genetic component of gene expression levels, a method known as transcriptome imputation [13]. This method allows the use of expression and phenotype data from different samples and it concurrently minimizes environmental confounding effects through the use of genetically predicted gene expressions. A recent implementation of this TWAS approach, called MetaXcan, only requires summary statistics to test gene expression associations with the outcome [14,15]. Another summary statistics-based TWAS approach is summary data-based Mendelian randomization (SMR) [16]. This method uses genetic variants as instrumental variables to test for the causative effects of gene expressions (exposure) on disease outcome. We take advantage of the two latter approaches, MetaXcan and SMR, in the current work. Furthermore, given that genetic variants associated with complex diseases, such as POAG, are predominantly located in non-coding regions, there is an existing hypothesis that epigenetic mechanisms (e.g., DNA methylation) mediate the effects of DNA on disease phenotypes via the regulation of gene expression [17,18].
Taken together, in the current post-GWAS study we combine genomic, transcriptomic, and epigenomic data from different sources to elucidate biological mechanisms underlying the pathophysiology of POAG. More specifically, we aimed to (i) prioritize the most likely causal genes and (ii) identify the underlying biological processes and tissues involved in glaucoma through functional enrichment analyses. Furthermore, pleiotropic variants and genetic correlations of POAG with other traits hypothesized to have a relationship with glaucoma (e.g., blood pressure and hypertension [19], type 2 diabetes [20], Alzheimer’s disease [21] and body mass index [22] were also examined).

2. Materials and Methods

We used GWAS summary statistics comprising 12,315 glaucoma cases and 227,987 controls, of which 7329 cases and 169,561 controls were from the UK Biobank, and 4986 cases and 58,426 controls were from the multi-ethnic GERA cohort [5]. We used a post-GWAS bioinformatics-based strategy that follows our previously published pipeline [23] with some modifications. This post-GWAS approach comprises two phases, each of which consists of multiple steps described below (see Figure 1).

2.1. Phase One: Gene Prioritization Analyses

2.1.1. In Silico Sequencing

Of the total of 84 identified or replicated gSNPs by Choquet et al. [5], 75 were selected for further analysis in our study based on the following criteria: (i) discovered in the UK Biobank and replicated (p < 0.05) in the GERA cohort or vice versa (n = 16), (ii) identified in previous glaucoma GWAS studies and replicated (p < 0.05) in the GERA cohort (n = 12), and (iii) discovered in the combined systematic meta-analysis (p < 5.0 × 10−8) of the UK Biobank and GERA cohort (n = 47; Table 1). In the case of the occurrence of a gSNP in more than one of the three criteria, the following order of priority of p-values was followed—meta-analysis > replication > discovery. The chromosomal locations and alleles of these gSNPs were verified using the Ensembl database and 1000 genomes project phase 3 data reference panel [24]. Next, we clumped the 75 gSNPs based on a physical distance of 1 Mb and r2 < 0.1 as LD metric, using PLINK (v1.9) [25].
The resulting 50 independent gSNPs (i.e., after clumping analysis; Table 1) were used as input for performing in silico sequencing and in silico look-ups of pleiotropic associations with other traits in the GWAS catalog. Using a 1Mb region on either side of the independent top gSNPs, a cut-off value for LD was set at r2 > 0.50 and the analysis was restricted to the European population. Annotation of the gSNPs together with their linked SNPs was carried out using the ANNOVAR software (version 23rd March 2019) [26]. The possible damaging effects of nonsynonymous SNPs on protein structure and function were predicted using the Sorting Intolerant From Tolerant (SIFT) [27] PROVEAN [28] and Polymorphism Phenotyping (PolyPhen) [29] scoring tools. Furthermore, the pleiotropic effect of POAG-associated loci was assessed using the GWAS catalog database (version 17 March 2019) [30].
Genes 13 01055 g001 550
Figure 1. Summary of the analysis pipeline: phase one shows the gene prioritization pipeline which was performed based on (i) 50 independent gSNPs (via nearest genes and coding consequence) and (ii) the full set of POAG GWAS summary statistics, using co-regulation, gene expression, and epigenetic regulation approaches. It also presents a summary of the quality control thresholds, reference panel used, data used for model predictions, method of data analysis, and the final number of genes prioritized for each part separately and for the entire pipeline. Similarly, phase two summarizes the functional and tissue enrichment analyses, which were performed using the 142 prioritized genes and the full set of POAG GWAS summary statistics. * Qi et al. (2018) [31]; and McRae et al. (2018) [32].
Figure 1. Summary of the analysis pipeline: phase one shows the gene prioritization pipeline which was performed based on (i) 50 independent gSNPs (via nearest genes and coding consequence) and (ii) the full set of POAG GWAS summary statistics, using co-regulation, gene expression, and epigenetic regulation approaches. It also presents a summary of the quality control thresholds, reference panel used, data used for model predictions, method of data analysis, and the final number of genes prioritized for each part separately and for the entire pipeline. Similarly, phase two summarizes the functional and tissue enrichment analyses, which were performed using the 142 prioritized genes and the full set of POAG GWAS summary statistics. * Qi et al. (2018) [31]; and McRae et al. (2018) [32].
Genes 13 01055 g001

2.1.2. Co-Regulated Genes within the POAG-Associated Loci

We ran Data-driven Expression Prioritized Integration for Complex Traits (DEPICT) [33] using the full set of POAG GWAS summary statistics (Figure 1). The significance threshold for index SNPs was set to p < 1 × 10−5 and clumping was based on r2 < 0.05 as the LD metric and a physical distance of 500 kb. DEPICT systematically prioritizes the most likely causal genes, gene sets and tissue enrichments based on gene function predictions, even for uncharacterized genes [33]. It performs functional predictions using 14,461 ‘reconstituted gene sets’ based on a curated data set of 77,840 human expression microarrays. DEPICT first identifies all genes at the trait-associated loci and then estimates their co-functionality to prioritize the most likely causal genes [33].

2.1.3. MetaXcan

In order to identify genes whose expression levels are associated with POAG, free of non-genetic confounders, we performed PrediXcan [13] analysis using the summary data-based pipeline named MetaXcan [15] (Figure 1). This pipeline integrates the summary statistics from genetically predicted transcriptome models with GWAS results to test for the association between the genetic component of gene expression levels and the outcome. We only included GWAS variants with minor allele frequency (MAF) ≥ 1% and, given their high LD ratio, excluding all variants and probes within the major histocompatibility complex (MHC) region. Our analysis was based on the transcriptome model of whole blood from the DGN cohort [34]. The significance level of association p-value was set to 4.39 × 10−6 regarding 11,397 genes (or variant sets) being tested in the DGN dataset (0.05/11,397 genes). Results were then retained if their prediction performance p-value was smaller than a Bonferroni corrected level (0.05/nsign; with nsign being the number of significant genes).
We then tested whether the associated variants for gene expression (eQTLs) and glaucoma are colocalized using the COLOC R package to address LD contamination concerns in significant genes [35]. The test is based on approximate Bayes factors on five hypotheses of (i) no causal variant (H0), (ii) causal variant for glaucoma only (H1), (iii) causal variant for gene expression only (H2), (iv) two distinct causal variants (H3) and (v) single causal variant for gene expression and glaucoma (H4). We took a probability of H3 < 0.5 and H4 > 0.5 as acceptable evidence of colocalized signals to filter out MetaXcan TWAS association results, which can be due to LD, as suggested by Barbeira et al. [15].

2.1.4. Summary Data-Based Mendelian Randomization (SMR) Based on Gene Expression Data

We performed SMR analysis combined with the Heterogeneity Independent Instruments (HEIDI) test (Figure 1), which jointly uses GWAS summary statistics and eQTL data from independent studies; providing more power to detect causal associations [16,36]. In SMR, the top cis-eQTL for each gene is used as ‘instrumental variable’ and ‘gene expression’ is considered the ‘exposure’ for the phenotype. This Mendelian randomization framework enables a test for the causal effect of the genetic variant (i.e., the eQTL) on the phenotype through gene expression [16]. Since the observation of a significant SMR association may be due to two distinct but linked causal variants, one affecting gene expression and the other influencing the phenotype, we also conducted the HEIDI test to ensure that the trait–gene expression associations are driven by the same genetic variant, not confounded by linkage [16].
We used two gene expression datasets: (i) blood eQTL summary data from the eQTLGen consortium (n~32,000) [37] and (ii) a large set of eQTL data from a meta-analysis of 10 brain regions (n effective ~ 1194) [31]. We used the 1000 Genomes Project phase 3 data panel [24] for LD calculations. The following exclusion criteria were applied after which we retained remaining eligible SNPs for the analysis: (i) rare genetic variants with MAF below 1%, (ii) variant and probes within the MHC region, and (iii) variants with inconsistent alleles or MAF differences > 0.20 amongst pairs of three datasets (GWAS, 1000 Genomes Project, and eQTL datasets). Bonferroni’s corrected significance level of p < 3.26 × 10−6 was set for the blood (0.05/15,352 genes) and p < 6.79 × 10−6 for the brain (0.05/7361 genes) SMR results. Similarly, a Bonferroni corrected significance level of p ≥ 2.78 × 10−3 (0.05/18 SMR significant probes) in blood and p ≥ 5.0 × 10−3 (0.05/10 SMR significant probes) in the brain dataset was set for HEIDI test.
SMR based on methylation data and 3xSMR (DNA→Methylation→Gene expression→POAG).
Given the likely role of epigenetics in complex disease, we also conducted epigenome-wide studies on POAG with SMR using both blood and brain mQTL data (MSMR, see below). That is, the SMR framework affords a test for the causal effect of the genetic variant (i.e., the mQTL) on the phenotype through methylation. We then mapped the significant methylation sites to their nearest genes using the FDb.InfiniumMethylation.hg19 R package [38] (Figure 1).
To further investigate the underlying mediating mechanism from DNA to POAG, we performed 3xSMR analyses on GWAS of POAG using blood and brain data, separately. The 3xSMR analysis was based on three sets of SMR results: (i) GWAS vs. mQTL (MSMR); in which methylation was the exposure and POAG was the outcome, (ii) mQTL vs. eQTL (MESMR); in which methylation was the exposure and gene expression was the outcome, and (iii) GWAS vs. eQTL (ESMR); in which gene expression was the exposure and POAG was the outcome, as described in the section above. The results of these three sets of analyses are integrated all into one causal model [31,36]. This enabled us to identify the associations between DNA, methylation, and gene expression, which consequently may led to the development of POAG (i.e., DNA→Methylation→Gene expression→POAG) (Figure 2).
We used the blood mQTL summary data from a meta-analysis (n = 1980) recently reported by Wu et al. [36] and performed MSMR analysis to find the likely causal DNA methylation sites for POAG. Then, to map those methylation sites to the subsequently regulated genes, we repeated SMR analysis on mQTL data [32] using blood eQTL data from the eQTLGen consortium (n = 32,000) [37] (MESMR). Finally, we used the blood ESMR results from the previous steps to retain genes with significant causal associations for all three steps (3xSMR).
For MSMR in the brain, we used the brain mQTL summary data from the Qi et al. meta-analysis (estimated effective n = 1160) [31] of ROSMAP [39], Hannon et al. [40] and Jaffe et al. [41]. We then performed MESMR using the same brain mQTL data against brain eQTL data reported by Qi et al. [31]. In the final step, we completed the 3xSMR analysis using the brain ESMR results from the previously performed analyses described above.
We used the same quality control applied in TWASs, i.e., GWAS variants with MAF below 1%, variants and probes within the MHC region, and variants with inconsistent alleles or MAF differences > 0.20 amongst pairs of four input datasets (GWAS, 1000 Genomes Project, eQTL and mQTL datasets) were excluded.
DNA: deoxyribonucleic acid; POAG: primary open-angle glaucoma; GWAS: Genome-wide association study; eQTL: expression quantitative trait loci; SMR: summary-data-based Mendelian randomization; mQTL: methylation quantitative trait loci; MSMR: SMR analysis of GWAS vs. mQTL; MESMR: SMR analysis of mQTL vs. eQTL; and ESMR: SMR analysis of GWAS vs. eQTL.

2.2. Phase Two: Functional Assessment

2.2.1. Functional and Tissue Enrichment Analysis

The biological pathways through which POAG-associated loci act were examined using two approaches. First, we used DEPICT to identify the enriched gene sets, alongside the corresponding functions, and tissue types for genes in POAG-associated loci. DEPICT uses the same data as for functional predictions and gene prioritization (see phase one), for gene set enrichment analysis, and a set of 37,427 human gene expression microarrays for tissue enrichment analysis of 209 tissue/cell type annotations. For gene set enrichment analysis, we applied the Affinity Propagation Clustering algorithm (APCluster R package [42]), as previously suggested and implemented by Ligthart et al. [43]. The clustering was conducted based on pairwise correlation of gene sets. Second, the prioritized gene lists based on phase one (i.e., nearest genes, genes with nsSNPs linked to POAG loci, and significant genes from DEPICT, ESMR, MSMR, and MetaXcan analyses) were merged and used as the input to run functional enrichment analysis with the GeneMANIA algorithm [44] as previously described [23] and with the Ingenuity Pathway Analysis (IPA; QIAGEN) software (Figure 1). IPA core-analysis yields the relationships, canonical pathways, diseases, and functions most relevant to the uploaded set of prioritized genes. In addition, we also performed a sensitivity functional enrichment analysis in GeneMANIA, using a subset of genes that showed two or more lines of evidence out of all five approaches that we used (see methods). Furthermore, we used the OTDB described by Wagner et al. [45] to assess whether prioritized genes are overrepresented in ocular tissues. The database contains microarray gene expression values of >20,000 genes in ten human ocular tissues (choroid, ciliary body, cornea, iris, lens, optic nerve, optic nerve head, retina, sclera, and trabecular meshwork [TM]). We performed right-tailed Fisher’s exact tests [46] to evaluate the enrichment of prioritized genes amongst the genes with gene expression values in the top 25% [47].

2.2.2. Genetic Correlation of POAG with Other Traits

We applied the (bivariate) LD score regression method [48] to estimate the genetic correlation of POAG with 597 UK Biobank traits with available GWAS summary statistics. POAG GWAS summary statistics data (~7.7 million SNPs) were uploaded to LDHub, an online tool dedicated to estimating the genetic correlation of traits of interest [49].

3. Results

3.1. Phase One: Gene Prioritization

3.1.1. In Silico Sequencing

In silico sequencing of 50 independent lead SNPs (gSNPs) from Choquet et al. [5] returned 3250 and 1493 SNPs that are, respectively, in moderate (r2 > 0.50) and high (r2 > 0.80) LD (Supplementary Table S1). One hundred and ten of the aforementioned 1493 SNPs were in complete LD (r2 = 1), indicating that these linked SNPs represent the same association signal. Annotation of these 3250 linked SNPs using the ANNOVAR software [26] detected nine nsSNPs. Two of the nine nsSNPs (rs3753841 and rs2274224) were in perfect LD (r2 = 1) with the corresponding linked gSNPs (rs993471 and rs3891783, respectively; Supplementary Table S1). These nine nsSNPs were located in seven genes, of which one (ACP2) was novel for POAG and four (ACP2, SH2B3, SIX6, and C14orf39) did not overlap with the nearest gene list of the 50 gSNPs (Table 2).

3.1.2. Co-Regulation Analyses Using DEPICT

DEPICT prioritized 86 co-regulated genes, out of 119 genes in POAG loci (r2 > 0.5), suggesting their related roles in the etiology of POAG (Supplementary Table S2A). Of the 86 prioritized genes, 41 were novel for POAG (Table 2). In addition, only 14 out of 86 (16%) overlapped with the nearest genes. Among these, five genes (FBXO32, PLCE1, ARHGEF12, LPP, and BICC1; p ≤ 8.75 × 10−7, FDR < 0.01) showed higher evidence of functional involvement in POAG pathogenesis (Supplementary Table S2A). Interestingly, DEPICT also prioritized three of the seven genes with linked nsSNPs (COL11A1, CAV2, and PLCE1), all at p ≤ 1.49 × 10−9 (Table 2).

3.1.3. MetaXcan

After the quality control steps, 11,397 variant sets were used to impute gene expression levels using the Depression Gene Network (DGN) transcriptome model. From the total of 14 significant genes, whose predicted gene expressions were associated with POAG, two genes with prediction performance p ≥ 3.57 × 10−3 (0.05/14) were filtered out and 12 significant genes remained, of which eight showed acceptable evidence of co-localization signals (Table 3). Three colocalized genes identified in MetaXcan analysis (NR1H3, LTBP3, and EHBP1L1) were also significant in SMR analysis (see below) of blood eQTL (Table 2).

3.1.4. SMR Based on Gene-Expression Data

After applying the quality control criteria, 15,352 genes were retained for the SMR analysis. Using blood gene expression profile as a source, our analysis yielded 18 genes previously implicated/associated with glaucoma. Thirteen out of these 18 have no significant evidence of linkage confounding (based on heterogeneity in dependent instruments [HEIDI] p ≥ 2.78 × 10−3; Table 4). Three (BICC1, LTBP3, and ABCA1) out of these 13 significant genes were also identified by at least two additional prioritization methods. Nine out of the 13 identified genes were novel for POAG (Table 2).
Similarly, in the brain eQTL, 7361 genes were eligible for SMR analysis. Our analysis returned 10 genes whose expression profile was significantly associated with glaucoma; eight of which have no significant evidence of linkage confounding (based on heterogeneity test, HEIDI p ≥ 5 × 10−3) (Table 4). Two of these eight genes, TXNRD2 and CDKN2B-AS1, overlapped with nearest genes and one (CDKN2B) with genes prioritized in the DEPICT analysis. Interestingly, the significant association of three HEIDI-passed genes (LRRC37A2, LRRC37A4P, and RP11-707O23.5) was also confirmed by SMR analysis of blood eQTL.

3.1.5. SMR Based on Methylation Data and 3xSMR

Methylation QTL (mQTL)-based SMR analysis (MSMR) of blood and brain data yielded, respectively, 14 (SMR p < 5.60 × 10−7) and 16 (SMR p < 5.39 × 10−7) methylation sites significantly associated with glaucoma, with no evidence of LD contamination (based on heterogeneity test; HEIDI p ≥ 1.28 × 10−3 and HEIDI p ≥ 1.39 × 10−3, respectively; Table 5). Collectively, our SMR analysis of POAG GWAS and mQTL of blood and brain identified 27 novel CpG sites in 15 genes. Of these, two genes are new (AFAP1-AS1 and TBKBP1), i.e., were not identified by the previous prioritization methods. Amongst significant MSMR genes, AFAP1-AS1 and NR1H3 are novel for POAG (Table 2). The integrative analysis of mQTL and eQTL data in the blood (MESMR) resulted in 32,420 DNA methylation (DNAm) sites significantly associated (SMRp < 1.85 × 10−8; regarding Bonferroni correction based on 2,697,257 tests) with expression levels of 10,680 genes not rejected by the HEIDI test (p ≥ 3.26 × 10−7). These results were used to link glaucoma-associated gene expressions to glaucoma-associated methylation levels, and identified the BICC1 gene with its genetic regulation in glaucoma to be explained by a likely causal chain (DNA→Methylation→Expression→Glaucoma). The integrative analysis of mQTL and eQTL data in the brain (MESMR) resulted in 10,685 DNAm sites significantly associated (SMRp < 1.44 × 10−8; regarding Bonferroni correction based on 3,482,629 tests) with expression levels of 3305 genes not rejected by the HEIDI test (p ≥ 2.16 × 10−6). We did not detect a likely causal chain from DNA to glaucoma through methylation and expression in the brain.

3.1.6. Integration of Results (Phase One)

Taken all these results together, AFAP1 and BICC1 were simultaneously highlighted in five of the six approaches and 34 of the 142 prioritized genes were supported by at least two lines of evidence (Table 2), suggesting that these genes are involved in glaucoma development by multiple mechanisms including coding, transcription, and manipulation of targeted gene expression. On the other hand, less than half of the nearest genes (n = 23) overlapped with prioritized genes identified in the pipeline (Table 2). That is, 92 likely causal genes identified by different prioritization approaches were not the genes nearest to the lead gSNPs.

3.2. Phase Two: Functional Assessment

3.2.1. Functional and Tissue Enrichment Analysis

We describe in this section the results of different functional and tissue enrichment analyses. The former include results from GeneMANIA Gene Ontology (GO) enrichment analysis, ingenuity pathways analysis (IPA) functional enrichment analysis of its manually curated biological pathways, DEPICT gene set enrichment (GSE) analysis, and affinity propagation clustering (APC) of its enriched gene sets. The latter, i.e., tissue prioritization, is based on DEPICT tissue enrichment analysis with its expression data from 209 cell types/tissues, and evaluating gene expression of the prioritized genes in 10 ocular tissues. The results of these analyses are described in order below.
GeneMANIA functional enrichment analysis revealed 246 significantly enriched gene ontology (GO) terms for 142 prioritized genes (Supplementary Table S3A). Sensitivity analysis in the 34 genes with two or more lines of evidence yielded 174 significantly enriched GO terms (Supplementary Table S3B). Along with several closely related pathways, most of the most significant GO terms were related to the extracellular matrix (ECM) (GO:0031012, q = 6.95 × 10−56), transforming growth factor-β (GO:0007179, q = 4.20 × 10−11), cardiovascular traits (e.g., blood vessel development, GO:0001568, q-value=2.40 × 10−7), heart development (GO:0007507, q = 2.52 × 10−6), Wnt signaling (GO:0016055, q = 2.52 × 10−6), retinoic acid receptor binding (GO:0042974, q = 1.18 × 10−4), and eye development (GO:001654, q = 3.09 × 10−3; Supplementary Table S3A,B). Several of these pathways, including heart development, regulation of protein phosphorylation, embryo development, and Wnt-protein signaling, were in line with the most prioritized terms from DEPICT. Similarly, IPA identified 64 canonical signaling pathways (Supplementary Table S3C). Based on the high percentage of focus molecules in our datasets, RAR activation (p = 1.32 × 10−5), leptin signaling (p = 7.94 × 10−4), and aryl hydrocarbon receptor signaling (p = 1.09 × 10−3) were the most strongly enriched canonical signaling pathways constructed in IPA.
Using the full set of POAG GWAS-summary statistics [5], DEPICT’s functional enrichment (FE) analysis identified 269 biological pathways enriched by glaucoma-associated loci (FDR < 0.05; Supplementary Table S2B). These pathways were based on five annotation categories (see Figure 3), which include Gene Ontology (GO), Kyoto Encyclopedia of Gene and Genomes (KE), REACTOME (RE), Mouse Phenotypes (MP), and Protein-Protein Interactions (PI). Cardiovascular-related terms, e.g., abnormal aorta morphology from the MP pathway resource, and Wnt and TGF-β signaling terms from the KE pathway resource, were significantly enriched at FDR < 0.05 (Figure 3), suggesting that dysregulation in any of these pathways may contribute to POAG development. For the sake of briefness, we only visualized the top five gene sets per each annotation category (Figure 3). Furthermore, affinity propagation clustering of enriched gene sets yielded 37 gene set clusters at FDR < 0.05, including artery morphology, vasculature development and cell motility regulation. The list of all cluster centers (nodes) and pairwise Pearson correlation (edges) between the nodes is summarized in Figure 4. In the DEPICT TE (tissue/cell type) enrichment analysis, 20 tissues were prioritized at FDR < 0.05 including tissues from the urogenital system and more specifically female reproductive organs, aortic and heart valves, and arteries (Supplementary Table S2C). Based on the assumed relevance for POAG, here we only show results for sense organ tissues (SO), exclusively those from the eye, as well as from the top ten nervous system tissues (Figure 3).
Our ocular tissue database (OTDB) investigations using Fisher’s exact test showed a statistically significant overrepresentation of prioritized genes in four (sclera, TM, ciliary body, and choroid) of the 10 ocular tissues (p-value < 0.005; i.e., 0.05/10) (Figure 5). More specifically, prioritized genes were overrepresented in the sclera, TM, and ciliary body by >1.5 fold.

3.2.2. In Silico Pleiotropy Look-Up and Genetic Correlation

In silico pleiotropy analysis using GWAS catalog with European subpopulation yielded 139 linked SNPs (LD, r2 > 0.80), that were previously associated with complex disease types including 41 with glaucoma itself. A complete list of all the traits identified by the in silico pleiotropic look-up along with the linked SNPs (r2 > 0.50) and their nearest genes is summarized in Supplementary Table S1. In Figure 6, we more specifically present the number of shared highly linked SNPs (r2 > 0.80) between glaucoma and six phenotypic categories that have been hypothesized in the literature to have a relationship with glaucoma, i.e., traits and diseases related to the eye (e.g., intraocular pressure) [50], anthropometry (e.g., body mass index) [22], blood pressure [19], lipids and type 2 diabetes [22], neurodegenerative disorder (e.g., Alzheimer’s disease) [21], and psychiatry (e.g., depression) [51]. Compared to other traits and diseases, POAG showed the highest genetic overlap with IOP and vertical cup-to-disc ratio, both of which are glaucoma endophenotypes with considerable heritability [2]. This can be considered as a proof of concept and internal validation of the pipeline.
We also visualized genetic correlations (rg) of POAG with traits from six phenotypic categories (Figure 7). POAG showed significant correlations with cardiometabolic disease (hypertension, rg = 0.08, p = 0.011), blood pressure (systolic blood pressure, rg = 0.06, p = 0.039; high blood pressure, rg = 0.08, p = 0.012) and eye diseases (diabetes-related eye disease, rg = 0.20, p = 0.025, other eye problems, rg = 0.44, p = 4.44 × 10−9, and senile cataract, rg = 0.27, p = 0.022; Figure 7). No significant genetic correlations were observed between glaucoma and Alzheimer’s disease or depression. Except for other eye problems, correlations were no longer significant after correcting for multiple testing of 597 UK Biobank traits.

3.2.3. Integration of Results (Phase Two)

Some signaling pathways, e.g., developmental (Wnt/β-catenin signaling, p = 2.82 × 10−3), retinoic acid receptor activation (p = 1.32 × 10−5), and cardiac-related (Cardiac Hypertrophy Signaling, p = 1.10 × 10−2), were in common with biological processes enriched in DEPICT and GeneMANIA. In line with functional enrichment results, tissue investigations also highlighted the eye, as well as cardiovascular tissues, to be the appropriate contexts of POAG genes. In completion, pleiotropy and genetic correlation analyses also revealed shared genetic loci between glaucoma and IOP, cup-to-disc ratio, as well as blood pressure.

4. Discussion

We aimed to prioritize the most likely causal genes and identify the underlying biological processes involved in POAG through post-GWAS bioinformatics analyses. Our systematic post-GWAS approach spotted 142 genes as the most likely causal and/or relevant genes for POAG, 64 of which were novel. Among the prioritized genes were seven genes with nsSNPs linked to POAG genomic loci (LD r2 > 0.5), and 34 (23.9%) of the prioritized genes were supported by at least two lines of evidence. With considerable overlap, DEPICT and GeneMANIA functional enrichment analysis revealed 269 and 246 signaling pathways associated with glaucoma, respectively. ECM was the most significant and repeatedly implicated pathway in GeneMANIA. Along with several closely related pathways, TGF-β signaling, blood vessel development, heart development, and retinoic acid receptor signaling were also significantly overrepresented. Furthermore, tissues from the female reproductive as well as the cardiovascular system were significantly enriched for POAG genes. POAG-prioritized genes were overrepresented in four ocular tissues: sclera, ciliary body, TM, and choroid.

4.1. Highlights of Separate Analyses

Six of the 64 novel POAG genes (NR1H3, ACP2, EHBP1L1, LRRC37A2, LRRC37A4P, and RP11-707O23.5) presented two or more lines of evidence (Table 2). NR1H3 is one of the top genes associated with IOP [52], a prominent risk factor for POAG, whereas ACP2 is found to be overexpressed in the cerebellum and brain stem in neuronal ceroid lipofuscinoses (CLN3) mice [53]. CLN3 disease is an inherited disorder that affects the nervous system, and children with this disorder are characterized by progressive neurological degeneration and vision loss. Using data derived from GWAS of large consortia, previous reports showed the association of EHBP1L1 with myopia and BP (diastolic BP and mean arterial pressure) [54,55]. Both myopia [56] and BP [19] are risk factors of glaucoma. Further studies are required to uncover how these novel genes are relevant to POAG.
Of the nine linked nsSNPs identified, two (rs3753841 and rs2274224) candidate variants were perfectly linked (LD r2 = 1) to POAG lead gSNPs, and rs3753841 mapping to COL11A1 gene was predicted to be ‘deleterious’ and ‘possibly damaging’ by SIFT [27] and Polyphen [29], respectively. Indeed, SNP rs3753841 has been reported previously to be associated with glaucoma [57,58]. This result provides evidence that rs3753841 alters an amino acid sequence in the collagen α chain precursor protein, a major component of the ECM, and contributes to susceptibility of glaucoma. This may be explained through the contribution of the modified collagen to IOP induction by generating outflow resistance in aqueous humor [59].
Furthermore, nsSNP rs8940 has previously been annotated in the coding region of CAV2 in Australian and Swedish glaucoma patients, but its association with POAG was not significant after adjusting for the effects of a correlated (LD r2 = 0.63) SNP rs4236601 [60]. CAV2 codes for caveolin protein family members, which is a specialized plasma membrane raft forming flask-shaped invaginations. It is involved in cell proliferation, transcellular transport, membrane lipid homeostasis, mechanotransduction, and signal transduction [61]. Caveolin 1 and caveolin 2 inhibit endothelial nitric oxide synthase enzyme activity within the caveolae, and alteration of this pathway has been associated with abnormal nitric oxide generation and TM function [62]. Caveolae are available in different retinal cell types including retinal vascular, retinal pigment epithelium, and Müller glia cells [61]. Cav-1, the principal protein of caveolae, modulates neuroprotective responses, and ablation of Cav-1 in mice and zebrafish was associated with defects in retinal pigment epithelium differentiation and STAT3 activation in the retina [63,64].
Of the 50 nearest genes identified in GWAS, only 23 (46%) genes were detected in the gene prioritization analyses, suggesting that not all nearest genes have a functional impact on protein coding, gene expression, or the regulation of transcription. This highlights the limited mapping resolution of GWAS results due to the complicated linkage disequilibrium structure of the genome, i.e., GWAS lead SNPs are not always the causal functional variants. This has been confirmed by a previous study which reported about 80% of the common GWAS variants are within 33.5 Kbp of the underlying causal variants [65]. Furthermore, our SMR analysis using methylation data also detected significant loci, strengthening the hypothesis that while SNPs located in the coding region may act by altering amino acid sequence and protein functions, genetic variants in the non-coding region may cause diseases through other mechanisms, including methylation and regulation of gene expression [66].
Whilst the role of epigenetics is well studied in mental disorders [67] and cancer [68], exploration of the role of DNA methylation in eye diseases is limited [69]. One study reported the association of a CpG site at CDKN2B in normal-tension glaucoma [70], another study found CpG sites at SKI and GTF2H4 in the retinal pigment epithelium of age-related macular degeneration patients [71]. Our combined SMR analysis of POAG GWAS data and mQTL of blood and brain data identified 27 novel DNA methylation sites for POAG in 15 genes, highlighting the role of epigenetics in gene expression and glaucoma. Moreover, our 3xSMR analysis, which was used for linking glaucoma-associated gene expression to glaucoma-associated methylation levels, enabled us to find two methylation sites (cg05938607 and cg12342675) at the BICC1 gene, confirming its likely causal role in methylation mediated genetic regulation in glaucoma (DNA→methylation→expression→glaucoma; Figure 2).

4.2. Post-GWAS Analyses Yielding Most Reliable Data

Three genetic loci (AFAP1, BICC1, and ABCA1) were identified in four or more approaches in our pipeline, confirming earlier evidence that these genes have a likely role in glaucoma pathogenesis. Actin filament associated protein 1 (AFAP1), also known as AFAP110 or AFAP-110, encodes the nonreceptor tyrosine kinase (Src) binding partner protein and affects actin filament organization in response to cellular signals [72]. Activation of Src signaling, in turn, has been implicated in the attenuation of ECM degradation via the inhibition of plasminogen activator expression, and the application of dasatinib, a potent Src signaling inhibitor, in rats improved the TGF-β2-induced adhesive and contractile characteristics of the TM and also attenuated ECM deposition [73]. BICC1, Bicaudal-C (BicC) family RNA binding protein 1, is involved in gene expression during embryonic development and is a negative regulator of the Wnt signaling pathway [74]. Previous studies reported the association of BICC1 with POAG [5,75], major depressive disorder [76], and high myopia [77]. The GTEx, CAGE, and FANTOM5 projects showed high average RNA expression levels in the eye [78]. BICC1 and AFAP1 were the two most highly prioritized genes; we further searched DEPICT gene set enrichment results in order to find the most likely relevant pathways through which these two genes act. Our co-regulation analyses predicted the highest functional similarity of BICC1 with an ECM-related cluster of gene sets. Within the cluster, the strongest correlation of BICC1 points towards abnormal tendon morphology, with corneal thinning also among the gene sets in the cluster (Figure 8A). In addition to cell motility, our analysis predicts AFAP1’s potential involvement in vasculature development as well as neuron differentiation (Figure 8B).
ATP-binding cassette transporter A1 (ABCA1) protein is a cholesterol and phospholipid transporter across the cell membrane, and mutation of this gene is associated with cancer and lipoprotein metabolism abnormality [79,80]. Although the association of AFAP1 and ABCA1 genes with cancer has been widely studied [81,82], there is a paucity of data showing the underlying mechanism in glaucoma. Cui et al. used reverse transcription polymerase chain reaction and immunolabeling approaches to examine the expression of AFAP1, ABCA1, and GMDS in human ocular tissues. Both AFAP1 and ABCA1 showed significant gene expression in the retina, retinal ganglion, optic nerve, and TM cells [83]. Our TWAS analyses showed that in contrast to the potential protective effect of increased AFAP1 expression, ABCA1 expression level is positively correlated with the risk of POAG. In line with this result, there is supporting evidence that ABCA1 is involved in the apoptosis of retinal ganglion cells in rat glaucoma models [84].

4.3. Functional Assessments

POAG is a multifactorial disease with multiple pathways involving different tissues. In normal human development, complex signaling pathways control cells’ proliferation, differentiation and fate, motility, and death; thus hypo/hyperactivation of either of these pathways may lead to the development of chronic diseases including POAG [85]. Functional enrichment analysis showed that the 142 prioritized causal genes act through several pathways, but most importantly, candidate genes were overrepresented in ECM, TGF-β signaling, blood vessel development, heart development, angiogenesis, and the retinoic acid receptor signaling pathway. Overrepresentation of the ECM and TGF-β pathways is in line with previous in vitro studies which reported TGF-β-induced morphological changes in the ocular structures, and possibly the development of POAG, using human TM cells as well as optic nerve astrocytes [86].
The ECM is a group of more than 300 complex and dynamic proteins regulating many biological processes and structures. The composition of ECM is unique in each organ, and it is continuously remodeled to regulate tissue homeostasis [87]. Previous studies have suggested that the cross-linking and deposition of ECM proteins in TM are responsible for aqueous humor outflow resistance and IOP elevation in glaucoma patients [86]. Additionally, glaucoma patients have elevated levels of TGF-β in their aqueous humor, and high TGF-β has been shown to increase the synthesis of ECM in human TM cells [86,88]. Few studies have described the mechanisms and cascade of events in how ECM leads to TM dysfunction and increased IOP. Prior experimental studies reported that the induction of ECM cross-linking LOX genes by three TGF-β isoform proteins change ECM stiffness in human glaucomatous TM cells. The authors inferred that TGF-β-mediated overexpression of LOX activity is partially responsible for the increased aqueous humor outflow resistance and IOP elevation [89].
TGF-β is a polypeptide whose signaling has a pleiotropic effect in several cell functions depending on the cellular context. For example, TGF-β receptor signaling, mediated via canonical SMAD pathways, controls the expression of hundreds of genes, while via non-canonical pathways, it regulates cell polarity and microRNA maturation [90,91]. Although TGF-β in normal tissues has pleiotropic roles in multiple biological processes and does not lead to fibrosis, it has a major impact on fibrosis and scarring following glaucoma surgery [92]. Of the three TGF-β isoforms (TGF-β1, TGF-β2, and TGF-β3), increased levels of TGF-β2 in the aqueous humor was associated with fibrosis of TM in POAG patients [93]. In addition, a recent study reported that the TGF-β-induced increase in collagen expression by TM cells is linked with phosphorylation of PTEN, and the manipulation of PTEN activity has been suggested as having therapeutic potential to prevent fibrosis of TM in POAG patients [94]. Other studies demonstrated that TGF-β2 increases expression of PAI-1 and prevents the activation of matrix metalloproteinases (enzymes responsible for the degradation of most ECM proteins) via both tissue-type (tPA) and urokinase-type urokinase (uPA) plasmin activators, thereby increasing the cross-linking of ECM components in TM cells via transglutaminase [95,96]. Another in vitro study demonstrated a cascade of events initiated by TGF-β2 could lead to ECM deposition. TGF-β2 injected into human TM cultured cells significantly increased fibronectin (ECM glycoprotein) levels. In non-glaucomatous eyes, TM cells secrete BMP-4 and antagonize the TGF-β2 pathway. In contrast, in POAG eyes, Gremlin (glycoprotein) is overexpressed in the TM cells, and this inhibits the antagonistic effect of BMP-4 on TGF-β2, thereby leading to increased ECM crosslinking and elevated IOP [97].
Overrepresentation of prioritized genes in cardiovascular-related pathways is consistent with the fact that abnormal blood vessel growth, through the activation of vascular endothelial growth factor (VEGF), has been implicated in neovascular glaucoma, age-related macular degeneration, and diabetic retinopathy [98]. Similarly, insufficient blood supply to the optic nerve head, either due to elevated IOP or low BP, is a prominent risk factor put forward to explain the development and progression of glaucoma through epidemiological studies [99]. Wnt signaling, combined with several signaling pathways, controls crucial tissues and organs during embryonic development [100]. Accordingly, a recent review showed that Wnt signaling has a vital role in angiogenesis and vascular morphogenesis in eye development, and increased activation of this signal was implicated in diabetic retinopathy, wet age-related macular degeneration, corneal neovascularization, and retinopathy of prematurity [101].
Our functional analysis implied the significance of retinoic acid (RA) and its signaling pathways in glaucoma pathogenesis. RA, a derivative of vitamin A, is essential for regulating genetic transcription that controls a wide range of biological processes including development and homeostasis, as well as cellular apoptosis and survival [102]. The identification of this pathway may strengthen the previous evidence that reported the involvement of CYP26A1 (which was included among the list of prioritized genes) and CYP26C1 in the regulation of RA metabolism, eye development, and maturation of visual function [103]. Moreover, RAs and their receptors (RARs and RXRs) have an important role in eye development and physiology [104], and in a recent study, Prat et al. demonstrated that myocilin (a pathogenic genetic biomarker for glaucoma) expression is regulated by RA via the RARE-DR2 promotor [105]. Furthermore, intraocular injection of RA receptor agonists significantly reduced the number of injured axons of the retinal ganglion cells in frogs, while the injection of antagonists and also the inhibition of RA synthesis with disulfiram had the opposite effect [106]. The role of RA signaling in the pathogenesis of POAG via myocilin transcription, or through its direct effect on retinal ganglion cells, may provide clues for new therapeutic approaches in glaucoma. Our findings highlight the urogenital system and, more specifically, female reproductive organs as tissues in which POAG genes are highly expressed. This is in line with the previous evidence of a protective role for female hormones in glaucoma [107,108,109]. Aortic and heart valves, which were among the most enriched tissues, support the results of our functional enrichment analysis regarding vasculature development and artery morphology, as well as previous studies [10]. These observations may also partly explain the shared genetics that we observed for glaucoma and blood pressure. Additional research is required to elucidate the relevance of digestive tract systems and to determine if this is only explained by the potential role of the gut microbiota in glaucoma progression, through the gut–retina axis [110]. Furthermore, the expression of POAG genes in sclera and choroid is in agreement with previous studies that showed an association between ocular rigidity (structural stiffness of the ocular tissues, including sclera, choroid, and cornea) and glaucoma [111,112]. For example, the biomechanical response of the optic nerve head to IOP stress is determined by the mechanical properties of the sclera, i.e., the sclera plays an important role in transmitting forces created by changes in IOP to the optic nerve head, thereby linking the pressure–glaucoma damage pathway [113,114].
In silico pleiotropy look-ups in the GWAS catalog confirmed the already-known genetic overlap between POAG and its related endophenotypes. For example, sixty-five linked glaucoma SNPs showed a pleiotropic effect on IOP, which is a modifiable risk factor for glaucoma (Figure 6). In addition, several glaucoma-linked SNPs were associated with optic disc parameters, including optic cup area, optic disc area, and vertical cup-to-disc ratio, a clinical metric of glaucoma [115].
Previous epidemiological studies have also suggested a close relationship between glaucoma and the two most common neurodegenerative diseases (Alzheimer’s and Parkinson’s disease), suggesting that glaucoma is the consequence of an age-related neurodegenerative process that affects the visual system [21,116]. Additionally, clinical studies have demonstrated that extracellular amyloid β (Aβ) accumulation, hyperphosphorylated tau protein (pTau) aggregation, and glial reaction are common pathological mechanisms in both glaucoma and Alzheimer’s diseases [117,118]. However, our pleiotropic look-ups and bivariate LD score regression analysis demonstrated no significant genetic correlation, highlighting that non-genetic factors (environmental factors), or undiscovered confounders, might account for the reported relationships. Our functional enrichment analysis outcomes agreed with previous comparable studies that reported the overrepresentation of ECM, TGF-β, lipid metabolism, developmental, and cardiovascular-related pathways in glaucoma, but could not confirm the importance of inflammation-related factors (e.g., nuclear factor-κB (NFκB), and tumor necrosis factor α (TNF-α)) [10,11]. The most plausible explanation for the apparent discrepancy, regarding the inflammation-related pathway, is the difference in the gene prioritization methodology used. We included genes that may have a role through different mechanisms (coding consequence, gene expression, co-regulation, and epigenetic regulation), whereas the data used as the input for the functional enrichment analysis in Janssen et al. [11] and Danford et al. [10] were significant genes curated from previous linkage, candidate-gene, and genome-wide association studies. IPA [10,11] and ConsensuspathDB [10] were used for functional annotation, whereas in addition to IPA, we used DEPICT as well as the GeneMANIA algorithm, along with its corresponding composite networks based on about 600 million interactions, which may also partly explain the difference in the identified functional pathways.

5. Strengths and Limitations

Despite the very large sample sizes of our input data sets and different statistical approaches used to find likely causal genes and pathways, there are still some limitations regarding our work. More specifically, we had to use expression data from the blood and brain to increase power for TWAS analyses, because only very limited sample sizes are available for eye tissues. Inevitably, as in any other research, our work is built on the basis of the previous studies in terms of data or methods, which may have biased our findings towards previous knowledge. However, we used agnostic methods and included data from large-scale studies, offering ample opportunities to uncover new aspects of the disease, while concurrently avoiding false interpretations by using conservative significance cut-off values by adjusting for multiple hypothesis testing. We acknowledge that while we were conducting this research, 19 (NR1H3, ACP2, EHBP1L1, ANGPTL2, CALD1, CLIC5, COL16A1, H1F0, KLF5, KREMEN1, LOC100147773, MAPT, MIR4778, MNAT1, NPEPPS, SLC2A12, THSD4, TRIB2, TRIOBP) of the 64 novel prioritized genes were detected in two recent POAG GWAS publications [119,120].

6. Conclusions

We shed light on the underlying biology of glaucoma in terms of likely involved genes and pathways using the most comprehensive recent GWAS on POAG. Our prioritized genes, particularly those with multiple lines of evidence, are eligible for further in vitro and in vivo studies. Such studies, together with improved knowledge on highlighted biological pathways including ECM organization in TM, retinoic acid signaling as well as blood vessel development, may ultimately yield new and more effective therapies.

Supplementary Materials

The following are available online at https://www.mdpi.com/article/10.3390/genes13061055/s1, Supplementary Table S1: In silico sequencing of 50 independent POAG gSNPs; Supplementary Table S2: DEPICT gene prioritization and tissue and cell enrichment output; Supplementary Table S3: GeneMANIA_GO_TERMS_and IPA canonical pathways using 142-Input-Genes.

Author Contributions

N.G.A. designed the study, performed pleiotropy and genetic correlation analyses, gene expression assessment and enrichment analysis of the ocular tissues, functional analyses using IPA, the interpretation of the results, and drafted the paper. Z.K. conducted in silico sequencing, DEPICT, TWASs in blood and brain, MSMR analyses in blood and brain, functional analysis using GeneMANIA, and was a major contributor in writing the manuscript. S.P. contributed to the initial review of published glaucoma papers and collecting GWAS loci. A.V., N.J., A.A.B., and H.S. performed critical reviews and supervision over the analyses and the draft. All authors have read and agreed to the published version of the manuscript.

Funding

This project has received funding from the European Union’s Horizon 2020 research and innovation programme under the Marie Skłodowska-Curie grant agreement No 661883. Additional funding has been provided by the Rotterdamse Stichting Blindenbelangen under grant number B20150036. The funders had no role in study design, data collection, and analysis, decision to publish, or preparation of the manuscript.

Institutional Review Board Statement

Not applicable.

Informed Consent Statement

Not applicable.

Data Availability Statement

The POAG GWAS dataset analyzed during the current study is available in the GWAS catalog repository (http://ftp.ebi.ac.uk/pub/databases/gwas/summary_statistics/ under Study accession GCST006065, accessed on 21 April 2019). Whole blood and brain eQTL datasets for TWAS analyses are available from eQTLGen website (https://www.eqtlgen.org/, accessed on 10 December 2019) and SMR data resource (https://cnsgenomics.com/software/smr/, accessed on 19 December 2019), respectively. Gene expression data for ocular tissues are available at Ocular Tissue DataBase (OTDB) (https://genome.uiowa.edu/otdb/, accessed on 23 January 2020). All results generated during this study are included in this published article and its Supplementary Information Files.

Acknowledgments

We would like to thank the Center for Information Technology of the University of Groningen for their support and for providing access to the Peregrine High Performance Computing cluster.

Conflicts of Interest

The authors declare that they have no competing interest.

References

  1. Wiggs, J.L.; Pasquale, L.R. Genetics of glaucoma. Hum. Mol. Genet. 2017, 26, R21–R27. [Google Scholar] [CrossRef]
  2. Asefa, N.G.; Neustaeter, A.; Jansonius, N.M.; Snieder, H. Heritability of glaucoma and glaucoma-related endophenotypes: Systematic review and meta-analysis. Surv. Ophthalmol. 2019, 64, 835–851. [Google Scholar] [CrossRef]
  3. Sheffield, V.C.; Stone, E.M.; Alward, W.L.; Drack, A.; Johnson, A.; Streb, L.M.; Nichols, B.E. Genetic linkage of familial open angle glaucoma to chromosome 1q21–q31. Nat. Genet. 1993, 4, 47–50. [Google Scholar] [CrossRef]
  4. Kumar, S.; Malik, M.A.; Sooraj, K.; Sihota, R.; Kaur, J. Genetic variants associated with primary open angle glaucoma in Indian population. Genomics 2016, 109, 27–35. [Google Scholar] [CrossRef]
  5. Choquet, H.; Paylakhi, S.; Kneeland, S.C.; Thai, K.K.; Hoffmann, T.J.; Yin, J.; Kvale, M.N.; Banda, Y.; Tolman, N.G.; Williams, P.A.; et al. A multiethnic genome-wide association study of primary open-angle glaucoma identifies novel risk loci. Nat. Commun. 2018, 9, 2278. [Google Scholar] [CrossRef] [Green Version]
  6. Gallagher, M.D.; Chen-Plotkin, A.S. The Post-GWAS Era: From Association to Function. Am. J. Hum. Genet. 2018, 102, 717–730. [Google Scholar] [CrossRef] [Green Version]
  7. Cano-Gamez, E.; Trynka, G. From GWAS to Function: Using Functional Genomics to Identify the Mechanisms Underlying Complex Diseases. Front. Genet. 2020, 11, 424. [Google Scholar] [CrossRef]
  8. Claussnitzer, M.; Dankel, S.N.; Kim, K.-H.; Quon, G.; Meuleman, W.; Haugen, C.; Glunk, V.; Sousa, I.S.; Beaudry, J.L.; Puviindran, V.; et al. FTO Obesity Variant Circuitry and Adi-pocyte Browning in Humans. N. Engl. J. Med. 2015, 373, 895–907. [Google Scholar] [CrossRef] [Green Version]
  9. Smemo, S.; Tena, J.J.; Kim, K.-H.; Gamazon, E.R.; Sakabe, N.J.; Gómez-Marín, C.; Aneas, I.; Credidio, F.L.; Sobreira, D.R.; Wasserman, N.F.; et al. Obesity-associated variants within FTO form long-range functional connections with IRX3. Nature 2014, 507, 371–375. [Google Scholar] [CrossRef] [Green Version]
  10. Danford, I.D.; Verkuil, L.D.; Choi, D.J.; Collins, D.W.; Gudiseva, H.V.; Uyhazi, K.E.; Lau, M.K.; Kanu, L.N.; Grant, G.R.; Chavali, V.R.M.; et al. Characterizing the “POAGome”: A bio-informatics-driven approach to primary open-angle glaucoma. Prog. Retin. Eye Res. 2017, 58, 89–114. [Google Scholar] [CrossRef] [Green Version]
  11. Janssen, S.F.; Gorgels, T.G.; Ramdas, W.D.; Klaver, C.C.; van Duijn, C.M.; Jansonius, N.M.; Bergen, A.A. The vast complexity of primary open angle glaucoma: Disease genes, risks, molecular mechanisms and pathobiology. Prog. Retin. Eye Res. 2013, 37, 31–67. [Google Scholar] [CrossRef]
  12. Ormel, J.; Hartman, C.A.; Snieder, H. The genetics of depression: Successful genome-wide association studies introduce new challenges. Transl. Psychiatry 2019, 9, 114. [Google Scholar] [CrossRef] [Green Version]
  13. Gamazon, E.R.; GTEx Consortium; Wheeler, H.E.; Shah, K.P.; Mozaffari, S.V.; Aquino-Michaels, K.; Carroll, R.J.; Eyler, A.E.; Denny, J.C.; Nicolae, D.L.; et al. A gene-based association method for mapping traits using reference transcriptome data. Nat. Genet. 2015, 47, 1091–1098. [Google Scholar] [CrossRef] [Green Version]
  14. Zhu, H.; Zhou, X. Transcriptome-wide association studies: A view from Mendelian randomization. Quant. Biol. 2020, 9, 1–15. [Google Scholar] [CrossRef]
  15. Barbeira, A.N.; GTEx Consortium; Dickinson, S.P.; Bonazzola, R.; Zheng, J.; Wheeler, H.E.; Torres, J.M.; Torstenson, E.S.; Shah, K.P.; Garcia, T.; et al. Exploring the phenotypic consequences of tissue specific gene expression variation inferred from GWAS summary statistics. Nat. Commun. 2018, 9, 1825. [Google Scholar] [CrossRef]
  16. Zhu, Z.; Zhang, F.; Hu, H.; Bakshi, A.; Robinson, M.R.; Powell, J.E.; Montgomery, G.W.; Goddard, M.E.; Wray, N.R.; Visscher, P.M.; et al. Integration of summary data from GWAS and eQTL studies predicts complex trait gene targets. Nat. Genet. 2016, 48, 481–487. [Google Scholar] [CrossRef]
  17. Huan, T.; Joehanes, R.; Song, C.; Peng, F.; Guo, Y.; Mendelson, M.; Yao, C.; Liu, C.; Ma, J.; Richard, M.; et al. Genome-wide identification of DNA methylation QTLs in whole blood highlights pathways for cardiovascular disease. Nat. Commun. 2019, 10, 4267. [Google Scholar] [CrossRef] [Green Version]
  18. Richardson, T.G.; Zheng, J.; Smith, G.D.; Timpson, N.J.; Gaunt, T.R.; Relton, C.L.; Hemani, G. Mendelian Randomization Analysis Identifies CpG Sites as Putative Mediators for Genetic Influences on Cardiovascular Disease Risk. Am. J. Hum. Genet. 2017, 101, 590–602. [Google Scholar] [CrossRef] [Green Version]
  19. Asefa, N.G.; Neustaeter, A.; Jansonius, N.M.; Snieder, H. Autonomic Dysfunction and Blood Pressure in Glaucoma Patients: The Lifelines Cohort Study. Investig. Ophthalmol. Vis. Sci. 2020, 61, 25. [Google Scholar] [CrossRef]
  20. Shiga, Y.; Akiyama, M.; Nishiguchi, K.M.; Sato, K.; Shimozawa, N.; Takahashi, A.; Momozawa, Y.; Hirata, M.; Matsuda, K.; Yamaji, T.; et al. Genome-wide association study identifies seven novel susceptibility loci for primary open-angle glaucoma. Hum. Mol. Genet. 2018, 27, 1486–1496. [Google Scholar] [CrossRef] [Green Version]
  21. Moon, J.Y.; Kim, H.J.; Park, Y.H.; Park, T.K.; Park, E.-C.; Kim, C.Y.; Lee, S.H. Association between Open-Angle Glaucoma and the Risks of Alzheimer’s and Parkinson’s Diseases in South Korea: A 10-year Nationwide Cohort Study. Sci. Rep. 2018, 8, 11161. [Google Scholar] [CrossRef]
  22. Pasquale, L.R.; Kang, J.H.; Manson, J.E.; Willett, W.C.; Rosner, B.A.; Hankinson, S.E. Prospective Study of Type 2 Diabetes Mellitus and Risk of Primary Open-Angle Glaucoma in Women. Ophthalmology 2006, 113, 1081–1086. [Google Scholar] [CrossRef]
  23. Vaez, A.; Jansen, R.; Prins, B.; Hottenga, J.-J.; de Geus, E.; Boomsma, D.; Penninx, B.W.; Nolte, I.M.; Snieder, H.; Alizadeh, B.Z. In Silico Post Genome-Wide Association Studies Analysis of C-Reactive Protein Loci Suggests an Important Role for Interferons. Circ. Cardiovasc. Genet. 2015, 8, 487–497. [Google Scholar] [CrossRef] [Green Version]
  24. The 1000 Genomes Project Consortium. A global reference for human genetic variation. Nature 2015, 526, 68–74. [Google Scholar] [CrossRef] [Green Version]
  25. Purcell, S.; Neale, B.; Todd-Brown, K.; Thomas, L.; Ferreira, M.A.R.; Bender, D.; Maller, J.; Sklar, P.; de Bakker, P.I.W.; Daly, M.J.; et al. PLINK: A tool set for whole-genome associ-ation and population-based linkage analyses. Am. J. Hum. Genet. 2007, 81, 559–575. [Google Scholar] [CrossRef] [Green Version]
  26. Wang, K.; Li, M.; Hakonarson, H. ANNOVAR: Functional annotation of genetic variants from high-throughput sequencing data. Nucleic Acids Res. 2010, 38, e164. [Google Scholar] [CrossRef]
  27. Sim, N.-L.; Kumar, P.; Hu, J.; Henikoff, S.; Schneider, G.; Ng, P.C. SIFT web server: Predicting effects of amino acid substitutions on proteins. Nucleic Acids Res. 2012, 40, W452–W457. [Google Scholar] [CrossRef]
  28. Choi, Y.; Chan, A.P. PROVEAN web server: A tool to predict the functional effect of amino acid substitutions and indels. Bioinformatics 2015, 31, 2745–2747. [Google Scholar] [CrossRef] [Green Version]
  29. Adzhubei, I.A.; Schmidt, S.; Peshkin, L.; E Ramensky, V.; Gerasimova, A.; Bork, P.; Kondrashov, A.S.; Sunyaev, S.R. A method and server for predicting damaging missense mutations. Nat. Methods 2010, 7, 248–249. [Google Scholar] [CrossRef] [Green Version]
  30. Buniello, A.; MacArthur, J.A.L.; Cerezo, M.; Harris, L.W.; Hayhurst, J.; Malangone, C.; McMahon, A.; Morales, J.; Mountjoy, E.; Sollis, E.; et al. The NHGRI-EBI GWAS Catalog of published genome-wide association studies, targeted arrays and summary statistics 2019. Nucleic Acids Res. 2019, 47, D1005–D1012. [Google Scholar] [CrossRef] [Green Version]
  31. Qi, T.; eQTLGen Consortium; Wu, Y.; Zeng, J.; Zhang, F.; Xue, A.; Jiang, L.; Zhu, Z.; Kemper, K.; Yengo, L.; et al. Identifying gene targets for brain-related traits using transcriptomic and methylomic data from blood. Nat. Commun. 2018, 9, 2282. [Google Scholar] [CrossRef] [PubMed] [Green Version]
  32. McRae, A.F.; Marioni, R.E.; Shah, S.; Yang, J.; Powell, J.E.; Harris, S.E.; Gibson, J.; Henders, A.K.; Bowdler, L.; Painter, J.N.; et al. Identification of 55,000 Replicated DNA Methylation QTL. Sci. Rep. 2018, 8, 17605. [Google Scholar] [CrossRef] [PubMed] [Green Version]
  33. Pers, T.H.; Genetic Investigation of ANthropometric Traits (GIANT) Consortium; Karjalainen, J.M.; Chan, Y.; Westra, H.-J.; Wood, A.R.; Yang, J.; Lui, J.C.; Vedantam, S.; Gustafsson, S.; et al. Biological interpretation of genome-wide association studies using predicted gene functions. Nat. Commun. 2015, 6, 5890. [Google Scholar] [CrossRef]
  34. Battle, A.; Mostafavi, S.; Zhu, X.; Potash, J.B.; Weissman, M.M.; McCormick, C.; Haudenschild, C.D.; Beckman, K.B.; Shi, J.; Mei, R.; et al. Characterizing the genetic basis of tran-scriptome diversity through RNA-sequencing of 922 individuals. Genome Res. 2014, 24, 14–24. [Google Scholar] [CrossRef] [PubMed] [Green Version]
  35. Giambartolomei, C.; Vukcevic, D.; Schadt, E.E.; Franke, L.; Hingorani, A.D.; Wallace, C.; Plagnol, V. Bayesian Test for Colocalisation between Pairs of Genetic Association Studies Using Summary Statistics. PLoS Genet. 2014, 10, e1004383. [Google Scholar] [CrossRef] [PubMed] [Green Version]
  36. Wu, Y.; Zeng, J.; Zhang, F.; Zhu, Z.; Qi, T.; Zheng, Z.; Lloyd-Jones, L.R.; Marioni, R.E.; Martin, N.G.; Mongomery, G.W.; et al. Integrative analysis of omics summary data reveals putative mecha-nisms underlying complex traits. Nat. Commun. 2018, 9, 918. [Google Scholar] [CrossRef] [PubMed]
  37. Võsa, U.; Claringbould, A.; Westra, H.-J.; Bonder, M.J.; Deelen, P.; Zeng, B.; Kirsten, H.; Saha, A.; Kreuzhuber, R.; Yazar, S.; et al. Large-Scale Cis- and Trans-EQTL Analyses Identify Thousands of Genetic Loci and Polygenic Scores That Regulate Blood Gene Expression. Nat. Genet. 2021, 53, 1300–1310. [Google Scholar] [CrossRef]
  38. Triche, T. Annotation package for Illumina Infinium DNA methylation probes. Bioconductor 2014. [Google Scholar] [CrossRef]
  39. Ng, B.; White, C.C.; Klein, H.-U.; Sieberts, S.K.; McCabe, C.; Patrick, E.; Xu, J.; Yu, L.; Gaiteri, C.; Bennett, D.A.; et al. An xQTL map integrates the genetic architecture of the human brain’s transcriptome and epigenome. Nat. Neurosci. 2017, 20, 1418–1426. [Google Scholar] [CrossRef] [Green Version]
  40. Hannon, E.; Spiers, H.; Viana, J.; Pidsley, R.; Burrage, J.; Murphy, T.M.; Troakes, C.; Turecki, G.; O’Donovan, M.C.; Schalkwyk, L.C.; et al. Methylation QTLs in the developing brain and their enrichment in schizophrenia risk loci. Nat. Neurosci. 2015, 19, 48–54. [Google Scholar] [CrossRef]
  41. Jaffe, A.E.; Gao, Y.; Deep-Soboslay, A.; Tao, R.; Hyde, T.M.; Weinberger, D.R.; Kleinman, J.E. Mapping DNA methylation across development, genotype and schizophrenia in the human frontal cortex. Nat. Neurosci. 2015, 19, 40–47. [Google Scholar] [CrossRef] [PubMed] [Green Version]
  42. Bodenhofer, U.; Kothmeier, A.; Hochreiter, S. APCluster: An R package for affinity propagation clustering. Bioinformatics 2011, 27, 2463–2464. [Google Scholar] [CrossRef] [PubMed]
  43. Ligthart, S.; Vaez, A.; Võsa, U.; Stathopoulou, M.G.; de Vries, P.S.; Prins, B.P.; Van der Most, P.J.; Tanaka, T.; Naderi, E.; Rose, L.M.; et al. Genome Analyses of >200,000 Individuals Identify 58 Loci for Chronic Inflammation and Highlight Pathways that Link Inflammation and Complex Disorders. Am. J. Hum. Genet. 2018, 103, 691–706. [Google Scholar] [CrossRef] [Green Version]
  44. Warde-Farley, D.; Donaldson, S.L.; Comes, O.; Zuberi, K.; Badrawi, R.; Chao, P.; Franz, M.; Grouios, C.; Kazi, F.; Lopes, C.T.; et al. The GeneMANIA prediction server: Bio-logical network integration for gene prioritization and predicting gene function. Nucleic. Acids Res. 2010, 38, W214–W220. [Google Scholar] [CrossRef]
  45. Wagner, A.; Anand, V.N.; Wang, W.-H.; Chatterton, J.E.; Sun, D.; Shepard, A.R.; Jacobson, N.; Pang, I.-H.; DeLuca, A.; Casavant, T.L.; et al. Exon-level expression profiling of ocular tissues. Exp. Eye Res. 2013, 111, 105–111. [Google Scholar] [CrossRef] [PubMed] [Green Version]
  46. Jin, L.; Zuo, X.-Y.; Su, W.-Y.; Zhao, X.-L.; Yuan, M.-Q.; Han, L.-Z.; Zhao, X.; Chen, Y.-D.; Rao, S.-Q. Pathway-based Analysis Tools for Complex Diseases: A Review. Genom. Proteom. Bioinform. 2014, 12, 210–220. [Google Scholar] [CrossRef] [PubMed] [Green Version]
  47. Hubens, W.H.G.; Breddels, E.M.; Walid, Y.; Ramdas, W.D.; Webers, C.A.B.; Gorgels, T.G.M.F. Mapping mRNA Expression of Glau-coma Genes in the Healthy Mouse Eye. Curr. Eye Res. 2019, 44, 1006–1017. [Google Scholar] [CrossRef] [Green Version]
  48. Bulik-Sullivan, B.; Finucane, H.K.; Anttila, V.; Gusev, A.; Day, F.R.; Loh, P.-R.; Duncan, L.; Perry, J.R.B.; Patterson, N.; Robinson, E.B.; et al. An atlas of genetic correlations across human diseases and traits. Nat. Genet. 2015, 47, 1236–1241. [Google Scholar] [CrossRef] [Green Version]
  49. Zheng, J.; Erzurumluoglu, A.M.; Elsworth, B.L.; Kemp, J.P.; Howe, L.; Haycock, P.C.; Hemani, G.; Tansey, K.; Laurin, C.; Pourcain, B.S.; et al. LD Hub: A centralized database and web interface to perform LD score regression that maximizes the potential of summary level GWAS data for SNP heritability and genetic correlation analysis. Bioinformatics 2017, 33, 272–279. [Google Scholar] [CrossRef] [Green Version]
  50. Khawaja, A.P.; Cooke Bailey, J.N.; Wareham, N.J.; Scott, R.A.; Simcoe, M.; Igo, R.P., Jr.; Song, Y.E.; Wojciechowski, R.; Cheng, C.-Y.; Khaw, P.T.; et al. Genome-wide analyses identify 68 new loci associated with intraocular pressure and improve risk prediction for primary open-angle glaucoma. Nat. Genet. 2018, 50, 778–782. [Google Scholar] [CrossRef]
  51. Zhang, X.; Olson, D.J.; Le, P.; Lin, F.-C.; Fleischman, D.; Davis, R.M. The Association Between Glaucoma, Anxiety, and Depression in a Large Population. Am. J. Ophthalmol. 2017, 183, 37–41. [Google Scholar] [CrossRef] [PubMed]
  52. Gao, X.R.; Huang, H.; Nannini, D.R.; Fan, F.; Kim, H. Genome-wide association analyses identify new loci influencing intraocular pressure. Hum. Mol. Genet. 2018, 27, 2205–2213. [Google Scholar] [CrossRef] [PubMed]
  53. Pohl, S.; Mitchison, H.M.; Kohlschütter, A.; van Diggelen, O.; Braulke, T.; Storch, S. Increased expression of lysosomal acid phosphatase in CLN3-defective cells and mouse brain tissue. J. Neurochem. 2007, 103, 2177–2188. [Google Scholar] [CrossRef]
  54. Simino, J.; Shi, G.; Bis, J.C.; Chasman, D.I.; Ehret, G.B.; Gu, X.; Guo, X.; Hwang, S.-J.; Sijbrands, E.; Smith, A.V.; et al. Gene-Age Interactions in Blood Pressure Regulation: A Large-Scale Investigation with the CHARGE, Global BPgen, and ICBP Consortia. Am. J. Hum. Genet. 2014, 95, 24–38. [Google Scholar] [CrossRef] [Green Version]
  55. Yoshikawa, M.; Yamashiro, K.; Miyake, M.; Oishi, M.; Akagi-Kurashige, Y.; Kumagai, K.; Nakata, I.; Nakanishi, H.; Oishi, A.; Gotoh, N.; et al. Comprehensive Replication of the Relationship Between Myopia-Related Genes and Refractive Errors in a Large Japanese Cohort. Investig. Ophthalmol. Vis. Sci. 2014, 55, 7343–7354. [Google Scholar] [CrossRef]
  56. Marcus, M.W.; de Vries, M.M.; Junoy Montolio, F.G.; Jansonius, N.M. Myopia as a risk factor for open-angle glaucoma: A systematic review and meta-analysis. Ophthalmology 2011, 118, 1989–1994.e2. [Google Scholar] [CrossRef]
  57. Wan, Y.; Li, S.; Gao, Y.; Tang, L.; Cao, W.; Sun, X. COL11A1 Polymorphisms Are Associated with Primary Angle-Closure Glaucoma Severity. J. Ophthalmol. 2019, 2019, 2604386. [Google Scholar] [CrossRef]
  58. Day, A.C.; Luben, R.; Khawaja, A.P.; Low, S.; Hayat, S.; Dalzell, N.; Wareham, N.J.; Khaw, K.-T.; Foster, P.J. Genotype-phenotype analysis of SNPs associated with primary angle closure glaucoma (rs1015213, rs3753841 and rs11024102) and ocular biometry in the EPIC-Norfolk Eye Study. Br. J. Ophthalmol. 2013, 97, 704–707. [Google Scholar] [CrossRef] [Green Version]
  59. Stamer, W.D. The Cell and Molecular Biology of Glaucoma: Mechanisms in the Conventional Outflow Pathway. Investig. Ophthalmol. Vis. Sci. 2012, 53, 2470–2472. [Google Scholar] [CrossRef] [Green Version]
  60. Thorleifsson, G.; Walters, G.B.; Hewitt, A.; Masson, G.; Helgason, A.; DeWan, A.; Sigurdsson, A.; Jonasdottir, A.; Gudjonsson, S.A.; Magnússon, K.P.; et al. Common variants near CAV1 and CAV2 are associated with primary open-angle glaucoma. Nat. Genet. 2010, 42, 906–909. [Google Scholar] [CrossRef] [Green Version]
  61. Gu, X.; Reagan, A.M.; McClellan, M.E.; Elliott, M.H. Caveolins and caveolae in ocular physiology and pathophysiology. Prog. Retin. Eye Res. 2016, 56, 84–106. [Google Scholar] [CrossRef] [PubMed] [Green Version]
  62. Chen, M.; Yu, X.; Xu, J.; Ma, J.; Chen, X.; Chen, B.; Gu, Y.; Wang, K. Association of Gene Polymorphisms With Primary Open Angle Glaucoma: A Systematic Review and Meta-Analysis. Investig. Ophthalmol. Vis. Sci. 2019, 60, 1105–1121. [Google Scholar] [CrossRef] [PubMed] [Green Version]
  63. Reagan, A.; Gu, X.; Hauck, S.M.; Ash, J.D.; Cao, G.; Thompson, T.C.; Elliott, M.H. Retinal Caveolin-1 Modulates Neuroprotective Signaling. Adv. Exp. Med. Biol. 2015, 854, 411–418. [Google Scholar] [CrossRef]
  64. Li, X.; Gu, X.; Boyce, T.M.; Zheng, M.; Reagan, A.M.; Qi, H.; Mandal, N.; Cohen, A.W.; Callegan, M.C.; Carr, D.J.; et al. Caveolin-1 Increases Proinflammatory Chemoattractants and Blood–Retinal Barrier Breakdown but Decreases Leukocyte Recruitment in Inflammation. Investig. Ophthalmol. Vis. Sci. 2014, 55, 6224–6234. [Google Scholar] [CrossRef]
  65. Wu, Y.; Zheng, Z.; Visscher, P.M.; Yang, J. Quantifying the mapping precision of genome-wide association studies using whole-genome sequencing data. Genome Biol. 2017, 18, 86. [Google Scholar] [CrossRef] [Green Version]
  66. Wang, H.; Lou, D.; Wang, Z. Crosstalk of Genetic Variants, Allele-Specific DNA Methylation, and Environmental Factors for Complex Disease Risk. Front. Genet. 2019, 9, 695. [Google Scholar] [CrossRef]
  67. Kumsta, R. The role of epigenetics for understanding mental health difficulties and its implications for psychotherapy research. Psychol. Psychother. 2019, 92, 190–207. [Google Scholar] [CrossRef]
  68. Fardi, M.; Solali, S.; Farshdousti, H.M. Epigenetic mechanisms as a new approach in cancer treatment: An updated review. Genes Dis. 2018, 5, 304–311. [Google Scholar] [CrossRef]
  69. Berdasco, M.; Gómez, A.; Rubio, M.J.; Mora, J.; Zanón-Moreno, V.; Lopez, M.; Hernández, C.; Yoshida, S.; Nakama, T.; Ishikawa, K.; et al. DNA Methylomes Reveal Biological Networks Involved in Human Eye Development, Functions and Associated Disorders. Sci. Rep. 2017, 7, 11762. [Google Scholar] [CrossRef] [Green Version]
  70. Burdon, K.P.; Awadalla, M.S.; Mitchell, P.; Wang, J.J.; White, A.; Keane, M.C.; Souzeau, E.; Graham, S.L.; Goldberg, I.; Healy, P.R.; et al. DNA methylation at the 9p21 glaucoma sus-ceptibility locus is associated with normal-tension glaucoma. Ophthalmic Genet. 2018, 39, 221–227. [Google Scholar] [CrossRef] [Green Version]
  71. Porter, L.F.; Saptarshi, N.; Fang, Y.; Rathi, S.; Hollander, A.I.D.; De Jong, E.K.; Clark, S.J.; Bishop, P.N.; Olsen, T.W.; Liloglou, T.; et al. Whole-genome methylation profiling of the retinal pigment epithelium of individuals with age-related macular degeneration reveals differential methylation of the SKI, GTF2H4, and TNXB genes. Clin. Epigenet. 2019, 11, 6. [Google Scholar] [CrossRef]
  72. Baisden, J.M.; Qian, Y.; Zot, H.M.; Flynn, D.C. The actin filament-associated protein AFAP-110 is an adaptor protein that modulates changes in actin filament integrity. Oncogene 2001, 20, 6435–6447. [Google Scholar] [CrossRef] [PubMed] [Green Version]
  73. Tsukamoto, T.; Kajiwara, K.; Nada, S.; Okada, M. Src mediates TGF-β-induced intraocular pressure elevation in glaucoma. J. Cell Physiol. 2019, 234, 1730–1744. [Google Scholar] [CrossRef]
  74. Kraus, M.R.-C.; Clauin, S.; Pfister, Y.; Di Maïo, M.; Ulinski, T.; Constam, D.; Bellanné-Chantelot, C.; Grapin-Botton, A. Two mutations in human BICC1 resulting in Wnt pathway hyperactivity associated with cystic renal dysplasia. Hum. Mutat. 2011, 33, 86–90. [Google Scholar] [CrossRef]
  75. MacGregor, S.; Ong, J.-S.; An, J.; Han, X.; Zhou, T.; Siggs, O.M.; Law, M.H.; Souzeau, E.; Sharma, S.; Lynn, D.; et al. Genome-wide association study of intraocular pressure uncovers new pathways to glaucoma. Nat. Genet. 2018, 50, 1067–1071. [Google Scholar] [CrossRef]
  76. Davidson, S.; Shanley, L.; Cowie, P.; Lear, M.; McGuffin, P.; Quinn, J.; Barrett, P.; MacKenzie, A. Analysis of the effects of depression associated polymorphisms on the activity of the BICC1 promoter in amygdala neurones. Pharm. J. 2015, 16, 366–374. [Google Scholar] [CrossRef] [PubMed]
  77. Jin, W.; Hepei, L.; Mingkun, X.; Li, W. Assessment of BicC family RNA binding protein 1 and Ras protein specific guanine nucleotide releasing factor 1 as candidate genes for high myopia: A case–control study. Indian J. Ophthalmol. 2017, 65, 926–930. [Google Scholar] [CrossRef] [PubMed]
  78. Uhlén, M.; Fagerberg, L.; Hallström, B.M.; Lindskog, C.; Oksvold, P.; Mardinoglu, A.; Sivertsson, A.; Kampf, C.; Sjostedt, E.; Asplund, A.; et al. Proteomics. Tissue-based map of the human proteome. Science 2015, 347, 1260419. [Google Scholar] [CrossRef]
  79. Ishigami, M.; Ogasawara, F.; Nagao, K.; Hashimoto, H.; Kimura, Y.; Kioka, N.; Ueda, K. Temporary sequestration of cholesterol and phosphatidylcholine within extracellular domains of ABCA1 during nascent HDL generation. Sci. Rep. 2018, 8, 6170. [Google Scholar] [CrossRef] [Green Version]
  80. Aguirre-Portolés, C.; Feliu, J.; Reglero, G.; Ramírez de Molina, A. ABCA1 overexpression worsens colorectal cancer prognosis by facilitating tumour growth and caveolin-1-dependent invasiveness, and these effects can be ameliorated using the BET inhibitor apabetalone. Mol. Oncol. 2018, 12, 1735–1752. [Google Scholar] [CrossRef] [Green Version]
  81. Esfandi, F.; Taheri, M.; Namvar, A.; Oskooei, V.K.; Ghafouri-Fard, S. AFAP1 and its naturally occurring antisense RNA are downregulated in gastric cancer samples. Biomed. Rep. 2019, 10, 296–302. [Google Scholar] [CrossRef] [PubMed] [Green Version]
  82. Pan, H.; Zheng, Y.; Pan, Q.; Chen, H.; Chen, F.; Wu, J.; Di, D. Expression of LXR-β, ABCA1 and ABCG1 in human triple-negative breast cancer tissues. Oncol. Rep. 2019, 42, 1869–1877. [Google Scholar] [CrossRef] [Green Version]
  83. Cui, X.-J.; Zhao, A.-G.; Wang, X.-L. Correlations of AFAP1, GMDS and PTGFR gene polymorphisms with intra-ocular pressure response to latanoprost in patients with primary open-angle glaucoma. J. Clin. Pharm. Ther. 2017, 42, 87–92. [Google Scholar] [CrossRef] [PubMed]
  84. Zhou, R.-R.; Li, H.-B.; You, Q.S.; Rong, R.; You, M.-L.; Xiong, K.; Huang, J.-F.; Xia, X.-B.; Ji, D. Silencing of GAS5 Alleviates Glaucoma in Rat Models by Reducing Retinal Ganglion Cell Apoptosis. Hum. Gene Ther. 2019, 30, 1505–1519. [Google Scholar] [CrossRef]
  85. Hunter, T. The Age of Crosstalk: Phosphorylation, Ubiquitination, and Beyond. Mol. Cell 2007, 28, 730–738. [Google Scholar] [CrossRef]
  86. Lütjen-Drecoll, E. Morphological changes in glaucomatous eyes and the role of TGFbeta2 for the pathogenesis of the disease. Exp. Eye Res. 2005, 81, 1–4. [Google Scholar] [CrossRef] [PubMed]
  87. Boyd, D.; Thomas, P.G. Towards integrating extracellular matrix and immunological pathways. Cytokine 2017, 98, 79–86. [Google Scholar] [CrossRef]
  88. Vranka, J.A.; Kelley, M.J.; Acott, T.S.; Keller, K.E. Extracellular matrix in the trabecular meshwork: Intraocular pressure regulation and dysregulation in glaucoma. Exp. Eye Res. 2015, 133, 112–125. [Google Scholar] [CrossRef] [Green Version]
  89. Sethi, A.; Mao, W.; Wordinger, R.J.; Clark, A.F. Transforming growth factor-β induces extracellular matrix protein cross-linking lysyl oxidase (LOX) genes in human trabecular meshwork cells. Investig. Ophthalmol. Vis. Sci. 2011, 52, 5240–5250. [Google Scholar] [CrossRef] [Green Version]
  90. Massagué, J. TGFβ signalling in context. Nat. Rev. Mol. Cell. Biol. 2012, 13, 616–630. [Google Scholar] [CrossRef]
  91. Akhurst, R.J.; Hata, A. Targeting the TGFβ signalling pathway in disease. Nat. Rev. Drug Discov. 2012, 11, 790–811. [Google Scholar] [CrossRef] [PubMed] [Green Version]
  92. Prendes, M.A.; Harris, A.; Wirostko, B.M.; Gerber, A.L.; Siesky, B. The role of transforming growth factor β in glaucoma and the therapeutic implications. Br. J. Ophthalmol. 2013, 97, 680–686. [Google Scholar] [CrossRef] [PubMed]
  93. Mead, A.L.; Wong, T.T.L.; Cordeiro, M.F.; Anderson, I.K.; Khaw, P.T. Evaluation of anti-TGF-beta2 antibody as a new postoperative anti-scarring agent in glaucoma surgery. Investig. Ophthalmol. Vis. Sci. 2003, 44, 3394–3401. [Google Scholar] [CrossRef] [PubMed] [Green Version]
  94. Tellios, N.; Belrose, J.C.; Tokarewicz, A.C.; Hutnik, C.; Liu, H.; Leask, A.; Motolko, M.; Iijima, M.; Parapuram, S.K. TGF-β induces phosphorylation of phosphatase and tensin homolog: Implications for fibrosis of the trabecular meshwork tissue in glaucoma. Sci. Rep. 2017, 7, 812. [Google Scholar] [CrossRef] [Green Version]
  95. Fuchshofer, R.; Welge-Lussen, U.; Lütjen-Drecoll, E. The effect of TGF-β2 on human trabecular meshwork extracellular proteolytic system. Exp. Eye Res. 2003, 77, 757–765. [Google Scholar] [CrossRef]
  96. Welge-Lüssen, U.; May, C.A.; Lütjen-Drecoll, E. Induction of tissue transglutaminase in the trabecular meshwork by TGF-beta1 and TGF-beta2. Investig. Ophthalmol. Vis. Sci. 2000, 41, 2229–2238. [Google Scholar]
  97. Wordinger, R.; Fleenor, D.L.; Hellberg, P.E.; Pang, I.-H.; Tovar, T.O.; Zode, G.S.; Fuller, J.; Clark, A.F. Effects of TGF-β2, BMP-4, and Gremlin in the Trabecular Meshwork: Implications for Glaucoma. Investig. Opthalmol. Vis. Sci. 2007, 48, 1191–1200. [Google Scholar] [CrossRef]
  98. Kim, M.; Lee, C.; Payne, R.; Yue, B.Y.; Chang, J.-H.; Ying, H. Angiogenesis in glaucoma filtration surgery and neovascular glaucoma: A review. Surv. Ophthalmol. 2015, 60, 524–535. [Google Scholar] [CrossRef] [Green Version]
  99. He, Z.; Vingrys, A.J.; Armitage, J.A.; Bui, B.V. The role of blood pressure in glaucoma. Clin. Exp. Optom. 2011, 94, 133–149. [Google Scholar] [CrossRef]
  100. Huelsken, J.; Birchmeier, W. New aspects of Wnt signaling pathways in higher vertebrates. Curr. Opin. Genet. Dev. 2001, 11, 547–553. [Google Scholar] [CrossRef]
  101. Wang, Z.; Liu, C.-H.; Huang, S.; Chen, J. Wnt Signaling in vascular eye diseases. Prog. Retin. Eye Res. 2018, 70, 110–133. [Google Scholar] [CrossRef]
  102. Piskunov, A.; Rochette-Egly, C. A retinoic acid receptor RARα pool present in membrane lipid rafts forms complexes with G protein αQ to activate p38MAPK. Oncogene 2011, 31, 3333–3345. [Google Scholar] [CrossRef] [PubMed] [Green Version]
  103. Luo, T.; Sakai, Y.; Wagner, E.; Dräger, U.C. Retinoids, eye development, and maturation of visual function. J. Neurobiol. 2006, 66, 677–686. [Google Scholar] [CrossRef] [PubMed]
  104. Samarawickrama, C.; Chew, S.; Watson, S. Retinoic acid and the ocular surface. Surv. Ophthalmol. 2015, 60, 183–195. [Google Scholar] [CrossRef] [PubMed]
  105. Prat, C.; Belville, C.; Comptour, A.; Marceau, G.; Clairefond, G.; Chiambaretta, F.; Sapin, V.; Blanchon, L. Myocilin expression is regulated by retinoic acid in the trabecular meshwork-derived cellular environment. Exp. Eye Res. 2017, 155, 91–98. [Google Scholar] [CrossRef]
  106. Duprey-Díaz, M.V.; Blagburn, J.; Blanco, R.E. Exogenous Modulation of Retinoic Acid Signaling Affects Adult RGC Survival in the Frog Visual System after Optic Nerve Injury. PLoS ONE 2016, 11, e0162626. [Google Scholar] [CrossRef] [Green Version]
  107. Vajaranant, T.S.; Nayak, S.; Wilensky, J.T.; Joslin, C.E. Gender and glaucoma: What we know and what we need to know. Curr. Opin. Ophthalmol. 2010, 21, 91–99. [Google Scholar] [CrossRef]
  108. Lee, A.J.; Mitchell, P.; Rochtchina, E.; Healey, P.R. Female reproductive factors and open angle glaucoma: The Blue Mountains Eye Study. Br. J. Ophthalmol. 2003, 87, 1324–1328. [Google Scholar] [CrossRef] [Green Version]
  109. Pasquale, L.R.; Kang, J.H. Female reproductive factors and primary open-angle glaucoma in the Nurses’ Health Study. Eye 2011, 25, 633–641. [Google Scholar] [CrossRef] [Green Version]
  110. Floyd, J.L.; Grant, M.B. The Gut–Eye Axis: Lessons Learned from Murine Models. Ophthalmol. Ther. 2020, 9, 499–513. [Google Scholar] [CrossRef]
  111. Hommer, A.; Fuchsjäger-Mayrl, G.; Resch, H.; Vass, C.; Garhofer, G.; Schmetterer, L. Estimation of Ocular Rigidity Based on Measurement of Pulse Amplitude Using Pneumotonometry and Fundus Pulse Using Laser Interferometry in Glaucoma. Investig. Opthalmol. Vis. Sci. 2008, 49, 4046–4050. [Google Scholar] [CrossRef] [PubMed] [Green Version]
  112. Schultz, D.S.; Lotz, J.C.; Lee, S.M.; Trinidad, M.L.; Stewart, J.M. Structural Factors That Mediate Scleral Stiffness. Investig. Opthalmol. Vis. Sci. 2008, 49, 4232–4236. [Google Scholar] [CrossRef] [PubMed] [Green Version]
  113. Coudrillier, B.; Tian, J.; Alexander, S.P.; Myers, K.; Quigley, H.A.; Nguyen, T.D. Biomechanics of the Human Posterior Sclera: Age- and Glaucoma-Related Changes Measured Using Inflation Testing. Investig. Opthalmol. Vis. Sci. 2012, 53, 1714–1728. [Google Scholar] [CrossRef] [PubMed]
  114. Downs, J.C.; Suh, J.-K.F.; Thomas, K.A.; Bellezza, A.J.; Hart, R.T.; Burgoyne, C. Viscoelastic Material Properties of the Peripapillary Sclera in Normal and Early-Glaucoma Monkey Eyes. Investig. Opthalmol. Vis. Sci. 2005, 46, 540–546. [Google Scholar] [CrossRef]
  115. Lavinsky, F.; Wu, M.; Schuman, J.S.; Lucy, K.A.; Liu, M.; Song, Y.; Fallon, J.; Ramos Cadena, M.D.L.A.; Ishikawa, H.; Wollstein, G. Can Macula and Optic Nerve Head Parameters Detect Glaucoma Progression in Eyes with Advanced Circumpapillary Retinal Nerve Fiber Layer Damage? Ophthalmology 2018, 125, 1907–1912. [Google Scholar] [CrossRef]
  116. Gupta, N.; Yücel, Y.H. Glaucoma as a neurodegenerative disease. Curr. Opin. Ophthalmol. 2007, 18, 110–114. [Google Scholar] [CrossRef]
  117. Criscuolo, C.; Fabiani, C.; Cerri, E.; Domenici, L. Synaptic Dysfunction in Alzheimer’s Disease and Glaucoma: From Common Degenerative Mechanisms Toward Neuroprotection. Front. Cell. Neurosci. 2017, 11, 53. [Google Scholar] [CrossRef] [Green Version]
  118. Sivak, J.M. The Aging Eye: Common Degenerative Mechanisms Between the Alzheimer’s Brain and Retinal Disease. Investig. Opthalmol. Vis. Sci. 2013, 54, 871–880. [Google Scholar] [CrossRef] [Green Version]
  119. Craig, J.E.; NEIGHBORHOOD Consortium; Han, X.; Qassim, A.; Hassall, M.; Bailey, J.N.C.; Kinzy, T.G.; Khawaja, A.P.; An, J.; Marshall, H.; et al. Multitrait analysis of glaucoma identifies new risk loci and enables polygenic prediction of disease susceptibility and progression. Nat. Genet. 2020, 52, 160–166. [Google Scholar] [CrossRef]
  120. Gharahkhani, P.; Jorgenson, E.; Hysi, P.; Khawaja, A.P.; Pendergrass, S.; Han, X.; Ong, J.S.; Hewitt, A.W.; Segrè, A.V.; Rouhana, J.M.; et al. Genome-wide meta-analysis identifies 127 open-angle glaucoma loci with consistent effect across ancestries. Nat. Commun. 2021, 12, 1258. [Google Scholar] [CrossRef]
Genes 13 01055 g002 550
Figure 2. Diagrammatic representation of 3xSMR analysis showing the cascade of causal associations from DNA methylation to POAG development.
Figure 2. Diagrammatic representation of 3xSMR analysis showing the cascade of causal associations from DNA methylation to POAG development.
Genes 13 01055 g002
Genes 13 01055 g003 550
Figure 3. Heatmap of DEPICT results: DEPICT prioritized genes (FDR < 0.05) in the context of most significant DEPICT gene sets and likely relevant tissues are visualized. Gray-scale colors represent each gene’s contribution to gene-set enrichment (GSE) or tissue enrichment (TE) described as a Z-score (only the top ten Z-scores per gene set/tissue are shown). For each of the five annotation categories (GO, KE, RE, MP, PI), only the top five gene sets are visualized. Sidebars represent p-values on the logit scale.
Figure 3. Heatmap of DEPICT results: DEPICT prioritized genes (FDR < 0.05) in the context of most significant DEPICT gene sets and likely relevant tissues are visualized. Gray-scale colors represent each gene’s contribution to gene-set enrichment (GSE) or tissue enrichment (TE) described as a Z-score (only the top ten Z-scores per gene set/tissue are shown). For each of the five annotation categories (GO, KE, RE, MP, PI), only the top five gene sets are visualized. Sidebars represent p-values on the logit scale.
Genes 13 01055 g003
Genes 13 01055 g004 550
Figure 4. DEPICT network plot: nodes represent enriched gene sets (FDR < 0.05) and links represent pairwise Pearson correlation coefficient (r) between gene sets. Only cluster centers from affinity-propagation clustering and only links with r > 0.3 are shown. The inner layer contains pathways with centrality degree ≥ 26, i.e., the 4th quartile of all gene sets, meaning the largest number of connections which imply their importance in network survival.
Figure 4. DEPICT network plot: nodes represent enriched gene sets (FDR < 0.05) and links represent pairwise Pearson correlation coefficient (r) between gene sets. Only cluster centers from affinity-propagation clustering and only links with r > 0.3 are shown. The inner layer contains pathways with centrality degree ≥ 26, i.e., the 4th quartile of all gene sets, meaning the largest number of connections which imply their importance in network survival.
Genes 13 01055 g004
Genes 13 01055 g005 550
Figure 5. Bar graph showing enrichment in ten ocular tissues of 112 available prioritized genes (out of the total of 142) amongst the genes in the ocular tissue database with gene expression values in the top 25%. Bars show the significance of overrepresentation (y-axis left), dashed horizontal line running through the bars is the threshold for p-value < 0.005 (adjusted for multiple testing), and the solid black line shows the fold of overrepresentation (y-axis right).
Figure 5. Bar graph showing enrichment in ten ocular tissues of 112 available prioritized genes (out of the total of 142) amongst the genes in the ocular tissue database with gene expression values in the top 25%. Bars show the significance of overrepresentation (y-axis left), dashed horizontal line running through the bars is the threshold for p-value < 0.005 (adjusted for multiple testing), and the solid black line shows the fold of overrepresentation (y-axis right).
Genes 13 01055 g005
Genes 13 01055 g006 550
Figure 6. Circular barplot: each bar shows the number of linked SNPs (LD r2 > 0.80) having pleiotropic effects between POAG-associated loci and six phenotypic categories. A: optic parameters and other eye problems; B: anthropometry-related traits; C: lipids and type 2 diabetes; D: blood pressure; E: psychiatric disorders; F: neurodegenerative disorders.
Figure 6. Circular barplot: each bar shows the number of linked SNPs (LD r2 > 0.80) having pleiotropic effects between POAG-associated loci and six phenotypic categories. A: optic parameters and other eye problems; B: anthropometry-related traits; C: lipids and type 2 diabetes; D: blood pressure; E: psychiatric disorders; F: neurodegenerative disorders.
Genes 13 01055 g006
Genes 13 01055 g007 550
Figure 7. Dot and whisker plot: the genetic correlations and 95% confidence intervals between POAG and six phenotypic categories were calculated using the LD score regression method.
Figure 7. Dot and whisker plot: the genetic correlations and 95% confidence intervals between POAG and six phenotypic categories were calculated using the LD score regression method.
Genes 13 01055 g007
Genes 13 01055 g008 550
Figure 8. Correlation of BICC1 (A) AFAP1 (B) with DEPICT reconstituted gene sets. Ovals represent gene set clusters. Edges represent correlation Z-score which for gene set clusters is the sum of correlation with their individual gene sets. Detailed correlation of genes with individual gene sets (circles) within clusters is shown as subnetworks.
Figure 8. Correlation of BICC1 (A) AFAP1 (B) with DEPICT reconstituted gene sets. Ovals represent gene set clusters. Edges represent correlation Z-score which for gene set clusters is the sum of correlation with their individual gene sets. Detailed correlation of genes with individual gene sets (circles) within clusters is shown as subnetworks.
Genes 13 01055 g008
Table
Table 1. Lead POAG gSNPs as reported by Choquet et al. [5].
Table 1. Lead POAG gSNPs as reported by Choquet et al. [5].
Ser. No.SNPChrPositionA1/A2GWAS p-ValueOR (95% CI)Nearest Gene
1rs41461152188227120T/G 4.1 × 10−91.09 (1.06–1.12) LMO4/PKN2-AS1
2rs9934711103385373G/A 2.0 × 10−81.08 (1.05–1.11) COL11A1
3rs28144711165739598C/T 2.0 × 10−621.37 (1.32–1.42) TMCO1
4rs75247551165694897T/C8.3 × 10−411.35 (1.27–1.44)TMCO1
5rs1192415191611540G/A2.4 × 10−51.12 (1.06–1.18)CDC7–TGFBR3
6rs46564611165717968G/A1.2 × 10−191.33 (1.25–1.41)TMCO1
7rs2627761255933014C/T 6.1 × 10−101.11 (1.07–1.15) PNPT1
8rs2860235266567896T/C 3.9 × 10−80.93 (0.90–0.95) MEIS1
9rs37891342111680155T/C 1.9 × 10−91.09 (1.06–1.12) ACOXL
10rs64340682153357541G/C 1.9 × 10−110.91 (0.89–0.94) FMNL2
11rs563355222213758234G/C 1.7 × 10−131.19 (1.13–1.24) IKZF2
12rs343390062234271522C/T 1.7 × 10−80.93 (0.90–0.95) DGKD
13rs561179022153304730A/C 7.3 × 10−50.88 (0.84–0.92)FMNL2
14rs1153606325574186A/G 3.4 × 10−81.10 (1.06–1.14) RARB
15rs9883252385138818T/C 6.1 × 10−141.11 (1.08–1.14) CADM2
16rs731624803150063454G/T 1.8 × 10−81.15 (1.09–1.20) TSC22D2
17rs98531153186131600T/A 1.9 × 10−131.10 (1.08–1.13) DGKG/TBCCD1
18rs67717363188066437C/G 4.9 × 10−80.92 (0.89–0.95) LPP
19rs34201102385137499A/G2.1 × 10−51.11 (1.08–1.15)CADM2
20rs685781447916540A/G 5.8 × 10−270.86 (0.84–0.89) AFAP1
21rs5952181147909772T/C9.8 × 10−170.86 (0.82–0.90)AFAP1
22rs933034847883887C/G 5.7 × 10−101.16 (1.12–1.20)AFAP1
23rs461989047851433G/A0.000251.08 (1.04–1.13)AFAP1
24rs76325372514837332A/C 6.7 × 10−131.12 (1.08–1.15) ANKH
25rs2553035172588027G/T 3.5 × 10−81.08 (1.05–1.11) BNIP1
26rs728359846642017C/T 1.7 × 10−150.86 (0.83–0.89) EXOC2
27rs939234861989604G/A 1.7 × 10−101.11 (1.08–1.15) GMDS
28rs17752199651406848A/G 4.7 × 10−91.14 (1.09–1.19) PKHD1
29rs37571556136458593C/T 9.5 × 10−151.12 (1.09–1.15) PDE7B
30rs47092126158976277C/A 7.8 × 10−91.09 (1.06–1.12) TMEM181
31rs30124556170448016A/G 9.3 × 10−91.14 (1.09–1.20) FAM120B
32rs94944576136474794T/A 7.4 × 10−61.16 (1.11–1.22)PDE7B
33rs20730066637465C/T1.2 × 10−60.86 (0.82–0.90)EXOC2
34rs274557261548134A/G5.2 × 10−71.12 (1.07–1.18)FOXC1
35rs1196998561922673G/A0.000171.13 (1.06–1.20)GMDS
36rs2526101711677452A/G 1.4 × 10−100.92 (0.89–0.94) THSD7A
37rs327636780848822A/G 7.0 × 10−91.08 (1.06–1.12) SEMA3C/CACNA2D1
38rs69697067116154831G/T 3.8 × 10−120.90 (0.88–0.93) CAV2/CAV1
39rs69476127117632975A/G 1.2 × 10−80.92 (0.90–0.95) CTTNBP2/LSM8
40rs624782457151505698C/T 3.3 × 10−80.82 (0.76–0.88) PRKAG2
41rs12699251711679113A/G0.0170.90 (0.87–0.93)THSD7A
42rs42366017116522675G/A1.0 × 10−40.91 (0.87–0.95)CAV2
43rs25148828108275781C/T 7.3 × 10−120.87 (0.84–0.91) ANGPT1
44rs625212878124552133C/T 8.3 × 10−90.86 (0.81–0.90) FBXO32
45rs25148848108276873C/T0.0140.84 (0.80–0.88)ANGPT1
46rs944801922051670G/C 2.1 × 10−490.81 (0.79–0.83) CDKN2B-AS1
47rs24724939107695848G/A 6.1 × 10−291.16 (1.13–1.19) ABCA1
48rs15369079129382560A/G 1.1 × 10−170.87 (0.84–0.90) LMX1B
49rs10811645922049656G/A1 × 10−180.77 (0.74–0.80)CDKN2B-AS1
50rs1333037922040765C/T2.9 × 10−290.84 (0.81–0.87)CDKN2B-AS1
51rs557703069129388033C/A 0.000190.86 (0.83–0.90)LMX1B
52rs4977756922068653G/A9.7 × 10−210.81 (0.77–0.84)CDKN2B-AS1
53rs23934551060374898C/A 7.1 × 10−91.08 (1.05–1.11) BICC1
54rs38917831096015793C/G 1.6 × 10−80.93 (0.90–0.95) PLCE1
55rs108386921147345100T/C 8.8 × 10−100.92 (0.89–0.94) MADD
56rs128083031165225319C/G 3.5 × 10−101.12 (1.08–1.16) NEAT1
57rs712641311120207405A/G 1.1 × 10−110.91 (0.89–0.94) ARHGEF12
58rs1280674011120203628G/A0.00170.90 (0.87–0.93)TMEM136
59rs5807304611120377784A/G4.3 × 10−50.87 (0.82–0.93)ARHGEF12
60rs123094671284038478C/G 1.9 × 10−91.08 (1.06–1.11) TMTC2/SLC6A15
61rs3247941283946450G/T 6.8 × 10−30.87 (0.83–0.91)TMTC2
62rs713782812111494996C/T0.00340.93 (0.89–0.98)ATXN2
63rs170758551322669058G/A 3.8 × 10−80.91 (0.88–0.94) LINC00540
64rs349355201461091401G/A 7.8 × 10−241.15 (1.12–1.18) SIX1/SIX6
65rs351550271461095174G/C 6.2 × 10−131.17 (1.12–1.23)SIX1/SIX6
66rs104837271460606157T/C1.6 × 10−121.17 (1.12–1.22)SIX1/SIX6
67rs16473811557086199C/G 8.2 × 10−90.90 (0.87–0.93) ZNF280D/TCF12
68rs22458991561952501G/A 5.1 × 10−91.08 (1.05–1.11) VPS13C
69rs25932211557501414A/G0.0250.86 (0.82–0.90)TCF12
70rs99139111710031183A/G 9.1 × 10−341.19 (1.15–1.22) GAS7
71rs98971231710117184C/T1.1 × 10−131.18 (1.13–1.23)GAS7
72rs6140010206473123A/G 7.6 × 10−101.09 (1.06–1.12) CASC20
73rs587149372219856710C/T 1.1 × 10−131.15 (1.11–1.20) TXNRD2
74rs57527742229105610C/T 1.8 × 10−110.91 (0.88–0.93) CHEK2
75rs359342242219885122C/T2.8 × 10−51.15 (1.08–1.22)TXNRD2
POAG: primary open-angle glaucoma; SNP: single-nucleotide polymorphism; gSNP: single-nucleotide polymorphism from genome-wide association study; OR: odds ratio; 95% CI: 95% confidence interval; A1: effect (risk) allele; A2: non-effect allele; SNPs in bold (n = 50): lead independent POAG gSNPs after clumping the 75 gSNPs.
Table
Table 2. Number of prioritized genes per pipeline and their novelty in POAG.
Table 2. Number of prioritized genes per pipeline and their novelty in POAG.
EntryGene SymbolTop SNPNearest Genes (n = 50)Genes with nsSNPs (n = 7)DEPICT (n = 86)SMR Blood (n = 13)SMR Brain (n = 8)MetaXcan (n = 8)MSMR Blood (n = 7)MSMR Brain (n = 12)Common in Different LinesNovelty *
1AFAP1rs68578141 1 1115Known
2BICC1rs23934551 11 115Known
3ABCA1rs24724931 1 114Known
4ARHGEF12rs71264131 1 1 3Known
5CAV2rs6969706111 3Known
6COL11A1rs993471111 3Known
7EXOC2rs728359841 113Known
8LTBP3rs12808303 11 1 3Known
9NR1H3rs326222 1 1 13Novel
10PLCE1rs3891783111 3Known
11ACOXLrs37891341 12Known
12ACP2rs2167079 1 1 2Novel
13ANGPT1rs25148821 1 2Known
14C14orf39rs34935520 1 12Known
15C22orf29rs58714937 1 1 2Known
16CDKN2Brs72652413 1 1 2Known
17CDKN2B-AS1rs9448011 1 2Known
18CTTNBP2rs69476121 1 2Known
19DGKDrs343390061 12Known
20EHBP1L1rs1346 1 1 2Novel
21FBXO32rs625212871 1 2Known
22GAS7rs99139111 12Known
23LMX1Brs15369071 12Known
24LPPrs67717361 1 2Known
25LRRC37A2rs112560196 11 2Novel
26LRRC37A4Prs112560196 11 2Novel
27NEAT1rs128083031 1 2Known
28PDE7Brs94944571 1 2Known
29RARBrs11536061 1 2Known
30RP11-707O23.5rs112073200 11 2Novel
31SIX6rs34935520 1 12Known
32THSD7Ars126992511 1 2Known
33TMTC2rs3247941 1 2Known
34TXNRD2rs359342241 1 2Known
35AC007038.1rs9646846 1 1Novel
36AC040170.1rs1687660 1 1Novel
37AFAP1-AS1rs62290601 1 1Novel
38ANGPTL2rs4837119 1 1Novel
39ANKHrs763253721 1Known
40ANTXR1rs6732795 1 1Known
41AP004608.1rs7942818 1 1Novel
42ARID2rs112405710 1 1Known
43ATXN2rs71378281 1Known
44BMP2rs2206916 1 1Known
45BNIP1rs2553031 1Known
46C17orf57rs4794057 1 1Novel
47C17orf69rs450751 1 1Novel
48C5orf41rs255303 1 1Novel
49CADM2rs342011021 1Known
50CALD1rs1026274 1 1Novel
51CASC20rs61400101 1Known
52CAV1rs2896175 1 1Known
53CCDC102Ars9921158 1 1Novel
54CDC14Ars2809823 1 1Novel
55CDKN2Ars72652413 1 1Known
56CHEK2rs57527741 1Known
57CLIC5rs3777588 1 1Novel
58COBLL1rs146065688 1 1Novel
59COL16A1rs10914457 1 1Novel
60COL8A2rs6664296 1 1Known
61CRHR1-IT1rs112560196 1 1Novel
62CYP26A1rs12260218 1 1Known
63CYR61rs821395 1 1Novel
64DCBLD2rs2439237 1 1Known
65DND1P1rs113991678 1 1Novel
66EP300rs139497 1 1Known
67FERMT1rs947465 1 1Novel
68FERMT2rs17125973 1 1Known
69FMNL2rs64340681 1Known
70FNDC3Brs62283814 1 1Known
71FOXC1rs2745572 1 1Known
72FOXCUTrs27455721 1Novel
73FOXL1rs11864330 1 1Novel
74GLIS1rs941125 1 1Known
75GLIS3rs28683166 1 1Known
76GMDSrs728359841 1Known
77GRHL2rs666026 1 1Novel
78H1F0rs5756813 1 1Novel
79KANSL1-AS1rs112073200 1 1Novel
80KIAA1267rs115690894 1 1Novel
81KIF1Crs7208035 1 1Novel
82KLF5rs4885062 1 1Novel
83KREMEN1rs16987271 1 1Novel
84LAMB4rs12670073 1 1Novel
85LINC00540rs170758551 1Known
86LINC01214rs731624801 1Novel
87LINC01364rs414611521 1Novel
88LINC02052rs98832521 1Known
89LINC02349rs22458991 1Novel
90LMO4rs7538446 1 1Known
91LOC100147773rs28144711 1Novel
92LOC102724511rs30124551 1Novel
93LOC105369146rs3276361 1Novel
94LRRC37Ars62641967 1 1Novel
95LTBP1rs34447926 1 1Novel
96LTBP2rs1077662 1 1Known
97MADDrs108386921 1Known
98MAP7D1rs6664296 1 1Novel
99MAPTrs62641967 1 1Novel
100MAST2rs499600 1 1Novel
101MECOMrs13062416; 1 1Known
102MEIS1rs2860235 1 1Known
103MIR4776-2rs563355221 1Novel
104MIR4778rs28602351 1Novel
105MNAT1rs34935520 1 1Novel
106MVB12Brs10122788 1 1Known
107MYL1rs9646846 1 1Novel
108NACAP1rs79905896 1 1Novel
109NPEPPSrs4794057 1 1Novel
110NUMBLrs10416308 1 1Novel
111PDGFRLrs4921799 1 1Known
112PIK3R3rs499600 1 1Novel
113PKHD1rs177521991 1Known
114PLEKHA7rs4141194 1 1Known
115PNPT1rs26277611 1Known
116PRKAG2rs624782451 1Known
117PRRX1rs76117299 1 1Novel
118ROR1rs2806545 1 1Novel
119RP11-259G18.1rs112073200 1 1Novel
120SALL3rs145025615 1 1Novel
121SALRNA1rs349355201 1Known
122SEMA3Crs327636 1 1Known
123SH2B3rs7137828 1 1Known
124SIX1rs34935520 1 1Known
125SIX4rs34935520 1 1Known
126SLC2A12rs2627231 1 1Novel
127SMG6rs2273984 1 1Known
128SOCS3rs73382006 1 1Novel
129STK38rs9380578 1 1Known
130TBKBP1rs4794057 11Known
131TCF12rs1647381 1 1Known
132TGFBR3rs11924151 1Known
133THSD4rs112786475 1 1Novel
134TMEM136rs12806740 1 1Known
135TMEM181rs47092121 1Known
136TRIB2rs35002856 1 1Novel
137TRIOBPrs5756813 1 1Novel
138TSC22D2rs59500396 1 1Known
139TTC28rs5752774 1 1Known
140UNC5Brs79416409 1 1Novel
141ZNF280Drs16473811 1Known
142ZNF516rs72973714 1 1Novel
POAG: primary open-angle glaucoma; nsSNP: non-synonymous single-nucleotide polymorphisms; DEPICT: data-driven expression-prioritized integration for complex traits; SMR: summary data-based Mendelian randomization; MSMR: SMR using mQTL data.; * We assessed novelty using four searching approaches: (i) Phenoscanner database (http://www.phenoscanner.medschl.cam.ac.uk/, accessed on 20 February 2020); (ii) GWAS catalog (https://www.ebi.ac.uk/gwas/, accessed on 3 February 2020); (iii) PubMed database; (iv) Glaucoma (genetics) review papers: [10,11].
Table
Table 3. Significant MetaXcan results of POAG analysis using the DGN whole blood transcriptome model.
Table 3. Significant MetaXcan results of POAG analysis using the DGN whole blood transcriptome model.
GeneChrPositionZ_ScoreEffect Sizep-ValueVar_gpred_perf r2pred_perf pval ¥pred_perf qvaln_snps in ModelCOLOC
AFAP147851047−7.7319−8.77 × 10−21.06 × 10−140.712420.720191.12 × 10−2561.10 × 10−25495Passed
CDKN2A921981525.5−6.51−3.07 × 10−17.51 × 10−110.043100.071531.45 × 10−163.79 × 10−1619Rejected
TMEM13611120200114.56.429127.36 × 10−11.28 × 10−100.006840.013753.59 × 10−45.59 × 10−412Passed
C22orf2922198380406.108181.20 × 10−11.01 × 10−90.263230.307082.46 × 10−752.16 × 10−7434Passed
FAM125B91291792245.712591.32 × 10−11.11 × 10−80.172920.233504.06 × 10−552.54 × 10−5448Rejected
EHBP1L111653518155.650375.88 × 10−11.60 × 10−80.009140.015961.20 × 10−41.92 × 10−410Passed
ACP21147265655−5.6276−8.67 × 10−21.83 × 10−80.357810.401121.57 × 10−1042.19 × 10−10330Passed
LTBP31165316338.5−5.4319−1.07 × 10−15.57 × 10−80.249760.263404.32 × 10−633.12 × 10−6242Passed
GAS7179957897−5.1001−1.43 × 10−13.40 × 10−70.113910.104009.33 × 10−243.01 × 10−23117Rejected
NR1H31147280123.5−4.9212−8.27 × 10−28.60 × 10−70.307980.284476.61 × 10−695.24 × 10−6866Passed
C17orf6917437116384.734457.05 × 10−22.20 × 10−60.443210.504601.77 × 10−1424.25 × 10−14128Passed
NPEPPS1745650475−4.6292−1.86 × 10−13.67 × 10−60.057240.075491.97 × 10−175.27 × 10−1722Rejected
POAG: primary open-angle glaucoma; Chr: chromosome; SNP: single-nucleotide polymorphisms; DGN: Depression Gene Network; COLOC: colocalization test; Var_g: variance of gene expression explained by SNPs in the model; pred_perf_r2: correlation coefficient of tissue model’s correlation to gene’s measured transcriptome (prediction performance); pred_perf_p-value: p-value of tissue model’s correlation to gene’s measured transcriptome (prediction performance); pred_perf_q-value: q-value of tissue model’s correlation to gene’s measured transcriptome (prediction performance); n_snps_in_model: number of SNPs in the model; COLOC passed results are those with probability < 0.5 at hypothesis three and probability > 0.5 at hypothesis four; ¥ the Bonferroni corrected significance threshold for MetaXcan p-value and prediction performance p-value were 4.34 × 10−6 and 3.57 × 10−3, respectively.
Table
Table 4. Significant SMR results of POAG GWAS using eQTLGen dataset (https://www.eqtlgen.org/, accessed on 10 December 2019) of whole blood and [31].
Table 4. Significant SMR results of POAG GWAS using eQTLGen dataset (https://www.eqtlgen.org/, accessed on 10 December 2019) of whole blood and [31].
TissueGene NameTop eQTL SNPTop SNP Chr:PositionA1/A2β GWASSE GWASP GWASβ eQTLSE eQTLP eQTLβ SMRSE SMRP SMRP HEIDIHEIDI
BloodRP11-466F5.8rs27900491:165743523A/G0.31370.0192.68 × 10−610.20140.01497.75 × 10−421.55740.14861.10 × 10−253.93 × 10−5Rejected
RP11-217B7.2rs29800839:107691362A/C0.10390.01351.51 × 10−140.20110.00893.24 × 10−1140.51660.0713.32 × 10−135.72 × 10−4Rejected
AFAP1rs622906014:7939008A/T−0.10240.01462.35 × 10−121.0770.00910.00E+00 #−0.09510.01362.57 × 10−126.34 × 10−12Rejected
MVB12Brs101227889:129206832A/G−0.07610.01351.52 × 10−8−0.51810.00850.00E+00 #0.14690.02611.75 × 10−88.54 × 10−1Passed
NR1H3rs32622211:47259668C/T−0.07810.01434.32 × 10−80.38780.00860.00E+00 #−0.20130.0375.40 × 10−81.39 × 10−1Passed
LTBP3rs1278902811:65326154A/G−0.09810.01815.97 × 10−80.47980.01180.00E+00 #−0.20450.03817.79 × 10−81.46 × 10−1Passed
BICC1rs1074073410:60364363A/G−0.07250.01346.78 × 10−80.17280.00893.58 × 10−84−0.41960.08071.99 × 10−72.18 × 10−1Passed
TMCO1rs46574731:165687151T/C−0.09410.01525.85 × 10−100.08720.00985.15 × 10−19−1.080.21233.66 × 10−79.08 × 10−6Rejected
EHBP1L1rs134611:65337251T/A−0.09790.0185.06 × 10−8−0.14730.01214.10 × 10−340.66480.13376.56 × 10−74.67 × 10−1Passed
TXNRD2rs11798572522:19860852C/T−0.1330.02675.98 × 10−7−0.44490.02091.76 × 10−1000.2990.06151.17 × 10−63.88 × 10−6Rejected
ABCA1rs24870529:107686405T/C−0.08580.01693.61 × 10−7−0.15880.01081.18 × 10−480.54020.11241.53 × 10−62.25 × 10−2Passed
KANSL1-AS1rs11207320017:44201791C/G0.07720.01632.22 × 10−60.90740.01280.00E+00 #0.08510.0182.34 × 10−61.09 × 10−2Passed
RP11-707O23.5rs11207320017:44201791C/G0.07720.01632.22 × 10−60.87220.01290.00E+00 #0.08850.01882.35 × 10−62.34 × 10−1Passed
RP11-259G18.1rs11207320017:44201791C/G0.07720.01632.22 × 10−60.4960.01371.90 × 10−2880.15560.03322.70 × 10−65.19 × 10−2Passed
DND1P1rs11399167817:43795634T/C0.07660.01632.65 × 10−60.75250.0130.00E+00 #0.10170.02172.85 × 10−61.72 × 10−1Passed
CRHR1-IT1rs11256019617:44200078T/A0.07640.01632.83 × 10−61.15280.01190.00E+00 #0.06630.01422.90 × 10−66.64 × 10−1Passed
LRRC37A4Prs11256019617:44200078T/A0.07640.01632.83 × 10−6−0.90980.01270.00E+00 #−0.0840.0182.97 × 10−62.45 × 10−1Passed
LRRC37A2rs11256019617:44200078T/A0.07640.01632.83 × 10−60.71040.01320.00E+00 #0.10760.02313.08 × 10−67.72 × 10−3Passed
BrainTXNRD2rs7314896522:19872935G/A0.14630.02113.99 × 10−121.30770.03624.05 × 10−2850.11190.01649.57 × 10−122.01 × 10−2Passed
RP11-466F5.8rs109182741:165714416C/T−0.3050.01885.96 × 10−59−0.34080.05886.66 × 10−90.89490.16394.77 × 10−84.72 × 10−3Rejected
CDKN2B-AS1rs5043189:22024023T/A−0.16320.01351.50 × 10−33−0.29040.05222.62 × 10−80.5620.11124.33 × 10−72.03 × 10−1Passed
CDKN2Brs4900059:22020493A/G−0.16310.01351.24 × 10−33−0.21260.03843.16 × 10−80.7670.15254.90 × 10−71.41 × 10−1Passed
RP11-217B7.2rs18009779:107690450G/A0.08460.01443.95 × 10−90.59580.0733.31 × 10−160.1420.02971.80 × 10−63.45 × 10−3Rejected
RP11-707O23.5rs1757550717:44134095G/A0.07710.01632.26 × 10−61.4650.03810.00E+00 #0.05260.01122.68 × 10−65.56 × 10−1Passed
LRRC37A2rs6264196717:44047216G/T0.07660.01632.65 × 10−61.30740.03310.00E+00 #0.05860.01263.11 × 10−65.66 × 10−2Passed
LRRC37Ars6264196717:44047216G/T0.07660.01632.65 × 10−61.31990.03390.00E+00 #0.0580.01243.13 × 10−65.31 × 10−2Passed
LRRC37A4Prs11274600817:44126650T/C0.0770.01632.46 × 10−6−1.37560.04444.34 × 10−211−0.0560.0123.19 × 10−63.28 × 10−1Passed
MAPTrs6264196717:44047216G/T0.07660.01632.65 × 10−6−1.17920.03412.77 × 10−262−0.06490.0143.26 × 10−65.28 × 10−3Passed
POAG: primary open-angle glaucoma; SNP: single-nucleotide polymorphisms; SMR: summary-data-based Mendelian randomization; eQTL: expression quantitative trait loci; GWAS: genome-wide association study; HEIDI: heterogeneity independent instruments; b: effect size; se: standard error; p: p-value; A1: effect (risk) allele; A2: non-effect allele; # p-values which are less than or equal to 3.27 × 10−310 are stored as 0.00E+00 due to arithmetic underflow condition.
Table
Table 5. Significant MSMR results of POAG GWAS using dataset of whole blood [36] and brain mQTLs [31].
Table 5. Significant MSMR results of POAG GWAS using dataset of whole blood [36] and brain mQTLs [31].
TissueProbe IDNearest GeneTop mQTL SNPTop SNP Chr:PositionA1/A2β GWASSE GWASP GWASβ mQTLSE mQTLP mQTL β MSMRSE MSMRP MSMRP HEIDI
Bloodcg17332705AFAP1-AS1rs622906014:7939008A/T−0.10240.01462.35 × 10−120.25690.03306.68 × 10−15−0.39860.07651.87 × 10−71.85 × 10−3
cg24250820AFAP1rs559381164:7933940A/C−0.10360.01482.66 × 10−120.26850.03172.17 × 10−17−0.38590.07156.78 × 10−81.08 × 10−2
cg12728606AFAP1rs28919284:7924802G/C0.14430.01377.16 × 10−26−0.50310.03296.38 × 10−53−0.28690.03314.34 × 10−182.45 × 10−3
cg24023194AFAP1rs562203814:7907636A/G−0.12940.01403.20 × 10−200.22200.03246.85 × 10−12−0.58270.10593.75 × 10−81.56 × 10−1
cg15957394AFAP1 (dist = 169)rs177714704:7931611C/G−0.11190.01435.39 × 10−15−0.31500.03242.32 × 10−220.35520.05831.11 × 10−91.83 × 10−3
cg19564367AFAP1 (dist = 198)rs560782224:7932073C/T−0.10370.01459.74 × 10−13−0.27400.03504.96 × 10−150.37860.07181.34 × 10−71.57 × 10−3
cg09806625EXOC2rs171352346:593109C/A0.14900.01904.04 × 10−151.48510.03170.00E+00 #0.10030.01309.52 × 10−151.84 × 10−2
cg21084119EXOC2rs171356796:614787C/T0.14620.01863.43 × 10−15−0.43990.03164.82 × 10−44−0.33230.04857.20 × 10−121.28 × 10−2
cg14812743PDE7Brs65700626:136388422T/G−0.09430.01391.32 × 10−11−0.33240.03295.49 × 10−240.28370.05051.89 × 10−81.18 × 10−2
cg14470647ABCA1rs18009779:107690450A/G−0.08460.01443.95 × 10−9−0.33470.03303.60 × 10−240.25280.04973.56 × 10−72.43 × 10−3
cg13430450ABCA1 (dist = 536)rs24224939:107690995A/G−0.10370.01362.06 × 10−14−0.23640.03274.55 × 10−130.43890.08351.47 × 10−78.62 × 10−3
cg05938607BICC1rs1074073110:60348886G/A0.07360.01343.85 × 10−80.87820.02734.86 × 10−2270.08380.01555.99 × 10−83.28 × 10−1
cg12342675BICC1rs747457010:60343085C/G0.07140.01349.69 × 10−81.10760.02240.00E+00 #0.06450.01221.15 × 10−76.11 × 10−1
cg10738003ARHGEF12rs711732111:120239051C/T0.08270.01359.00 × 10−10−1.27810.01610.00E+00 #−0.06470.01061.01 × 10−97.19 × 10−3
Braincg15605172ACOXLrs67205032:111665137A/G0.08220.01426.79 × 10−91.26230.03411.00 × 10−3000.06510.01141.02 × 10−81.55 × 10−3
cg25107522DGKDrs74222722:234268308A/C−0.07590.01373.17 × 10−8−1.26050.03401.00 × 10−3000.06020.01104.47 × 10−82.31 × 10−2
cg20312457AFAP1rs356090194:7847892G/C0.11480.01488.56 × 10−150.48710.06036.28 × 10−160.23560.04212.17 × 10−84.60 × 10−3
cg24023194AFAP1rs134357304:7923991G/A0.14180.01374.81 × 10−250.42590.06259.67 × 10−120.33300.05851.29 × 10−86.70 × 10−1
cg07406289AFAP1rs23859024:7921634A/G0.14190.01372.97 × 10−250.37450.06323.05 × 10−90.37890.07362.62 × 10−76.79 × 10−1
cg08544002EXOC2rs177567126:625071A/G−0.12080.01699.46 × 10−13−1.45340.03760.00E+00 #0.08310.01182.22 × 10−121.35 × 10−2
cg21084119EXOC2rs177567126:625071A/G−0.12080.01699.46 × 10−13−1.42240.03793.47 × 10−3080.08490.01212.34 × 10−121.62 × 10−3
cg14470647ABCA1rs22433129:107690124A/G0.08250.01438.49 × 10−90.50310.04403.10 × 10−300.16400.03192.71 × 10−76.13 × 10−2
cg01159576LMX1Brs625781279:129386860C/T0.12790.01536.45 × 10−17−0.38950.04773.33 × 10−16−0.32820.05625.27 × 10−93.29 × 10−2
cg09696939BICC1 (dist = 823)rs1074073110:60348886G/A0.07360.01343.85 × 10−81.21460.03281.00 × 10−3000.06060.01115.38 × 10−81.18 × 10−2
cg00554250NR1H3rs105237311:47354787C/T−0.08430.01401.97 × 10−91.32020.03561.00 × 10−300−0.06380.01083.15 × 10−93.34 × 10−3
cg14056187C14orf39 (dist = 1445)rs715544814:60937851C/T−0.11840.01413.98 × 10−170.38080.04523.70 × 10−17−0.31100.05222.64 × 10−92.22 × 10−2
cg01170045SIX6 (dist = 4306)rs6199155114:60982515A/G−0.11870.01413.46 × 10−17−0.41620.06481.38 × 10−100.28510.05583.28 × 10−71.48 × 10−1
cg10299448GAS7rs1768700617:10019797C/T0.11020.01674.36 × 10−11−0.64650.05338.03 × 10−34−0.17050.02947.01 × 10−91.67 × 10−2
cg15661389GAS7rs1768700617:10019797C/T0.11020.01674.36 × 10−11−0.52670.05486.75 × 10−22−0.20920.03855.42 × 10−83.14 × 10−2
cg19784903TBKBP1rs479405717:45786452C/T−0.06920.01374.14 × 10−71.09260.02770.00E+00 #−0.06340.01265.14 × 10−72.93 × 10−3
POAG: primary open-angle glaucoma; SNP: single-nucleotide polymorphisms; SMR: summary-data-based Mendelian randomization; eQTL: expression quantitative trait loci; GWAS: genome-wide association study; HEIDI: heterogeneity in dependent instruments; b: effect size; se: standard error; p: p-value; A1: effect (risk) allele; A2: non-effect allele; P GWAS: p-value of SNP association with POAG; # p-values which are less than or equal to 3.27 × 10−310 are stored as 0.00E+00 due to arithmetic underflow condition. Only HEIDI-passed results are shown: pHEIDI ≥ 2.78 × 10−3 (0.05/18 SMR significant probes) in blood and pHEIDI ≥ 5.0 × 10−3 (0.05/10 SMR significant probes) in the brain.
Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Share and Cite

MDPI and ACS Style

Asefa, N.G.; Kamali, Z.; Pereira, S.; Vaez, A.; Jansonius, N.; Bergen, A.A.; Snieder, H. Bioinformatic Prioritization and Functional Annotation of GWAS-Based Candidate Genes for Primary Open-Angle Glaucoma. Genes 2022, 13, 1055. https://doi.org/10.3390/genes13061055

AMA Style

Asefa NG, Kamali Z, Pereira S, Vaez A, Jansonius N, Bergen AA, Snieder H. Bioinformatic Prioritization and Functional Annotation of GWAS-Based Candidate Genes for Primary Open-Angle Glaucoma. Genes. 2022; 13(6):1055. https://doi.org/10.3390/genes13061055

Chicago/Turabian Style

Asefa, Nigus G., Zoha Kamali, Satyajit Pereira, Ahmad Vaez, Nomdo Jansonius, Arthur A. Bergen, and Harold Snieder. 2022. "Bioinformatic Prioritization and Functional Annotation of GWAS-Based Candidate Genes for Primary Open-Angle Glaucoma" Genes 13, no. 6: 1055. https://doi.org/10.3390/genes13061055

APA Style

Asefa, N. G., Kamali, Z., Pereira, S., Vaez, A., Jansonius, N., Bergen, A. A., & Snieder, H. (2022). Bioinformatic Prioritization and Functional Annotation of GWAS-Based Candidate Genes for Primary Open-Angle Glaucoma. Genes, 13(6), 1055. https://doi.org/10.3390/genes13061055

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop