Update on the Genetics of Primary Open-Angle Glaucoma

Abstract

Affecting nearly 80 million individuals, glaucoma is the number one cause of irreversible blindness in the world. This ocular disease describes a set of optic neuropathies of which primary open angle glaucoma (POAG) is the most common. POAG is associated with progressive visual field deterioration resulting from damage to the optic nerve and loss of retinal ganglion cells. Risk factors for POAG include elevated intraocular pressure, aging, African and Hispanic ancestry, and a positive family history of POAG. Multiple genes have been found to contribute to POAG. Much of POAG genetics and pathology has yet to be explained. Recent genome-wide association studies have identified a large number of novel loci associated with POAG and its endophenotypes. Genomic and proteomic profiling of biofluids has contributed to our knowledge of differential gene expression in POAG. Functional studies both in cell culture and animal models have confirmed the effects of variants and differential gene expression on ocular physiology while in silico analyses have increased our understanding of disease risk and progression so that we might better diagnose and treat this complex genetic illness.

1. Introduction

Glaucoma is an optic neuropathy characterized by retinal ganglion cell (RGC) loss and optic nerve damage resulting in progressive visual field loss (Liu and Allingham, 2017; Weinreb et al., 2014; Wiggs and Pasquale, 2017). With a predicted 80 million cases by the year 2020, it is the leading cause of irreversible blindness in the world (Quigley and Broman, 2006; Tham et al., 2014). Seventy-five to ninety percent of glaucoma cases can be attributed to primary open angle glaucoma (POAG), a type of glaucoma that shares the characteristics of other glaucoma phenotypes but lacks an obvious physical cause (Gao et al., 2019; Liu and Allingham, 2017; Quigley and Broman, 2006; Tham et al., 2014). For example, POAG can be distinguished from primary angle closure glaucoma (PACG) on the basis of the degree of closure of the iridocorneal angle (MacGregor et al., 2018b; Wiggs and Pasquale, 2017).

Although the pathogenesis of POAG has not been fully elucidated, there are known risk factors for the disease including elevated intraocular pressure (IOP), age, ethnicity, and a positive family history (Bonnemaijer et al., 2018; Gupta and Chen, 2016; Liu and Allingham, 2017; Rudnicka et al., 2006; Weinreb et al., 2014; Wiggs and Pasquale, 2017). Of these risk factors, only IOP can be modified clinically, and therefore it is the primary target for both medical and surgical treatment (Boland et al., 2013; Liu and Allingham, 2017; Weinreb et al., 2014). Although elevated IOP is a risk factor for POAG, it is not a necessary characteristic for diagnosis. While individuals with sustained or untreated elevated IOP (≥22 mmHg) are said to have high tension glaucoma (HTG), a subset of POAG cases occur at low to normal IOP levels (≤21 mmHg) and are therefore classed as normal tension glaucoma (NTG) (Bailey et al., 2018; Liu and Allingham, 2017; Pasquale et al., 2017; Weinreb et al., 2014). RGC atrophy, optic nerve cupping, and visual field loss are common characteristics of POAG regardless of IOP levels and are the basis of diagnosis (Gupta and Chen, 2016; Liu and Allingham, 2017; Weinreb et al., 2014).

Age is also a risk factor for POAG although age of onset can vary (Gupta and Chen, 2016; Rudnicka et al., 2006; Wiggs and Pasquale, 2017). Generally, POAG is diagnosed after the age of 40 (Rudnicka et al., 2006; Tham et al., 2014; Wiggs and Pasquale, 2017). In certain populations, POAG prevalence has been shown to quadruple from age 40 to age 80 (Budenz et al., 2013; Gupta and Chen, 2016). However, because POAG patients often do not seek treatment until significant vision loss has occurred, age of onset may occur many years before diagnosis (Gupta and Chen, 2016; Hennis et al., 2007; Leite et al., 2011; Liu and Allingham, 2017; Rotchford et al., 2003; Sathyamangalam et al., 2009; Weinreb et al., 2014). More recently, there has been evidence suggesting that sex may also be a risk factor for glaucoma. POAG prevalence has been shown to be higher in males than females in several studies (Budenz et al., 2013; Khachatryan et al., 2019; Rudnicka et al., 2006). However, a difference in risk based on sex has not been fully accepted as of yet due to conflicting reports (Budenz et al., 2013; Khachatryan et al., 2019; Rudnicka et al., 2006).

The prevalence of POAG differs between ethnic groups (Budenz et al., 2013; Liu and Allingham, 2017). East Asian and Hispanic populations have slightly higher prevalence rates than populations of non-Hispanic European ancestry while POAG is almost twice as prevalent in individuals of African descent (16% vs. 7–10%) and is especially prevalent among West Africans (Budenz et al., 2012; Budenz et al., 2013; Choquet et al., 2018; Mwanza et al., 2019; Rotchford et al., 2003). In fact, the risk for developing POAG is 3–5 times greater in individuals of African ancestry (Bonnemaijer et al., 2018; Budenz et al., 2013; Khachatryan et al., 2015; Kyari et al., 2013; Mwanza et al., 2019; Taylor et al., 2019). Furthermore, individuals of African descent are likely to have a more severe phenotype and lose their vision entirely (Bonnemaijer et al., 2018; Cook, 2009; Kyari et al., 2013).

Because positive family history is a risk factor (POAG occurs in more than one immediate family member in 4–16% of cases), POAG is considered to have a genetic, heritable basis (Sheffield et al., 1993). However, this genetic basis is complex and does not demonstrate a distinct means of inheritance (Drewry et al., 2018). Despite the discovery of many contributing genetic elements by linkage analyses and GWAS, more than 90% of POAG genetics remains unexplained (Drewry et al., 2018; Gong et al., 2019; Liu and Allingham, 2017). Several previous reviews have summarized these genetic discoveries (Liu and Allingham, 2011, 2017; Sakurada and Mabuchi, 2018; Wiggs and Pasquale, 2017). The most well-characterized genetic elements include MYOC, OPTN, WDR36, CAV1, TMCO1, CDKN2B-AS1, SIX6, ABCA1, TXNRD2, and FOXC1 among others (Liu and Allingham, 2017). Tables 1 and 2 from Liu et al., 2017 may be referenced for more comprehensive information on these and other genetic elements.

Table 1.

Novel SNPs associated with POAG identified by GWAS. All loci passed genome-wide significance in discovery GWAS (P < 5.00E-08) and nominal significance in discovery + replication meta-analysis (P < 0.05). Odds ratio (OR) values were taken from discovery + replication meta-analysis.

Leading SNP	Neighboring Genes	OR	Ethnicity/Ancestry	References
rs141186647	EXOC4	0.90	African	(Bonnemaijer et al., 2018)
rs56335522	IKZF2	1.19	Multi-ethnic	(Choquet et al., 2018)
rs76325372	ANKH	1.12	Multi-ethnic	(Choquet et al., 2018)
rs9494457	PDE7B	1.12	Multi-ethnic	(Choquet et al., 2018)
rs34201102	CADM2	1.11	Multi-ethnic	(Choquet et al., 2018)
rs9853115	DGKG	1.10	Multi-ethnic	(Choquet et al., 2018)
rs324794	TMTC2-SLC6A15	1.08	Multi-ethnic	(Choquet et al., 2018)
rs56117902	FMNL2	0.91	Multi-ethnic	(Choquet et al., 2018)
rs55770306	LMX1B	0.87	Multi-ethnic	(Choquet et al., 2018)
rs2073006	EXOC2	0.86	Multi-ethnic	(Choquet et al., 2018)
rs2935057	LOC101929614-LOC105378153	1.15	European	(MacGregor et al., 2018b)
rs2073006	EXOC2	1.14	European	(MacGregor et al., 2018b)
rs1013278	CTTNBP2-CFTR	1.09	European	(MacGregor et al., 2018b)
rs7924522	ETS1	1.09	European	(MacGregor et al., 2018b)
rs4141671	BICC1	0.91	European	(MacGregor et al., 2018b)
rs61394862	ANKH	0.90	European	(MacGregor et al., 2018b)
rs9284802	CADM2	0.90	European	(MacGregor et al., 2018b)
rs11710139	LOC107986141-LOC107986142	0.90	European	(MacGregor et al., 2018b)
rs12699251	THSD7A	0.90	European	(MacGregor et al., 2018b)
rs10505100	ANGPT1	0.84	European	(MacGregor et al., 2018b)
rs73174345	MECOM	0.84	European	(MacGregor et al., 2018b)
rs10819187	LMX1B	1.21	Japanese, East Asian, European	(Shiga et al., 2018)
rs28480457	MEIS2	1.14	Japanese	(Shiga et al., 2018)
rs61275591	ANKRD55-MAP3K1	1.13	Japanese, European	(Shiga et al., 2018)
rs1048661	LOXL1	1.13	Japanese, European	(Shiga et al., 2018)
rs7636836	FNDC3E	1.12	Japanese, East Asian, European	(Shiga et al., 2018)
rs343093	HMGA2	1.11	Japanese	(Shiga et al., 2018)
rs12262706	LHPF	1.11	Japanese, East Asian	(Shiga et al., 2018)
rs185815146*	ENO4	Not reported	African	(Taylor et al., 2019)

^*Replication and OR data was not supplied.

Table 2.

Novel SNPs associated with POAG endophenotypes IOP, VCDR, and CCT as identified by GWAS. All loci passed genome-wide significance in the meta-analysis (P < 5.00E-08).

Endophenotype	Novel SNPs	Ethnicity/Ancestry	References
Infraocular Pressure (IOP)	161	European	(Khawaja et al., 2018; MacGregor et al., 2018a)
Vertical Cup/Disk Ratio (VCDR)	2	Multi-ethnic	(Ghanbari et al., 2017b)
Central Cornea Thickness (CCT)	19	Cross-ancestry, European, Asian	(Iglesias et al., 2018)

In the last few years, several additional studies have been conducted on glaucoma genetics. These studies have identified novel variants, analyzed gene expression in biofluid samples, and used cell culture and animal models for functional validation. This paper seeks to provide an update to previous reviews by focusing on the glaucoma genetics work conducted over the last three years.

2. Genome-Wide Association Studies (GWAS)

Genome-wide association studies (GWAS) seek to identify genetic variants (e.g. single nucleotide polymorphisms, SNPs) commonly found in individuals with a given phenotype or disease by analyzing large patient genomic datasets (Liu and Allingham, 2017). Such common variants are said to be associated with the phenotype/disease (Liu and Allingham, 2017). More than 80 loci have been found to be associated with either POAG or one of its endophenotypes, many of which have been verified by multiple studies. In the past, the majority of POAG genome-wide association studies were conducted in sample populations with European or Asian ancestry, leading to little explanation of POAG pathogenesis in other populations where POAG is more prevalent (Bonnemaijer et al., 2018). In recent years, however, there has been an increase in the number of GWAS in African American and multi-ethnic populations. Not only do these multi-ethnic GWAS help to explain variation in these populations, but they also contribute to a better understanding of POAG pathogenesis in the population as a whole by discovering additional loci (Taylor et al., 2019).

2.1. Primary Open Angle Glaucoma (POAG)

Multiple genes associated with POAG have been highlighted in previous reviews (Liu and Allingham, 2011, 2017; Sakurada and Mabuchi, 2018; Wiggs and Pasquale, 2017). The most well-established genes include CAV1, TMCO1, CDKN2B-AS1, SIX6, ABCA1, GMDS, AFAP1, GAS7, TGFBR3, TXNRD2, ATXN2, and FOXC1 (Bailey et al., 2016; Bonnemaijer et al., 2018; Burdon et al., 2011; Chen et al., 2014; Choquet et al., 2018; Gharahkhani et al., 2014; Hysi et al., 2014; Liu and Allingham, 2017; Luo et al., 2015; Thorleifsson et al., 2010; Wiggs et al., 2013; Wiggs et al., 2012). Many of these genes’ association has been replicated in multiple studies and in a variety of ethnic populations. In addition to replicating the genetic associations mentioned, GWAS conducted in the last few years have discovered 29 novel variants associated with POAG at the level of genome-wide significance (P < 5.00E-08) (Table 1). This review seeks to address these novel variants in particular. In addition to finding variants in well-studied European and Asian populations, these recent GWAS have expanded to include previously neglected African and Hispanic populations by either studying these populations individually or by incorporating them into an inclusive multi-ethnic population.

A multi-ethnic GWAS of a diverse population of African, Asian, Hispanic, and non-Hispanic European ancestry from the Genetic Epidemiology Research in Adult Health and Aging (GERA) study identified five novel variants that significantly associated with POAG (Choquet et al., 2018). Three of the novel variants discovered in the GERA cohort were confirmed to be significant by a replication GWAS in the UK Biobank (UKBB) cohort (P < 0.05) (Choquet et al., 2018). These three loci were located in or near the following genes with their leading variant: PDE7B (rs9494457), TMCTC2 (rs324794), and FMNL2 (rs56117902) (Choquet et al., 2018). When UKBB was used as the discovery cohort, nine additional novel loci were identified of which six replicated significantly in GERA (Choquet et al., 2018). These six loci were located in or near the following genes: IKZF2 (rs56335522), ANKH (rs76325372), CADM2 (rs34201102), DGKG (rs9853115), LMX1B (rs55770306), and EXOC2 (rs2073006) (Choquet et al., 2018). This multi-ethnic study was able to increase number of variants associated with POAG in non-European populations, thereby explaining more of the variation seen in these populations (Choquet et al., 2018).

In addition to the variants found in the GERA multi-ethnic study, the African Descent and Glaucoma Evaluation Study (ADAGES) III identified a novel variant (rs185815146) at ENO4 as being significantly associated with POAG (Taylor et al., 2019).

Another GWAS meta-analysis conducted in an African and African American cohort discovered a novel variant (rs141186647) that was associated with POAG at genome-wide level of significance (Bonnemaijer et al., 2018). This variant was located within EXOC4, a gene coding SEC-8, a protein involved in vesicle transport and secretion, a function which suggests the variant’s contribution to POAG occurs through faulty vesicular transport in the trabecular meshwork (TM), leading to reduced AH outflow (Bonnemaijer et al., 2018). Two novel SNPs (rs9475699 and rs62023880) located between COL21A1-DST and MNS1-ZNF280D respectively were also found to be associated with POAG although they did not pass the genome-wide significance threshold (Bonnemaijer et al., 2018). Interestingly, none of these variants were prominent in European populations (Bonnemaijer et al., 2018). Furthermore, although some of the variants found in European and Asian GWAS were also identified in African populations (TXNRD2, CDKN2B-AS1, and TMCO1 at nominal significance), many of them were not (Bonnemaijer et al., 2018). Moreover, the variants that were identified had different allele frequencies in European/Asian and African populations (Bonnemaijer et al., 2018).

In addition to these studies conducted in multi-ethnic and African populations, there have been several GWAS conducted in Asian and European cohorts. Seven novel variants in or near the genes LMX1B, MEIS2, ANKRD55-MAP3K1, LOXL1, FNDC3E, HMGA2, and LHPF were found to be associated with POAG at a level of genome-wide significance in a Japanese cohort (Shiga et al., 2018). Of these, three variants were associated with POAG in a Chinese population and four were associated with POAG in a European population (Shiga et al., 2018). In addition to their association with POAG, these variants were also associated with type 2 diabetes and cardiovascular disease, suggesting that POAG may have some relationship with these diseases as well (Shiga et al., 2018).

A GWAS meta-analysis conducted using data combined from the UKBB and the Australian and New Zealand Registry of Advanced Glaucoma (ANZRAG) discovered 24 variants associated with POAG, 11 of which were novel (MacGregor et al., 2018b). These included SNPs near the genes CADM2, THSD7A, ANGPT1, ANKH, LOC101929614-LOC105378153, EXOC2, BICC1, MECOM, CTTNBP2-CFTR, ETS1, and LOC107986141-LOC107986142 (Table 1) (MacGregor et al., 2018b).

2.2. Intraocular Pressure (IOP)

While finding variants associated with POAG provides a relatively direct method for determining genetic contributions to pathogenesis of the disease, an alternative approach is to find variants associated with POAG endophenotypes. This approach is especially relevant to POAG because it is a progressive disease with a broad range of trait expression. IOP is the most significant and only treatable POAG risk factor and endophenotype. Even a 1 mmHg increase in pressure can significantly increase an individual’s risk for developing POAG (>15%) (de Voogd et al., 2005; Khawaja et al., 2018). Despite this fact, the EPIC-Norfolk Eye Study has shown that IOP is not an accurate or precise means of POAG diagnosis (Chan et al., 2017). Rather diagnosis must be based on vertical cup/disk ratio (VCDR) and visual field measurements (Chan et al., 2017). Regardless, IOP does provide a quantifiable trait that may be used to discover variants associated with POAG.

In a European GWAS, sixty-eight novel variants significantly associated with IOP (Table 2; Supplemental Table S1) (Khawaja et al., 2018). Several of the genes neighboring these variants have developmental functions that may contribute to glaucoma while others have roles in mitochondrial and lipid metabolic processes, pathways that are frequently associated with POAG (Khawaja et al., 2018). Although the association of these variants with IOP was novel, several had been previously associated with POAG or its endophenotypes. Variants in or near AFAP1, FOXC1, TXNRD2-GNB1L, and ATXN2-SH2B3 among others have been known to be associated with POAG while HGF, PLEKHA7, FERMT2, and GLIS3 are associated with PACG and variants in BCAS3, EFEMP1, and RARB are associated with either cup or disk area (Bailey et al., 2016; Gharahkhani et al., 2014; Jiang et al., 2013; Khawaja et al., 2018; Khor et al., 2016; Springelkamp et al., 2017). Another eighty-five novel variants were found to be associated with IOP at a level of genome-wide significance in a GWAS meta-analysis of a European cohort (UKBB and the International Glaucoma Genetics Consortium, IGGC) (Table 2; Supplemental Table S1) (MacGregor et al., 2018b).

2.3. Vertical Cup/Disk Ratio (VCDR)

VCDR is a diagnostic trait for POAG that assesses damage to the optic nerve head by measuring the degree of cupping in the optic disk (Charlesworth et al., 2010; Foster et al., 2002; Ghanbari et al., 2017a; Liu and Allingham, 2017). The larger the ratio of the cup diameter to the diameter of the entire optic disk, the more significant the damage. Individuals may be diagnosed with POAG if their VCDR exceeds 0.7 or if one eye has a VCDR differing more than 0.2 from the other eye (Ghanbari et al., 2017a; Springelkamp et al., 2017). Similarly to IOP variants, variants associated with VCDR may contribute to POAG pathogenesis. Several novel variants associated with VCDR have been identified recently (Table 2; Supplemental Table S2).

In a microRNA (miRNA) variant analysis of IGGC data, a SNP (rs12803915) in the 3’ tail of miR-612 is associated at genome-wide significance with VCDR (Ghanbari et al., 2017a). This variant upregulates the expression of miR-612, a miRNA that regulates expression of USH2A, FAM101A, JRK, and SIX4 (Ghanbari et al., 2017a). Additionally, a variant (rs2273626) in CARD10-regulating miR-4707 showed association with VCDR, but not at the level of genome-wide significance (Ghanbari et al., 2017a). In addition to the variants in miRNAs, 47 variants in miRNA target genes were said to be associated with POAG endophenotypes (Ghanbari et al., 2017a). Of these variants, one SNP in the 3’UTR of PSCA was identified as passing the genome-wide significance threshold for VCDR (Ghanbari et al., 2017a).

Another eight novel variants (in or near LOC105376196-ABCA1, GAS7, LOC102723944-GMDS, CAV1/2, EXOC2, MYOF-XRCC6P1, BICC1, and CTTNBP2-CFTR) were identified as being nominally associated with VCDR (P < 0.01) in an IOP GWAS in a European cohort (MacGregor et al., 2018b). However, none of these variants passed the genome-wide significance threshold (MacGregor et al., 2018b).

2.4. Central Cornea Thickness (CCT)

Central corneal thickness (CCT) is another measurable trait related to POAG (Herndon et al., 2004; Iglesias et al., 2018). Severe glaucoma has been associated with thin CCT (Herndon et al., 2004). African Americans and Japanese who have an increased risk for developing glaucoma have been shown to have a thinner CCT than Caucasians, possibly explaining some of the added risk for POAG in these populations (Aghaian et al., 2004; Herndon et al., 2004; King et al., 2018; Sng et al., 2016). SNPs in genes contributing to corneal structural development have been considered as possible variants contributing to POAG (Iglesias et al., 2018).

A recent GWAS in a European and Asian cross-ancestry cohort identified 19 new loci associated with CCT (Iglesias et al., 2018). These included variants near ADAMTS8, DCN, NDUFAF6, STAG1, LTBP1, COL6A2, LOXL2, HABP2, FBN1, FGF1, STON2, SAMD9, GLIS3, TGFB2, RUNX2, ARVCF, COL12A1, ADAMTS2, and THBS2 (Iglesias et al., 2018) (Table 2; Supplemental Table S3). Of these, ADAMTS8, DCN, NDUFAF6, STAG1, and LTBP1 were significantly associated with CCT in the isolated European cohort while ADAMTS8 was the only variant associated with POAG in the isolated Asian cohort at a level of genome-wide significance (Iglesias et al., 2018). Many of the identified variants lay in regulatory regions, and a fifth of the variants were located near (≤ 1 Mb) genes responsible for corneal diseases or diseases of connective tissue (Iglesias et al., 2018). Although proximity does suggest that the gene contributes to the disease, it cannot be taken as conclusive (Iglesias et al., 2018). None of the CCT variants in this study passed the genome-wide significance threshold for POAG (Iglesias et al., 2018). However, three variants did significantly associate with the corneal disease keratoconus (Iglesias et al., 2018).

3. In silico disease modeling

In order to synthesize the abundant and often complex data obtained from GWAS into a cohesive and biological meaningful whole, pathway and network analyses are often undertaken. In this way, the interaction of POAG candidate genes can be studied as well as the biological and molecular processes in which they are involved. Genes recently identified as being associated with IOP and POAG implicate pathways related to EGFR signaling, cell movement, vascularization, toll-like receptor signalling, long-term potentiation, focal adhesion, tight junctions, collagen, and the extracellular matrix (ECM) (Drewry et al., 2018; MacGregor et al., 2018b; Shiga et al., 2018).

Several other studies have supported the involvement of EGFR signaling in glaucoma pathogenesis. Both PI3K-AKT and MAPK signaling are controlled by EGFR and have been evidenced as being either neuroprotective against glaucoma in the case of PI3K-AKT or activated in glaucomatous eyes in the case of MAPK (Harder et al., 2017; Shiga et al., 2018). In addition, EGFR can increase harmful RGC oxidative stress through its regulation of NOS-2 (Liu and Neufeld, 2003; Shiga et al., 2018). Furthermore, EGFR signaling, which can be increased under high pressure, has been implicated as contributing to RGC loss as well as harm to optic nerve head astrocytes (Liu and Neufeld, 2003, 2004; Shiga et al., 2018).

The contribution of collagen and ECM pathways to glaucoma pathogenesis is further supported by a recent CCT GWAS that found that CCT associated pathways included ECM, collagen, basement membrane, connective tissue disease, endoplasmic-reticulum-associated protein degradation (ERAD), cancer, and TGF-β pathways (Iglesias et al., 2018). Due to the importance of collagen in cornea structure and development, changes in the collagen and ECM composition of the cornea could certainly affect corneal thickness and result in subsequent changes to IOP regulation leading to the development of POAG.

The most recent study on POAG genetics discovered additional pathways affected by long non-coding RNA (lncRNA) genes via a Kegg pathway enrichment analysis (Xie et al., 2019). These included pathways involving protein metabolism, phosphate metabolism, transcription/translation, melanogenesis, bicarbonate homeostasis, cancer, DNA repair, pertussis, anemia, taste, calcium signaling, and prion diseases (Xie et al., 2019). In addition, pathways associated with lipid metabolism and innate immune response have been implicated as being involved in POAG (Sharma et al., 2018).

Pathways related to the metabolism of sex hormones have also been suggested to play a role in POAG pathogenesis (Agapova et al., 2006; Akar et al., 2004; Bailey et al., 2018; Deschenes et al., 2010; Drewry et al., 2018; Newman-Casey et al., 2014; Vajaranant et al., 2016). A panel of 16 genes containing variants identified in the NEIGHBORHOOD and ANZRAG datasets representative of the testosterone pathway were tested for association with POAG (Bailey et al., 2018). The testosterone pathway SNP panel was found to be consistently associated with POAG in men and with HTG (Bailey et al., 2018). However, there was no consistent association with POAG in women or in both sexes combined (Bailey et al., 2018). Neither was there consistent association of the testosterone pathway with NTG (Bailey et al., 2018). Furthermore, no one variant was significantly associated with POAG and there were no common significant genes between dataset analyses (Bailey et al., 2018).

In addition to pathway analysis, polygenic risk scores (PRSs) provide another means of using abundant GWAS data to understand POAG pathogenesis (Gao et al., 2019). PRSs take into account the combined contributions of associated variants in order to produce a predictive measure of disease risk(Gao et al., 2019). While variants associated with a disease can be used to assemble a PRS, sometimes it is more useful to construct a PRS based on variants associated with an endophenotype of the disease (Gao et al., 2019). IOP variants were used to construct a PRS that has shown significant association for both IOP and POAG in the UK Biobank samples (P ~ 10²⁰⁰ and P ~ 10⁷⁷ respectively) (Gao et al., 2019). According to this PRS, POAG was 6 times more likely to occur in individuals having a score in the top 25% (Gao et al., 2019). Variants were included if they had a significance greater than 5.00E-05, instead of the genome-wide significance value of 5.00E-08, thereby increasing the number of variants used to design the PRS (Gao et al., 2019). Although these variants would not have traditionally passed the threshold for genome-wide significance, including these variants allowed an additional 4% of IOP variation to be explained that previous studies had not been able to account for (Gao et al., 2019). Therefore, in the future, perhaps more attention should be paid to nominally associated variants, especially in complex genetic diseases such as glaucoma where many genes make small contributions to its pathogenesis (Gao et al., 2018).

4. In vitro and in vivo functional work

When genomic variants are identified, their expression in ocular tissues must be verified (Abecasis and Cookson, 2000; Shiga et al., 2018; Xie et al., 2019). Cellular studies can be used to test which genes are affected by a variant as well as to examine the effects of a variant in a known gene (Ghanbari et al., 2017a; Shiga et al., 2018). For example, both the major (C) and minor (A) alleles of the POAG-associated variant rs2273626 were incorporated into the CARD10-regulating miR-4707, and transfected into HEK293 cells to test the effects of the SNP on the ability of miR-4707 to regulate the expression of CARD10 (Ghanbari et al., 2017a). Cells that had been transfected with miR-4707 containing the POAG-associated minor (A) allele of rs2273626 expressed higher levels of CARD10, suggesting that this miRNA variant had a reduced ability to degrade CARD10 mRNA (Ghanbari et al., 2017a).

Cellular studies can also be used to verify the impact of gene expression on the cellular and biological processes uncovered by pathway analysis. BMP2 has been suggested to play a role in the calcification of TM, thereby reducing AH outflow and increasing IOP (Borras and Comes, 2009; Xie et al., 2019). The expression of the long noncoding RNA ENST00000607393 has been shown recently to correlate with BMP2 expression, suggesting that it might have some role in calcification of the TM (Xie et al., 2019). Both calcification and ENST00000607393 expression could be induced concurrently by treating primary human TM cells with peroxide (Xie et al., 2019). Furthermore, knockdown of this lncRNA in primary human TM cells resulted in reduced calcification, suggesting that ENST00000607393 is indeed involved in TM calcification (Xie et al., 2019).

In a separate study, a variant in FMNL2 was found to be associated with POAG (Choquet et al., 2018). The protein product FMNL2 is known to be involved in cytoskeletal rearrangement and cell motility (Block et al., 2012; Choquet et al., 2018; Kage et al., 2017). A siRNA knockdown of FMNL2 in human TM cells resulted in cells with abnormal shape, suggesting that FMNL2 plays a role in cellular morphology and may in turn impact AH outflow and therefore IOP (Choquet et al., 2018).

Animal models are necessary to confirm findings in cell culture. The BXD mouse strain, a recombinant inbred cross between DBA/2J and C57BL/6J strains, has been a useful model for IOP-induced POAG (Chintalapudi et al., 2017; King et al., 2018; Peirce et al., 2004; Taylor et al., 1999). This model was used to identify the function of the gene CACNA2D1 (identified by GWAS) in IOP regulation (Chintalapudi et al., 2017). Pregabalin, a drug that binds to the protein product of CACNA2D1, which codes for a calcium channel subunit, was administered to BXD mice with different haplotypes for the gene (Chintalapudi et al., 2017). Pregabalin was able to lower IOP more significantly in the B6/BXD14 CACNA2D1 haplotype than in the D2/BXD48 haplotype (Chintalapudi et al., 2017). This drug works by decreasing calcium transport through the calcium channel of which CACNA2D1 is a subunit, thereby lowering ion transport and muscle contractility (Chintalapudi et al., 2017). It is thought that this reduction in TM contractility might increase AH outflow (Chintalapudi et al., 2017).

BXD mice were also used to identify CCT quantitative trait loci (QTL) (King et al., 2018). The NEIGHBORHOOD GWAS datasets were then used to verify that the QTLs found to be associated with CCT in mice were also associated with POAG in human patients (King et al., 2018). The gene POUF6F2 contained a variant (rs76319873) that was nominally significant in the NEIGHBORHOOD cohort (King et al., 2018). This gene is expressed in RGCs as well as corneal cells (King et al., 2018). In addition to effects on corneal thickness as evidenced by POU6F2 knockout mice, POU6F2 variants may also effect RGC health through interaction with SLC30A3 (King et al., 2018; Li et al., 2017).

Other mouse models have also been used. Six pathogenic variants (Glu39Asp, Glu54Lys, Phe103Ser, Asn113Lysfs*10, Glu143Lys, and Ser171Arg) in the RAMP2 (receptor activity-modifying protein 2) gene were identified by sequencing the whole exome of POAG patients in Indian, German, and Han Chinese populations (Gong et al., 2019). A heterozygous Ramp2 knockout mouse model was created to test whether Ramp2 haploinsufficiency could contribute to POAG pathogenesis (Gong et al., 2019). The knockout mouse model demonstrated RAMP2 accumulation, a dysfunctional AM-RAMP2/CRLR-cAMP pathway, decreased cAMP levels, and subsequent RGC loss (Gong et al., 2019), functionally confirming the role of RAMP2 in POAG pathogenesis that was suggested by whole exome sequencing (Gong et al., 2019).

In another study, mouse models with mutations (V265D and Q82X) in LMX1B were developed in C57BL/6J (B6), 129S6/SvEvTac (129), and DBA/2J (D2) strains (B6.Lmx1bV265D/+, 129.Lmx1bV265D/+, and D2. Lmx1bQ28X/+) in order to test the gene’s effects on IOP (Choquet et al., 2018). The mutations resulted in elevated IOP and retinal damage (Choquet et al., 2018). The V265D mutated mice had more significantly altered eye development than the Q82X mutant which evidenced no developmental defects, but had age-related changes in pupil size (Choquet et al., 2018). Despite the 129.Lmx1bV265D/+ model having the most severe developmental defects, D2. Lmx1bQ28X/+ and B6.Lmx1bV265D/+ mice had more advanced retinal damage (Choquet et al., 2018).

Transforming growth factor β2 (TGF-β2) is known to be involved in the pathogenesis of glaucoma due to its role in the TM where it signals for the production of various ECM proteins (Raychaudhuri et al., 2018). One such protein, tissue transglutaminase (TGM2), is responsible for crosslinking ECM proteins (Raychaudhuri et al., 2018). Abnormal accumulation of ECM proteins in the TM have been shown to reduce AH outflow, thereby leading to increased IOP and a POAG phenotype (Braunger et al., 2015; Raychaudhuri et al., 2018; Tamm, 2009). After observing an IOP increase in response to TGM2 overexpression in the TM, a recent study developed a knockout mouse model to test the effects of TGM2 on AH outflow and IOP (Raychaudhuri et al., 2017, 2018). IOP decreased in TGM2 knockout mice (Raychaudhuri et al., 2018). Furthermore, knocking out TGM2 increased AH outflow and decreased IOP in TGFβ2-overexpressing glaucoma models (Raychaudhuri et al., 2018).

5. Analysis of biofluids

Biofluids such as aqueous humor and serum can be collected for analysis from patients during invasive procedures, and have been subjected to proteomic analysis to examine POAG-dependent changes in protein expression. Knowledge about differential expression not only provides clues about disease pathology, but also offers a source of diagnostic and prognostic biomarkers (Sharma et al., 2018).

Aqueous humor (AH) is produced by the ciliary epithelium and drains through the TM and Schlemm’s canal to the episcleral vein (Drewry et al., 2018; Sharma et al., 2018). The balance between the rate of production and its drainage establishes the level of intraocular pressure (IOP). When AH outflow is reduced intraocular pressure increases. In fact, elevated IOP is most commonly caused by decreased AH outflow facility (Drewry et al., 2018). In this way, AH contributes to POAG pathogenesis (Sharma et al., 2018). Therefore, examining differences between the constituents of glaucomatous and normal AH provides some clues about the mechanisms of reduced AH outflow in POAG. Due to ethical considerations, AH can only be collected during a necessary procedure such as cataract surgery. Therefore, studies examining the composition of glaucomatous AH must use the AH of cataract patients without other ocular diseases as controls (Drewry et al., 2018; Jayaram et al., 2017; Sharma et al., 2018; Xie et al., 2019). This may bias the differential expression analysis of AH constituents (Liu et al., 2018; Sharma et al., 2018). Despite these limitations, several differentially expressed proteins, miRNAs, and lncRNAs have been identified.

A recent proteomic study identified 33 proteins that were differentially expressed in POAG AH (Sharma et al., 2018). Some of the most significant proteins included Niemann-Pick disease type C2 (NPC2), collagen type XVIII α1 (COL18A1), NACHT and WD domain containing 1 (NWD1), serine/cysteine proteinase inhibitor clade F member 2 (SERPINF2), tetraspanin 14 (TSPAN14), Ig K chain C region (IGKC), succinate CoA ligase GDP-forming subunit β (SUCGL2), inter-α-trypsin inhibitor heavy chain 4 (ITIH4), and isocitrate dehydrogenase 3 (NAD) subunit α (IDH3A) (Sharma et al., 2018). These proteins are involved with pathways associated with lipid metabolism, cellular transport, proteolysis, immune response, glycoproteins, and cellular structure, pathways that are associated with TM outflow capacity (Sharma et al., 2018). Due to their differential expression and their possible role in POAG pathogenesis, the identified proteins may serve as biomarkers for disease progression (Sharma et al., 2018).

In addition to proteins, miRNAs are of special interest in genetic studies of POAG because of their role in regulating gene expression, as well as their presence within exosomes in the AH (Dismuke et al., 2015; Guo et al., 2010; Jayaram et al., 2017). Several miRNAs have been identified over the years. Recent studies have discovered novel miRNAs, further contributing to knowledge on their involvement in POAG pathogenesis. Novel differentially expressed miRNAs include miR-143, miR-518d, miR-302d-3p, miR-451a, and miR-125b-5p (Drewry et al., 2018; Jayaram et al., 2017). Many of the genes targeted by these miRNAs are expressed in the anterior segment including the TM (Drewry et al., 2018). Furthermore, some of them are differentially expressed between POAG and control tissues and are involved in pathways involving inflammation, Wnt signaling, proteolytic processes, endocytosis, the ECM, and cell-cell adhesion suggesting that they may play a role in the outflow facility of the TM (Drewry et al., 2018). Wnt signalling is known to play a role in POAG pathogenesis through IOP effects (Jayaram et al., 2017; Mao et al., 2012). Changes in proteolysis and endocytosis could affect the ability of TM cells to uptake and degrade unnecessary or dysfunctional macromolecules, leading to waste accumulation and reduced AH outflow (Jayaram et al., 2017; Liton, 2016). Furthermore, changes in the ECM and in cell-cell adhesion could similarly reduce AH outflow and therefore increase IOP (Jayaram et al., 2017).

Not only are miRNAs differentially expressed in POAG and control AH, but some such as miR-16–5p, miR-200a-3p, miR-200b-3p, miR-205–5p, and miR-205, further show different expression levels between mild and advanced POAG (Liu et al., 2018). Although miRNAs in AH are generally thought to be expressed in anterior tissues, they are also known to migrate from posterior tissues (i.e., the retina) (Jonas et al., 2012; Liu et al., 2016; Liu et al., 2018; Matsuo et al., 1993; Oh et al., 2010).

Another type of regulatory RNA, lncRNAs, have also been the subject of POAG studies in AH. The lncRNAs ENST00000607393, T342877, and T267384 have evidenced differential expression in AH as well as in plasma and iris tissue (Xie et al., 2019). POAG endophenotype retinal nerve fiber layer (RNFL) thickness is associated with variants in lncRNA ENST00000607393 and T342877 (Xie et al., 2019). Differential expression of these lncRNAs corresponded with differential expression of BMP2, a protein that has been implicated in reducing AH outflow and increasing IOP via calcium mineralization in ECM of TM tissue (Borras and Comes, 2009; Buie et al., 2013; Li et al., 2014; Wordinger et al., 2002; Xie et al., 2019).

Although AH is the primary biofluid used to study POAG due its involvement in disease pathology, serum has also been used. This approach was used to isolate protein biomarkers for POAG by identifying the targets of circulating antibodies using LC-MS proteomics on the Fab region of immunoglobulins (Schmelter et al., 2017). The production of antibodies against POAG proteins suggests that POAG causes a systemic response (Schmelter et al., 2017).

6. Conclusion

In the past several years, many additional loci associated with POAG and its endophenotypes have been identified, contributing to our knowledge of the genetic basis and pathology of the disease. GWAS conducted in multi-ethnic and African populations have been of special importance as these populations are at particular risk for developing POAG. In addition, the rising use of proteomic and genomic analyses have increased our knowledge about the differential expression of genes in POAG while also providing diagnostic and prognostic biomarkers. Many of the genes identified by GWAS and other methods have been examined by pathway analysis in order to understand their roles in POAG pathogenesis, and these roles have then been tested using a variety of animal models as well as cell culture.

Despite the great progress made over the last several years, there are many questions still left to be explained about this complex disease. The outcomes of these genetic analyses could be improved by increasing the number of patient samples, including more diverse populations, and identifying better controls for proteomic and genomic comparisons by developing less invasive means of sample collection. Furthermore, many of the HTM cells used today lack validation and require further testing. Lastly, there is still very little known about the crosstalk between the anterior and posterior portions of the eye, a critical component in the pathology of POAG.

Highlights.

• Brief summary of recent GWAS findings in POAG and related endophenotypes (IOP, CCT, VCDR)

• Summary of in silico pathway and network analysis in POAG

• Summary of recent cellular and animal functional studies in POAG

• Summary of recent genomic and proteomic studies in aqueous humor

Abbreviations

POAG

primary open angle glaucoma

RGC

retinal ganglion cell

PACG

primary angle closure glaucoma

IOP

intraocular pressure

HTG

high tension glaucoma

NTG

normal tension glaucoma

GWAS

genome-wide association studies

SNP

single nucleotide polymorphism

GERA

Genetic Epidemiology Research in Adult Health and Aging

UKBB

UK Biobank

ADAGES

African Descent and Glaucoma Evaluation Study

ANZRAG

Australian and New Zealand Registry of Advanced Glaucoma

VCDR

vertical cup/disk ratio

IGGC

International Glaucoma Genetics Consortium

miRNA

microRNA

CCT

central cornea thickness

ECM

extracellular matrix

lncRNA

long non-coding RNA

PRS

polygenic risk score

QTL

quantitative trait locus

TM

trabecular meshwork

SC

Schlemm’s canal

RNFL

retinal nerve fiber layer