Analysis of ASB10 variants in open angle glaucoma

Abstract

Glaucoma is a common cause of visual disability and affects ∼1.6% of individuals over 40 years of age ( 1). Non-synonymous coding sequence variations in the ankyrin repeat and SOCS box containing gene 10 (ASB10) were recently associated with 6.0% of cases of primary open angle glaucoma (POAG) in patients from Oregon and Germany. We tested a cohort of POAG patients (n= 158) and normal control subjects (n= 82), both from Iowa, for ASB10 mutations. Our study had 80% power to detect a 4.9% mutation frequency in POAG patients. A total of 11 non-synonymous coding sequence mutations were detected in the cohort, but no association with POAG was detected when analyzed individually or as a group (P > 0.05). Furthermore, a survey of the National Heart, Lung, and Blood Institute's (NHLBI's) Exome Sequencing Project revealed that non-synonymous ASB10 mutations are present in the general population at a far higher frequency than the prevalence of POAG. These data suggest that non-synonymous mutations in ASB10 do not cause Mendelian forms of POAG.

INTRODUCTION

Glaucoma is a common disease of the optic nerve that is a leading cause of vision impairment and blindness worldwide. Glaucoma has been categorized as primary (idiopathic) or secondary (due to another observed ocular condition). Glaucoma has been further classified by the gross appearance of the eye as ‘open angle’ in cases where the drainage structures are unobstructed and ‘closed angle’ in cases where the drainage structures are clearly blocked. Primary open angle glaucoma (POAG) is the most common form of glaucoma in the USA and many other nations. In the Framingham study, 1.6% of individuals over 40 years of age are afflicted with POAG (1). Heredity plays an important role in POAG (2). Several genes that cause glaucoma primarily on their own have been identified by family based positional cloning studies including myocilin (3), optineurin (4), WDR36 (5) and TBK1 (6). Together these genes account for around 5% of cases of POAG. Other genes have also been discovered that contribute to more complex genetic forms of glaucoma in which the small effects of many genes and environmental factors combine to produce glaucoma (2).

Recently, Pasutto et al. reported that ankyrin repeat and SOCS box containing 10 (ASB10) is the glaucoma-causing gene in the GLC1F locus that they previously mapped to chromosome 7q35-q36 (7). They found that non-synonymous mutations in the ASB10 gene were associated with 6.0% of POAG cases in cohorts from the USA and Germany (8). Additionally, Pasutto et al. (8) demonstrated that the silencing of ASB10 transcripts in a perfused human anterior segment organ culture resulted in a reduced outflow of aqueous humor, suggesting that defects in ASB10 may influence the fluid dynamics of the eye to cause increased intraocular pressure.

In this report, we investigated the role of the ASB10 gene in POAG patients from Iowa by testing for coding sequence variations.

RESULTS

A cohort of 158 POAG patients and 82 control subjects from Iowa were tested for mutations in ASB10. A total of 20 DNA sequence variants were identified (Table 1). Eleven of the sequence variants altered the predicted ASB10 protein (non-synonymous sequence changes), while five were synonymous sequence variants. An additional four variants were located within intron 1 of ASB10. Five of the 11 non-synonymous sequence changes were not previously reported in the initial association between ASB10 and POAG by Pasutto et al. (8).

Table 1.

ASB10 variants

Non-synonymous ASB10 mutations (mutation frequency)	POAG (n= 158)	Normal (n= 82)	P-value
Cys17Trp	1 (0.63%)	0	>0.99
Pro19Arg	1 (0.63%)	0	>0.99
Arg72His*	1 (0.63%)	1 (1.2%)	>0.99
Ala142Val#	4 (2.5%)	0	0.30
Arg222Gly	1 (0.63%)	1 (1.2%)	>0.99
Asp276Gly	1 (0.63%)	0	>0.99
Arg289Cys	2 (1.3%)	1 (1.2%)	>0.99
Arg345His	1 (0.63%)	0	>0.99
Arg357Cys#	20 (13%)	6 (7.3%)	0.27
Ser359Asn	1 (0.63%)	0	>0.99
Pro387Thr	8 (5.1%)	6 (7.3%)	0.56
Total	41 (26%)	15 (18.3%)	0.20
Total excluding Arg357Cys and Pro387Thr	13 (8.2%)	3 (3.7%)	0.27
Synonymous ASB10 mutations (allele frequency)	POAG (n= 316 alleles)	Normal (n= 164 alleles)	P-value
Pro68Pro	1 (0.32%)	0	>0.99
Ala251Ala	21 (6.6%)	6 (3.7%)	0.21
Ala275Ala#	130 (41%)	52 (31.7%)	0.048
Ser300Ser	0	1 (0.61%)	0.34
Leu327Leu	2 (0.63%)	2 (1.2%)	0.61
Intronic ASB10 mutations (allele frequency)	POAG (n= 316 alleles)	Normal (n= 164 alleles)	P-value
IVS1 + 28 (A>T)	1 (0.32%)	0	>0.99
IVS1 + 10 G>A#*	24 (7.6%)	6 (3.7%)	0.11
IVS1 + 7 G>A*	1 (0.32%)	0	>0.99
IVS1 − 15 G>A#	85 (26.9%)	50 (30.5%)	0.45

A total of 20 variants were detected including 11 non-synonymous mutations, 5 synonymous mutations and 4 variants within the introns. The variants are enumerated using ASB10 isoform #3 (NM_080871.3) except those that are not contained within this transcript, which are indicated with an asterisk and enumerated using isoform #1 (NM_001142459.1). Homozygous Ala142Val and Arg357Cys were each detected in a single POAG subject. The nucleotide position of each intervening sequence (IVS) variant is indicated relative to the nearest exon (+ is upstream of the exon and – is downstream). All variants were detected in the heterozygous state except those indicated with a ‘#’ symbol which were also seen in the homozygous state. Those variants that were not previously reported by Pasutto et al. (8) are in bold. Homozygous mutations IVS1 + 10 G>A, Ala142Val and Arg357Cys were detected a single time in different POAG subjects, while homozygous Ala275Ala and homozygous intervening sequence IVS1 − 15 G>A mutations were more frequently observed and mutation genotypes were analyzed at these sites (Table 2). P-values in the table were calculated using Fisher's exact test prior to correction for multiple measures.

Of the five novel ASB10 mutations that were detected, one (Ala142Val) was present in four (2.5%) of our POAG patients, one of which was homozygous, and was absent from the control subjects (Table 1). However, when the allele frequency of each of the non-synonymous ASB10 coding sequence variants was compared between the POAG patients and normal control subjects, none was statistically more common in patients (P > 0.05). Similarly, when the frequency of the non-synonymous ASB10 variants were compared between POAG patients and control subjects as a group, there was no association with glaucoma (P = 0.20).

Pasutto et al. (8) reported that a single synonymous mutation in ASB10, Thr255Thr, was linked to disease in one large POAG pedigree. This variant was not observed in our cohort from Iowa; however, five other synonymous ASB10 mutations were detected (Table 1). When the frequency of each synonymous ASB10 mutation was compared between POAG patients and normal controls, one common variant Ala275Ala was marginally associated with glaucoma (uncorrected P = 0.048). The frequencies of the three possible genotypes of this variant were also only marginally different between patients and controls (P = 0.06, Table 2). When corrected for multiple measures, these P-values were not statistically significant (P > 0.05).

Table 2.

ASB10 variants (genotype frequency)

ASB10 common variants (genotype analysis)	POAG (n= 158)	Normal (n= 82)	P-value
Ala275Ala homozygous (wild-type)	54 (34%)	41 (50%)	0.06
Ala275Ala heterozygous	78 (49%)	30 (37%)
Ala275Ala homozygous (mutant)	26 (16%)	11 (13%)
IVS1 − 15 G>A homozygous (wild-type)	81 (51%)	42 (51%)	0.12
IVS1 − 15 G>A heterozygous	69 (44%)	30 (37%)
IVS1 − 15 G>A homozygous (mutant)	8 (5.1%)	10 (12%)

Two variants Ala275Ala and IVS1 − 15 G>A were frequently observed and the genotype frequencies of these variants were compared between POAG patients and normal controls subjects. P-values were calculated using Fisher's exact test prior to correction for multiple measures.

Finally, four intervening sequence variants in ASB10 were detected in our cohort (Table 1). When the frequencies of these variants were compared between POAG patients and control subjects, no association with glaucoma was detected (P> 0.05).

DISCUSSION

In the last 20 years, 17 loci for open angle glaucoma (GLC1A-GLC1Q) have been mapped by studying large pedigrees. This family based approach has led to the discovery of several glaucoma genes. There is consensus that myocilin and optineurin are the glaucoma genes at the GLC1A and GLC1E loci, respectively (9). However, there is some controversy about reports that WDR36 and neurotrophin 4 (NTF4) are the glaucoma genes at the GLC1G and GLC1O loci (9–11) and the first study replicating the discovery that TANK-binding kinase 1 (TBK1) is the glaucoma gene at the GLC1P locus (6) has only just been reported (12).

Recently, Pasutto et al. reported that ASB10 is the glaucoma gene at the GLC1F locus and that non-synonymous ASB10 mutations are associated with POAG. Ocular expression studies and a human anterior segment ocular culture system provided further support for a role for ASB10 in glaucoma pathogenesis (8). We set out to see if mutations in this gene are present in our POAG patients from Iowa. However, our study of POAG patients and controls from Iowa identified no association between non-synonymous mutations in ASB10 and glaucoma (P = 0.20), even when selected missense variations (Arg357Cys and Pro387Thr) were omitted from our analysis as was done by Pasutto et al. (8) (P = 0.27). It remains possible that some of these variants may have a role in the pathogenesis of glaucoma, but we did not find any statistically significant evidence supporting this hypothesis in our cohort.

Of the ASB10 missense variants that were analyzed by Pasutto et al., the two most common were Arg72His and Arg222Gly, which were each detected in 10 (1.0%) of 977 German POAG patients. The Arg222Gly variation was also detected in 2 (1.0%) of 195 Oregon POAG patients. However, each of these variations was also previously observed in 1.2% of the Oregon controls. The Arg72His variation was absent from German controls subjects, but the Arg222Gly variation was present in 0.8% of German controls. In our study, we detected both the Arg72His and Arg222Gly variations in 0.63% of our POAG patients from Iowa and we also observed both of these variations in 1.2% of our control subjects (Table 1). Based on these observations, the most common previously reported ASB10 mutations are present at similar frequency in multiple control populations, which argues against their pathogenicity.

One synonymous ASB10 coding sequence variant, Ala275Ala, was observed slightly more often in POAG patients than in controls when allele frequencies (P = 0.048) and genotypes (P = 0.06) were compared. This variant does not alter the encoded ASB10 protein. In addition, the high frequency of Ala275Ala in control populations, 32% in our study and 37% in the study by Pasutto et al. (8), suggests that it does not cause highly penetrant forms of open angle glaucoma, which has population prevalence 10-fold lower than the frequency of the Ala275Ala variant. Also, when analyses of the Ala275Ala variant are adjusted for multiple measures, the allele and genotype frequencies are not statistically different between patients and controls (P > 0.05). When the frequency of ASB10 synonymous mutations was similarly analyzed as a group, there was a marginal association with glaucoma (P = 0.02) that was not significant once corrected for multiple measures (P > 0.05).

Pasutto et al. focused their studies on the ASB10 gene after a synonymous coding sequence variation, Thr255Thr, in this gene was seen to be co-inherited with POAG in one large pedigree. Subsequent studies identified the Thr255Thr variation in 1 (0.085%) of 1172 POAG subjects and in none of their 461 control subjects from Oregon and Germany (8). We did not detect the Thr255Thr variation in our smaller cohort of patients and controls (n = 240). These results suggest that the Thr255Thr variation is rare among patients with glaucoma.

Our cohort of patients and controls from Iowa has 80% power to detect a 4.9% mutation rate in ASB10 at a significance level of 0.05 when the mutation rate in control subjects is low (0.1%). However, we observed a high frequency of ASB10 non-synonymous coding sequence mutations in our control subjects (18.3%) as did Pasutto et al. (13.2%). Pasutto et al. chose to exclude the most frequent non-synonymous coding sequence variations (Arg357Cys and Pro387Thr) and analyze only those mutations whose prevalence supported an association with glaucoma. Even with such exclusions, non-synonymous ASB10 coding sequence mutations are observed in control populations at frequencies (5.9% in Oregon, 2.1% in Germany and 3.7% in Iowa) that are greater than the population prevalence of glaucoma, 1.6% in the Framingham Study (1). Detection of non-synonymous ASB10 variants at a frequency much higher than the prevalence of POAG in multiple control populations weakens the argument that these variants are associated with Mendelian (single-gene) forms of glaucoma.

There are a number of other possible explanations for the difference between our results analyzing non-synonymous ASB10 mutations and those of Pasutto et al. (8). The previous report included two important selection biases. The first potential bias involves the exclusion of ASB10 variants from analyses based on their frequency in patient and control cohorts. While we compared the frequency of all non-synonymous ASB10 coding sequence variants between patients and controls, Pasutto et al. chose to exclude two non-synonymous ASB10 variants from their analysis because they were commonly observed and because they were detected in patients and control subjects at similar frequency. This omission may introduce a bias that favors a significant result.

The second potential selection bias in the report by Pasutto et al. is due to their use of a much larger (2.5-fold larger) cohort of POAG patients (n= 1172) than control subjects (n= 461). This unequal cohort size favored the identification of a larger number of rare ASB10 variants in the POAG cohort (6.0%) than in the control subject cohort (2.8%) in their study. In fact, of the 30 non-synonymous coding sequencing variants in ASB10 reported by Pasutto et al. (8), 16 were each identified in one POAG patient and no control subjects, suggesting that some might be found at similar frequencies in both groups if the cohort sizes were the same. To examine this possibility, we searched for ASB10 mutations in the exome data from over 3500 Caucasian subjects in the National Heart, Lung, and Blood Institute's (NHBLI's) Exome Sequencing Project (http://evs.gs.washington.edu/EVS/), whose participants represent a cross-section of the general population. In the NHBLI's general population cohort, a total of 39 non-synonymous ASB10 coding sequence variations were detected including a nonsense mutation, Arg296Stop (Table 3). The sum of the frequency of these non-synonymous ASB10 mutations in the NHLBI cohort was 5.5%, nearly twice the rate of mutations detected in the smaller cohort of controls studied by Pasutto et al. (8) (2.8%). These data suggest that Pasutto et al. may have detected the same frequency of ASB10 mutations in patients and controls if they had studied a similar number of control subjects as POAG patients.

Table 3.

Non-synonymous ASB10 mutation frequency in the NHLBI Exome Sequencing Project

	Variant	Instances of variant	Number of exomes	Frequency of variant (%)
1	Glu36Lys*	1	3508	0.029
2	Ile46Val*	1	3509	0.028
3	Arg72His*	19	3510	0.541
4	Ala75Glu*	3	3510	0.085
5	Val86Ile*	1	3510	0.028
6	Asp91Tyr*	3	3510	0.085
7	Arg111Cys	1	3438	0.029
8	Thr135Ile	2	3499	0.057
9	Ala136Thr	2	3500	0.057
10	Glu139Lys	1	3503	0.029
11	Ala142Val*	13	3501	0.371
12	Arg174Trp	5	3486	0.143
13	Ala190Val	1	3489	0.029
14	Val192Leu	14	3499	0.400
15	Arg195Trp	1	3505	0.029
16	Arg209Trp	1	3506	0.029
17	Arg209Pro	1	3507	0.029
18	Arg222Gly	55	3510	1.567
19	Gly233Val	1	3510	0.028
20	Arg257His	2	3508	0.057
21	Asp276Asn	1	3504	0.029
22	Gln280Leu	1	3506	0.029
23	Arg296Stop*	1	3505	0.029
24	His299Tyr*	1	3506	0.029
25	Arg289Cys	26	3504	0.742
26	Ala305Thr	1	3508	0.029
27	Thr329Met*	7	3507	0.200
28	Pro331Ser	1	3503	0.029
29	His341Arg	1	3482	0.029
30	Val344Ile	1	3465	0.029
31	Arg345His	5	3464	0.144
32	Val349Ala*	1	3494	0.029
33	Arg351Trp*	1	3491	0.029
34	Arg357Cys#	143	1266	11.295
35	Thr375Arg*	1	1266	0.079
36	Glu379Lys	1	1266	0.079
37	Pro387Thr	59	1266	4.660
38	Arg438Cys	9	3492	0.258
39	Arg441Leu	1	3503	0.029
	Total			21.4
	Total (excluding Arg357Cys and Pro387Thr)			5.5

Four subjects were found to be homozygous for one ASB10 variation, Arg357Cys indicated with a ‘#’ symbol. The codon numbering system was based on one isoform of the ASB10 gene (NM_080871.3) in most cases. Those variants with codon assignments based on another isoform (NM_001142459.1) are indicated with an asterisk. The frequencies of the non-synonymous ASB10 variants were examined by searching the exomes of the >3500 Caucasian subjects in the NHLBI's Exome Sequencing Project using the Exome Variation Server (14). The ASB10 mutations that were detected in the Exome Sequencing Project that were not observed in the study by Pasutto et al. (8) are in bold. The sum of the frequencies of non-synonymous ASB10 variants is greater than the frequency of glaucoma, even when the most common variants, Arg357Cys and Pro387Thr, are excluded from analysis as done by Pasutto et al. (8).

Our study of subjects from Iowa is subject to the same selection bias from examining a larger cohort of patients (n= 158) than controls (n= 82). Despite this potential bias, which would tend to favor an association with disease, our study still did not detect an association between ASB10 mutations and glaucoma.

We further explored the role of ASB10 variants in glaucoma by comparing the frequency of ‘case-specific’ and ‘control-specific’ non-synonymous variants (Table 4). Case-specific variants were more common than control-specific variants in both our cohort from Iowa (P = 0.03) and in the cohort from Pasutto et al.'s study (8) (P = 0.0044). Some of the case-specific variants (Pro19Arg and Arg245His) were present in both cohorts of POAG patients. This analysis provides some support for an association between glaucoma and ASB10 variants.

Table 4.

Comparison of POAG-specific and control-specific ASB10 variants

Iowa cohort (current study)			Pasutto et al.'s cohort (8)
ASB10 variant	POAG n= (158)	Normal (n= 82)	ASB10 variant	POAG (n= 1172)	Normal (n= 461)
Cys17Trp	1	0	Pro19Ser	6	0
Pro19Arg	1	0	Arg32Ser	1	0
Ala142Val	4	0	Arg39Gln	1	0
Asp276Gly	1	0	Thr48Ser	1	0
Arg345His	1	0	Gly65Glu	1	0
Ser359Asn	1	0	Val67Met	1	0
Total	9 (5.70%)	0	Asp91Tyr	0	1
	P = 0.030		Arg94Gln	1	0
			Asp97Glu	1	0
			Ala157Val	0	1
			Arg168Cys	1	0
			Cys173X	1	0
			Arg182Val	1	0
			Val192Leu	6	0
			Arg257His	1	0
			Gln280Leu	1	0
			Ala305Thr	1	0
			His317Gln	1	0
			His341Tyr	1	0
			Arg345His	2	0
			Ser425Gly	1	0
			Total	30 (2.56%)	2 (0.43%)
				P = 0.0044

The frequency of case-specific and control-specific ASB10 missense variants was compared for the Iowa cohort and for subjects of the previously reported study by Pasutto et al. (8).

More case-specific ASB10 mutations were detected in both cohorts from Iowa and from the study by Pasutto et al. (Table 4). However, the results from this case-specific/control-specific analysis may also be biased by the large sample size difference between cases and controls. To address this potential bias, we repeated the case-specific and control-specific mutation analysis using data from a larger control set, the cohort of ∼3500 European Americans from the NHLBI's Exome Sequencing Project. When the cohort of POAG patients from Iowa was examined, mutations unique to the POAG patients occurred at the same frequency (2.53%) as mutations unique to the NHBLI control cohort (2.34%), P = 0.79 (Supplemental Material, Table S1). Similarly, ASB10 mutations unique to POAG patients in the study by Pasutto et al. (8) occur at the same frequency 1.62% as mutations unique to the NHLBI control cohort 1.49%, P = 0.78 (Supplemental Material, Table S2). Different mutation detection methods were used in each of these studies [exome sequencing by NHLBI; Sanger sequencing by Pasutto et al.; and high-resolution melt (HRM), single-strand conformation polymorphism (SSCP) and Sanger sequencing by our study]. It is possible that the use of different mutation detection methods with different sensitivities may have had some influence on these results. There is also the possibility that population stratification may also have affected the results of the analysis. However, overall these data suggest that when larger control cohorts are examined, ASB10 variants are not more common in POAG patients than in control subjects.

In summary, our study did not detect a higher frequency of non-synonymous ASB10 coding sequence variations in POAG patients than in control subjects from Iowa. In contrast, Pasutto et al. previously reported that the frequency of non-synonymous ASB10 variations was significantly higher in their cohort of POAG patients than in their cohort of controls (8). We suggest that the difference between our results may be due in part to an underestimation of the frequency of ASB10 variations in control subjects in the original report due to the different sizes of their POAG patient and control cohorts and due to the exclusion of common missense variations from their analysis. These ASB10 data are complex and subject to alternative interpretations; however, our data and observations suggest that non-synonymous ASB10 variations occur at the same frequency in POAG patients and control subjects.

MATERIALS AND METHODS

Patient cohorts

Patients were diagnosed with glaucoma if they had excavation of their optic nerve head with corresponding glaucomatous visual field loss in at least one eye. Glaucomatous cupping of the optic nerve was defined by a threshold of a cup-to-disc ratios of >0.7 with thinning of the neural rim, asymmetry of the optic nerve cup-to-disc ratio of >0.2 or photographic documentation of progressive loss of the neural rim. Patients were 40 years of age or older at diagnosis and had open iridocorneal angles on gonioscopy (angle greater than Shaffer grade II). Patients were also required to have an IOP of >21 mmHg on at least one occasion. Normal control subjects were a minimum of 50 years old and were examined and judged to have normal optic nerve head appearance and IOP of ≤21 mmHg by board-certified ophthalmologists. All POAG and control subjects were examined by clinicians at the University of Iowa Hospitals and Clinics and ascertained in Iowa. Written informed consent was given by all participants and the study was approved by the human subject review committee at the University of Iowa.

ASB10 mutation screening

DNA samples were prepared from peripheral blood samples extracted from patients in the clinic by standard procedures. The entire coding region of ASB10 was polymerase chain reaction (PCR) amplified using standard PCRs (primer sequence available on request). Amplified DNA was analyzed using either SSCP, HRM or DNA sequence analysis as described previously (3,6). Abnormal PCR products identified were then bi-directionally sequence verified using Applied Biosystems (ABI) model 3730 automated sequencer.

Statistical analysis

The cohort of POAG (n= 158) and control subjects (n= 82) from Iowa has 80% power to detect a mutation rate of 4.9% at a significance level of α = 0.05 when the frequency of mutations in controls is 0.1%. The allele frequency of DNA sequence variants in the ASB10 coding sequence that altered the encoded protein was compared between POAG patients and controls using Fisher's exact test with a threshold for significance of P < 0.05. Multiple measure adjustments were made to P-values for three comparisons (genotype frequency of non-synonymous coding sequence mutations, allele frequency of synonymous coding sequence variations and genotype frequency of synonymous coding sequence variations) using the Bonferroni correction.

Frequency analysis of ASB10 variants

The general population frequency of non-synonymous ASB10 variants was evaluated with the NHLBI's Exome Variation Server to search their database of the exomes from over 3500 Caucasian subjects (13). The subjects in this database are participants in prospective studies and have a population frequency of glaucoma.

SUPPLEMENTARY MATERIAL

Supplementary Material is available at HMG online.

FUNDING

This research was supported in part by the National Institutes of Health (NEI R01EY018825).