Evaluation of the β2-Adrenergic Receptor Gene as a Candidate Glaucoma Gene in 2 Ancestral Populations FREE

Glaucoma is an important worldwide public health problem.¹ Hence, the identification of clinically informative and inexpensive tests to improve glaucoma diagnostics and treatment outcomes is a high priority. Using glaucoma clinical phenotype ascertainment, Mendelian genetic approaches have led to the identification of 15 genes and 30 loci (http://www.ncbi.nlm.nih.gov/mapview/ using glaucoma as the search term). However, these monogenic forms of glaucoma are not common, and the search for glaucoma genes will increasingly require approaches designed to account for genetic complexity^{2 - 5} and contributions of environmental risk factors.

Another approach is based on the study of important phenotypes that contribute to the pathophysiology of glaucoma such as intraocular pressure (IOP),^{6 - 9} IOP fluctuation,^{10 - 12} and optic disc factors.^{13 - 14} The genetic contribution to IOP is supported by results of family,^{15 - 17} twin,^{18 - 20} and epidemiologic studies.^{13 - 14 ,21 - 24} The genome screening used in the Beaver Dam Eye Study²⁵ and the linkage used by Charlesworth et al¹⁷ show promise in identifying genes whose products play a role in IOP regulation.

Another potentially relevant approach is to identify genetic markers for variation in IOP response to glaucoma medical treatment as an application of pharmacogenomics.²⁶ Given our understanding of the signaling pathways that regulate aqueous humor dynamics, the polymorphic β₂-adrenergic receptor (ADRB2) gene is a logical choice for study because it is clear that the mechanism of action for β-adrenergic receptor antagonists or β-blockers is reducing aqueous flow^{27 - 31} by binding to β-adrenergic receptors in the ciliary body.^{32 - 40} The effects of ADRB2 single nucleotide polymorphisms (SNPs) (Figure 1) on pharmacology and physiology have been studied in asthma, hypertension, congestive heart failure, obesity, and acute coronary syndrome.^{41 - 43} Although the ADRB2 gene is not a susceptibility locus for these diseases,^{44 - 46} the gene is believed to modify the disease course owing to variations in sympathetic tone and response to β-adrenergic receptor treatments.⁴⁴ The effects of the common nonsynonymous SNPs on ADRB2 function have been studied by means of site-directed mutagenesis.⁴⁷ A change from the more commonly expressed or wild-type Arg16 to Gly16 enhances isoproterenol-mediated down-regulation. A change from Gln27 to Glu27 was associated with attenuated isoproterenol-mediated down-regulation when coexpressed with the Arg16. The less common change from Thr164 to Ile164 caused lower binding affinity to isoproterenol.

Based on this evidence in systemic adrenergic receptor–mediated physiology and pharmacology, we hypothesize that the polymorphic ADRB2 gene might modify glaucoma pathophysiology by contributing to IOP fluctuation and to variation in IOP response to β-blocker treatment. Given that investigation of the role of ADRB2 variants in pharmacological studies of glaucoma would be confounded if these variants also played a role in the disease process itself, we set out to determine whether the ADRB2 gene is a potential susceptibility locus for primary open-angle glaucoma (POAG).

Informed consent was obtained from all subjects according to study protocols approved by the institutional review boards at the University of Michigan, Duke University Medical Center, and Ghana Ministry of Health, Accra, all of which complied with the Declaration of Helsinki. The design was a comparative study with a phenotype assignment for subjects with POAG and control subjects. The phenotype was ascertained by results of the ophthalmologic examination on the basis of slitlamp biomicroscopy, IOP applanation findings, and results of gonioscopy and fundus evaluation including the optic disc. Subjects with POAG underwent visual field testing.

Eligibility criteria for subjects with POAG included being 35 years or older and having open filtration angles, IOP of greater than or equal to 22 mm Hg, glaucomatous optic neuropathy, and glaucomatous field loss. Eligibility criteria for controls included no evidence of glaucoma or ocular hypertension and no family history of POAG. Racial information was ascertained by self-report.

We isolated genomic DNA from blood using a commercially available kit (Puregene; Gentra Systems, Minneapolis, Minn) according to the manufacturer's instructions. The entire ADRB2 gene, including the promoter with the 5′ cistron leader, was amplified using sequence-specific primers A and I (Table 1) designed on the basis of the reference sequence (GenBank entry J02960). The polymerase chain reaction (PCR) analysis was conducted in a 50-μL volume using a 2-μL aliquot of the DNA template (50 ng/μL), 2 U of platinum Taq DNA polymerase (Invitrogen Corp, Carlsbad, Calif), 1.5mM magnesium chloride, 0.2mM of each deoxyribonucleotide triphosphate, 1× Q solution (Qiagen Inc, Valencia, Calif), 0.2μM primers (Table 1), 20mM Tris hydrochloride (pH, 8.4), and 50mM potassium chloride. The optimal PCR cycling settings for this reaction were 55°C for 30 cycles. The quality of the predicted 1.616-kilobase (kb) PCR products was evaluated using agarose gel electrophoresis.

Genotyping was performed by means of fluorescence-based sequencing on the amplified PCR gene product with a commercially available sequencing kit (BigDye ABI terminator; PerkinElmer, Boston, Mass) and semiautomated sequencing technology (Applied Biosystems, Foster City, Calif) using primers designed to span the intronless gene every 300 to 400 bases to ensure a readable sequence (Table 1). The amplified ADRB2 gene and the promoter were sequenced in the forward and reverse directions from an initial 275 subjects, and in the forward direction only in the remaining 308 subjects. Sequences were read at least twice independently without knowledge of the diagnosis and compared with the reference sequence (GenBank entry J02960). All polymorphisms were compared with published SNPs,^{48 - 50} the allelic variant list in the Online Mendelian Inheritance in Man database (http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=OMIM), and the National Center for Biotechnology Information database (http://www.ncbi.nlm.nih.gov/).

Given the linkage disequilibrium among the closely spaced SNPs of the ADRB2 gene, there are a limited number of haplotypes for the ADRB2 gene.^{44 - 45 ,51} Twelve haplotypes have been described for the ADRB2 SNPs using the nomenclature described by Drysdale et al.⁴⁵ The most common haplotypes were 2, 4, and 6. Haplotype 2 corresponds to the CGG SNPs in the β₂-adrenergic receptor upstream peptide, codon Gly16, and codon Glu27; haplotype 4, to the TAC SNPs in the β₂-adrenergic receptor upstream peptide, Arg16, and Gln27; and haplotype 6, to the TGC SNPs in the β₂-adrenergic receptor upstream peptide, Gly16, and Gln27.

To perform phasing in subjects who were double heterozygotes at the 2 common functional SNPs at codons 16 and 27, we used a combination of allele-specific PCR with the sense primer B for the Arg16 allele and the sense primer C for the Gly16 allele (Table 1), the antisense primer E, and restriction enzyme digestion (Figure 2). The PCR settings were conducted in a 50-μL volume using a 2-μL aliquot of a 10⁴ dilution of the amplified 1.616-kb DNA product as template, 2 U of Taq DNA polymerase (Promega Corp, Madison, Wis), 1.5mM magnesium chloride, 0.2mM of each deoxyribonucleotide triphosphate, 0.2μM primers, 10mM Tris hydrochloride (pH, 9.0), 0.1% Triton X-100, and 50mM potassium chloride. For reactions using primer B for the Arg16 allele, the annealing temperature was 70°C for 30 cycles. For primer C based on the Gly16 allele, the same annealing temperature was used for 25 cycles. An 8-μL aliquot of the expected 168-base pair (bp) product was treated with the restriction enzyme BbvI (New England Biolabs Inc, Beverly, Mass) at 2 U per reaction (of 50 μL) in the manufacturer's buffer system at 37°C for 16 hours. Because BbvI recognizes a restriction site that is 8 nucleotides downstream from the Gln27 allele, 2 fragments with sizes of 105 and 63 bp can be detected by gel electrophoresis (Figure 2). In contrast, BbvI does not cut in the presence of the Glu27 allele, so the 168-bp fragment remains intact (Figure 2). Using this simple, reproducible method, we were able to determine the haplotypes for subjects who were double heterozygous with the Arg16Gly and Gln27Glu alleles.

To evaluate ADRB2 as a possible POAG gene, the sample size was determined from power calculations using previously published ADRB2 allele frequencies.⁴⁸ If we posit that the homozygous Arg16 allele is predictive of a glaucoma phenotype, then a sample size of 528 subjects is needed, with 132 subjects in each of the 4 groups stratified by ancestry and disease. Alternatively, if we posit that the homozygous Gly16 allele is associated with glaucoma, then 4328 subjects are needed. The power calculations become more complex in consideration of haplotypes.

We analyzed the ADRB2 SNPs in our study population using several approaches. In the first approach, we examined the individual allele combinations for Arg16Arg, Arg16Gly, Gly16Gly, Gln27Gln, Gln27Glu, and Glu27Glu. The total allele frequencies were determined for Arg16, Gly16, Gln27, and Glu27. The resulting allele frequencies were compared between the POAG and control groups within the ancestry divisions, among the controls by ancestry, and among subjects with POAG by ancestry. Observations were statistically tested using the Fisher exact test, χ² test, and an examination of post hoc standardized Pearson residual.⁵² For the residuals, we used an absolute value of 3.0 as a cutoff for significance. The allele frequencies were tested to determine whether the Hardy-Weinberg equilibrium held for each group.⁵³ All calculations were performed in R version 2.3 statistical software.⁵⁴

In the second approach, we assembled the haplotype combination for each subject. Subjects who were double heterozygous at Arg16Gly and Gln27Glu underwent phasing using the method described in the preceding subsection. Haplotype combinations were counted and frequencies were determined for each of the 4 groups. Haplotype combinations were compared between the POAG and control groups within the ancestry groups, among the controls by ancestry, and among subjects with POAG by ancestry using the same statistical tests as described in the preceding paragraph.

A total of 583 subjects met eligibility criteria for this comparative association study stratified by ancestry (white vs black African) and disease (POAG vs control group). The black African ancestry group included 9 subjects with POAG from Ghana and 14 controls from Ghana. Among these 4 groups, there were 156 white subjects with POAG (91 women and 65 men; mean ± SD age, 56.5 ± 12.2 years [1 SD]), 143 subjects with POAG who were of black African ancestry (60 women and 83 men; mean ± SD age, 55.7 ± 11.6 years), 148 white controls (70 women and 78 men; mean ± SD age, 65.7 ± 14.9 years), and 136 controls of black African ancestry (61 women and 75 men; mean ± SD age, 57.8 ± 12.7 years).

In the analysis of the common ADRB2 alleles, we studied the SNP diplotypes encoding codons 16 and 27 and their frequencies (Figure 3 and Table 2). At codon 16, the SNP diplotypes encoded for homozygous Arg16Arg, heterozygous Arg16Gly, or homozygous Gly16Gly. At codon 27, the combinations were Gln27Gln, Gln27Glu, and Glu27Glu. When the total allele frequencies for codons 16 and 27 were examined, there were no differences comparing Arg16, Gly16, Gln27, and Glu27 between the subjects with POAG and controls within the ancestry groups using the Fisher exact test (Table 2). Results of tests for association indicate that the Gly16Gly combination is statistically significant (P = .003) in the comparison of white with black African subjects with POAG (Figure 3A and Table 2). All other ancestry comparisons for codon 16 allele combination frequencies were no different. Statistical testing verifies the observed differences (Figure 3B and Table 2) between white and black African subjects within the POAG (P<.001, χ² test) and control (P<.001, χ² test) groups with none of the frequencies, except for Gln27Glu in the POAG group, matching statistical expectations as determined by Pearson residual.

With the exception of the black African POAG group for codon 16 (P = .05), the common allele frequencies were consistent with Hardy-Weinberg equilibrium. This deviation may imply that patients referred to our tertiary care center, as a group, had a skewed ADRB2 genotype that was not observed in the black African controls.

For the other nonsynonymous SNPs at codons 34 and 164 (Figure 1), all subjects were homozygous for the common alleles encoding for Val34 and Thr164. We did not identify any subject who had SNPs encoding for homozygous Met34 or Ile164 or for heterozygous Val34Met or Thr164Ile.

In summary, our analysis of the individual common ADRB2 SNPs showed that there were no differences in allele combinations or total allele frequencies between the subjects with POAG and the controls within the ancestry groups. Thus, the common ADRB2 SNPs were not associated with POAG in our study population. However, there were ancestral differences. We found a statistically significant ancestral difference between the white and black African POAG groups that appears to be localized to the homozygous Gly16Gly allele (Figure 3A and Table 2). At codon 27 (Figure 3B), we found ancestral differences within the POAG and control groups. Among the subjects with POAG, all 3 combined genotypes encoding for Glu27Glu, Gln27Glu, and Gln27Gln were different between the white and black African groups. In controls, the differences appeared between homozygous Gln27Gln and Glu27Glu.

Based on these individual allele combinations, we assembled the ADRB2 haplotypes for each subject using the nomenclature described in the methods by Drysdale et al⁴⁵ (Figure 4 and Table 3). For subjects with compound heterozygous alleles at codons Arg16Gly and Gln27Glu, we phased these alleles using allele-specific PCR and restriction enzyme digestion (Figure 2) as described in the “Molecular Methods for Sequencing and Haplotyping” subsection of the “Methods” section. In 1 white subject with POAG, we identified a haplotype combination of C in the β₂-adrenergic receptor upstream peptide, A for Arg16, and G for Glu27 that had not been previously reported. Thus, for this group, there were a total of 155 subjects in this haplotype combination analysis in contrast to the 156 subjects in the SNP analysis. There were no significant differences in haplotype combinations 2, 4, and 6 between the subjects with POAG and the controls within each of the 2 ancestry groups. Thus, the ADRB2 gene was not associated with POAG in our study population.

However, there were ancestral differences between populations for certain haplotype combinations. Among the subjects with POAG, there were significant differences in haplotypes 2/2 (P<.001), 2/4 (P=.002), and 4/6 (P<.001) between the white and black African ancestry groups. The controls showed significant differences in the same haplotypes (2/2 [P=.001], 2/4 [P=.01], and 4/6 [P<.001]) when the white and black African ancestry groups were compared.

In our association study, the ADRB2 gene is not a susceptibility gene for POAG, whether analyzed by individual common SNPs or haplotypes in each of the 2 populations of white and black African ancestries. In addition, we confirm previously reported differences in the frequencies of the common ADRB2 SNPs between groups by different ancestry⁴⁸ and identify a novel haplotype not seen before.⁴⁵ Given these ancestral differences between the common SNPs, it is not surprising that we identified differences in the ADRB2 haplotypes frequencies by ancestry (Table 3 and Figure 4). Genetic studies have not previously identified the ADRB2 gene, localized to 5q31-q32, as a glaucoma locus. Thus, the role of the ADRB2 gene in IOP fluctuation and IOP response to β-blockers can be studied without the confounding effect as a POAG disease susceptibility gene.

A retrospective analysis of our subjects with POAG did not allow us to study the association between IOP response to β-blockers and ADRB2 haplotypes. Among the first 165 subjects with POAG (105 in the white and 60 in the black African ancestry groups) who underwent sequencing, 58 of those with available records had used β-blockers. Even with available records, problems with analysis of the retrospective data included the absence of baseline IOP, lack of information on the peak vs trough effect of β-blockers on IOP, the use of multiple classes of glaucoma medications, and the confounding effect of systemic β-blockers. Thus, the role of ADRB2 haplotypes on β-blocker IOP response variations could not be appropriately studied using the retrospective data available in our population.

Previous studies have investigated the role of ADRB2 SNPs on glaucoma susceptibility and IOP. Güngör et al⁵⁵ reported no differences in the common ADRB2 allele frequencies between 30 cases of congenital glaucoma, 105 cases of POAG, and 92 controls in a Turkish population. Güngör et al⁵⁶ also investigated the potential effect of ADRB2 SNPs on IOP in 19 healthy young men with a mean ± SD age of 22.6 ± 2.8 years. They found that recovery from exercise-induced lowering of the IOP was prolonged in those who were homozygous for the Gly16 allele compared with those who were heterozygous for the Arg16Gly allele. The implication of their study is unclear because of the small sample size, the unclear role of β₂-adrenergic receptors on exercise-induced lowering of the IOP, the use of individual SNPs as opposed to haplotypes, and the study design in young subjects.

In another study, Fuchsjager-Mayrl et al⁵⁷ screened 270 healthy subjects and identified 24 subjects with homozygous combinations for Arg16/Gln27, 18 with homozygous combinations for Gly16/Gln27, and 47 with homozygous combinations for Gly16/Glu27. All 3 groups showed a comparable IOP response to 0.5% timolol, which led them to conclude that factors other than ADRB2 SNPs caused the “intersubject variability seen with timolol in glaucoma subjects.”⁵⁷ However, this conclusion may be premature because their findings in young healthy men should not be extrapolated to explain pharmacodynamic variation in an older population with glaucoma, and they did not design a clinical pharmacology study for efficacy³⁰ to measure peak and trough responses to timolol.⁵⁸

More recently, Inagaki et al⁵⁹ reported that the Gly16 allele was associated with a younger age at diagnosis, and that the Glu27 allele was associated with higher IOP at diagnosis in Japanese subjects with open-angle glaucoma compared with controls. They suggest that these ADRB2 polymorphisms may influence the pathophysiology of glaucoma in Japanese subjects.

The ADRB2 gene was not a POAG susceptibility gene in our study population. Thus, we can focus on the role of the ADRB2 SNPs and haplotypes in glaucoma without the confounding effects as a glaucoma disease susceptibility locus. Future association studies may be undertaken to determine whether the ADRB2 gene modifies the pathophysiology of glaucoma by contributing to the known risk factor of IOP fluctuation.^{10 - 12} The role of this polymorphic gene and the roles of other biologically relevant polymorphic genes involved in the regulation of aqueous humor dynamics may be studied to determine their relative contribution to IOP fluctuation. Furthermore, pharmacogenomic studies may be designed to study the potential association between the ADRB2 haplotypes and IOP response to β-blockers in patients with POAG. Such studies will lead to the identification of clinically informative and inexpensive tests to improve glaucoma diagnostics and treatment outcomes that will lead to personalized treatment for glaucoma.²⁶

Correspondence: Sayoko E. Moroi, MD, PhD, Department of Ophthalmology and Visual Sciences, University of Michigan, 1000 Wall St, Ann Arbor, MI 48105 (smoroi@umich.edu).

Submitted for Publication: June 30, 2006; final revision received August 23, 2006; accepted August 25, 2006.

Financial Disclosure: None reported.

Funding/Support: This study was supported by grants EY00353 (Dr Moroi), EY11671 (Dr Richards), EY015543 (Dr Allingham), and EY014939 (Dr Allingham) from the National Institutes of Health (NIH); University of Michigan Vision Core grant EY07003 from the NIH; Midwest Eye-Banks and Transplantation Center (Dr Moroi); a Career Development Award (Dr Moroi) and an unrestricted departmental research grant (Dr Musch) from Research to Prevent Blindness; the University of Michigan Office of the Vice President for Research (Dr Moroi); and the Glaucoma Research Foundation (Dr Moroi).

Acknowledgment: We thank Samir Shah, MD, and Neetaben Patel, MD, for extensive medical chart reviews on some of these subjects.