Glucocorticoid Receptor Gene Variant in the 3′ Untranslated Region Is Associated with Multiple Measures of Blood Pressure

Abstract

Context: The glucocorticoid receptor (GR) is a key hormone in the hypothalamus-pituitary-adrenal axis that regulates many pathways including blood pressure homeostasis. Thus, GR gene variation may influence interindividual differences in blood pressure in human populations.

Objective: We resequenced individual GR alleles for comprehensive discovery of GR variants and their chromosomal phase in three major American ethnic groups. We examined the influence of GR variants on blood pressure in large numbers of families using family-based association methods.

Design and Participants: For association studies, we genotyped GR variants in family members from the Genetic Epidemiology Network of Arteriopathy (GENOA) study that were measured for multiple blood pressure traits. The GENOA families consisted of African-Americans, Mexican-Americans, and European-Americans.

Main Measurements: The blood pressure measurements for association studies included systolic blood pressure, diastolic blood pressure, mean arterial pressure, and pulse pressure.

Results: Single-nucleotide polymorphisms (SNPs) identified by resequencing were tested for associations with blood pressure measures in GENOA families. Analysis of individual SNPs identified significant associations of rs6198 A/G in exon 9β with multiple blood pressure measures in European-Americans. Analysis of GR haplotypes found significant associations of a haplotype that is distinguished by rs6198 A/G.

Conclusions: Significant associations of blood pressure with rs6198 A/G likely reflect allelic effects on GR signaling. This SNP disrupts a 3′ untranslated region sequence element in exon 9β that destabilizes mRNA, resulting in increased production of the inactive GRβ isoform. Excess heterodimerization with the active GRα isoform may reduce GR signaling with subsequent physiological effects on blood pressure regulation.

Resequencing of the glucocorticoid receptor gene identified extensive variability, but limited haplotype diversity among three American ethnic groups. Family-based studies identified associations between a functional polymorphism in the 3’ untranslated region with multiple measures of blood pressure.

Essential hypertension (HT) is a complex disease that is influenced by the action of many genes as well as interactions among genetic and environmental determinants. Comprehensive studies of interindividual variability in biological candidate genes selected for their etiological roles in hypertension may help finding the genetic components of this disease. The glucocorticoid receptor gene (GR or NR3C1) on chromosome 5q encodes a nuclear transcription factor that mediates glucocorticoid signaling. GR is a central component of the hypothalamus-pituitary-adrenal axis, with major effects on numerous physiological pathways including blood pressure homeostasis. Hypertension is a major clinical feature of Cushing’s disease that is characterized by abnormal elevation of glucocorticoids mostly due to pituitary adenoma (1). Glucocorticoid-mediated pathways are involved in regulation of vascular tone by modulating vasoconstrictors (2,3) and vasodilators (4,5,6,7) via GR signaling.

Previous studies of selected GR polymorphisms have reported associations with hypertension and related physiological measures (8,9,10,11). However, few associations have proven consistent across multiple studies, in part due to interrogation of only one or a small number of GR variants, variation in patterns of linkage disequilibrium (LD) in different populations, and hidden population substructure (12). In this study, we sought to address these common problems in association studies by extensive resequencing to comprehensively characterize GR sequence variation in three major American ethnic groups including African-Americans (AA), Mexican-Americans (MA), and European-Americans (EA). We resequenced GR in hybrid cell lines that are monosomic for human chromosome 5 to unequivocally determine molecular haplotypes. To ensure adequate sample size, we genotyped single-nucleotide polymorphisms (SNPs) discovered by resequencing in the Genetic Epidemiology Network of Arteriopathy (GENOA) study that includes 3204 individuals from 1351 sibships (AA, MA, and EA). To accurately characterize LD patterns for each race, we used our genotypic data from the GENOA sibships to identify bins of correlated SNPs spanning the GR locus. To avoid hidden population substructure, we used transmission-based methods for association analyses of blood pressure measures using individual SNPs as well as haplotypes across the GR locus.

Subjects and Methods

Resequencing of GR (NR3C1) in hybrid cell lines monosomic for human chromosome 5

As previously described, we developed mouse-human somatic cell lines that are monosomic for human chromosomes for direct determination of chromosomal phase in separated GR alleles (molecular haplotypes) (13). Briefly, lymphoblastoid cell lines from representative individuals for three major American ethnic groups (20 AA from Atlanta, GA, 18 MA from Starr County TX, and 19 EA from Rochester, MN) were electrofused with a mouse host cell line (embryonic fibroblasts). The resulting fusion cell products retain all mouse chromosomes (geneticin selection) but randomly lose human chromosomes except the X chromosome (hypoxanthine-aminopterin-thymidine selection) (14,15). The fusion cell lines that contained single copies of chromosome 5 were identified by genotyping microsatellite markers that were heterozygous in the original individuals. For each chromosome 5 monosomic cell line (two per individual for a total of 114 cell lines), DNA was extracted for amplification of GR functional regions comprising approximately 30 kb of the entire human GR gene (158 kb). Functional regions included all exons (plus at least 0.5 kb of flanking introns), 5′ promoter regions (1.3 kb), and 3′ flanking regions (1.7 kb). A total of 31 overlapping PCR fragments (1–3 kb) were sequenced on both strand using an automated capillary DNA sequencer (ABI Prism 3730XL DNA analyzer; Applied Biosystems, Foster City, CA). Fluorescent sequence traces were assembled and aligned to identify GR variants (Lasergene sequence analysis software; DNASTAR, Madison WI).

GENOA study subjects and GR genotyping

GENOA consists of sibships representing three major American ethnic groups (1228 AA from 553 sibships from Jackson MS, 954 MA from 363 sibships from Starr County TX, and 1,022 EA from 435 sibships from Rochester, MN). For AA and EA, sibships were recruited that contained at least two siblings with essential hypertension diagnosed before the age of 60 yr. Hypertensive status was based on systolic blood pressure (SBP) greater than 140 mm Hg or diastolic blood pressure (DBP) greater than 90 mm Hg, or taking HT medication. For MA, sibships were recruited that contained at least two siblings with type 2 diabetes (16). GENOA provided demographic and pedigree information, medical histories, and various biological measures and biomarkers for study participants. All GENOA participants provided written informed consents after approval by appropriate institutional review boards for the protection of human subjects.

We used lymphocyte DNA samples provided by GENOA for automated multiplexed genotyping with the SNPlex platform (Applied Biosystems) that is based on oligonucleotide ligation assays (17). A total of 41 GR variants identified by resequencing in monosomic cell lines were selected for genotyping in GENOA sibships based on minor allele frequencies (MAFs; >0.1) and potential functionality (coding and promoter regions). Blind duplicate DNA samples with random plate positions were included for quality control. Fluorescent traces from capillary electrophoresis of SNPlex products (ABI Prism 3730XL DNA analyzer) were used for automated allele designation by GeneMapper software (version 3.7; Applied Biosystems, Foster City, CA).

Statistical analyses

Allele frequencies for GR variants were computed by allele counting, and Hardy-Weinberg equilibrium was evaluated using Fisher’s exact P value within each population. Correlation among variants was estimated using Tagger (Haploview) (18) with a threshold of r² > 0.8. Haplotypes for GENOA association studies were constructed by combining individual tagSNPs from each correlation bin as well as SNPs with potential functionality, high frequency, or that showed significant associations with HT-related traits in previous studies. For association studies, quantitative blood pressure (BP) traits consisted of the average of three measurements for each participant including SBP, DBP, pulse pressure (PP) and mean arterial pressure (MAP). The average BP measures were log normalized and adjusted for significant covariates including gender, age, body mass index, total cholesterol, triglycerides, creatinine, smoking, and alcohol. We also adjusted BP measures for glucose and insulin levels, of particular importance because GENOA ascertainment resulted in enrichment for type 2 diabetes in MA families. In addition, we adjusted for antihypertensive medication due to the enrichment of hypertensive subjects in EA and AA families. Residuals were used for association analysis. Family-based association tests (FBATs) that allow for missing parental genotypes were used for association analyses with an additive model for genetic effects (19). Haplotype association tests were performed by haplotype-based association tests (20). Empirical P values were calculated by permutation (100,000 times) as part of FBAT analysis. The P values for individual SNP associations were corrected for multiple testing based on false discovery rates (FDRs) (21).

Results

Characterization of GR (NR3C1) variation by resequencing in monosomic hybrid cell lines

Figure 1 shows our strategy for resequencing functional regions of the GR gene (NR3C1). Alignment of GR sequences from the monosomic hybrid cell lines yielded 115 variants, including 95 SNPs, 19 biallelic insertion/deletions (indels), and one microsatellite. Table 1 shows the identity and position of each variant as well as MAFs in the ethnic groups. The variants identified by resequencing include previously known ones from other studies (22,23). Overall, GR variability was much higher in AA (93 variants vs. 49 in MA and 41 in EA), 58 of which (62%) were found only in AA. Two SNPs were identified in GR promoter sequences that altered sp1 transcription factor binding sites including 30900 A/G as well as 31827 C/T that has not been previously reported (23,24,25). Two nonsynonymous SNPs (rs6192 Phe65Val and rs6195 Asn363Ser) were identified in exon 2. Supplemental Table 1, published as supplemental data on The Endocrine Society’s Journals Online Web site at http://jcem.endojournals.org, shows molecular haplotypes identified by resequencing of individual GR alleles in the monosomic cell lines. Overall, a total of 56 molecular haplotypes were observed, 14 of which were observed at least twice. GR haplotype variability was higher in AA (32 haplotypes vs. 21 in MA and 14 in EA). However, the three most frequent haplotypes were shared by all three ethnic groups.

Figure 1.

GR gene resequencing strategy. The resequenced genomic area and excluded intronic gaps are indicated on the schematic drawing of the gene. Eleven exons with splice variants of exon 1A, exon 3 resulting in GRγ, and exon 9 producing GRα and GRβ are drawn above the map. Open blocks indicate UTRs and black blocks indicate coding regions.

Table 1.

List of 115 GR polymorphisms from resequencing 114 chromosome 5 monosomics

Positiona	dbSNPb	Gene regionc	Major/minord	Minor allele frequency
				Monosomics			GENOA
				EA	AA	MA	EA	AA	MA
−1225		Distal promoter	A/G		0.103			0.111	0.004
−921	Hawkins 04	Distal promoter	G/A			0.111	0.054	0.015	0.203
−645	Hawkins 04	Distal promoter	+/−		0.103
−400		Distal promoter	T/C		0.025
−307		Distal promoter	G/A			0.028
−226	rs6868190	Distal promoter	C/T		0.125			0.109	0.005
1015	Hawkins 04	Intron 1A	G/A		0.025	0.028	0.003	0.032	0.004
1050	Hawkins 04	Intron 1A	T/C		0.150
1439		Intron 1A	G/A			0.028
30330	rs35722527	Proximal promoter	G/C		0.025
30464	rs10482603	Proximal promoter	G/A			0.057
30836		Proximal promoter	−/+		0.025
30838		Proximal promoter	G/T		0.025
30900	Koyano 05	Proximal promoter (sp1)	A/G		0.025
30948	rs3806855	Proximal promoter	T/G	0.132	0.175	0.057
30950	rs3806854	Proximal promoter	T/C	0.132	0.175	0.057
30963	Koyano 05	Proximal promoter	+/−		0.025
31122		Exon 1B, 5′ UTR	C/G		0.025
31128	rs5871845	Exon 1B, 5′ UTR	−/+	0.026	0.025
31475	rs10482604	Proximal promoter	A/G		0.025		0.001	0.021
31556	rs10482605	Proximal promoter	T/C	0.054	0.125	0.083	0.198	0.110	0.085
31807		Proximal promoter	+/−		0.025
31809	rs10482606	Proximal promoter	T/C	0.053
31827		Exon 1C, 5′ UTR (sp1)	C/T		0.025
31990	rs10482609	Exon 1C, 5′ UTR	T/G			0.056	0.009	0.001	0.024
32033		Exon 1C, 5′ UTR	G/A	0.026
32062	rs10482610	Exon 1C, 5′ UTR	+/−		0.100
32064	rs10482611	Exon 1C, 5′ UTR	C/T		0.100
32190	Koyano 05	Exon 1C, 5′ UTR	10/ (9,11) MS		0.125
32506	rs4406157	Intron 1C	C/A		0.100
32675	rs10482614	Intron 1C	G/A	0.132	0.175	0.057
32877		Intron 1C	T/G	0.053			0.007	0.004
33045	rs9324923	Intron 1C	C/T		0.025
33198		Intron 1C	A/G			0.028
33227	rs9324922	Intron 1C	C/A		0.025
33510	rs10482616	Intron 1C	G/A	0.132	0.225	0.222	0.141	0.217	0.260
33909	rs10482617	Intron 1C	A/G		0.026
34232		Intron 1C	C/T			0.111			0.007
34380	rs4634384	Intron 1C	A/G	0.500	0.525	0.278
34382	rs4682314	Intron 1C	T/G	0.500	0.525	0.278
34545	rs10482620	Intron 1C	C/T		0.026			0.018
34865	rs6192	Exon 2	T/G, Phe>Val		0.026			0.021
35296		Exon 2	G/A		0.026			0.003
35551	rs10482622	Exon 2	G/A		0.079
35760	rs6195	Exon 2	A/G, Asn>Ser	0.079	0.026	0.028	0.042	0.006	0.025
36188		Intron 2	+/−		0.026
36502	rs41423247	Intron 2	C/G, BclI RFLP	0.447	0.225	0.194	0.388	0.252	0.305
120776	rs10482665	Intron 2	T/C	0.053	0.125	0.083	0.190	0.109	0.081
121098		Intron 2	+/−	0.054
121099		Intron 2	A/C		0.026
121231	rs4986593	Intron 2	T/C	0.316	0.050	0.139	0.240	0.057	0.213
121283		Intron 2	G/A	0.026
121621	rs10482667	Intron 3	A/G	0.053			0.024	0.003	0.009
121679	rs852979	Intron 3	A/G	0.184	0.308	0.143	0.340	0.299	0.189
121933		Intron 3	G/A			0.028
122045	rs10482669	Intron 3	A/C			0.056	0.002	0.007	0.146
122115		Intron 3	C/T			0.028
122544	rs10482672	Intron 3	C/T	0.139	0.231	0.176	0.140	0.214	0.127
123324		Intron 3	G/T	0.056	0.125	0.083	0.190	0.109	0.081
(Continued)

Table 1A.

(Continued)

Positiona	dbSNPb	Gene regionc	Major/minord	Minor allele frequency
				Monosomics			GENOA
				EA	AA	MA	EA	AA	MA
123336		Intron 3	C/T		0.025
123559		Intron 3	+/−	0.054	0.125	0.083
123693		Intron 3	G/A			0.029
124379		Intron 3	T/C		0.025
124481	rs852978	Intron 3	A/G	0.135	0.175	0.056	0.171	0.216	0.108
124578	rs9654496	Intron 3	C/T		0.025
124682		Intron 3	T/G		0.025
124698		Intron 3	G/A			0.111
125046	rs10482674	Intron 3	T/C		0.051			0.021
125315		Exon 4	A/G		0.026			0.002
125772		Intron 4	+/−			0.028
125860	rs10214015	Intron 4	G/A		0.025
125864		Intron 4	G/T		0.025
133990	rs10482675	Intron 4	−/+	0.342	0.100	0.472
134350		Intron 4	+/−		0.025
134475		Intron 4	A/G		0.025
134733	rs6188	Intron 4	G/T	0.184	0.350	0.139	0.338	0.326	0.181
135116		Intron 5	A/T		0.025
135654	rs10482681	Intron 5	T/G	0.027
135680	rs10482682	Intron 5	G/A	0.378	0.175	0.222	0.403	0.153	0.283
135842		Intron 5	C/T		0.025
135870		Intron 5	G/A		0.025
135925		Intron 5	C/G	0.316	0.050	0.139
136078	rs2918420	Intron 5	G/A		0.026
137297	rs9324914	Intron 6	G/A		0.025
137375	rs9324913	Intron 6	A/G		0.025
137410		Intron 6	G/A		0.025
138069	rs10482686	Intron 6	A/T		0.026
138238		Intron 6	−/+		0.026
138415		Intron 6	+/−		0.026
138606		Intron 6	C/T		0.026
138607	rs2963145	Intron 6	G/A		0.103
138673	rs258814	Intron 6	C/A	0.189	0.308	0.111
138893	rs2307674	Intron 6	+/−	0.135	0.179	0.056
139126	rs10482688	Intron 6	+/−	0.135	0.026	0.056
139254	rs10482689	Intron 6	G/A	0.053	0.132	0.088	0.189	0.109	0.080
140387	rs258813	Intron 7	C/T	0.184	0.350	0.139	0.339	0.326	0.181
152154	rs9324912	Intron 7	T/C		0.025
152585		Intron 7	T/G		0.025
152712	rs10482703	Intron 7	−/+			0.028
152797	rs258751	Exon 8	C/T		0.125		0.001	0.147	0.004
153188	rs258750	Intron 8	T/C	0.184	0.350	0.139	0.344	0.327	0.192
153235	rs9324911	Intron 8	C/G		0.025
153587	rs6196	Exon 9α	T/C	0.132	0.175	0.056	0.175	0.216	0.120
154033	rs10482707	Exon 9α, 3′ UTR	−/+		0.025
154459		Exon 9α, 3′ UTR	C/A		0.025		0.001	0.011
155503	rs10043662	Exon 9α, 3′ UTR	A/G		0.025			0.021
155811	rs6193	Exon 9α, 3′ UTR	A/G		0.075			0.067
156921	rs6191	Exon 9β, 3′ UTR	G/T	0.500	0.525	0.278	0.450	0.442	0.377
157456	rs6198	Exon 9β, 3′ UTR	A/G, AUUUA motif	0.053	0.025	0.083	0.190	0.066	0.074
157865	rs17209237	3′ Flank	T/C	0.389	0.075	0.139	0.262	0.061	0.225
158056	rs17287758	3′ Flank	C/T	0.056	0.125	0.083	0.189	0.109	0.078
158264	rs258747	3′ Flank	T/C	0.500	0.500	0.278	0.451	0.424	0.381
158571		3′ Flank	−/+			0.056
158623	rs258748	3′ Flank	G/C	0.111	0.175	0.056	0.171	0.214	0.105
158979		3′ Flank	T/C		0.025

The poly-A-microsatellite polymorphism is shown as MS. 

^aEach SNP is numbered from the first base pair of exon 1A (human sequence build: NT_029289.10). The SNPs genotyped in GENOA are underlined. 

^bPolymorphisms without dbSNP rs-numbers or previous references are novel ones discovered from this study. 

^c5′ flanking regions for exon 1A and exon 1B/1C are designated as distal and proximal promoter regions, respectively. 

^dMajor/minor allele designation was based on overall allele frequencies from the combined ethnic groups. 

GR single SNP associations with BP measures in GENOA families

For association studies, we selected 41 SNPs identified by resequencing in the monosomic cell lines for genotyping in GENOA families. We included SNPs with high frequencies (MAF >0.1), potential functionality (promoter regions, nonsynonymous SNPs in coding regions), and previously reported associations with HT and related phenotypes. Table 1 shows the identity, gene position, and MAFs for the three ethnic groups. Figure 2 shows correlation patterns for GR variants determined by genotyping in GENOA. We identified a total of eight multi-SNP correlation bins (r² > 0.8) (numbered 1–8), as well as seven uncorrelated SNPs (numbered 9–15).

Figure 2.

Allelic correlation pattern and SNPs used for haplotype analysis. Twenty-four SNPs that constitute correlation bins (r² > 0.8) are plotted (numbered 1–8). SNPs within each bin are connected with a line. Solid circles are tag SNPs and open circles are correlated SNPs. The open circles with asterisks are extra SNPs needed to tag bin 2 in MA and bin 6 in AA. Seven uncorrelated SNPs (solid triangles numbered 9–15) that were included to construct GENOA haplotypes are also shown (SNPs with potential functionality and/or high MAF).

Supplemental Table 2 shows the number of GENOA subjects and families for each racial group as well as other general characteristics including anthropometrics, BP-related measurements, HT status, and various plasma biomarkers. The AA and EA groups included higher numbers of subjects with hypertension due to ascertainment of sibships with at least two hypertensive subjects (16). Figure 3 shows the results of single SNP association tests for SBP, DBP, MAP, and PP for each racial group. Table 2 shows the P values and Z-scores for all SNPs that were significantly associated with BP traits (P < 0.05). In AA, rs10482674 T/C (intron 3) showed associations with DBP and MAP. In EA, rs10482682 G/A (intron 5) showed association with PP and SBP. We found associations with PP in EA for rs6191 G/T and rs258747 T/C, two correlated SNPs that comprise bin 5 (Fig. 3). The strongest and most consistent associations across the BP traits were found in EA for two correlated SNPs (bin 6) including rs6198 A/G and rs17287758 C/T that showed associations with DBP, SBP, and MAP. Three additional correlated SNPs in bin 6 also showed associations with SBP and MAP (rs10482665 T/C, 123324 G/T, and rs10482689 G/A).

Figure 3.

Results of single SNP association tests in GENOA for 41 GR SNPs. The solid horizontal lines indicate significance levels of 0.05 and 0.01 (empirical −log10 P values). SNPs with informative families less than 10 were not tested. SNPs with P < 0.05 are connected to their correlated SNPs with a dashed line and their bin is labeled. EA, black square; AA, gray diamond; MA, open triangle.

Table 2.

FBAT single SNP analysis results (P < 0.05, additive model)

Trait	Race	Bin	SNPa	Allele	Frequency	Families, nb	Z-scorec	Pd	P (FDR)e
PP	EA		rs10482682	A	0.399	177	−2.146	0.033	0.264
	EA	5	rs6191	G	0.465	186	2.374	0.016	0.193
	EA	5	rs258747	T	0.467	184	2.561	0.011	0.262
DBP	AA		rs10482674	C	0.023	14	2.313	0.020	0.639
	EA	6	rs6198	G	0.190	110	−1.943	0.044	1.000
	EA	6	rs17287758	T	0.190	110	−1.943	0.051	0.615
MAP	AA		rs10482674	C	0.023	14	1.991	0.043	1.000
	EA	6	rs10482665	C	0.191	111	−2.489	0.013	0.077
	EA	6	123324	T	0.191	109	−2.467	0.013	0.102
	EA	6	rs10482689	A	0.190	110	−2.462	0.015	0.074
	EA	6	rs6198	G	0.190	109	−2.613	0.010	0.115
	EA	6	rs17287758	T	0.190	109	−2.613	0.009	0.206
SBP	EA	6	rs10482665	C	0.191	111	−2.875	0.004	0.029
	EA	6	123324	T	0.191	109	−2.822	0.005	0.027
	EA		rs10482682	A	0.399	177	−1.984	0.044	0.175
	EA	6	rs10482689	A	0.190	110	−2.789	0.005	0.025
	EA	6	rs6198	G	0.190	109	−2.931	0.003	0.037
	EA	6	rs17287758	T	0.190	109	−2.931	0.003	0.069

^aNovel SNPs are indicated with genomic position from the first base pair of exon 1A. 

^bFamilies, n indicate the number of informative families used by FBAT. 

^cThe sign of the Z-score indicates whether the allele increases (+) or decreases (−) blood pressure. 

^dP values were 100,000 times Monte-Carlo permuted. 

^eP values were adjusted for multiple testing using FDR. 

GR SNP haplotype associations with BP measures in GENOA families

To construct GR haplotypes, we chose tag SNPs that represent correlation bins determined from GENOA genotypes (bins numbered 1–8). We included additional SNPs based on potential functionality, high frequencies (MAF >0.1), and previously reported associations with HT and related phenotypes (SNPs numbered 9–15). As a result 10, 13, and 12 SNPs were chosen to comprise haplotypes in EA, AA, and MA, respectively (Fig. 2). Although we did not identify significant associations in AA and MA, we found several associations of GR haplotypes with BP measures in EA (Fig. 4). Haplotype 2 (GAGCCGATTA) was significantly associated with PP (P = 0.025, Z = −2.20). Haplotype 3 (GACTCTATTG) was associated with multiple BP measures including SBP (P = 0.008, Z = −2.62), DBP (P = 0.055, Z = −1.88), and MAP (P = 0.016, Z = −2.41). Haplotype 3 (Fig. 4) exactly corresponds to molecular haplotype 5 as determined by direct sequencing of GR alleles in the monosomic cell lines (supplemental Table 1). Haplotype 3 is distinguished by rs6198 A/G in the 3′ untranslated region (UTR) (bin 6) that also showed associations with multiple BP measures in single SNP analysis (Fig. 3 and Table 2). In haplotype analyses that address multiple testing, we found that global haplotype tests were significant in EA for SBP (P = 0.0270), DBP (P = 0.0413), and MAP (P = 0.0343). Haplotype-specific tests did not show significance, although haplotype 3 (lowest individual P value) approached significance for SBP (P = 0.0576).

Figure 4.

Results of GR haplotype association tests in GENOA EA. The 10 SNPs that comprise GR haplotypes in EA for association tests are labeled below the table, and their approximate locations are shown on the gene map (open blocks indicate UTRs; black blocks indicate coding regions). Significant P values (P < 0.05) for haplotype associations are underlined in the table.

Discussion

Resequencing revealed extensive variability across the GR locus (NR3C1) including 115 variants among three major ethnic groups (AA, MA, and EA) (Table 1). Despite our focus on GR functional regions (exons and promoter regions), the majority of SNPs (61%) were found in intronic sequences that immediately flanked the sequenced exons. We found only two nonsynonymous substitutions (Phe65Val and Asn363Ser) with relatively rare frequencies, reflecting evolutionary conservation of GR protein structure and function. By resequencing separated GR alleles in hybrid cell lines that contain a single copy of chromosome 5 (monosomics), we unambiguously identified common GR haplotypes at the molecular level (supplemental Table 1). Despite extensive variability, GR exhibits remarkably low haplotype diversity with only 14 common haplotypes (MAF >0.05) from 115 variants, and the ethnic groups share the three most common haplotypes that account for 54% of the combined common haplotypes. Characterization of allelic correlation patterns among SNPs reveals overlapping bins that span large regions (up to 128 kb) of the GR locus. Shared haplotype diversity and extensive LDs across this large locus likely reflect the influence of common selective pressures among the ethnic groups. GR bins 2–8 are located downstream from exon 1B, likely reflecting the presence of a recombinational hot spot between exons 1A and 1B (HapMap on NCBI B35) (Fig. 2).

We selected SNPs from GR resequencing for genotyping in GENOA families to investigate effects on blood pressure traits. The strongest and most consistent associations were for correlated SNPs that comprise bin 6 in EA with multiple BP measures including MAP, SBP, and DBP (Fig. 3 and Table 2). The SNP in bin 6 that showed the strongest evidence for association was rs6198 A/G in the 3′ UTR of exon 9β that remained significant after accounting for multiple testing (FDR adjusted, P = 0.037). In addition, haplotype 3 that is distinguished by rs6198 A/G showed a strong association with SBP in EA (P = 0.008). In subjects from informative families, homozygotes for the common rs6198 allele (AA) had a mean adjusted SBP of 138 mm Hg (n = 155) compared with 133 mm Hg in AG heterozygotes (n = 161), and 122 mm Hg in GG homozygotes (n = 14) (supplemental figure, published as supplemental data on The Endocrine Society’s Journals Online Web site at http://jcem.endojournals.org). Thus, GG homozygotes had an 11% reduction in SBP relative to AA homozygotes (difference of 16 mm Hg, P = 0.006).

Our study included other GR variants that have been used in previous association studies of hypertension and BP measures. For example, the nonsynonymous rs6195 A/G polymorphism (Asn363Ser) introduces an additional phosphorylation site in the GR domain that controls transcriptional activation, and a previous study reported association with increased glucocorticoid sensitivity (26,27). Despite the potential functionality of Asn363Ser, most previous studies have not found associations with BP traits (8,28,29,30). However, a recent study reported association of Asn363Ser with steroid-induced ocular hypertension (31). We did not find associations of Asn363Ser with blood pressure measures in this study in any of the three ethnic groups.

Human GR is found in two major isoforms [GRα and GRβ that share exons 1–8 but contain different forms of the terminal exon (exon 9α or exon 9β) due to alternative splicing] (32). The GRα isoform binds glucocorticoids in the cytoplasm and moves to the nucleus to regulate transcription of downstream genes via GR response elements. In contrast, the GRβ isoform does not bind glucocorticoids or regulate downstream gene expression. Previous studies have shown that the GRβ isoform can inhibit glucocorticoid signaling in a dominant-negative fashion by heterodimerization with GRα (33,34). Increased GRβ production has been associated with glucocorticoid resistance in immune system diseases including asthma and acute lymphoblastic leukemia (35,36). Interestingly, the rs6198 A/G polymorphism that shows strong associations with BP traits in EA is located in exon 9β, disrupting a consensus sequence element in the 3′ UTR that is known to destabilize mRNA. Therefore, the A>G substitution (AUUUA>GUUUA) may lead to increased stabilization of GRβ mRNA, resulting in increased production of the GRβ isoform. In vitro studies of rs6198 A/G reported increased stability of GRβ mRNA for the G allele (GUUUA) as well as increased protein production in transfected COS-1 cells (37,38). These findings offer support for rs6198 A/G as the functional variant responsible for associations of the correlated SNPs in bin 6 by influencing blood pressure via physiological pathways related to glucocorticoid mediated gene expression.

Previous studies of GR variants have also found associations of rs6198 A/G with disease-related phenotypes. Derijk et al. (37) reported associations of the rs6198 G allele (GUUUA) with rheumatoid arthritis and systemic lupus erythematosus, perhaps due to enhanced proinflammatory response from reduced glucocorticoid sensitivity mediated by increased GRβ levels. Similarly, van den Akker et al. (39) reported associations of the rs6198 G allele with increased Staphylococcus aureus nasal carriage, presumably also due to enhanced proinflammatory response due to increased levels of GRβ that may act as a dominant-negative inhibitor of GRα activities. In addition, a recent report by van den Akker et al. (40) found that homozygotes for a GR haplotype distinguished by the rs6198 G allele was associated with increased risk of myocardial infarction and coronary heart disease in the Dutch population, likely due to enhancement of proinflammatory response associated with reduced glucocorticoid sensitivity.

Because these rs6198 A/G associations were detected in EA families enriched for hypertensives, our results should be of particular interest to individuals in at-risk populations of EA origin with elevated BPs. It is not clear why these associations with this potentially functional SNP are not found in AA and MA. Future studies in other cohorts will be necessary for both confirmation of rs6198 A/G associations with BP measures in EA and the lack of associations in AA and MA. In addition, further studies will directly examine carriers of rs6198 A/G for differences in GRβ mRNA stability as well as levels of GRβ isoforms. If these findings prove robust, future studies will need to include other interacting genes and environmental factors that may explain different effects of GR variants among different ethnic groups.