Dopamine D1 receptor (DRD1) genetic polymorphism: pleiotropic effects on heritable renal traits

Abstract

Because dopamine D₁ receptors (DRD1) influence renal sodium transport and vascular hemodynamics, we examined whether genetic polymorphisms play a role in renal function. We conducted polymorphism discovery across the DRD1 open reading frame and its 5′-UTR and then performed association studies with estimated glomerular filtration rate (eGFR), plasma creatinine (pCr), and fractional excretion of uric acid (FeUA). We used a twin/family group of 428 subjects from 195 families and a replication cohort of 677 patients from the Kaiser health-care organization sampled from the lower percentiles of diastolic blood pressures. Although the coding region lacked common non-synonymous variants, we identified two polymorphisms in the DRD1 5′-UTR (G-94A, A-48G) that occurred with frequencies of 15 and 30%, respectively. In the twin/family study, renal traits were highly heritable, such that DRD1 G-94A significantly associated with eGFR, pCr, and FeUA. Homozygotes for the G-94A minor allele (A/A) exhibited lower eGFR, higher pCr, and lower FeUA. No effects were noted for DRD1 A-48G. Patients in the Kaiser group had similar effects of G-94A on eGFR and pCr. Kidney cells transfected with the -94A variant but not the wild type vectors had increased receptor density. Because the -94A allele is common and may reduce glomerular capillary hydrostatic pressure, G-94A profiling may aid in predicting survival of renal function in patients with progressive renal disease.

Dopamine and its receptors regulate sodium transport and hemodynamics through alterations in vascular smooth muscle contractility, including interactions with other neurohormonal factors such as the renin–angiotensin–aldosterone and sympathetic nervous systems; accordingly, the dopaminergic system has been extensively studied in hypertension.¹ Five dopamine receptor subtypes, all G-protein-coupled receptors, have been identified (D₁, D₂, D₃, D₄, and D₅), and classified into two functional subgroups: D₁-like receptors, which stimulate adenylyl cyclase, and D₂-like receptors, which inhibit adenylyl cyclase. The D₁ dopamine receptor (DRD1), a member of the D₁-like receptor subfamily, is expressed in the kidney, largely in the proximal tubule as well as in the vasculature (both afferent and efferent arterioles), and increases urine flow rate and sodium excretion when activated systemically or intrarenally.²

DRD1 has been associated with hypertension in animal studies, including in D1a-knockout mice,³ and in human studies.^3–5 Genetic polymorphisms of DRD1 have been implicated in human essential hypertension^6,7 and renal tubular sodium disposition.⁷ As dopamine is produced intrarenally and may act locally in the kidney, multiple effects of DRD1 receptors may contribute to the patho genesis of hypertension through alterations in renal function.⁸

Here we sought to identify DRD1 genetic polymorphisms through systematic polymorphism discovery, to evaluate such variants' ability to influence glomerular and tubular function, and to assess the effects of trait-associated polymorphism on receptor expression and function in transfected cells.

RESULTS

DRD1 polymorphism discovery: 5′-UTR variants

We undertook polymorphism discovery at DRD1 by PCR amplifying a 1572-bp fragment (Figure 1) spanning the open reading frame (ORF) and 187 bp of 5′-UTR (untranslated region) and 45 bp of 3′-UTR. When the amplicon was sequenced in 27 white individuals (54 chromosomes), no variants that altered encoded amino acids were noted in the ORF. However, we found two common variants in the 5′-UTR (numbering upstream of the ATG initiation codon): G-94A and A-48G.

Figure 1. Human DRD1: coding region and interspecies sequence conservation.

The human dopamine D1 receptor (DRD1) coding exon is shown with indication of the PCR amplicon and open reading frame sequenced in this study. Noted are the 5′-UTR polymorphisms, G-94A and A-48G, confirmed through re-sequencing, and evaluated in association with cardiorenal traits in this study. Plot created at http://pipeline.lbl.gov/cgi-bin/vistatrack.

We then evaluated these two variants in a larger sample (see below twin/family study) of 428 white subjects to determine allele and haplotype frequencies. Taking one subject per family (n=195), the minor allele frequency for G-94A was 15% and for A-48G was 30%. The naturally occurring 5′-UTR haplotypes (across G-94A→A-48G) were GA (at 60% of chromosomes), GG (at 26%), AA (at 13%), and AG (at 0.5%). In these white subjects, the two alleles were in strong (though not complete) linkage disequilibrium (D′ =0.81, r²=0.05, from 195 subjects, taking one per family).

Twin/family study: association of DRD1 5′-UTR polymorphism G-94A with renal function

Table 1 describes the twin/family study. The 428 subjects, self-identified as white, from 195 families, were predominantly female (74%), with a mean age of 40.3 years. Overall, 78% were twins. Mean systolic blood pressure (SBP) was in the pre-hypertensive range. Based on a physician's diagnosis or on the use of BP medications, only 8.9% reported being hypertensive, but on the classification of BP status based on our measurements, 24.3% were in the hypertensive range (SBP≥140 or diastolic blood pressure (DBP)≥90), whereas 49.8% were in the pre-hypertensive range (either 120≤SBP<140 or 80≤DBP<90). Plasma creatinine (pCr) and estimated glomerular filtration rate (eGFR) were within normal ranges. Twin (monozygotic versus dizygotic) correlations (evaluated using SOLAR, Sequential Oligogenic Linkage Routines) established that the traits under consideration showed substantial heritability (h² values of ~41 to 78%), with especially high h² for eGFR at 77.6±_3.4% (P=7.67E−26, Table 1).

Table 1.

Demographic and genotypic data for the twin/family study

Twin/family study (n 428)	Mean	Standard error	Heritability (h2) % ±s.e.m. (P-value)
Trait
Age (years)	40.3	0.80	N/A
SBP (mm Hg)	130.7	0.76	46.3±6.4 (P = 1.68E−09)
DBP (mm Hg)	70.9	0.47	52.7±5.9 (P = 2.77E−12)
eGFR (by MDRD ml/min per 1.73 m2)	96.4	1.3	77.6±3.4 (P = 7.67E−26)
Plasma creatinine (mg/dl)	0.80	0.0075	76.3±3.7 (P = 3.04E−23)
Plasma uric acid (mg/dl)	4.2	0.056	0.34±0.074 (P = 1.79E−5)
Urine study
Uric acid/creatinine (g/g)	0.35	0.013	44.8±6.8 (P = 1.49E−08)
FeUA	0.072	0.0033	41.5±7.6 (P = 1.6E−07)
Participants: 195 families	Twins: 78.7%	Parents/siblings: 21.3%
Sex	Female: 78%	Male: 22%
Ethnicity by self-report	White/European-American: 100%
Blood pressure status by self-report	Normotensive: 91.1%	Hypertensive: 8.9%
Blood pressure category by measurements	Normotensive: 25.9%	Pre-hypertensive: 49.8%	Hypertensive: 24.3%
Family history of hypertension	Yes: 42.8%	No: 48.6%	Unknown: 8.6%
DRD1 polymorphisms: verification of SNP genotyping results by re-sequencing
DRD1 G-94A, rs 5326	G/G: 320	G/A: 57	A/A: 26
HWE χ2 = 35, P < 0.001	Re-sequencing of 48 individuals: 100% concordance
DRD1 A-48G, rs4532	A/A: 193	A/G 125	G/G: 63
HWE χ2 = 6.9, P = 0.009	Re-sequencing of 48 individuals: 97.9% concordance

DBP, diastolic blood pressure; eGFR, glomerular filtration rate; FeUA, fractional excretion of uric acid; HWE, Hardy-Weinberg equilibrium; MDRD, Modification of Diet in Renal Disease; SBP, systolic blood pressure; SNP, single-nucleotide polymorphism.

Means and standard errors are reported for continuous traits. Trait heritability values are from a subset of this study (n = 374) and have been partially reported previously.⁹ As the polymorphisms were out of HWE with results reported using one founder per family (HWE), re-sequencing was performed to confirm genotyping quality. Polymorphism genotyping was carried out on the Sequenom MALDI-TOF platform. Re-sequencing was carried out on an ABI-3100 capillary device. rs, RefSNP (http://www.ncbi.nlm.nih.gov).

Table 2 indicates that multiple markers of renal function were significantly associated with the DRD1 G-94A polymorphism: eGFR, pCr, urine uric acid: creatinine ratio, and fractional excretion of uric acid (FeUA), all of which maintained significance with permutation testing, which also adjusted for the multiple genetic models tested. Figure 2a illustrates an apparently recessive effect of the DRD1 G-94A A allele on eGFR. Subjects with wild-type (G/G) or heterozygous (G/A) genotypes had higher eGFR than those homozygous for the minor allele (A/A, P=0.013). pCr showed a similar effect of the genotypes: subjects with minor allele homozygosity (A/A) had higher pCr levels (0.88±0.19 mg/dl) than did those carrying the G allele (G/A heterozygotes 0.81±0.20 mg/dl and wild-type G/G homozygotes 0.84±0.013 mg/dl, P=0.041, Figure 2b), with sex and age as covariates.

Table 2.

Results of association study of DRD1 G-94A polymorphism with traits associated with hypertension and renal function in the twin/family study

	DRD1 G-94A diploid genotype			Significance (P-value)
Twin/family study	G/G n=320	G/A n=57	A/A n=26	Additive model	Recessive model	Permutation test	Adjusted for population stratification
Trait
SBP (mm Hg)	133.0±1.0	132.7±1.7	136.4±3.8	>0.06	>0.06	-	-
DBP (mm Hg)	72.0±2.4	71.3±1.2	75.1±2.4	>0.06	>0.06	-	-
eGFR (by MDRD, ml/min per 1.73 m2)	100.3±2.2	104.3±3.4	93.2±2.7	0.024*	0.013*	0.014*	0.019*
Plasma creatinine (Cr, mg/dl)	0.84±0.013	0.81±0.020	0.88±0.019	0.056	0.041	0.042	0.090
Plasma uric acid (mg/dl)	4.44±0.083	4.64±0.19	4.72±0.41	>0.06	>0.06	-	-
Urine study
Uric acid/Cr (g/g)	0.35±0.018	0.30±0.027	0.20±0.030	0.020a	0.0088a	0.006a	0.0083a
FeUA	0.069±0.0043	0.056±0.0062	0.041±0.0071	0.039a	0.029a	0.026a	0.025a

No associations were noted for the DRD1A-48G polymorphism. For abbreviations, see Table 1. Additive model: three groups: G/G, G/A, and A/A. Recessive model: G/G and G/A groups combined. Permutation testing was carried out on significant results, without any significant changes; P-values from adjustment for population stratification and permutation testing are reported for the recessive models. Age and sex were covariates in all analyses except where noted by * (for eGFR).

^aSquare root transformation carried out for normalization with presentation of geometric means. Bold P-values denote P <0.05.

Figure 2. Effects of the DRD1 G-94A polymorphism on human renal traits.

(a) DRD1 G-94A effects on estimated glomerular filtration rate (eGFR) by MDRD algorithm in the twin/family and Kaiser normal blood pressure studies. The results of the twin/family study are on the x-axis, and those of the Kaiser normal blood pressure study are on the y-axis. Subjects homozygous for the minor allele (A/A) had lower eGFR than those who were homozygous for the major allele (G/G) or were heterozygous (G/A), indicating an A-allele recessive effect. Thus, P-values are reported for the recessive model (G/G and G/A versus A/A) in the twin/family study (P 0.013) and the Kaiser study (P 0.020). MDRD, Modification of Diet in Renal Disease. (b) DRD1 G-94A on plasma (pCr) or serum (sCr) creatinine concentration in the twin/family and Kaiser normal blood pressure studies. The results of the twin/family study are on the x-axis and the Kaiser study on the y-axis. Subjects who are minor allele homozygous (A/A) at the allele had higher pCr than those who were major allele homozygous (G/G) or heterozygous (G/A), again illustrating an A-allele recessive effect. P-values are thus reported for the recessive model (G/G and G/A versus A/A) in the twin/family study (P = 0.041) and in the Kaiser study (P = 0.007). Sex and age are included as covariates in the model. (c) Effects of DRD1 G-94A on urinary uric acid excretion (urine uric acid:creatinine ratio and fractional excretion of uric acid (FeUA)) in the twin/family study. FeUA is on the left y-axis, whereas urine uric acid:creatinine ratio is on the right y-axis. An additive effect is noted for both urinary uric acid excretion measures, with the minor allele homozygotes (A/A) showing the least excretion, major allele homozygotes (G/G) with the most excretion, and heterozygotes (G/A) with an intermediate effect. For the additive models: urine uric acid:creatinine ratio with P = 0.020 and FeUA with P = 0.039. P-values are reported from square-root transformed data for normalization. Sex and age were covariates in these analyses.

Urinary uric acid excretion (P=0.020) and FeUA (P=0.039) were also associated with the G-94A polymorphism, but showed an apparent additive effect (A/A<G/A<G/A) on uric acid transport and excretion (Figure 2c).

Table 2 also shows the results of the recessive model with adjustment for population stratification. pCr levels, which showed marginal significance before adjustment, was no longer statistically significant, suggesting population admixture in these subjects who self-identified as European-Americans. Significance was maintained for eGFR, FeUA, and urinary uric acid excretion.

The DRD1 A-48G polymorphism did not associate with BP or markers of renal function, perhaps consistent with only partial linkage disequilibrium (D′ =0.81, r²=0.05) between the A-48G and G-94A loci. Results after permutation testing were unchanged.

Replication study: the Kaiser normal blood pressure study

Table 3 describes the Kaiser study used to replicate the findings from the twin/family study, and shows the results of G-94A polymorphism association studies in the Kaiser subjects. The 677 subjects consisted of 57% females, all self-identified as white and without a history of hypertension. Their mean age was 56.8 years; mean SBP, DBP, pCr, and eGFR were within the normal ranges. These subjects had substantially higher mean ages and (consistent with the effect of age) lower GFR values than those in the twin/family study (Table 1, Figure 2a).

Table 3.

Demographic data for the Kaiser normal blood pressure study and associations with the DRD1 G-94A

	Trait statistic		DRD1 G-94A diploid genotype (HWE χ2 = 0.13, P = 0.71)			Significance (P-value)
Kaiser normal blood pressure study (n = 677)	Mean	Standard error	G/G (n = 466)	G/A (n = 165)	A/A (n = 13)	Additive model	Recessive model	Permutation test	Adjusted for population stratification
Trait
Age (years)	56.8	0.58				N/A	N/A	-	-
SBP (mm Hg)	108	0.57	108.5±0.6	108.0±1.1	109.0±3.8	>0.06	>0.06	-	-
DBP (mm Hg)	56	0.16	56.9±0.2	56.3±0.3	56.4±1.1	>0.06	>0.06	-
eGFR (by MDRD, ml/min per 1.73 m2)	77.3	0.70	77.5±0.82	79.2±1.4	66.1±5.4	0.044*	0.020*	0.027*	0.034*
Serum creatinine (mg/dl)	0.95	0.0079	0.97±0.0083	0.95±0.014	1.1±0.054	0.012	0.007	0.010	0.020
Sex	Male: 43%	Female: 57%
Ethnicity by self-report	100% White/European-American

DBP, diastolic blood pressure; eGFR, glomerular filtration rate; HWE, Hardy-Weinberg equilibrium; MDRD, Modification of Diet in Renal Disease; SBP, systolic blood pressure. Means with standard error are reported. Permutation testing was then carried out on significant results, without any significant changes. P-values for adjustment for population stratification and permutation testing are reported for the recessive models. Sex and age adjusted for all analyses except where indicated * (for eGFR).

Additive model: three groups: G/G, G/A, and A/A; recessive model: G/G and G/A combined. Bold P-values denote P <0.05.

As we found in the twin/family study, eGFR was predicted by the DRD1 G-94A polymorphism, with lower eGFR in minor allele homozygotes (A/A, 66.1±0.9 ml/min per 1.73 m², compared with G/A (79.2±1.4) and G/G (77.5±0.8, P=0.020; Table 3 and Figure 2a). Table 3 and Figure 2b illustrate the similar association of pCr with genotype; those homozygous for the minor allele (A/A) had the highest pCr levels (P=0.007). The results of permutation testing and adjustment for population stratification in the recessive models, indicating no changes in statistical significance, are shown in Table 3.

When analyzing the twin/family study with the Kaiser study in a combined analysis, significance was maintained for the DRD1 G-94A polymorphism and pCr (P=0.011, covariates: sex, age, study group) and for eGFR (P=0.0031, covariate: study group).

Heterologous expression and binding of the DRD1 receptor in transfected cells: 5′-UTR variant -94A increases receptor number

Three common haplotypes occurred across the 5′-UTR, capturing >99% of human chromosomes (from G-94A→A-48G), including G→A (60% of chromosomes), G→G (26%), and A→A (13%). We thus studied these three combinations in transfected cells to probe the mechanism whereby the G-94A variant influenced renal traits. Three plasmids were generated, under the control of the CMV promoter, which expressed the DRD1 coding region and common upstream 5′-UTR sequence: wild type (WT; G-94→A-48 alleles) and each 5′-UTR variant, -94A (with A-48) and -48G (with G-94).

The three constructs were transiently transfected into COS-7 kidney cells followed by radioligand binding assays. In order to quantitate receptor number in the transfected cells, we incubated them with a fixed concentration of [³H]SCH23390, a dopamine receptor antagonist, and with increasing concentrations of SCH23390 (7-chloro-3-methyl-1-phenyl-1,2,4,5-tetrahydro-3-benzazepin-8-ol). Figure 3a indicates that cells transfected with the -48G sequence had similar binding properties (number of binding sites, B_max, and affinity for radioligand, K_d) to those transfected with the WT expression plasmid. In contrast, cells transfected with the -94A sequence had similar K_d values (~0.36±0.05 nM) but a 2.5-fold higher B_max than those transfected with either WT or -48G. Studies conducted with replicate transfections are shown in Figure 3b: cells transfected with the -94A sequence showed significantly higher expression relative to cells that expressed the WT sequence (7093±318 vs 2580±420 fmol per mg protein, n=3, P<0.01), whereas cells transfected with the -48G sequence showed a similar number of receptors as did WT (2269±370 fmol per mg protein). To control for transfection efficiency, we also co-transfected a luciferase reporter plasmid (pGL3-Promoter). Even with such normalization, cells transfected with the -94A sequence showed higher receptor expression than did those transfected with the WT sequence (1.51 vs 0.89 × 10⁻⁴ fmol/RLU (relative light units), n=3, P<0.05; data not shown). DRD1 receptors in cells transfected with the three plasmids had similar K_d values for [³H]SCH23390 (WT, 0.42±0.19 nM; -48G, 0.38±0.18 nM; and -94A, 0.36±0.05 nM).

Figure 3. DRD1 variant G-94A alters receptor density.

(a) Illustrative experimental data points. Competitive binding assay for each DRD1 SNP construct (WT, -48G, and -94A) with increasing concentration of D1-specific radioligand [³H]SCH23390. Cells transfected with the -48G sequence (▲) had similar binding properties (B_max and K_d) to cells transfected with the wild-type (WT) DRD1 sequence (▼). Cells transfected with the -94A sequence (■) had a similar K_d but ~ 2.5-fold higher B_max than either WT or -48G-expressing cells. Expression of both WT and variant DRD1 were governed by the strong CMV promoter in the vector pcDNA3.1/Hygro(+). (b) Summary data for binding capacity (B_max). DRD1 expression in cells transfected with each polymorphism construct (WT, -94A, and -48G). DRD1 expression was determined by radioligand binding assay with [³H]SCH23390 and presented as fmol receptor per mg membrane protein. Expression of the three variants DRD1s was governed by the strong CMV promoter in the vector pcDNA3.1/Hygro(+). The construct from the -94A variant yielded the greatest amount of receptor per mg protein, **P<0.01.

mRNA structure: bioinformatics of DRD1 G-94A

G-94A is located in an mRNA region highly conserved across primate species (Figure 4), with the G allele ancestral. A search of the local 5′-UTR region did not show mRNA motifs disrupted by G-94A, in particular microRNA recognition sites, nor did G-94A substantially alter predicted local RNA stem/loop folding patterns or free energies of folding (equilibria).

Figure 4. Primate interspecies sequence conservation in the region of DRD1 G-94A.

Results from Clustal-W alignment indicate the highly conserved region in the 5′-UTR of DRD1 surrounding position -94 across primate species. The ancestral (non-human primate) allele is G.

DISCUSSION

Overview

This study reports an association between the DRD1 5′-UTR polymorphism G-94A and eGFR in predominantly normotensive or pre-hypertensive individuals, with replication in an independent population sample. We found that subjects with minor allele homozygosity (A/A) at G-94A have decreased eGFR (Figure 2a) and increased pCr (Figure 2b), suggesting a recessive action of the A allele on glomerular traits. We previously reported that DRD1 G-94A also correlates with albuminuria in this twin/family study.⁹ Thus, coordinate actions of the A/A diploid genotype to lower both GFR and albuminuria suggest diminished glomerular capillary hydrostatic pressure as a potential mechanism. We also found effects of the genetic variant on renal uric acid transport traits (Table 2, Figure 2c), suggesting that D₁ receptors may influence other aspects of renal function, as has been previously suggested.^10,11 In addition, we carried out assays with plasmid constructs of WT and variant DRD1 genes transfected into heterologous cells and found that the -94A variant, but not the -48G variant, had increased maximal binding capacity. To our knowledge, this is the first report of correlation between DRD1 polymorphisms and renal function in human subjects and also the first evidence that the -94A variant alters the number of DRD1 receptors.

DRD1 and renal function

As dopamine and its receptors have a broad spectrum of actions, including roles in sodium transport, renal hemodynamics, and interactions with other vasoactive substances, several hypotheses can be posed as to how an increased DRD1 receptor may affect renal function, although the literature is inconclusive regarding the comprehensive effects of dopamine in the kidney, including the roles of both DRD1 and DRD2 receptors. One study in experimental animals suggested that DRD2 receptors, which typically inhibit adenylyl cyclase activity, influence glomerular filtration as opposed to DRD1 receptors, which stimulate adenylyl cyclase activity.¹²

Dopamine receptors are present on afferent and efferent glomerular arterioles,^13,14 and thus receptor number may affect the hemodynamics required for filtration. Increased dilation of the efferent arteriole would be expected to decrease glomerular capillary hydrostatic pressure and, hence GFR, although rat studies have shown that GFR is maintained despite the vasodilation.¹⁵ This is likely attributable to equivalent declines in both pre- and postglomerular resistance.¹⁵ Animal studies have yielded conflicting results regarding whether the actions of DRD1 receptors on proximal Na⁺ reabsorption may trigger reciprocal changes in GFR through tubuloglomerular feedback.^15–18

Although DRD1 activation in smooth muscle myocytes is generally believed to result in vasodilation, studies in some animals show interactions between DRD1 and the angiotensin type I¹⁹ and type II²⁰ receptors, which affect natriuresis in the proximal tubule and also vasoconstriction. The equivalent receptor (DRD1a) in rats may also trigger secretion of renin from juxtaglomerular cells,²¹ particularly when COX2 expression is suppressed.²² Human studies with the DRD1 agonist, fenoldopam, have also suggested increases in renin–angiotensin activity.^23,24 Such an altered milieu of vasomotor tone modulators may contribute to decreased eGFR.²⁵ Filtration may be affected by the presence of DRD1 receptors, as documented on podocytes²⁶ and mesangial cells.²⁷ In addition, mesangial cells may produce catecholamines in vitro,^28,29 though further studies are needed in this area.

Using selective DRD1 agonists such as fenoldopam, human and animal studies have not shown consistent or substantial changes in GFR after DRD1 receptor activation.^5,23,30 In seven humans with normal BP, an increase in GFR was noted,²³ but a similar study in six hypertensive subjects did not yield a significant difference in GFR.²⁴ Animal studies have shown that during diabetes- or amino acid-induced elevation of GFR, fenoldopam may inhibit or reduce such hyperfiltration.^31,32 Our study subjects, particularly in the twin/family study with mean SBP in the prehypertensive range, might exhibit glomerular hyperfiltration, which could then be attenuated with DRD1 stimulation.

Our heterologous receptor expression results suggest that increased DRD1 receptor density produced by the -94A variant may contribute to changes in GFR. Heterozygous D1a receptor knockout mice exhibited intermediate receptor density between the WT and knockout,³ suggesting that the WT (+/+) genotype may be dominant in the mouse, a finding consistent with the dominant (G) and recessive (A) actions of G-94A on eGFR and pCr (Figure 2a and b) that we observed.

Previous studies of DRD1 5′-UTR polymorphisms and their association with hypertension, particularly with A-48G, have yielded mixed results in various populations.^6,7,33,34 Our study of white subjects, similar to that by others,³³ did not show correlations of the A-48G or G-94A polymorphisms with BP. The FLEMENGHO group evaluated the G-94A single-nucleotide polymorphism (SNP) on both BP and renal traits, although their findings included haplotype analyses that we did not perform.⁷ They reported a decrease in SBP and DBP in subjects with a DRD1 haplotype (-48A, -94G, and −800C). As in our twin/family and Kaiser normal BP studies, the individual SNPs were not significantly associated with BP. Given that the C-800T SNP was not scored in our study, we are unable to evaluate this haplotype in our cohorts. GFR was not determined in the FLEMENGHO study, and thus it is uncertain how the BP in subjects with the DRD1 AGC haplotype may influence renal function. This group also reported that carriers of the recessive allele -94A had an increased fractional distal reabsorption of sodium and decreased fractional excretion of sodium (FeNa).⁷ As FeNa and FeUA appear to be positively associated,^35,36 our results of increased FeUA in the WT (G/G) homozygotes at DRD1 G-94A are consistent with the FLEMENGHO study, in which FeNa was greater in WT (G/G).

Precise causal relationships between uric acid and renal disease remain unresolved, although associations have been shown.³⁷ Uric acid may have a function in the regulation of intrarenal vessel function, including interactions with dopamine^10,38 or the renin–angiotensin system.³⁹ Although genetic influences on the excretion of uric acid have been established,⁴⁰ our findings indicate that decreased uric acid excretion in minor allele homozygotes of G-94A (A/A) may be linked to a decrease in GFR noted in the same subjects. Decreased GFR may be mechanistically tied to the decline in microalbumin excretion in G-94A in minor allele homozygotes (A/A), which has been noted for a subset of this twin/family study.⁹ Both changes can be consequences of diminished glomerular capillary hydrostatic pressure, a therapeutic goal in patients with glomerular disease.⁴¹ Thus, minor allele homozygosity (A/A) may modify the temporal course of progressive renal disease. Studies that address this possibility would be of interest, as positive results would suggest that profiling patients with chronically declining renal function by G-94A genotyping might yield a useful predictor of renal survival. Further studies are necessary to define the role of DRD1 in renal uric acid transport.

mRNA structure: interspecies conservation

The conservation of the ancestral G allele of DRD1 G-94A across non-human primates (Figure 4) suggests a region of functional significance, but the basis of such action(s) is/are uncertain. Further investigation is necessary to determine how G-94A variation results in increased receptor density. Possible mechanisms include alteration of a transcriptional control motif (e.g., an enhancer element), disruption of the interaction of the 5′-UTR with as-yetundefined trans-acting factors influencing mRNA stability or translatability, or perhaps effects on other proposed regulators of translation.^42,43

Advantages and limitations of these studies

Twins

Twin pair studies allowed us to quantify the heritability (h²) of each cardiorenal trait under consideration, several of which studied here displayed substantial and significant h² (at h² ~41 to 78% of trait variance, Table 1), thereby indicating the likelihood of genetic effects on such traits, and setting the stage for tests of particular polymorphisms, such as those at DRD1.

Genomics

We systematically scanned the DRD1 coding exon for common polymorphism (Figure 1) and focused on the 5′-UTR. We found trait associations for a variant in the DRD1 5′-UTR and excluded common non-synonymous variation in the coding region, but we did not systematically examine variants in other gene regions, such as the promoter or the 3′-UTR, leaving open the possibility of other functional variants in control of renal traits.

The frequencies of both DRD1 5′-UTR polymorphisms studied deviated from Hardy–Weinberg equilibrium (HWE) in the twin/family study (Table 1) but not in the Kaiser normal BP study. Although >97% concordance with re-sequencing across the loci renders genotyping error unlikely, other factors known to perturb HWE might be operative, including evolutionary forces such as migration or increased mutation rates. This HWE deviation may also be a sampling effect of a small population in San Diego, though it was not noted in the Kaiser group. The polymorphism that yielded the significant results, G-94A, has been evaluated in previous association studies,^7,33 one of which also found deviation from HWE in hypertensives.³³ The pre-hypertension status within the twin/family study (almost 50% of the sample) may also contribute to the deviation from HWE, which was not seen in the normotensive subjects in the Kaiser cohort.

Even among subjects who self-report as European-Americans, admixture may confound associations, as ancestral groups may harbor genetic variation that is not associated with changes in gene expression. Thus, we assessed genetic heterogeneity based on additional genotyping data available from biallelic markers distributed across the autosomes, and calculating an identity-by-state (IBS) distance matrix. The first three multidimensional scaling components were extracted and used as covariates for a population structure. As noted in Tables 2 and 3, this adjustment had minimal influence on statistical significance, with the exception of pCr in the twin/family study; such sensitivity of the pCr trait to differential admixture may reinforce that eGFR (by Modification of Diet in Renal Disease (MDRD) equation) is a more effective estimator of glomerular filtration than pCr alone.

Multiple comparisons and replication

With recent increases in genome-wide association studies, concern has arisen that multiple comparisons may result in false-positive results. Here, we were able to replicate our findings by studying two distinct populations: a younger twin/family cohort with mean age of 40.3±0.8 years and eGFR of 96.4±1.3 ml/min per 1.73 m² and a Kaiser cohort with a mean age of 56.8±0.6 years and eGFR of 77.3±0.7 ml/min per 1.73 m². In addition, in this candidate gene study, we used SNPSpD, proposed by Nyholt,⁴⁴ to determine that the two SNPs evaluated (G-94A and A-48G) were in sufficiently strong linkage disequilibrium that effectively only ~1 SNP was tested, eliminating the need to choose a more stringent P-value for statistical significance. Results of the cardiorenal trait associations were also maintained after permutation testing, which minimizes type I errors due to multiple comparisons.⁴⁵

Multiple comparisons of phenotypes must also be considered, although our study involves highly correlated traits. As a conservative measure, we may consider a Bonferroni correction based on three major groups of traits in our twin/family study: blood pressure (SBP and DBP), uric acid excretion (urine uric acid: creatinine ratio, FeUA, and plasma uric acid), and renal function (eGFR and pCr). The conservative target P-value for significance would then be 0.017 (=0.05/3) in the twin/family study. Thus, pCr and FeUA would no longer reach statistical significance. However, we opted to test for replication of the findings in a second population. In the Kaiser study, two major groups of traits were studied, BP and renal function. The conservative target P-value for significance would then be 0.025 (=0.05/2), with both eGFR and pCr maintaining their significance.

Estimation of renal function

We found that eGFR by MDRD is highly heritable in twin pairs (77.6±3.4%, P=7.67E-26). The MDRD equation is most accurate in subjects with GFR<60 ml/min per 1.73 m², tending to underestimate GFR in healthy individuals.⁴⁶ Standardized pCr calibrations either to the MDRD reference laboratory or to the Cleveland Clinic Research Laboratory measurements have improved accuracy in healthy individuals.⁴⁷ Our pCr levels were evaluated with customary quality control external standards in calibrated automated analyzers, by the Kinetic Jaffe Reaction, with traceable isotope dilution-mass spectrometry.⁴⁸ In limiting the study to subjects with pCr<1.2 mg/dl, within the normal laboratory range, small underestimations of GFR likely had minimal effect on associations that we assessed.

Pleiotropy

Our study illustrates that a single SNP, particularly in a highly expressed gene in the kidney such as DRD1, may have several consequences and affect multiple traits.¹ Dopamine has multiple renal actions, including interactions with other vasoactive substances and their receptors. It is unclear whether the pleiotropic effects of G-94A observed here result from actions in parallel or in series. For example, if DRD1 G-94A alters glomerular hydrostatic pressure, this may result in simultaneous changes in eGFR and albuminuria, while the effect on FeUA is more likely because of receptors in a different region, such as the renal tubular epithelia.

Functional studies

In addition to our genomic studies, we assessed the functional consequence of the 5′-UTR genetic polymorphisms in a heterologous system. This is the first report regarding the role of DRD1 genetic variants on the expression of cell surface DRD1 receptors. On the basis of preliminary studies (data not shown), we found that the heterologous system used in these studies to assess receptor density is not sensitive enough to display downstream signaling effects (e.g., cAMP response) of a two- to threefold increase in receptor number, although such differences may be sufficient to affect signaling in vivo. Defining the precise molecular mechanisms by which cell surface receptor density is altered by the polymorphism will require future efforts; such mechanisms may include transcriptional and/or post-transcriptional (including post-translational) events that may be altered by the polymorphisms. One potential mechanism is through a `leader cistron', by which the expression of another G-protein-coupled receptor, the β-2 adrenergic receptor, is regulated.⁴² According to this mechanism, the 5′ leader region of the DRD1 receptor messenger RNAs containing the −94 polymorphism may act as a short ORF preceding the receptor cistron and encoding a small peptide that inhibits receptor translation. Genetic alteration in this ORF may alter the expression of the downstream receptor gene. We also acknowledge that differential receptor expression (especially as determined in a heterologous system) may not necessarily translate into differential function on either a cellular or whole animal level.

Conclusions and perspectives

The DRD1 5′-UTR G-94A polymorphism is reproducibly associated with the highly heritable trait of eGFR in subjects with predominantly normal or pre-hypertensive blood pressure. The association with -94A may be a result of decreased glomerular capillary hydrostatic pressure, as a consequence of increased receptor expression, but the precise location in the kidney (tubules, arterioles, podocytes, or mesangial cells) of the receptors that account for the effect, and how changes in receptor expression translate into altered GFR, will require further investigation. As the -94A allele is relatively common (~15% in our populations), this variant might be useful in prognostic profiling of subjects with progressive renal disease.

MATERIALS AND METHODS

Systematic polymorphism discovery at DRD1

Genomic DNA was isolated from blood leukocytes, as has been previously described⁴⁹ using PureGene DNA Extraction Kit (Gentra Systems, Minneapolis, MN, USA). An anonymized sample of 27 unrelated white subjects was used for polymorphism discovery.

PCR and sequencing reactions

PCRs were carried out in MJ Research Dyad Thermal Cyclers (Waltham, MA, USA). Two sets of oligonucleotide primers were designed to generate overlapping fragments of the DRD1 sequence. PCRs of 25 μl were carried out using 25 ng DNA (or H₂O for negative controls), 25 mM MgCl₂, 10 mM dNTPs, 20 μM primers, and 0.5 U per reaction of AmpliTaq Gold DNA Polymerase (Applied Biosystems, Foster City, CA, USA). The reaction was cycled 40 times with a denaturation step of 95 °C for 30 s, an annealing step of 65.8 °C for 1 min, an elongation step of 72 °C for 1 min, and a final elongation step of 72 °C for 8 min. PCR product of 10 μl was used to verify amplification (or the absence of amplified product in negative controls) on a 1% agarose gel using electrophoresis. PCR product of 15 μl was purified with Exonuclease I (3 U per reaction) and Shrimp Alkaline Phosphatase (0.8 U per reaction) by incubation at 37 °C for 30 min, then 85 °C for 15 min. Sequencing reaction was performed according to Applied Biosystems Big Dye Terminator v3.1 Cycle Sequencing Kit Protocol (2002) and product was purified using Sephadex G-50 DNA Grade beads (Sigma Scientific, St Louis, MO, USA). Hi-Di Formamide of 10 μl was added to the sequencing reaction. Sequencing was carried out on an ABI-3100 capillary device (Applied Biosystems).

Twin/family study

The twin/family UCSD study has previously been described.⁹ A total of 428 white (by self-identification) subjects from 195 families, including 168 twin pairs and their siblings met criteria for this study with a baseline pCr level <1.2 mg/dl. Overall, 78% were female and 8.9% reported that they had been diagnosed with hypertension or were on medications. Brachial cuff BPs (mm Hg) were measured in seated subjects in triplicate using a DynaPulse oscillometric device (PulseMetric, San Diego, CA, USA) as previously described and validated.⁵⁰ Using these triplicate measurements, we classified 24.3% in the hypertensive range (SBP≥140mm Hg or DBP ≥90 mm Hg) and 49.8% in the pre-hypertensive range (120≤SBP<140mm Hg or 80≤DBP <90mm Hg). Normotension was defined as SBP<120mm Hg and DBP<80mm Hg. Plasma and urine samples were obtained for biologic phenotyping. Urine and pCr values were measured by autoanalyzer (Beckman-Coulter, Brea, CA, USA) using the Kinetic Jaffe Reaction with traceable isotope dilution-mass spectrometry.⁴⁸ Urine uric acid was determined by colorimetric autoanalyzer and normalized to urine creatinine from the same sample, whereas FeUA was calculated by the equation FeUA=(U_{Uric acid}*P_Creatinine)/(P_Uricacid*U_Creatinine). GFR was estimated by the MDRD equation, using pCr, age, sex, and ethnicity of each subject.⁴⁶

Subjects were volunteers from Southern California, and each gave informed, written consent; the protocol was approved by the UCSD San Diego Human Research Protection Program.

Replication study: Kaiser normal BP group

A replication study was performed on subjects sampled from 53,078 individuals (27,475 women and 25,538 men) whose medical information was obtained through routine, yearly health appraisal visits to the Kaiser Permanente Medical Group, a subscription-based, primary care, health maintenance organization located in San Diego (CA, USA), which has previously been described.⁵¹ We studied 677 white subjects without a history of hypertension or history of taking anti-hypertensive medications, sampled from the lower quantiles of DBP, because twin and family studies have shown that DBP is substantially heritable.^52,53 The mean age of the study group was 56.8 years, with 57% females. BP was measured in seated subjects using aneroid sphygmomanometry. If DBP was elevated, repeat measurement was obtained. sCr was determined in the Kaiser-Permanente laboratory by spectrophotometric autoanalyzer. eGFR by MDRD was determined with sCr, age, sex, and ethnicity of each subject.

Genotyping for association studies

The two SNPs determined by the systematic polymorphism discovery at the 5′-UTR of the DRD1 locus (G-94A rs5326 and A-48G rs4532) were scored by the Sequenom (San Diego, CA, USA) MassARRAY matrix-assisted laser desorption ionization time-of-flight (MALDI-TOF) mass spectrometry system.⁵⁴ HWE, by χ², was not maintained for either DRD1 polymorphism in the twin/family study even after using only one member of each family with P<0.05 (Table 1). Thus, to confirm genotyping quality in that study, 48 subjects were selected and re-sequenced across the DRD1 locus (ABI-3100 capillary sequencer; Applied Biosystems) to evaluate the concordance with a reference standard. This yielded 100 and 97.9% concordance between the ABI-3100 sequencer and the MALDI-TOF system at the G-94A and A-48G loci, respectively. HWE was maintained in the Kaiser study.

Statistics

For the twin/family study, descriptive and inferential statistics were computed with generalized estimating equations (GEE), PROC GENMOD, in Statistical Analysis System (SAS, Cary, NC, USA), which consider intra-family correlations by forming an exchangeable correlation matrix.⁵⁵ Trait heritability (h²), or the fraction of trait variance (V_P) accounted for by additive genetic variance (V_G), that is, h²=V_G/V_P, was estimated from twin (monozygotic versus dizygotic) data by SOLAR (Sequential Oligogenic Linkage Routines; http://www.sfbr.org/solar).⁵⁶ Heritability results are from a subset (n=374) and have partially been reported previously.⁹ Univariate analysis of variance (ANOVA) was carried out using PROC GLM in SAS for the Kaiser replication study. A combined study with the twin/family and Kaiser groups was performed with GEEs. In the instance that inspection of the data suggested a recessive model for the effect of the minor allele on a trait, statistics were performed for the major allele homozygotes and the heterozygotes combined. When data were found to be non-normally distributed, such as with the measures of urinary uric acid excretion, normalization was performed by square-root transformation, with presentation of geometric means.

To adjust for the possibility of multiple comparisons when testing the effect of two DRD1 SNPs on renal function and neurohormonal traits, we used the method of SNP Spectral Decomposition (SNPSpD) proposed by Nyholt⁴⁴ and implemented at http://genepi.qimr.edu.au/general/daleN/SNPSpD to yield an `effective' number of markers within a block of LD. For this purpose, we used SNP data from one member of each family. This method takes into account intermarker correlations in calculating a new experiment-wide threshold to keep the type I error rate at ≤0.05 for a single phenotype. In this study, statistical significance was maintained at P<0.05 given that the two polymorphisms (position −94 and −48) are in substantial linkage disequilibrium with D′=0.81 (n=195 considering only one member per family), as determined by Haploview⁵⁷ (http://www.broad.mit.edu/mpg/haploview/).

Permutation testing was performed using SAS for significant associations with the DRD1 polymorphisms, which adjusted for the two genetic models tested. For the twin/family study, the permutation was performed, accounting for LD between the two SNPs. The polymorphisms were permuted and analyzed 1000 times, and the observed test statistic was then ranked to determine significance using GEE for the twin/family study and ANOVA for the Kaiser normal BP study.

Statistical power was determined using the online instrument G*Power 3 (http://www.psycho.uni-duesseldorf.de/abteilungen/aap/gpower3/literature).⁵⁸ More than 92% power to detect an effect size of 0.18 would be achieved with a sample size of 425. For three groups, the sample size of 428 in the twin/family study required an effect size (difference in means between two groups over the common standard deviation) of 0.19 to reach α=0.05 and β=0.8, whereas the sample size of 677 in the Kaiser study could detect an effect size of 0.12. When a recessive model is tested, the sample size in the twin/family study could detect an effect size of 0.14, whereas the Kaiser study could detect an effect size of 0.11.

Population stratification

To address potential population stratification in these cohorts of subjects with self-identified European ancestry, we assessed genetic heterogeneity of both the twin/family and Kaiser cohorts based on additional genotyping data available from selected biallelic markers distributed across the autosomes (19 chromosomes), which were scored by the Sequenom MassARRAY MALDI-TOF mass spectrometry system.⁵⁴ We carried out a multidimensional scaling analysis using PLINK 1.6, (http://pngu.mgh.harvard.edu/purcell/plink)⁵⁹ including 155 markers for the twin/family study and 174 markers for the Kaiser normal BP study, with overall genotyping rates of 94.5 and 92.1%, respectively. The analysis excluded subjects with <80% genotypes carried out and SNPs that were out of HWE. For each cohort, we calculated an IBS allele sharing distance matrix between all subjects and extracted the first three multidimensional scaling components to account for contributions from other major ancestral backgrounds. To control for potential confounding because of population stratification, we then used these components as covariates in the association analyses. Compared with a non-corrected analysis, the genomic inflation factor (based on the median χ²) was reduced from 1.405 to <1.001, thus showing adequate power to correct for population stratification.

DRD1 mutant and WT receptors: expression and binding

Plasmid constructs

A human DRD1 was generated by PCR using human genomic DNA as template with sense primer 5′-GATTGACTTGGATTGCCACTC-3′ and antisense primer 5′-TCTAGCTTTTGGGATGAGCATG-3′, resulting in a DRD1 clone containing the ORF, a 5′-UTR of 187 nucleotides upstream of the start codon, and 3′-UTR of 45 nucleotides downstream from the stop codon. The resultant fragment was cloned into pGEM-T Easy (Promega, Madison, WI, USA) and sequenced to verify the insert. NotI fragments were subcloned from these plasmids into pcDNA3.1/Hygro(+) (Invitrogen, Carlsbad, CA, USA) to construct DRD1 expression plasmids, WT (WT receptor sequence), -94A and -48G, under the control of the CMV promoter.

Cell culture and transient transfection

African green monkey kidney SV40-T-antigen-transformed COS-7 cells were obtained from ATCC (Manassas, VA, USA). COS-7 cells were cultured in Dulbecco's modified Eagle's medium (Cellgro, Mediatech, VA, USA), supplemented with 10% fetal bovine serum (Omega Scientific, Tarzana, CA, USA) at 37 °C in a humidified atmosphere containing 5% CO₂. Transient transfections were carried out in a 100-mm dish using Lipofectamine Plus reagent (GibcoBRL, Invitrogen, Carlsbad, CA, USA) with 4 μg of each construct DNA, and 0.1 μg of pGL3-Promoter vector (Promega), a transfection efficiency control plasmid in which luciferase expression is driven by the SV40 early promoter.

Binding assay

DRD1 expression was quantitated by binding of the radiolabeled DRD1-selective antagonist [³H]SCH23390 (NEN, Boston, MA, USA). Transfected cells were collected by scraping with a rubber policeman into ice-cold buffer A: 50 mM Tris-HCl pH 7.4, 1mM EDTA, 5 mM KCl, 1.5 mM CaCl₂, 4mM MgCl₂, 120 mM NaCl, with protease inhibitor cocktail (Sigma, St Louis, MO, USA). Cells were lysed using a Dounce homogenizer and the homogenate centrifuged at 300g for 10 min at 4 °C. The supernatant was centrifuged at 40,000g for 20 min at 4 °C, and the pellet resuspended in buffer A. These membrane fractions were stable at least 3 months after aliquoting and storage at −70°C. For the binding assays, 2 μg of membranes were incubated in triplicate with increasing concentration of the unlabeled ligand (1 μM–1 pM of SCH23390; Sigma) and 0.65 nM [³H]SCH23390 (75.5 Ci/mmol) in 0.2 ml buffer A for 2 h at 25 °C. The binding was terminated by vacuum filtration through Whatman (Maidstone, Kent, UK) GF/C filters. The filters were rapidly washed twice with 10 ml of 50 mM Tris-HCl (pH 7.4), transferred to scintillation vials with 10 ml EcoLume liquid scintillation cocktail (ICN, Biomedicals, Irvine, CA, USA), and quantitated in a liquid scintillation counter (Beckman).

Data analyses for binding assays and curve fitting (for B_max and K_d) were carried out by non-linear regression using Prism software (GraphPad, San Diego, CA, USA). Data shown for competitive binding assays represent triplicate measurements.

mRNA structure: bioinformatics

SNP information and interspecies homology information were extracted from http://genome.ucsc.edu. Interspecies sequence alignments were performed by Clustal-W (http://www.ebi.ac.uk/Tools/es/cgi-bin/clustalw2). RNA motif predictions were accomplished by RegRNA (http://regrna.mbc.nctu.edu.tw/index.php). RNA secondary structure predictions and folding energy calculations were performed by GeneBee (http://www.genebee.msu.su/services/rna2_reduced.html).