Variations in the Potassium Channel Genes KCNK3 and KCNK9 in Relation to Blood Pressure and Aldosterone Production: An Exploratory Study

Abstract

Context:

Two potassium (K) channel genes, Kcnk3 and Kcnk9, when deleted in mice, produced a model of hyperaldosteronism and hypertension.

Objective:

Our objective was to explore genetic variation [single-nucleotide polymorphisms (SNP)] in KCNK3 and KCNK9 in relation to blood pressure (BP) and aldosterone production in humans.

Subjects and Study Design:

Two groups of healthy European Americans (EA) and African Americans (AA) were studied: 1) a longitudinal study group (age ∼14 yr when enrolled, 444 EA and 351 AA) and 2) an inpatient cross-sectional study group (age ∼23 yr, 85 EA and 109 AA). Plasma renin activity, plasma aldosterone concentration, and level of serum K were measured cross-sectionally; BP was measured semiannually in the longitudinal study. SNP were selected to provide coverage of the genes for both EA and AA (15 in KCNK3 and 74 in KCNK9).

Results:

No associations with KCNK3 were observed. In the longitudinal study, multiple SNP in KCNK9 associated with systolic BP in AA, whereas associations were primarily with aldosterone production in EA. The direction of the changes was the same for aldosterone production and BP, whereas serum K changed in the opposite direction. In the cross-sectional study, associations were observed only in AA. Combining the two studies, one SNP in particular, rs888345, was strongly associated with BP in AA and with indices of aldosterone production in AA and EA.

Conclusion:

Results of an exploratory study suggest that BP and aldosterone production may be affected by variations in KCNK9. The findings could have relevance to risk for hypertension.

Aldosterone stimulates reabsorption of sodium (Na) and secretion of potassium (K) in the kidney's distal nephron (1). These separate although interrelated actions of aldosterone serve to, respectively, defend against volume depletion due to Na loss and prevent hyperkalemia. The regulation of aldosterone production likely evolved during an early ancestral period when diets contained little Na and a surfeit of K. Today, aldosterone production may exceed what is required for achieving Na and K homeostasis, becoming instead pathogenic by participating in the development of hypertension. Indeed, approximately 10% of individuals with hypertension have primary aldosteronism, the consequence of either adrenal tumor development or adrenal hyperplasia (2), and 20% or more of those with resistant hypertension [high blood pressure (BP) refractory to multiple medications] have either primary aldosteronism (3) or normal aldosterone levels but respond to treatment with antagonists of aldosterone (4). Moreover, among individuals without hypertension, a higher but normal plasma aldosterone concentration (PAC) was shown to be predictive of an elevation in BP within 4 yr (5). If indeed, given today's dietary practices, aldosterone production becomes relatively dissociated from a need to retain Na, then identification of other determinants of aldosterone production would make an important contribution to an understanding of the causes of hypertension. In an earlier study, we demonstrated a heritability of aldosterone excretion rates (6), suggesting that genetic studies could reveal important additional determinants of human hyperaldosteronism.

Two pore domain K channel subunits, TWIK-related acid-sensitive K+ channel 1 (TASK1) and TASK3 (encoded by KCNK3 and KCNK9, respectively) are prominently expressed on the plasma membrane of rodent cell zona glomerulosa (ZG) (7, 8). The subunits form homo- or heterodimeric leak K channels and are important ionic conductances that restrain electrical excitability and limit the production of aldosterone. Extracellular K within the physiological range of 3–5 mm acts both independently (9) and synergistically with angiotensin II (10) to stimulate aldosterone production in ZG cells. Its principle mechanism of action is to depolarize the membrane potential of the ZG cell, which permits the opening of voltage-dependent calcium channels and the consequent increase of extracellular calcium entry that is essential for sustaining the production of aldosterone. Indeed, the exquisite sensitivity of aldosterone production in vivo to incremental changes in serum K (0.5–1.0 mmol/liter) (9) is a direct measure of the strong dependence of the membrane potential of the ZG cell on K permeability pathways. Thus, modulation of K channel activity is a powerful means for controlling the production of aldosterone. Deletion of Kcnk3 in the mouse results in disruption of adrenal zonation with hyperaldosteronism and elevated BP (11). The tandem inactivation of Kcnk3 and Kcnk9 genes from the mouse results in a 20-mV depolarization of the ZG cell plasma membrane and a phenotype that closely resembles human idiopathic primary aldosteronism: elevated aldosterone levels, low renin levels, an increased aldosterone to renin ratio (ARR), failure to suppress aldosterone production with high salt or normalize with candesartan, and hypertension (12). Thus, in the present study, we explored the genetic association of single-nucleotide polymorphisms (SNP) in KCNK3 and KCNK9 with BP and indices of aldosterone production in young normotensive subjects that had over a period of years repeated measurements of BP and had cross-sectional assessments of aldosterone production.

Subjects and Methods

Subjects and study design

All subjects were healthy (normotensive) and either European American (EA) or African American (AA). They were participants in a longstanding cohort study of BP regulation (13). Race was based on the subject's self-report. There were two groups that differed from each other by the study protocols employed. The first group consisted of individuals who had measurements made longitudinally in an outpatient setting (13). The second group consisted of individuals who had measurements made in an inpatient facility [General Clinical Research Center (GCRC)] cross-sectionally. They are here referred to as the longitudinal study and the cross-sectional study. In the longitudinal study, 444 EA and 351 AA had measurements made over the age range of 5.6–25 yr. In the cross-sectional study, there were 106 AA and 85 EA who ranged in age from 18–36 yr. The latter group had been participants in an inpatient study of racial differences in Na reabsorption (14); only data collected at baseline were used in the present study. Nineteen subjects participated in both the longitudinal study and the cross-sectional study. The studies were approved by the Indiana University-Purdue University of Indianapolis Institutional Review Board. Each subject provided consent or assent with a parent or guardian providing informed consent.

Measurements

In the longitudinal study, weight, height, and BP were measured every 6 months, in the morning, either at the school they attended or in the GCRC. Blood samples were collected after subjects had been in the sitting position for 10 min. After an additional 10 min of sitting, BP was measured in the right arm three times (with intervals of 2 min between readings) using a random zero sphygmomanometer (Hawksley and Sons, Lancing, West Sussex, UK); the average of the last two readings was used in the analyses. The number of measurements ranged from two to 27 per subject with a mean of 16. Plasma renin activity (PRA), PAC, and serum K were measured once but in some cases twice.

In the cross-sectional study, subjects were admitted to the GCRC the afternoon of the day before when measurements were made. They consumed a standardized diet for dinner and evening snack and for breakfast the following morning. After breakfast, subjects were supine for 1 h before measurements were made and blood samples drawn. BP was measured three times (2 min between readings) in the right arm using a Dynamap Pro Series 300 machine.

Assay procedures

K was measured using a COBAS MIRA analyzer (Roche Diagnostics, Indianapolis, IN); PRA was measured using a RIA for generated angiotensin I (Clinical Assays, GammaCoat RIA kit from Diasorin, Stillwater, MN); and PAC was measured by RIA (Coat-A-Count kit from Diagnostic Products Corp., Los Angeles, CA). The intraassay and interassay coefficients of variation were, respectively, for PRA, 4.6 and 7.6%, and for PAC, 5.4 and 13.1%.

SNP selection and genotyping

KCNK3 spans 40 kb on chromosome 2p23, and KCNK9 spans 92 kb on chromosome 8q24.3. Tagging SNP of KCNK3 and KCNK9 were selected from Single Nucleotide Polymorphism Database (http://www.ncbi.nlm.nih.gov/projects/SNP/) using the criteria of R² = 80% and minor allele frequency over 10% based on HapMap YRI (Yoruba from Nigeria) data. Seventy-four tagging SNP of KCNK9 and 15 of KCNK3 were selected. All SNP were in Hardy-Weinberg Equilibrium (P value >0.05). SNP location was determined from the annotations in the NCBI human genome assembly Build 36.2.

Genotyping was carried out using a modified single-nucleotide extension reaction with allele detection by mass spectrometry (Sequenom MassArray system; Sequenom, San Diego, CA). The assays were designed and run with either of two formats, hME or iPLEX. The genotyping success rate for each genotype was over 95%, and minor allele frequencies were over 10%. SNP coverage across both genes was evaluated using the program Haploview version 4.2, which examines the extent of linkage disequilibrium (LD) between pairs of SNP to estimate how well the selected SNP represented the genetic information contained in nongenotyped SNP.

Quality control and population stratification

Due to potential differences in minor allele frequencies and LD patterns, EA and AA were studied separately. We genotyped 57 ancestry-informative SNP markers (12, 13) 10 on the X-chromosome to correct for population stratification and exclude individuals who were misclassified as to ancestry. In addition, we performed principal component analysis on the ancestry-informative SNP markers using EIGENSTRAT and obtained the first 10 eigenvectors.

Statistical analysis

Longitudinal study

All BP data, which spanned the ages 5–25 yr, were used in the analyses. A semiparametric mixed-effect model for repeated measurement analysis of the longitudinal data were used to test for associations of each SNP with systolic and diastolic BP. The model incorporates genetic (SNP) effects on BP as a fixed effect at the population level and subject-specific growth curves of BP level as random effects. Nonparametric growth curve models with random subject effects were fitted using penalized splines, and additive genetic effects were tested after adjusting for sex, age, body mass index (BMI), and four principal component scores (15, 16) for accommodation of subpopulation differences. The code for these analyses was written using R software (version 2.13.1). For biochemical measurements such as PRA, PAC, ARR, and serum K, we used a parametric mixed-effect model with a random effect for data analysis. Genetic variations were included in the models of main factors of interest. We adjusted PAC for PRA and serum K as well as for sex, age, BMI, and principal component score (reflecting the magnitude of population stratification for each subject). The mixed-effect model analysis was implemented using SAS PROC MIXED (version 9.2). Many of the SNP were in LD, and thus, we estimated a gene-wide significance threshold using Nyholt's method (17) that takes into account LD among SNP and calculates an effective number of independent tests for a given gene. The estimated significance thresholds for KCNK9 that took into account the multiplicity of comparisons was P = 0.0011 for AA and 0.0014 for EA.

Cross-sectional study

There were nine AA sib pairs and 15 EA sib pairs. No significant correlation of phenotypes between siblings was observed based on the Spearman correlation test, and thus data collected from siblings were used in the analyses. A mixed-effect model was employed to test for additive genetic effects of each SNP with PRA, PAC, ARR, and serum K after adjusting for sex, BMI, and age including principle component scores as well as relevant phenotypes described above in Longitudinal study. Logarithms of PRA, PAC, and ARR were used because normality assumptions were not met.

Results

Characteristics of subjects

In the longitudinal study (Table 1), AA were heavier (P < 0.0001) and had higher BP (P < 0.0001) than EA. PRA, PAC, and ARR were lower in AA (P < 0.0001 to 0.002). For subjects in the cross-sectional study (Table 2), female AA had higher BMI (P < 0.0001) and higher systolic BP (P = 0.03) than female EA, and again PRA and PAC were lower in AA than in EA (P = 0.0002 to 0.03), but ARR was higher in AA than in EA (P = 0.03). Male AA had a higher serum potassium level than male EA (4.2 vs. 4.0; P < 0.01).

Table 1.

Characteristics of subjects at time of enrollment in the longitudinal study group (mean ± sd)

Characteristic	Males			Females
	EA (n = 229)	AA (n = 150)	P value	EA (n = 215)	AA (n = 201)	P value
Age (yr)	14.0 ± 4.1	14.7 ± 4.2	0.71	14.5 ± 4.4	15.0 ± 4.7	0.52
BMI (kg/m2)	20.9 ± 4.8	23.8 ± 6.2	<0.0001	21.2 ± 4.6	24.5 ± 6.9	<0.0001
SBP (mm Hg)	107.1 ± 12.7	109.4 ± 12.5	<0.0001	102.8 ± 10.1	105.2 ± 10.2	<0.0001
DBP (mm Hg)	62.2 ± 10.7	64.1 ± 10.8	<0.0001	61.7 ± 9.7	62.9 ± 9.7	0.0004
PAC (ng/dl)	14.0 ± 9.0	8.8 ± 6.5	0.0001	14.7 ± 9.6	9.2 ± 6.8	0.0001
PRA (ng/liter · sec)	3.3 ± 2.3	2.6 ± 2.0	0.002	3.3 ± 2.5	2.8 ± 2.1	0.0008
ARR	5.30 ± 3.6	4.9 ± 6.8	0.001	5.8 ± 4.5	5.1 ± 7.9	0.001
K (mmol/liter)	4.2 ± 0.4	4.3 ± 0.7	0.88	4.2 ± 0.4	4.3 ± 1.0	0.32

DBP, Diastolic BP; SBP, systolic BP.

Table 2.

Characteristics of subjects in cross-sectional study group (mean ± sd)

Characteristic	Males			Females
	EA (n = 38)	AA (n = 49)	P value	EA (n = 47)	AA (n = 60)	P value
Age (yr)	22.3 ± 3.8	22.8 ± 4.0	0.6	23.5 ± 4.6	24.0 ± 4.9	0.6
BMI (kg/m2)	25.4 ± 4.5	26.9 ± 5.6	0.17	25.9 ± 5.4	30.8 ± 8.6	0.0004
SBP (mm Hg)	122.0 ± 9.3	124.1 ± 10.8	0.36	110.0 ± 8.8	113.7 ± 8.7	0.03
DBP (mm Hg)	68.1 ± 6.6	67.9 ± 6.7	0.88	65.0 ± 7.1	64.2 ± 5.4	0.54
PAC (ng/dl)	9.0 ± 4.5	5.2 ± 3.1	0.0002	8.3 ± 5.3	5.9 ± 3.4	0.0004
PRA (ng/liter · sec)	1.4 ± 1.0	1.0 ± 0.9	0.03	1.5 ± 1.0	1.0 ± 0.8	0.0002
ARR	9.4 ± 6.8	12.6 ± 21.9	0.67	7.0 ± 6.0	12.0 ± 13.4	0.03
K (mmol/liter)	4.0 ± 0.2	4.2 ± 0.4	0.01	4.0 ± 0.3	4.1 ± 0.3	0.47

DBP, Diastolic BP; SBP, systolic BP.

Association studies using individual SNP

The pattern of LD differed between AA and EA for both KCNK3 and KCNK9 (Fig. 1). For KCNK9, EA have a large LD block consisting of 18 SNP (from rs7833765 to rs1619829, R² > 70%, D′ >80%, where D′ is the magnitude of linkage disequilibrium between two SNPs) as well as five pair-wise LD blocks. By contrast, AA had a smaller LD block containing four pair-wise LD blocks. Because of the LD, the 74 genotyped SNP provided information across the entire gene region of KCNK9, including 10 kb upstream and 10 kb downstream of the coding region. The average R² for CEU (Utah residents with Northern and Western European ancestry) was 0.976, capturing 83% (60 of 72) of known HapMap alleles with minor allele frequency (MAF) higher than 0.1 in the region at R² higher than 0.7. The average R² for YRI (Yoruba in Ibadan, Nigeria) was 0.938, capturing 67% (53 of 78) of known HapMap alleles with MAF higher than 0.1 at R² higher than 0.5. The 15 SNP of KCNK3 captured a similar amount of the variation for CEU (nine SNP capturing 89%) and YRI (11 SNP capturing 65%). Because a number of SNP we genotyped were not in the HapMap database (HapMap Data Phase III), the actual coverage of the region with SNP is underestimated.

Fig. 1.

Pairwise LD structure of KCNK9 (R²). Top panel, AA LD; bottom panel, EA LD. The KCNK9 gene structure is illustrated in the middle in scale; the black box represents exons; the arrow represents the direction of transcription. Red dots indicate the locations of nine SNP significantly associated with systolic BP in AA in the longitudinal study.

We found no association for any SNP in KCNK3 with BP or with any of the indices of aldosterone production in either AA or EA, and thus, only results of association studies with KCNK9 are reported in detail. The P values of individual SNP that showed associations with KCNK9 are presented in Tables 3 and 4. The direction of the arrow accompanying the superscript nucleotide depicts either an increase or decrease in the measured parameter.

Table 3.

Associations of selected SNPs in KCNK9 in subjects from the longitudinal study group (P values)

SNP	NT	AA (n = 351)						EA (n = 444)
		MAF	SBP	DBP	PAC	ARR	Serum K	MAF	SBP	DBP	PAC	ARR	Serum K	PRA
rs2542424	C/T	0.30	0.07	0.026C↑	0.20	0.12	0.56	0.44	0.05	0.09	0.16	0.74	0.87	0.02C↓
rs2545462	T/G	0.47	0.02T↑	0.11	0.12	0.64	0.08	0.40	0.33	0.21	0.92	0.67	0.66	0.23
rs888345	G/A	0.15	0.009G↓	0.02G↓	0.01G↓	0.003G↓	0.02G↑	0.20	0.29	0.15	0.005G↓	0.009G↓	0.56	0.19
rs13256087	A/G	0.27	0.90	0.91	0.70	0.40	0.56	0.37	0.04G↑	0.01G↑	0.37	0.46	0.58	0.50
rs1469039	A/G	0.19	0.03A↓	0.26	0.99	0.87	0.22	0.17	0.78	0.87	0.0009A↓	0.0005A↓	0.53	0.15
rs2545460	A/G	0.39	0.24	0.026A↓	0.87	0.89	0.15	0.47	0.73	0.52	0.10	0.03A↓	0.76	0.04A↑
rs12114521	A/G	0.49	0.02A↓	0.11	0.89	0.46	0.62	0.04	0.43	0.98	0.01A↓	0.08	0.31	0.82
rs3780045	T/C	0.13	0.01T↑	0.05	0.50	0.05	0.38	0.03	0.07	0.58	0.93	0.71	0.22	0.31
rs3824281	A/G	0.33	0.004A↑	0.10	0.51	0.23	0.05	0.06	0.45	0.87	0.01G↑	0.010G↑	0.54	0.73
rs7004779	T/C	0.15	0.31	0.63	0.16	0.11	0.01T↓	0.03	0.92	0.29	0.006T↑	0.004T↑	0.27	0.15
rs3780037	T/G	0.30	0.01T↑	0.02T↑	0.86	0.91	0.02T↑	0.23	0.11	0.17	0.73	0.29	0.04T↓	0.46
rs10110946	C/T	0.23	0.01C↑	0.11	0.73	0.91	0.11	0.24	0.10	0.14	0.77	0.36	0.03C↓	0.50
rs7828107	A/C	0.21	0.004A↓	0.83	0.49	0.09	0.29	0.49	0.11	0.37	0.50	0.38	0.01A↑	0.30

Superscript letters depict nucleotides associated with risk alleles; direction of the arrow depicts whether level of phenotype increases or decreases. DBP, Diastolic BP; NT, nucleotides with minor allele followed by major allele; SBP, systolic BP.

Table 4.

Associations of selected SNP in KCNK9 in subjects from the cross-sectional study group (P values)

SNP	NT	AA (n = 106)						EA (n = 83)
		SBP	DBP	PAC	ARR	Serum K	Urine K	SBP	DBP	PAC	ARR	Serum K	Urine K
rs4736287	G/C	0.01G↑	0.06	0.40	0.53	0.04G↓	0.83	0.94	0.95	0.22	0.96	0.28	0.38
rs2545432	A/G	0.01A↑	0.01A↑	0.37	0.89	0.13	0.90	0.70	0.62	0.46	0.34	0.03	0.12
rs741463	C/G	0.96	0.62	0.25	0.01C↓	0.04C↑	0.94	0.08	0.55	0.77	0.60	0.94	0.36
rs888345	G/A	0.03G↓	0.10	0.67	0.57	0.82	0.32	0.85	0.84	0.92	0.05	0.26	0.44
rs13256087	A/G	0.46	0.95	0.04A↑	0.006A↑	0.08	0.72	0.13	0.67	0.85	0.46	0.90	0.16
rs1469039	A/G	0.04A↓	0.53	0.97	0.56	0.03A↓	0.35	0.76	0.69	0.98	0.09	0.20	0.34
rs882555	T/C	0.57	0.32	0.02T↓	0.17	0.50	0.04T↓	0.98	0.40	0.06	0.79	0.71	0.07
rs2542481	A/G	0.03A↓	0.67	0.27	0.15	0.01A↑	0.30	0.21	0.52	0.67	0.66	0.85	0.99
rs888349	G/T	0.04G↓	0.39	0.50	0.38	0.02G↓	0.39	0.59	0.27	0.22	0.67	0.33	0.05
rs3780037	T/G	0.02T↑	0.29	0.45	0.23	0.89	0.005T↑	0.69	0.68	0.19	0.22	0.37	0.42
rs10110946	C/T	0.02C↑	0.09	0.20	0.10	0.13	0.01C↑	0.53	0.61	0.53	0.19	0.95	0.35
rs7828107	A/C	0.02A↓	0.27	0.22	0.35	0.36	0.73	0.83	0.29	0.95	0.56	0.53	0.69

Superscript letters depict nucleotides associated with risk alleles; direction of the arrow depicts whether level of phenotype increases or decreases. DBP, Diastolic BP; NT, nucleotides with minor allele followed by major allele; SBP, systolic BP.

Longitudinal study

We found associations of several SNP with each of the phenotypes measured, although often an association was found only in EA or only in AA (Table 3). Nine SNP in KCNK9 were associated with systolic BP in AA (in four instances, the SNP associated with a lower BP); none of them associated with BP in EA. An additional SNP was marginally associated (rs13256087, P = 0.04). Two of those nine SNP were also associated with diastolic BP in AA, along with two other SNP; in EA, the single SNP associated with systolic BP was also associated with diastolic BP. In AA, rs888345 showed an association with all five measures. In contrast to AA where associations were primarily with BP, the associations in EA were related primarily to parameters of aldosterone production, primarily PAC, ARR, and serum K. Among the five SNP associated with PAC in EA, four were also associated with ARR.

Cross-sectional study

In this study group, significant associations were observed only in AA (Table 4). Five of nine SNP that associated with systolic BP in AA from the longitudinal study also associated with systolic BP in subjects in the cross-sectional study (rs888345, rs1469039, rs3780037, rs10110946, and rs7828107). The direction of the change in BP for all five SNP was the same in the two study groups. Five SNP (rs4736287, rs741463, rs1469039, rs2542481, and rs888349) associated with serum K, four of which also associated with BP. Five SNP (rs741463, rs13256087, rs882555, rs3780037, and rs10110946) showed associations with PAC and/or ARR. Because the study protocol for the cross-sectional study included collection of urine samples, urinary K was also analyzed. Two SNP showed the direction of the association with urinary K excretion to be the same as the change in systolic BP (rs3780037 and rs10110946).

Effect size on BP

Effect size was estimated separately in EA and AA using a semiparametric mixed-effect model unadjusted for population stratification. The effect size for systolic BP of rs888345 was approximately 1.5 mm Hg with two copies of the risk allele (homozygous for the risk allele, A), compared with homozygous and heterozygous groups of the nonrisk allele, G. The effect size of rs888345 on diastolic BP was approximately 2–3 mm Hg with at least one copy of the risk allele.

Gene interactions

Although we observed no associations with KCNK3 in the current study, deletion of both Kcnk3 and Kcnk9 from the mouse genome resulted in a greater state of hyperaldosteronism than removal of either gene alone. We therefore looked for evidence of a genetic interaction that might enhance associations with BP, serum K, PAC, or ARR. In an exploratory analysis, we tested whether certain polymorphisms in KCNK9 might permit and/or augment the phenotypic expression of genetic variations in KCNK3. The three SNP in KCNK9 that showed the most significant associations (rs888345, rs3780037, and rs3824281) and the SNP in KCNK3 were studied. The analysis was limited to AA. Using an additive interaction model, we found that rs34292597 in KCNK3 and rs888345 and rs3780037 in KCNK9 were associated with serum K interactively (P < 0.005), whereas rs34292597 alone was not. In addition, there was evidence to suggest that any of eight SNP in KCNK3 and rs3824281 in KCNK9 interacted to enhance associations with BP, serum K, PAC, and ARR. Thus, there was suggestive evidence for an interaction of the two channel subunits in humans.

Additive effects

To look for additive effects, we incorporated the nine SNP into a single model as additive terms. In so doing we found that two of the nine SNP remained significant: rs3780037 (P = 0.0023) and rs7828107 (P = 0.046). The remaining SNP appeared to function nonadditively. The individual effect sizes of rs3780037 and rs7828107 were 1.7 and 2.8 mm Hg, respectively. We estimated the additive effect size of the two SNP when homozygous for the risk alleles (TT for rs3780037 and CC for rs7828107) would be 3.8 mm Hg higher than those heterozygous (GT for rs3780037 and AC for rs7828107). The heterozygous group was estimated to be 3.5 mm Hg higher than the group homozygous for the nonrisk alleles (GG for rs3780037 and AA for rs7828107).

Discussion

Although it is widely accepted that BP variability, including the development of hypertension, has a strong genetic basis (6, 18), BP genes have been exceptionally elusive (19). Most hypertension is accompanied by an exaggerated retention of Na (20). It therefore follows that genetic contributions frequently encode for proteins involved with the reuptake of Na in the kidney. What has become increasingly apparent is that aldosterone with its Na-retaining properties participates in the pathogenesis of hypertension including common forms (4, 5). Production of aldosterone is highly sensitive to the transmembrane potential of the ZG cells, a characteristic coupled to the activity of K channels (21, 22). We report on translational studies, taking findings in the mouse (deletion of Kcnk3 and Kcnk9 led to hyperaldosteronism and hypertension) (12) and exploring those genes in human subjects. Our sample sizes were not of optimal size, but this was mitigated in part by the refined phenotypic measurements and the choice of young subjects. In the longitudinal study group (n = 795), BP was measured biannually over a period of years, and in the cross-sectional study group (n = 189), assessments were carried out in the GCRC under tightly controlled and standardized conditions.

We found common variations in KCNK9 that associated with BP and production of aldosterone. In mice, deletion of Kcnk9 alone also produces hyperaldosteronism, albeit with a milder phenotype [aldosterone excretion rates increased 37%, whereas they increased 400% in Kcnk9/Kcnk3 double-knockout mice (23); BP also increased more in the double knockout than in the Kcnk9]. In the present study, variations in KCNK3 showed no associations; however, an interaction of the genes was suggested in that selected SNP from each gene showed stronger associations with BP, serum K, PAC, and ARR than did individual SNP.

Associations were observed in both the longitudinal and the cross-sectional study groups and in AA and EA with P values <0.05. We analyzed 74 SNP to capture most of the variability in KCNK9, in keeping with the exploratory intent of the study. One SNP (rs1469039 in EA in the longitudinal group) met the stringent threshold for significance given the multiple testing. Several others were nominally significant and worth targeted exploration in other samples.

The findings were also noteworthy in that individual SNP associated with multiple phenotypes whose directional change was consistent with the known biology (PAC and ARR increased or decreased as BP increased or decreased with serum K going in the opposite direction; an exception was rs3780037 in AA, Table 3). In the longitudinal group, rs888345 showed associations with five phenotypes in AA and two in EA. Additionally, six SNP that showed an association in the longitudinal study group showed an association in the cross-sectional study. There were, therefore, elements of replication when one considers associations with the same SNP in two separate race groups as well as in separate study groups. These internally and externally consistent findings suggest that the associations go beyond simply mere coincidence, despite their only nominal significance given the low sample size. The observations should be tested in more directed future studies.

In the longitudinal study group, where BP had been measured repeatedly over time and thus from the sheer numbers of observations could be more informative, multiple SNP in KCNK9 showed an association with BP in AA (nine SNP associated with systolic BP) (Table 3). At the same time, there was no similar degree of association with aldosterone production. However, levels of aldosterone (and PRA) were measured cross-sectionally with fewer overall actual assessments, and thus as phenotypes, they may have been less robust. In addition, because BP in AA is characteristically salt sensitive (24), minor increases in Na reabsorption from small differences in the level of aldosterone may have been enough to increase BP. Evidence for a genetic influence on aldosterone production was, on the other hand, apparent in EA but without a comparable change in BP. In this case, BP may simply have been less responsive to the Na retention evoked by aldosterone. In the cross-sectional study, no SNP in EA associated with any of the phenotypes. The findings are in keeping with AA residing closer to a threshold wherein only a small increment in Na uptake may be sufficient to affect BP.

Although the genetic variations in KCNK9 reported here in healthy individuals are not contained within the coding sequences of the gene, it is possible that they are in LD with coding variants or that intronic variants modify the expression of TASK3 subunits to produce changes in the K conductance of the ZG cell. If the reduction in K conductance is small, the baseline membrane potential of the ZG cell may not change significantly and the expressed phenotype, aldosterone production, may be normal. However, a reduced K conductance would render the ZG cell more susceptible to depolarizing influences (such as angiotensin II and ACTH) and, thus, we predict may modestly amplify responses to physiological concentrations of aldosterone secretagogues and thus increase the risk for hypertension. By contrast, mutations in the coding region of KCNJ5 channels found in a subset of patients with tumorigenic primary aldosteronism that change the ion selectivity of the channel pore itself are responsible for large overproduction of aldosterone observed in some patients with severe primary aldosteronism (25). The same group also found KCNK9 and KCNK3 expressed in aldosterone-producing adrenal adenomas (25).

Our subjects were healthy and for the most part too young to have hypertension. Although this could be considered a limitation, at the same time, it provided the advantage of having eliminated the confounding influences that accompany age and/or the hypertensive state, including the attendant use of antihypertensive medications. Although genetic associations were observed in AA and EA, providing some measure of replication, confirmation of our findings in other populations will be necessary to fully establish a role for KCNK9 in causing BP variability.

In summary, genetically derived differences in the actions of a K channel that is expressed in ZG cells and that normally functions to restrain aldosterone production showed associations with BP and level of plasma aldosterone in this exploratory study. The findings will hopefully encourage additional testing of associations in other population groups.