|Home | About | Journals | Submit | Contact Us | Français|
Endogenous ouabain (EO) has been linked with long-term changes in sodium balance and cardiovascular structure and function. The biosynthesis of EO involves, cholesterol side-chain cleavage (CYP11A1), 3-β-hydroxysteroid dehydrogenase (HSD3B) with sequential metabolism of pregnenolone and progesterone. Furthermore, the renal excretion of cardiac glycosides is mediated by the organic anion transporter (SLCO4C1) at the basolateral membrane and the P-glycoprotein (PGP) (encoded by MDR1) at the apical membrane of the nephron.
Average 24-h ambulatory blood pressures were recorded in 729 untreated essential hypertensives. Aldosterone (Aldo), EO, urinary Na+, and K+ excretions were determined. Single-nucleotide polymorphism (SNP) and haplotype-based association study was performed with a total of 26 informative SNPs.
Plasma EO was significantly directly related to both day (r = 0.131, P < 0.01) and nighttime diastolic blood pressure (DBP) (r = 0.143, P < 0.01), and remained significantly related after correction for confounders (sex, body mass index, age). Genotype analysis for EO levels and daytime DBP gave significant results for CYP11A1 rs11638442 and MDR1 rs1045642 (T/C Ile1145) in which the minor allele tracked with higher EO levels (T/T 210.3 (147–272) vs. C/C 270.7 (193–366) pmol/l, P < 0.001). Association was found between HSD3B1 polymorphisms and/or haplotypes with blood pressure (systolic blood pressure (SBP) 140.3 (11.7) vs. 143.8 (11.2) mm Hg, P < 0.01) and plasma Aldo (P < 0.05). Haplotype-based analyses support the data of SNP analysis.
Among patients with essential hypertension, cholesterol side-chain cleavage and MDR1 loci are related to circulating EO and DBP, most likely by influencing EO synthesis and transmembrane transport, respectively. In contrast, variants in HSD3B1 are related with SBP probably via Aldo.
Endogenous ouabain (EO), a mammalian counterpart to the plant-derived cardiac glycoside ouabain, is synthesized and released from the adrenal glands and possibly from the hypothalamus. EO is a versatile modulator of the ubiquitously expressed Na+ pump and has been linked to human essential hypertension and cardiac hypertrophy.1–6 Also in genetically hypertensive Milan rats (MHS), Na/K-ATPase activity and blood pressure levels are associated (M. Ferrandi, personal communication) to elevated hypothalamic and plasma levels of EO compared to control rats (MNS).7,8 The molecular basis for the elevated EO in MHS was probed in the hypothalamus and adrenal using bioinformatics and genomic techniques. Elevated transcripts for cholesterol side-chain cleavage (also known as cytochrome P450scc, CYP11A1) and β-hydroxysteroid dehydrogenase/δ5-4 isomerase genes (HSD3B) were detected in hypothalamus but not in the adrenal.9 Other studies have suggested that, as with aldosterone (Aldo), the biosynthesis of cardiac glycosides by the adrenal gland likely involves cholesterol side-chain cleavage to form pregnenolone with further metabolism of progesterone.10–12
Recently, we observed marked increases in circulating EO in uremic patients.13 These data imply that, in addition to secretion, renal clearance is one of the major determinants of plasma EO. The recovery of the cardiac glycosides digoxin and ouabain at the basolateral membrane of the proximal tubular cell is mediated by an organic anion transporting polypeptide (SLCO4C1).14 In addition, the secretion of digoxin at the apical membrane is mediated by the P-glycoprotein (PGP) efflux transporter, a protein encoded by the multidrug resistance 1 gene (MDR1).15 The MDR1 has a broad substrate specificity and is modulated transcriptionally by steroid hormones including ouabain.16–18 Recent work shows that MDR1 variants are associated with blood pressure in humans.19
Accordingly, the CYP11A1, HSD3B, SLCO4C1, and MDR1 genes may modulate blood pressure via changes in circulating EO and Aldo. To address this, we probed for the association of common single-nucleotide polymorphisms (SNPs) and haplotypes of these genes with blood pressure, Aldo, and EO levels in patients with never-treated mild essential hypertension. A total of 26 SNPs were genotyped. These SNPs were selected by tagging or functional criteria and accounted for most of the variants for these genes in the Caucasian population.
A cohort of 729 (548 males and 181 females) newly discovered and never-treated hypertensive patients were enrolled in the study. These patients were consecutively referred to the outpatient clinic at the San Raffaele Hospital, Milan, between January 1997 to January 2008, and informed consent was obtained from each individual. Patients with other significant concomitant illness were excluded as previously described.20 Ambulatory blood pressure monitoring was performed using Spacelabs 90207 (Space Labs, Redmond, WA) devices. Subjects were monitored on a day chosen for typical weekly activity, and day and nighttime were defined according to patient’s diary. A total of 430 participants agreed to 24-h urinary collection for sodium (UNa), potassium (UK), and blood was withdrawn for plasma renin activity (PRA), plasma Aldo, plasma EO, plasma sodium (PNa), and potassium (PK), on the day of ambulatory blood pressure monitoring recording.
To investigate the relationship between genotypes and the mentioned phenotypes as categorical variables, the patients were divided into those with mild hypertension, that is, day-time BP <140/90 mm Hg, and a second group with moderate hypertension and daytime BP >140/90 mm Hg. Levels greater than the median values defined patients with high plasma EO, that is, 280 pmol/l in men and 220 pmol/l in women.
Urinary and plasma Na and K were determined by flame photometry. Plasma EO was determined by radioimmunoassay on extracted samples using a specific antiserum as previously described.20 PRA and Aldo were measured by commercial radioimmunoassay (Sorin Laboratories, Saluggia, Italy).
A total of 26 SNPs were selected from public databases available at the NCBI dbSNP and at the International HapMap Project websites,21 http://www.ncbi.nlm.nih.gov and http://www.hapmap.org, respectively, and are listed in Table 1.
Five tagSNPs for CYP11A1 and nine tagSNPs for SLCO4C1 spanned the entire genic regions of interest. These SNPs tagged for each locus all the common SNPs in linkage disequilibrium (LD) (r2 > 0.8) among them, based on CEU HapMap (release 19/phaseII October 2005, on NCBI B34 assembly, dbSNP b124) and allowed a complete linkage block analysis.
Furthermore, for HSDβ3 locus, we considered both type 1 and type 2 isoforms encoded by two distinct genes (HSD3B1 and HSD3B2) because they are arranged in tandem 84 Kb apart on chromosome 1 and they are expressed in a tissue-specific pattern. Among the eight informative SNPs spanning the chromosomal region for both genes, SNP rs6203 (C/T Leu338) was previously analyzed in association with blood pressure in two different studies.22,23 Finally, four exonic SNPs in the MDR1 locus were analyzed, including rs1045642 (T/C Ile1145) and rs2032582 (G/T Ala893Ser), which influence the expression of MDR1-encoded PGP and alter the transport of many drugs and modulate digoxin kinetics.24,25
SNP genotyping was performed on genomic DNA from blood samples, with three alternative methods: (i) real-time PCR SNP genotyping with allele-specific MGB probes; (ii) allele-specific PCR SNP with universal energy transfer primers; and (iii) PCR-RFLP analysis, according to the manufacturers standard PCR protocols.
Reported data are means ± s.d. We normalized the distributions of plasma EO, Aldo, PRA, and of the urinary excretion rates of sodium and potassium by a logarithmic transformation. We reported the geometric mean and the inter-quartile intervals. Hardy–Weinberg equilibrium was assessed by χ2-analysis. LD was calculated using Haploview 3.2 software package (http://www.broad.mit.edu/mpg/haploview).26 The dominant or recessive allele effect has been tested using one-way analysis of variance and the Bonferroni post hoc multiple comparisons analysis. Pooled data are presented when the dominant or recessive allele effect was found significant P < 0.05. For all gene regions, the selected SNPs were combined as a single block by definition of Gabriel et al.,27 and subjects with >25% missing genotypes were excluded from analysis. To reconstruct the haplotypes and to estimate their frequency, we used the expectation–maximization algorithm as implemented in the SAS Genetic (SAS Institute, Cary, NC), version 9.1 and in Haploview. Single locus and haplotype association tests for categorical variables were computed with Haploview, with reported P values corrected for multiple testing by 10,000 permutations. Metric variables were compared between groups (genotypes) by analysis of variance. In this case we used SNPSpD interface (http://gump.qimr.edu.au/general/daleN/SNPSpD) to perform intragenic correction for multiple testing of SNPs in LD on the basis of the Spectral Decomposition of matrices of pairwise LD between SNPs.28
The clinical characteristics of the subjects (548 males and 181 females) are shown in Table 2. There were significant differences between men and women in some variables, including body mass index, and systolic blood pressure (SBP) and diastolic blood pressure (DBP). A subgroup of 335 males and 95 females was further characterized for plasma EO, Aldo, PRA, and urinary Na and K excretion. Plasma EO, urinary Na, and K excretion were higher in males. Similarly, cholesterol, without reaching statistical significance, high-density lipoprotein, and triglycerides were lower in females.
On univariate analysis, plasma EO was significantly directly related to both day (r = 0.131, P < 0.01) and nighttime DBP (r = 0.143, P < 0.01). In multiple linear regression models, including those with a series of potentially confounding variables (age, body mass index, sex), plasma EO maintained an independent association with day and nighttime DBP (b = 0.12, P < 0.05).
The allele frequency for each SNP in the present population (Table 1) is in agreement with that reported by HapMap. In addition, genotype data for all 26 SNPs were consistent with the Hardy–Weinberg equilibrium (P > 0.05).
When the patients divided into mild and moderate hypertensive groups were analyzed, Haploview analysis gave significant results (P < 0.05) for three SNPs (rs 2236780, rs3765945, rs6203) in the chromosomal region that included the HDS3B1 gene. (Supplementary Table S2 online). A significant result was also obtained for one SNP rs1045642 (T/C Ile1145) in MDR1 when patients were clustered in high- and low-plasma EO groups. In both analyses, significant levels were adjusted by 10,000 permutations and the association of HSD3B1 rs2236780 with mean blood pressure, and MDR1 rs1045642 with plasma EO, remained evident (Supplementary Table S2 online).
In addition, the continuous phenotypes, SBP, DBP, and plasma EO, were evaluated in the entire population by the above-mentioned genotypes after adjustment for the confounding risk variables. After correction for multiple testing,28 HSD3B1 SNPs, rs2236780, rs3765945, and rs1047303, remained significantly associated with SBP with a global increment of blood pressure from homozygotes for major allele (11) to homozygotes for minor allele (22). Higher plasma Aldo and lower plasma potassium were associated with the minor allele of rs3765945 only (209.8 (148–344) vs. 175.0 (120–267) ng/dl and 4.15 ± 0.33 vs. 4.23 ± 0.38 mEq/l respectively). The significant P values observed prior to correction are likely to be relevant in any case because these gene polymorphisms have plausible roles in determining the levels of BP and EO as previously suggested.7,8 Therefore, under a Bayesian framework,29 the P value should be interpreted with reference to the a priori probability of association. As a consequence, HSD3B1 SNPs, rs2236780, rs3765945, and rs6203, and SLCO4C1 rs841922 are likely associated with DBP (Table 3).
Genotype analysis for EO levels and daytime DBP gave significant results for CYP11A1 rs11638442, and MDR1 rs1045642 (T/C Ile1145) in which the minor allele tracked with higher EO levels. Under the Bayesian hypothesis,29 both CYP11A1 rs11638442 and MDR1 were correlated with plasma EO levels (Table 3). In addition, a different CYP11A1 SNP rs1484215 was associated with plasma cholesterol (Table 3).
No association between any of the SNPs and the PRA/Aldo ratio was found.
Haplotype analysis included only haplotypes having >5% total frequency in the population (major haplotypes). If we considered patients clustered into mild and moderate hypertensive groups, Haploview haplotype analysis (corrected by permutations) gave significant results for HSD3B1 haplotype 2; in addition, when the patients were divided into high- and low-EO levels, a borderline result was obtained for the MDR1 haplotype 2 (Supplementary Table S3 online).
For continuous SBP, DBP, and EO phenotypes, analysis of variance of the haplotypes was performed in the entire population (not shown). The patients were grouped, using the post hoc analysis to verify the dominant effect, in individuals with 2 (+/+), 1 (+/−), or 0 (−/−) copies of the described haplotype and the significant results are reported in Table 4. The HSD3B1 haplotype 2 showed significance for SBP and DBP with the highest blood pressure values in the carrier of two copies of this haplotype, whereas an opposite trend was displayed by HSD3B1 haplotype 1. Haplotype analysis for plasma EO levels were significant for both MDR1 haplotype 1 and haplotype 2 with the lowest EO levels in homozygous carriers of haplotype 1 and the highest EO levels in the carriers of two copies of haplotype 2.
The major new results of the present study are as follows. First, the SNPs, rs11638442 and rs1045642, in the CYP11A1 and in MDR1 gene, respectively, affect circulating EO. A small set of informative tagSNPs and haplotype analysis for the entire region of these genes in our hypertensive population, which were characterized also for EO levels, allowed us to identify these associations. Haplotype-based analyses of these genes did not increase the power of association with respect to SNP analysis. Second, variants in HSD3B1 are associated with blood pressure and variations in plasma Aldo and potassium among patients with essential hypertension. Third, our investigation provides no evidence for an association of the specific gene variants, including at-risk common haplotypes of the HSD3B2, CYP11A1, and SLCO4C1 genes, with blood pressure or plasma EO in our study population. Nevertheless, under a Bayesian framework with reference to the a priori probability of association, SLCO4C1 and CYP11A1 were associated significantly with DBP and plasma EO levels, respectively. Fourth, the present study shows that plasma EO was significantly related with day and nighttime DBPs among patients with never-treated essential hypertension. This observation reinforces previous reports from our laboratory.
The 3b-hydroxysteroid dehydrogenase/δ5-4 is a key rate-limiting enzyme in steroid biosynthesis pathways that produce estradiol, testosterone, cortisol, and Aldo. It is expressed as two tissue-specific isoforms (HSD3B1 and HSD3B2) with different substrate affinities.30 This enzyme has been shown to be in the biosynthetic pathway for EO based on studies with cultured adrenocortical cells10 and in the MHS genetic model of hypertension.9 We investigated both isoforms considering the following: (i) These two genes are arranged in tandem on the same chromosome; (ii) Although the quantitative analysis in rats identified differential expression for HSD3B2, the oligonucleotide used for gene silencing with RNAi does not distinguish between the two high homologous isoforms;9 (iii) A quantitative trait locus for blood pressure was identified in the HSD3B2 chromosomal region in genome-wide linkage analysis;31 and (iv) In a population of Swedish normotensives, Rosmond et al.22 found a positive association of blood pressure with HSD3B1 rs6203 (C/T Leu338) with higher SBP and DBP in carrier of the C allele. This SNP was, however, not associated with hypertension in a cohort of Caucasian Australian hypertensives,23 although the study revealed C allele tracked with a higher DBP.23 Our analysis confirms that the C allele is associated with a higher DBP; however, the statistical significance was lost after correction for multiple testing.
TagSNPs and haplotype analysis in our population add further support for the involvement of HSD3B1 in blood pressure regulation, with rs3765945 and rs1047303 showing association with SBP. In contrast, variants in the HSD3B2 gene, located only 84 Kb away from HSD3B1, were not associated with blood pressure in our study. Further tagging of the intergenic region not containing known genes, with two additional informative SNPs, excluded association with the described phenotypes. Furthermore, the observation that the HSD3B1 gene was associated with increased plasma Aldo and lower plasma K+ suggests that a gain of function mutation in HSD3B1 gene in LD with the intronic rs3765945 augments the biosynthesis of this hormone; however, there was no association between these or other SNPs, when both the PRA/Aldo ratio and the urinary K excretion were considered, which seems surprising given that these two parameters may be considered to be a more sensitive index of Aldo action than that measured using the plasma Aldo only.
Functional polymorphism rs1045642 (T/C Ile1145) in the MDR1 gene has been previously identified24,25 as determinant of transporter activity and variability in drug disposition and efficacy. Hoffmeyer et al.24 showed that individuals homozygous for the T allele had significantly lower intestinal MDR1 and higher plasma levels of digoxin. In contrast, Nakamura et al.32 and Sakaeda et al.33 observed higher MDR1 expression in duodenal enterocytes and lower plasma digoxin in association with homozygosity of the T allele. Thus, there is an inverse relationship between MDR expression and serum digoxin levels in humans. In the present study, we restricted our analysis to 4 mutations in the coding region for MDR1 instead of 14 tagSNPs required to tag the entire gene region (http://www.hapmap.org). We found an association between rs1045642 (T/C Ile1145) and plasma EO with the highest circulating levels in subjects carrying the CC genotype.
It is also known that ouabain, at nanomolar concentrations, can modulate MDR1 gene transcription and PGP synthesis.17 MDR1 is expressed in liver, intestine, kidney, brain, and the adrenal gland and regulates the transport (i.e., intestinal absorption and renal elimination) of many hydrophobic compounds including digoxin34,35 and other steroids. However, ouabain and EO are highly polar steroids; therefore, at present, it seems less likely that PGP would have a significant impact on circulating EO via an excretory pathway. With this in mind, it is of interest that PGP exerts a profound influence on the regulation of the hypothalamic–pituitary–adrenocortical system.34,36 Thus, PGP variants might affect plasma adrenocorticotropic hormone, which is a known stimulator to EO secretion,37 and/or PGP itself may facilitate the transmembrane secretion of EO from the adrenal cortex. Our findings show that functional polymorphisms of the MDR1 gene influence the basal circulating level of EO in hypertensive humans. Further work will be needed to distinguish the mechanism by which MDR genotypes affect plasma EO.
P450scc, the mitochondrial cholesterol side-chain cleavage enzyme, is the only enzyme that catalyzes the conversion of cholesterol to pregnenolone; thus, it is the common pathway leading to the production of progesterone, androgens, estrogen, and other steroid hormones. Moreover, the adrenal biosynthesis of EO is thought to involve cholesterol side-chain cleavage with sequential metabolism of pregnenolone and progesterone.11,37,38 Therefore, it is of interest that we found an association of the CYP11A1 gene with serum cholesterol levels, circulating EO, and DBP in our study population. Plasma cholesterol and high-density lipoprotein-cholesterol are typical multifactorial phenotypes affected by genetic and environmental factors. Accordingly, the influence of a common polymorphism in CYP11A1 genes directly involved in the cholesterol biosynthesis pathway is surprising and suggests, at its simplest level, a substrate-product relationship with EO that will require further studies to define.
Recently, we showed39 that EO in cooperation with the citoskeleton protein adducin, increases tubular Na reabsorption in hypertensives through their action on the Na-K ATPase. Accordingly, we tested the interaction between those genes, involved EO metabolism, and adducin polymorphisms; however, the results were not conclusive because of the small subgroups of patients. Further studies should be carried out in a larger cohort of patients.
In conclusion, among a large cohort of patients with mild-to-moderate hypertension, we employed a candidate-gene approach using a small set of highly informative SNPs to probe for significant associations of blood pressure and related parameters. We observed a significant association of HSD3B1 and CYP11A1 variants with blood pressure, likely based upon contemporary understanding, and to be mediated through plasma Aldo and EO, respectively. Furthermore, we provide the first evidence that variations in the MDR1 gene are linked with plasma EO among patients with never-treated essential hypertension. Our results reinforce the view that inherited variations in steroid biosynthetic pathways as well those potentially involved with EO transmembrane transport have significant effects on the circulating levels of Aldo, EO, and blood pressure in patients with mild-to-moderate essential hypertension.
Research included in this report was supported by the European Union (grants IC15-CT98-0329-EPOGH, LSMH-CT-2006-037093 InGenious HyperCare, and Ministero Universitá e Ricerca Scientifica of Italy (PRIN grant 2006065339_01) and in part by USPHS HL75584 and HL78870 (J.M.H.). We acknowledge the expert technical assistance of Cinzia Scotti.
Disclosure: G.B. is advising Sigma-Tau (Pomezia, Italy). G.T., M.F., R.M. are employees of pharmaceutical company research institute (Prassis Sigma-Tau). Other authors declare that they have no conflict of interest.