Blood pressure and human genetic variation in the general population

Abstract

Purpose of review

Hypertension is a complex trait with multiple environmental and genetic contributors. Linkage studies of rare Mendelian disorders of hypertension and hypotension have produced the most notable progress toward understanding the heritable basis of blood pressure. Association studies to identify common variants have been limited in the past by small sample sizes and most findings lack replication.

Recent Findings

Recently, well-powered, targeted candidate gene and genome-wide association studies have reported reproducible associations between common genetic variants and blood pressure.

Summary

Identification of novel blood pressure genes will lead to an improved understanding of blood pressure regulation and the potential for novel therapies.

Introduction

Elevated blood pressure (BP) or hypertension (HTN) increases the risk for stroke, congestive heart failure, coronary heart disease, and end-stage renal disease [1–3]. Lifestyle changes and antihypertensive therapy are associated with reductions in morbidity and mortality [4–6]. However, current population surveys indicate that control rates are generally poor [7]. An improved understanding of the environmental and genetic determinants of BP variation could point to novel avenues to prevent HTN and its complications.

The genetic basis of HTN at the population level has only recently begun to yield to intense investigation. Prior Current Opinion in Cardiology reviews from Binder et al. in 2007 and Hamet et al. in 2007 covered genome-wide linkage approaches in family- and population-based studies and early candidate gene studies. The current review focuses on genetic discoveries in 2008 and 2009.

Population burden of hypertension

More than a quarter of the world’s adult population is hypertensive [8]. In North America roughly 1 in 3 adults have HTN. Most individuals eventually become hypertensive if they live long enough, with a lifetime risk of ~90% in individuals who are normotensive at age 55 [9]. Despite the high prevalence, almost one third of all hypertensive individuals are unaware, half untreated, and three fourths uncontrolled [3].

Hypertension is a dichotomous trait, but BP shows a continuous relationship to outcomes. A meta analysis including one million adults found increasing risk of cardiovascular mortality with increasing BP starting as low as 115/75 mm Hg [1]. The Framingham Heart Study found high-normal BP (SBP 130 to 139 mm Hg, DBP 85 to 89 mm Hg) to be associated with increased cardiovascular risk [10]. These observations are consistent with a graded and continuous relationship of increasing BP with risk of cardiovascular disease.

Observational studies have reported the association of elevated BP with higher salt intake [11], excessive alcohol consumption [12], higher BMI [13] and reduced physical activity [14]. Interventional trials have established that modification of each of these risk factors lowers BP effectively, in some cases in excess of the BP-lowering effect of single-medication therapy [3–5,15,16].

Genetic factors influence BP and hypertension

Intra-individual BP variation, measurement error, strong environmental contributions, and anti-hypertensive therapy of varying intensity pose challenges to the genetic study of BP. Despite a large number of covarying factors, family studies have shown significant genetic contributions to inter-individual differences in BP [17–20]. Long-term BP estimates have been shown to be highly heritable, with 50–60% of long-term systolic or diastolic BP (SBP, DBP) attributable to additive genetic factors [20].

The genetic architecture—a function of the number of variants as well as their frequency and strength of effect—that underlies this heritability has been unknown. Genome-wide linkage studies in rare families with Mendelian forms of HTN and hypotension successfully identified rare amino acid-altering variants that have relatively strong effects on salt-handling genes (Table 1). These have been well reviewed elsewhere [21]. While these rare variants clearly play a major role in BP variation within the families that harbor them, what role they play in overall BP variation has been unclear. Studies in the past two years have added importantly to our understanding of this genetic architecture.

Table 1. Mendelian syndromes of hypertension and hypotension.

Shown below are the genes that are well described to harbor rare mutations that cause Mendelian forms of hypertension and hypotension. Different mutations in some of the same genes (^*) can lead to both hyper- or hypo-tension depending on whether they result in gain or loss of protein function.

gene	gene name (alias)	disease	gain(+) or loss(−) of function	refs
Hypertension
CYP11B1	cytochrome P450, family 11, subfamily B, polypeptide 1	11 β hydroxylase deficiency	−	[22]
CYP11B2*	cytochrome P450, family 11, subfamily B, polypeptide 2 (aldosterone synthase)	glucocorticoid remediable aldosteronism	+	[23]
CYP17A1	cytochrome P450, family 17, subfamily A, polypeptide 1	combined 17α-hydroxylase/17,20-lyase deficiency	−	[24,25]
CYP17A1	cytochrome P450, family 17, subfamily A, polypeptide 1	isolated 17,20-lyase deficiency	−	[26]
HSD11B2	hydroxysteroid (11-beta) dehydrogenase 2	apparent mineralocorticoid excess	−	[27,28]
NR3C2*	mineralocorticoid receptor (MCR)	mineralocorticoid receptor mutation	+	[29]
SCNN1B* SCNN1G*	sodium channel, nonvoltage-gated 1 beta and gamma subunits	Liddle’s Syndrome	+	[30– 32]
WNK1	WNK lysine deficient protein kinase 1	pseudohypoaldosteronism type II (Gordon Syndrome)	+	[33]
WNK4	WNK lysine deficient protein kinase 4	pseudohypoaldosteronism type II (Gordon Syndrome)	−/+	[33– 35]
Hypotension
BSND	Bartter syndrome, infantile, with sensorineural deafness	Bartter’s syndrome Type 4	−	[36]
CASR	calcium-sensing receptor	Bartter’s syndrome Type 5	+	[37,38]
CLCNKB	chloride channel Kb	Bartter’s syndrome Type 3	−	[39]
CYP11B2*	cytochrome P450, family 11, subfamily B, polypeptide 2	corticosterone methyloxidase type I and II deficiency	−	[40,41]
CYP21A2	cytochrome P450, family 21, subfamily A, polypeptide 2	21-hydroxylase deficiency	−	[42,43]
HSD3B2	hydroxy-delta-5-steroid dehydrogenase, 3 beta- and steroid delta-isomerase 2	3 β hydroxysteroid dehydrogenase deficiency type II	−	[44]
KCNJ1	potassium inwardly-rectifying channel, subfamily J, member 1 (ROMK)	Bartter’s syndrome Type 2	−	[45]
NR3C2*	mineralocorticoid receptor	pseudohypoaldosteronism type I	−	[46]
SCNN1A SCNN1B* SCNN1G*	sodium channel, nonvoltage-gated 1 alpha, beta and gamma subunits	pseudohypoaldosteronism type I	−	[47]
SLC12A1	solute carrier family 12 member 1 (Na-K-2Cl cotransporter, NKCC2, loop diuretic target)	Bartter’s syndrome Type 1	−	[48]
SLC12A3	solute carrier family 12 member 3 (Na-Cl cotransporter, NCCT, thiazide diuretic target)	Gitelman’s syndrome	−	[49]

^*gene associated with both hypertension and hypotension

Do rare variants at the population level influence BP?

Richard Lifton’s group and others have identified many disease-causing mutations in renal salt-handling genes, placing these genes and pathways centrally in BP regulation (Table 1). Whether genetic variation in these genes at the population level contributes to inter-individual differences in BP was an open question. It is possible that rare mutations in a gene lead to a detectable phenotype in a handful of families, but that low prevalence (due to negative selection) makes variation in these genes a negligible contributor to differences in BP in general. A series of reports on a gene responsible for Gitelman’s syndrome, an autosomal recessive syndrome characterized by hyplochloremic metabolic alkalosis, hyopokalemia and hypocalciuria, have expanded our understanding of the genetic architecture of BP.

Homozygous mutations in families

In a linkage study in 12 families, Simon et al. mapped Gitelman’s syndrome to the SLC12A3 gene encoding the thiazide-sensitive Na-Cl cotransporter (NCCT) [49]. They identified homozygous loss-of-function mutations in SLC12A3 that result in salt wasting from the distal convoluted tubule. The authors speculated that heterozygous carriers of mutant alleles (one mutated copy, one normal copy of the gene) might be protected against development of HTN.

Heterozygous mutations in a large family with Gitelman’s syndrome

Cruz et al. examined an Amish kindred with Gitelman’s syndrome, including 60 individuals with zero, 113 with one and 26 with two copies of the mutant allele [50]. Blood pressure was significantly lower in heterozygous children but not in adults, perhaps due to limited power, age effects or improved dietary compensation in adults. These observations in a single family raised the intriguing possibility that heterozygous mutations in renal salt-handling genes at the population level might affect BP, not just in families selected for Gitelman’s syndrome.

Heterozygous mutations in the general population

In 2008, Ji et al. reported a screen in 3,125 individuals (many in families) from the community-based Framingham Heart Study for mutations in 3 candidate genes in which homozygous mutations cause Gitelman’s and Bartter’s syndromes: NCCT (SLC12A3); NKCC2 (SLC12A1); and ROMK (KCNJ1) (Table 1) [51] ••. Using criteria of complete conservation across multi-species comparisons with rare allele frequency (<1 per 1,000) or known Gitelman’s or Bartter’s mutations, they identified a final set of 30 putative functional mutations (15 in SLC12A3, 10 in SLC12A1 and 5 in KCNJ1) found in 49 subjects. 80% of the carriers of at least one mutation had long-term SBP values below the mean of the entire cohort (P=0.001). Carriers had 6.3 mm Hg lower SBP (P=0.0009) and 3.4 mm Hg lower DBP (P=0.0003) compared to non-carriers. The BP reduction among mutation carriers was observed in all age groups, with a ~60% reduction in the risk of developing HTN by age 60 (P<0.0003).

Collectively, these studies chart the identification of a gene influencing BP through linkage in rare families with recessive hypotension syndromes, use of large kindreds to examine within-family effects of heterozygosity and ultimately extension to the population level. Along with the finding that rare mutations in ABCA1 result in marked reduction in high-density lipoprotein levels in the general population [52], these are the first demonstrations that rare mutations of strong effect contribute to variation in common cardiovascular traits. Whether similar unbiased (cf genome-wide linkage) or targeted (cf sequencing of candidate genes) approaches could identify common variants that contribute to BP variation at the population level was unclear until recently. The literature is full of reports of small numbers of variants in modest-sized studies that failed to be replicated and led to a general sense that association studies of common variants were hopeless.

Common variants at the population level

Linkage studies are well suited to detect rare variants with strong effects that are conclusively scored in individual families for conditions that are relatively rare but has been an utter failure for common complex diseases [53]. By contrast, association studies are better suited to detect modest genetic effects of common variants [54]. Association studies were plagued by poor reproducibility, stemming from false positive reports due to inappropriately permissively significance thresholds and the inability to validate suggestive findings in underpowered replication attempts [55]. The creation of a reference sequence of the human genome, generation of large databases of common genetic variation, growth of high-throughput genotyping platforms and development of rigorous methodological approaches set the stage for the discoveries that followed.

First validated common BP variants identified

Atrial and B-type natriuretic peptides (ANP, BNP) have direct vasodilatory and natriuretic effects and antagonize the renin-angiotensin-aldosterone system [56–58]. In mouse models, knockout of the NPPA (encoding ANP) or natriuretic peptide receptor A (NPR-A) genes [59,60] result in salt-sensitive HTN and ANP overexpression [61] lowers BP. The known physiologic effects in humans and animal models make the natriuretic peptide system a compelling candidate to influence interindividual differences in BP, but until recently there was no clear evidence that ANP or BNP influence BP regulation in humans.

In 2009, Newton-Cheh et al. reported a candidate gene association study at the locus on chromosome 1 where NPPA and NPPB lie in tandem (Table 2) [62] ••. Thirteen single nucleotide polymorphisms (SNPs) were genotyped in a discovery sample from the Framingham Heart Study (n=1,705) and tested for association with plasma ANP and BNP, with subsequent replication in samples from Sweden and Finland (total n=14,743). Three non-coding SNPs showed consistent association with ANP or BNP at significance levels that were unambiguous (P ranging from 10⁻¹⁰ to 10⁻⁷⁰), presumably through shared regulatory effects given the concordant effects on both peptides.

Table 2. Common variants implicated in recent candidate gene and GWAS studies with p<5×10⁻⁸.

Fourteen variants at 13 independent loci are shown; the first two variants are at the same locus but are incompletely correlated and may represent different signals of association. P-values for variants observed in more than one study (or a close proxy) are based on meta-analysis of reported studies.

Nearby genes	SNP	Chr	Allele frequency	Trait	Effect (mm Hg)	p-value	Reference
NPPA/NPPB	rs5068	1	0.94	SBP	1.10	3×10−9	[62,63]
MTHFR/CLCN6/NPPA/NPPB	rs17367504	1	0.86	SBP	0.85	2×10−13	[63]
CYP17A1	rs11191548	10	0.91	SBP	1.16	7×10−24	[63,64]
PLCD3	rs12946454	17	0.28	SBP	0.57	1×10−8	[63]
FGF5	rs16998073	4	0.21	DBP	0.50	1×10−21	[63]
c10orf107	rs1530440	10	0.81	DBP	0.39	1×10−9	[63]
SH2B3	rs3184504	12	0.47	DBP	0.46	3×10−18	[63,64]
CYP1A2/CSK/LMAN1L	rs1378942	15	0.36	DBP	0.43	1×10−23	[63,64]
ZNF652	rs16948048	17	0.39	DBP	0.31	5×10−9	[63]
PLEKHA7	rs381815	11	0.26	SBP	0.65	2×10−9	[64]
ATP2B1	rs2681492	12	0.80	SBP	0.85	6×10−17	[64,65]*
ULK4	rs9815354	3	0.17	DBP	0.49	3×10−9	[64]
CACNB2	rs11014166	10	0.66	DBP	0.37	1×10−8	[64]
TBX3/TBX5	rs2384550	12	0.65	DBP	0.35	4×10−8	[64]

^*A correlated SNP was reported in Cho et al just missed genome-wide significant association with SBP (P=1×10⁻⁷).

With the addition of further samples from Sweden (final n=29,717), we found that two alleles associated with higher ANP and BNP were also associated with lower SBP (0.9–1.5 mm Hg) and DBP (0.3–0.8 mm Hg) and decreased odds of HTN (0.85–0.90). BP associations had much more modest significance than those of natriuretic peptide concentrations (P ranging from 1×10⁻⁶ to 6×10⁻⁵) [62]. The identification of cis-acting non-coding common genetic variants that influence plasma concentrations of ANP and BNP as well as BP provided the first in vivo evidence in humans of the relevance of endogenous variation in the natriuretic peptide system in BP regulation in humans.

It is worth pointing out that these weak BP effects could not be detected in such a small discovery sample size (n=1,705) were it not for the use intermediate ANP/BNP traits which showed much stronger association. Unbiased screens of the genome for BP variants, requiring stringent p-value thresholds to distinguish false- from true-positive associations (e.g. P<5×10⁻⁸) would require even larger sample sizes to detect such an effect.

Candidate gene studies are like the drunk searching under the lamppost for his keys—one can only look where biologic involvement is already suspected—and thus do not expand our understanding of human biology as much as unbiased surveys that test thousands of genes comprehensively. Genome-wide association studies (GWAS) have only borne fruit for BP when very large sample sizes allowed stringent P-value thresholds to be met.

GWAS in small to moderately large studies do not reach genome-wide significance

The first few published GWA studies published by us and others in 2007, including the Wellcome Trust Case Control Consortium (WTCCC, 2000 hypertensive cases, 3000 population-based controls) [66], the Diabetics Genetics Initiative (DGI, 1,464 cases with type 2 diabetes and 1467 matched controls) [67] and the Framingham Heart Study 100K project (1,327 individuals for SBP and DBP) [68] failed to identify any loci meeting stringent genome-wide significance thresholds. Genome-wide significance has been variably defined but is generally accepted at <5×10⁻⁸, which represents Bonferroni correction for the estimated 1 million independent common variant tests in the genome for individuals of European ancestry (0.05/1,000,000) [69]. In 2009, further GWA studies of BP traits and/or HTN in populations of different ancestries and of moderate size have subsequently been reported but none achieved genome-wide significance [65,70–74]. A few are highlighted.

In 2009, Wang et al. [72] reported a GWAS in 542 Amish subjects and found a SNP at 2q24.3 with modest association with SBP (P=8×10⁻⁵). They genotyped an additional 1,347 individuals and looked up results from published GWA studies for a total sample size of 7,125, but did not reach genome-wide significance (P=2×10⁻⁷). The SNP lies within an intron of the serine threonine kinase 39 (STK39) gene which encodes the SPAK protein, known to regulate NCCT and NKCC2 in the kidney [75,76]. No replication of this association has been published.

In 2009, Cho et al. [65]• reported a GWAS from 2 population-based cohorts of 8,842 individuals with replication in 7,861 individuals all of Korean ancestry. SNP rs17249754 near ATP2B1 was associated with SBP with borderline significance (P=1×10⁻⁷). This SNP is located near the ATPase calcium ion transporting plasma membrane 1 (ATP2B1) gene, which encodes a protein involved in calcium movement from cytosol to the extracellular compartment [77]. A correlated SNP at ATP2B1 was subsequently reported to be associated with SBP and HTN in a GWAS in European samples (see below).

Adeyemo et al. reported the first GWAS for BP and HTN in an African-American sample n=1,017 individuals, some in families) [74]. The authors selected 17 SNPs for follow-up replication in 980 West Africans. Of these, none were genome-wide significant upon joint analysis with the original GWAS result (all P>3×10⁻⁶).

Large GWAS meta-analyses and replication identify 13 novel loci

In 2009, two meta-analyses of genome-wide association studies of BP, Global BPgen [63]•• and CHARGE BP [64] ••, reported in parallel the first genome-wide significant associations between common genetic variants and BP.

Global BPgen consortium identifies 8 BP loci

Newton-Cheh et al. reported findings of the Global BPgen consortium including a stage 1 GWAS meta-analysis in 34,433 individuals from 17 cohorts of European ancestry ascertained through population-based sampling and controls from case-control collections. In total, 2.5 million genotyped and imputed SNPs were tested for association with SBP and DBP under an additive genetic model, adjusting for age, age², sex, body mass index, with imputation in individuals treated with anti-hypertensives. Twelve independent SNPs were selected from the stage 1 GWAS meta-analysis and directly genotyped in up to 71,225 European-derived and 12,889 Asian ancestry individuals (stage 2a). Twenty SNPs associated with SBP or DBP were looked up in the CHARGE BP results (stage 2b).

Eight loci met genome-wide significance for association with BP (3 SBP, 5 DBP) on meta-analysis of stage 1+2a+2b (P range 10⁻²³ to 10⁻⁸, n≤134,258, Table 2) [63]. Effect sizes were quite modest, generally ≤ ≈1 mm Hg systolic or ≈0.5 mm Hg diastolic. All variants were associated with both SBP and DBP, even if ascertained on association with one trait, and all were associated with dichotomous HTN. Of the 8 loci, two harbor genes known to be involved in BP physiology and six do not.

Two loci contain known BP genes

The MTHFR/CLCN6/NPPA/NPPB locus included a SNP rs17367504 associated with BP (Table 2). The SNP is partially correlated to the top SNP from our prior candidate gene study and is associated with plasma ANP concentration making it likely to mediate its effect through regulation of the natriuretic peptide system. However, rs17367504 is also strongly associated with hepatic expression of CLCN6 [78], which encodes a renally-expressed chloride channel, suggesting that a simple SNP→ANP→BP model may not hold. The Global BPgen study did confirm the association of the rs5068 SNP from the candidate gene study with SBP (joint P=3×10⁻⁹, Table 2) [62,63]. Whether multiple independent variants at the locus exist, potentially acting through different mechanisms, will require much more work.

One locus contains a gene involved in Mendelian HTN, CYP17A1 (Table 1, Table 2). CYP17A1 encodes a protein with both 17α-hydroxylase activity—mediating a key step in the biosynthesis of glucocorticoids—and 17,20-lyase activity involved in sex steroidogenesis [26,79]. Missense mutations in CYP17A1 cause adrenal hyperplasia with HTN and hypokalemia. There are many other genes in the associated interval at this locus, but the well-annotated Mendelian HTN gene certainly stands out as a strong candidate to underlie the common variant association.

Thus, strong candidate genes at these loci appear to be good places to start to explain these common variant associations. But ultimately, it will be necessary to determine the causal variants and to elucidate their relationships to gene expression for regulatory noncoding variants or to gene function via altered proteins for coding or splice variants. Unfortunately, the genetic code for regulatory machinery is only beginning to be understood and smoking-gun amino-acid altering variants are a minority of common variant associations. Of the remaining 6 loci identified by Global BPgen, the 12q24 locus includes an amino-acid altering variant.

A missense SNP in SH2B3 is associated with multiple traits

The missense SNP rs3184504 in the SH2B3 (SH2B adaptor protein 3) gene codes for an arginine to tryptophan at amino acid 262 (R262W) in the lymphocyte-specific adapter protein (LNK). A relationship of this gene to BP physiology has not previously been recognized. In earlier GWA studies, the minor T allele (262W) of this SNP was found to be associated in humans with increased odds of autoimmune diseases including type 1 diabetes [80], celiac disease [81], myocardial infarction [82,83] as well as higher eosinophil and other blood cell counts [82,83] (Table 3). The mouse knockout of LNK is associated with higher eosinophil, lymphocyte and platelet counts, suggesting by extension that the 262W allele results in loss of function (Table 3). The increased risk of autoimmune disease seems consistent with a potentially augmented immunologic activity. In fact, strong signatures of positive selection on the 262W allele in European-derived individuals suggest a selective advantage in human history, perhaps through fighting infection.

Table 3. SH2B3 262W allele in humans mirrors the mouse knockout.

Shown are the effects in a mouse knockout and in humans for the minor 262W allele.

	eosinophils	lymphocyte	platelets	type 1 diabetes risk	celiac disease risk	myocardial infarction	diastolic blood pressure	selection
mouse knockout [85]	↑	↑	↑
SH2B3 262W allele	↑ [82,83]	↑ [82]	↑ [82]	↑ [80]	↑ [81]	↑ [82,83]	↑ [63,64]	positive [63,83]

The role of SH2B3 in BP regulation or myocardial infarction pathogenesis is less clear but could be due to an effect outside of the hematopoietic lineage. In an endothelial cell model, exposure to TNF-α has been found to upregulate LNK, inhibiting VCAM1 expression and ERK1/2 and increasing eNOS activity via the PI3K/AKT pathway [84]. Given the role of the endothelial cell in atherothrombosis and smooth muscle tone, it seems possible that loss of LNK function from the 262W allele could lead to increased BP and risk of myocardial infarction through altered endothelial function.

CHARGE BP consortium

In a parallel publication, the CHARGE BP consortium reported results using a sample consisting of 29,136 European-derived individuals [64]•• with validation of top results in the Global BPgen GWAS. CHARGE BP identified 8 genome-wide significant loci, including 3 in common with Global BPgen and 5 unique to CHARGE BP (Table 2). The 5 additional loci have not previously been annotated to harbor genes involved in BP regulation, with the exception of CACNB2 which encodes a beta subunit of voltage-gated calcium channels, including the L-type dihydropyridine-sensitive channel.

Conclusions and questions raised

Until recently only Mendelian families yielded robust genetic BP findings. Large sample sizes, comprehensive genetic screens and rigorous p-value thresholds have allowed human genetics to expose additional novel pathways involved in BP regulation through the study of unselected population samples. However, in total the common and rare variants identified to date that influence BP explain less than a percent or two of overall population variation in BP [51,63,64], far less than age or body mass index. It seems unlikely, but requires formal testing, that genetic testing will add clinical utility to predicting future HTN in patients, beyond that available from a scale, a BP cuff and their age.

Why are common BP variant effects so weak? BP regulation requires that organ perfusion be achieved through a very dynamic range of daily activities and environments, from sleeping to running through feast or famine. This has likely led to many redundant systems that allow tight regulation of BP and poor tolerance of genetic variants of strong effect that are unable to rise to high frequency due to negative selection.

Why are so few of the antihypertensive targets represented among the common BP variants identified to date? It is possible that evolution has been particularly intolerant of genetic variation in the targets of antihypertensive therapies. However, it is equally possible that hundreds of genes are involved in BP, and that antihypertensive targets represent a small fraction of the total number of BP genes.

What do the weak common variant effects tell us about the potential of novel targets to have large effects on BP control? Not much. The minor allele of the SNP rs3846663 has been associated with a trivial 2.5 mg/dL higher low-density lipoprotein (LDL) cholesterol in a genome-wide association study [86]. If one did not already know that HMG-CoA reductase is involved in LDL cholesterol synthesis, then finding this SNP in an intron of the HMGCR gene could allow identification of this target for cholesterol-lowering therapy.

Next steps

Even larger GWA studies are likely to identify additional common genetic variants of even weaker effects. The ongoing ICBP-GWAS consortium, including the Global BPgen and CHARGE BP consortia as well as others, is one such example, with >70,000 individuals with genome-wide genotyping. Sequencing will enable fine mapping of association signals and can potentially identify less common, stronger effect alleles (more “Mendelian”) that may be more easily recognized by alteration of amino acids and more tractable for physiologic study because of their greater effect sizes. High-resolution phenotyping of individuals selected on genotype can potentially home in on the mechanisms underlying association of a particular genetic variant, as has recently been applied to the IL2RA gene [87]. Identification of novel genetic influences on BP is just a start, as was identification of the first genes underlying familial hypercholesterolemia in the 1970s, with the promise of an improved understanding of BP regulation and identification of novel therapeutic targets.