Tyrosine Hydroxylase, the Rate-Limiting Enzyme in Catecholamine Biosynthesis
Discovery of Common Human Genetic Variants Governing Transcription, Autonomic Activity, and Blood Pressure In Vivo
Background— Tyrosine hydroxylase (TH) is the rate-limiting enzyme in catecholamine biosynthesis. Does common genetic variation at human TH alter autonomic activity and predispose to cardiovascular disease? We undertook systematic polymorphism discovery at the TH locus and then tested variants for contributions to sympathetic function and blood pressure.
Methods and Results— We resequenced 80 ethnically diverse individuals across the TH locus. One hundred seventy-two twin pairs were evaluated for sympathetic traits, including catecholamine production, reflex control of the circulation, and environmental (cold) stress responses. To evaluate hypertension, we genotyped subjects selected from the most extreme diastolic blood pressure percentiles in the population. Human TH promoter haplotype/reporter plasmids were transfected into chromaffin cells. Forty-nine single-nucleotide polymorphisms were discovered, but coding region polymorphism did not account for common phenotypic variation. A block of linkage disequilibrium spanned 4 common variants in the proximal promoter. Catecholamine secretory traits were significantly heritable (h2), as were stress-induced blood pressure changes. In the TH promoter, significant associations were found for urinary catecholamine excretion and for blood pressure response to stress. TH promoter haplotype 2 (TGGG) showed pleiotropy, increasing both norepinephrine excretion and blood pressure during stress. Coalescent simulations suggest that TH haplotype 2 likely arose ≈380 000 years ago. In hypertension, 2 independent case-control studies (1266 subjects with 53% women and 927 subjects with 24% women) replicated the effect of C-824T in the determination of blood pressure.
Conclusions— We conclude that human catecholamine secretory traits are heritable, displaying joint genetic determination (pleiotropy) with autonomic activity and finally with blood pressure in the population. Catecholamine secretion is influenced by genetic variation in the adrenergic pathway encoding catecholamine synthesis, especially at the classically rate-limiting step, TH. The results suggest novel pathophysiological links between a key adrenergic locus, catecholamine metabolism, and blood pressure and suggest new strategies to approach the mechanism, diagnosis, and treatment of systemic hypertension.
Received January 18, 2007; accepted May 8, 2007.
Tyrosine hydroxylase (TH) is the rate-limiting enzyme in catecholamine biosynthesis.1,2 Substantial loss of TH enzymatic activity as a consequence of rare inactivating mutations has profound consequences in humans1 and in mice with targeted ablation of the TH locus.3
Editorial p 970
Clinical Perspective p 1006
The human TH locus also displays more common natural allelic variation such as the tetranucleotide repeat [or microsatellite polymorphism (TCAT)n] in its first intron,4 which has been used to probe the role of TH in psychiatric illnesses.5–7 This microsatellite also may be associated with essential hypertension.4 Previously, we found that the common hepta-allelic (TCAT)n polymorphism predicted alterations in human autonomic function.8 However, is this intronic variant in itself functional? In transfections,9 the (TCAT)n repeat silences transcription in a copy number–dependent way; in contrast, we observed in vivo directionally opposite associations of common (TCAT)n alleles with autonomic function: (TCAT)10i with activation and (TCAT)6 with diminution of sympathetic outflow. Therefore, we sought additional causative/functional variation at the TH locus.
The spectrum of allelic variation at human TH is currently unknown; therefore, we undertook systematic polymorphism discovery at the locus. To probe the impact of TH variation on stress-induced disease pathways, we resequenced ≈1.2 kbp of the 5′ promoter and all 13 exons and adjacent intronic regions in 80 ethnically diverse subjects and 422 twins. Twin pairs enabled us to study whether TH allelic variation contributed to heritable control of the circulation.
Our results suggest that common variation in the TH proximal promoter contributes to heritable alteration in multiple autonomic traits, biochemical and physiological, and the ultimate disease trait of hypertension.
The following categories are described extensively in the Methods section of the online Data Supplement: subjects and clinical characterization, genomics, biochemical phenotyping in twin pairs (catecholamines), physiological/autonomic phenotyping in twin pairs in vivo, and TH promoter haplotype activity in vitro. Further details on subjects and clinical characterization also are given in the online Methods section.
Subjects and Clinical Characterization
Subjects (n=80) resequenced across the TH locus are described in the online Data Supplement. Their biogeographic ancestries were as follows: 23 European (white), 25 sub-Saharan African (black), 16 east Asian, and 16 Mexican American (Hispanic).
The 172 twin pairs (119 monozygotic, 53 dizygotic; age, 15 to 84 years) are described extensively in the Data Supplement.
Hypertension and Population Blood Pressure Extremes
Subjects with hypertension, both the initial and replication samples, are described in the online Data Supplement. In the initial sample, 1266 subjects with the highest (4.9th percentile) and lowest (4.8th percentile) diastolic blood pressures (DBPs) in the population were evaluated; 53% were female. In the follow-up study, 927 subjects (essential hypertension versus normal BP) were evaluated; 24% were female.
Statistical Genetic Analyses
Descriptive statistics (mean, SE) were computed across all of the twins with generalized estimating equations (PROC GENMOD) in SAS (SAS, Cary, NC) to account for correlated trait values within each twinship using an exchangeable correlation matrix.10
Heritability of Phenotype Expression in Twin Pairs In Vivo
Heritability (h2) is the fraction of phenotypic variance accounted for by genetic variance (h2=VG/VP). Estimates of h2 were obtained by using the variance component method implemented in the Sequential Oligogenic Linkage Analysis Routines (SOLAR) package.11 This method maximizes the likelihood assuming a multivariate normal distribution of phenotypes in twin pairs (monozygotic versus dizygotic), with a mean dependent on a particular set of explanatory covariates. The null hypothesis (H0) of no heritability (h2=0) is tested by comparing the full model, which assumes genetic variation, and a reduced model, which assumes no genetic variation, using a likelihood ratio test. Covariates (sex and age) significant at P<0.05 were retained in the heritability models.
Haplotypes and Linkage Disequilibrium
Haplotypes were inferred from unphased diploid genotypes with the software package PHASE,12 assigning the 2 most likely haplotypes to each diploid individual. We inferred the TH promoter haplotypes using 10 single-nucleotide polymorphisms (SNPs) discovered by resequencing 293 unrelated individuals (n=586 chromosomes) chosen to span 4 diverse ethnic groups: white (European ancestry), black (sub-Saharan African ancestry), Hispanic (Mexican American), and east Asian. Blocks of pairwise (SNP-by-SNP) linkage disequilibrium (LD) were displayed using graphical overview of LD.13
Association studies were performed for the common TH promoter alleles (minor allele frequency >5%). Each study subject was categorized according to diploid genotype at a biallelic SNP locus or carrier status (2, 1, or 0 copies) of a particular TH SNP, haplotype, or diploid haplotype (diplotype). Unpaired t tests evaluated the significance of the in vitro haplotype-specific TH promoter activity.
Pleiotropy: Bivariate Genetic Analyses
Pleiotropy (genetic covariance for 2 correlated, heritable traits14) was estimated as the parameter ρG in SOLAR.15 As a test of pleiotropy, bivariate analyses in SOLAR11 (www.sfbr.org/solar, chapter 9.2) were done to test whether genotype coordinately influenced 2 dependent variables (traits), biochemical (eg, catecholamines) and hemodynamic (eg, stress BP responses), using nested log-likelihood values for the bivariate model in the presence or absence of the genotype: −2(Δlog likelihood)=χ2 at df =1. Hardy-Weinberg equilibrium was assessed with a χ2 goodness-of-fit test using 1 individual from each twin pair.
Coalescent and Phylogeny Reconstruction
An approximation of the likely age of an SNP mutation (within a haplotype) in the human lineage was generated by constructing a coalescent tree16 for SNPs in genetic regions that showed a significant association between SNP haplotype and autonomic function with the coalescent software package GENETREE.17 GENETREE permits the construction of coalescent trees inferring the time to most recent common ancestors of sets of SNP haplotypes. The “root” of the coalescent tree (the ancestral alleles/haplotype) was specified by the chimpanzee variant17 because the chimpanzee is the contemporary nonhuman primate with the closest evolutionary ties to humans (divergent lineages, ≈4 to 6 million years ago. The resulting coalescent units were transformed into estimates of years in the past (time to most recent common ancestors) through adjustment for expected single base mutation rate per generation, effective human population size, generation length, and ploidy.16 The approximate time at which each haplotype arose (from its most recent common ancestors) was calculated from this equation and assumptions: age (years)=(coalescent units from GENETREE)×(20 years/generation)×[2(diploid genome)]×(effective population size of 10 000).
To adjust for the possibility of multiple comparisons when testing the effect of 4 TH promoter SNPs on autonomic traits, we used the method of SNP spectral decomposition (SNPSpD) proposed by Nyholt18 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 TH promoter SNP data from 1 member of each twinship (ie, 1 founder per family). This method takes into account intermarker correlations in calculating a new experiment-wide threshold to keep the type I error rate at ≤5%, for a single phenotype.
To enable model-free analyses without reliance on standard asymptotic assumptions, we also used more computationally intensive permutation tests through the use of the recursive algorithm of Mehta and Patel19 as implemented by Clarkson et al.20 Trait values were dichotomized (or converted into 2 quantiles) to allow construction of 3×2 contingency tables (diploid genotype by trait); the test is implemented online at http://www.physics.csbsju.edu/stats/exact.html.
The authors had full access to and take full responsibility for the integrity of the data. All authors have read and agree to the manuscript as written.
TH Genomics: Systematic Human Polymorphism Discovery
To identify genetic variants in TH, we resequenced all 13 exons and adjacent intronic regions and 1.2 kbp of 5′ promoter (online Table I) from 80 ethnically diverse subjects. Later, we resequenced the promoter from an additional 213 twin pairs.
Online Figure I illustrates sequence tracings for 4 common (minor allele frequency >10%) SNPs discovered in the TH proximal promoter: C-824T, G-801C, A-581G, and G-494A.
Figure 1A shows the local genomic region resequenced. Forty-nine SNPs and 1 tetranucleotide repeat were identified. Ten SNPs were located in the ≈1-kbp proximal promoter, 14 in coding regions, and 25 in untranslated regions or exon-adjacent intronic regions. Of the 49 SNPs, 13 were common (minor allele frequency >5%), including 4 SNPs in the proximal promoter, 2 in coding regions, 4 in untranslated regions, and 4 in introns. Among 14 coding region polymorphisms, 11 specify amino acid substitutions. Val81Met is the most common, with a minor allele frequency at 37.4%; synonymous Lys240Lys is the second most common at 23.6%. Unusual/rare coding SNPs (all <1.2%, most at 0.3%) were the balance: Ala6Thr, Arg15His, Ala17Val, Ala23Thr, Arg89Arg, Ala143Thr, Thr225Asn, Arg410Trp, Asp467Asp, Val468Met, Ala492Val, and Gly497Asp.
LD Across the TH Locus
Pairwise LD between each common SNP across TH was quantified as D’, scaled from 0 to 1. Visual inspection (Figure 1B) of plots of the LD structure across TH reveals 2 blocks of particularly high LD (D’ >0.9): at the 5′ (promoter) and mid to 3′ (intron A→exon 12) regions of the gene.
TH Promoter Variants
Ten biallelic variants were discovered (online Table I and Figure 1A; major allele/position upstream of the ATG/minor allele): C-833T, C-824T, G-814A, G-801C, T-741C, A-641G, A-581G, G-494A, C-388T, and G-94T. Alleles at each SNP were in Hardy-Weinberg equilibrium (determined within the largest ethnic group, white). Four of 10 SNPs (C-824T, G-801C, A-581G, and G-494A; shown in online Figure I) had relatively common minor alleles (>10% frequency); these 4 occurred within a span of only 331 bp in the proximal promoter. Five of 6 uncommon (<5%) promoter SNP alleles were each found in only 1 ethnic group (C-833T, G-814A, T-741C, C-388T, and G-94T; online Table I). Even the more common SNP genotype frequencies differed by ethnicity, with the greatest differences between Asians and blacks at C-824T (16.7% versus 60.3%), A-581G (10% versus 59%), and G-494A (71.7% versus 17.9%). Eight of 10 promoter SNPs (all but G-801C and G-94T) were either purine/purine (A/G) or pyrimidine/pyrimidine (C/T) transitions.
Twin Phenotyping: Descriptive Statistics and Heritability
Online Table II describes the twin subject population (n=344 individuals). Female subjects (n=267) had lower basal SBP (P=0.0438), plasma epinephrine (P=0.0065), urinary epinephrine (P=0.0284), and norepinephrine (P=0.0192) values than male subjects (n=77), consistent with previous reports.21
Older subjects (age ≥40 years, n=180) had higher basal SBP (P<0.0001) and DBP (P<0.0001) and poststress SBP (P<0.0001) and DBP (P=0.0008) values than younger subjects (n=164). Older subjects had lower baroreceptor slope, during both upward (P<0.0001) and downward (P<0.0001) deflections. Plasma norepinephrine was increased (P=0.0045) in older subjects.
Online Table III presents correlations between variables. In general, similar correlations were obtained with the parametric and nonparametric methods. Because of the effects of sex and age (online Table II), further inferential statistics were performed on age- and sex-adjusted data. Urinary norepinephrine excretion correlated directly with basal SBP, DBP, and poststress SBP and inversely with baroreceptor slopes, both downward (ρ=−0.278, P<0.001) and upward deflections (ρ=−0.303, P<0.001).
Heritability (h2; see Methods) estimates from twin pairs are shown in online Table IV. Both plasma catecholamine and urinary catecholamines were significantly heritable, with the most prominent values for plasma norepinephrine (h2=69.9±4.4%; P<0.0001) and urinary epinephrine excretion (h2=66.7±5.9%; P<0.0001).
Basal BP and heart rate displayed significant heritability, with heart rate substantially more heritable (at h2=61±6%; P<0.0001) than either systolic BP (SBP; h2=26±8%; P=0.0016) or DBP (h2=18±9%; P=0.0359). Stress-induced changes in vital signs also were heritable whether expressed as maximal values, absolute changes (maximal minus basal), or percent changes. Heritability was significant for baroreceptor slope, both upward BP deflections with reflex bradycardia (h2=33.3±9.2%; P=0.0004) and downward BP deflections with reflex tachycardia (h2=43.0±7.3%; P<0.0001).
The h2 of traditionally heritable reference traits (weight at h2=87±2%, P<0.0001; height at h2=93±1%, P<0.0001) in our sample is consistent with previous observations.14
Twin TH Promoter Genotypes: Marker-on-Trait Mapping/Associations
Univariate Analyses Across the Promoter
Significant associations were found for urinary catecholamine excretion (epinephrine and norepinephrine; Figure 2A) and the BP response to environmental (cold) stress (ΔSBP and ΔDBP; Figure 2B) with variants at C-824T and A-581G. Variants at G-801C and G-491A did not associate (Figure 2). The common TH coding region/exon 2 nonsynonymous variant Val81Met did not associate with catecholamines or stress BP changes, nor did the less common (minor allele frequency <5%) coding region variants.
Multivariable Analyses Within the Promoter
Because both C-824T and A-581G seemed to influence catecholamine secretion (Figure 2A) and the BP response to stress (Figure 2B), we performed a multivariable analysis of all 4 common TH promoter polymorphisms (C-824T, G-801C, A-581G, and G-494A) in SOLAR in an attempt to discern the most important variant(s). In this analysis, C-824T became the most significant predictor of ΔDBP during cold stress (P=0.000164), although G-494A also achieved significance (P=0.0276). We did not detect SNP-by-SNP interactions (epistasis) within the TH promoter on adrenergic traits (all P>0.2).
Multiple Comparisons Within the TH Promoter
Because we tested the phenotypic effects of 4 different promoter SNPs, we reevaluated the α threshold to avoid false-positive conclusions. The SNPSpD method of Nyholt18 tested intermarker correlations within the promoter block of LD to yield an “effective” number of markers at 2.92, indicating that an appropriate α threshold to maintain the type I error rate at ≤5% for a single phenotype would be P=0.0171. If we include the common (≈37%) coding region (exon 2) polymorphism Val81Met in the SNPSpD analysis, the effective number of markers rises to 3.73, and the appropriate α threshold falls to P=0.0134. The significance levels for TH promoter polymorphism effects on catecholamine secretion and stress-induced BP changes exceeded even these more stringent thresholds (Figure 2). We also tested TH promoter SNPs against 2 general classes of autonomic phenotypes, biochemical and physiological (Figure 2); because these traits are significantly correlated (online Table I), a full Bonferroni correction would be inappropriately conservative22; alternatively, Sidak’s adjustment for correlated traits would yield a revised α threshold at P=0.025323 for a single genotype. On the basis of the principles of the less conservative false discovery rate,24 an appropriate α threshold for 2 phenotypic categories would be P=0.0375.
Another approach to multiple genotype–phenotype comparisons is the use of model-free exact (permutation) tests. We dichotomized continuous traits about the median value and then constructed 3×2 contingency tables (3 diploid genotypes by 2 trait quantiles). After permutation, the effect of C-824T remained significant on both urinary norepinephrine excretion (P=0.00633) and the DBP response to cold stress (P=0.0157).
Because several traits differed by sex and age (online Table II), we performed twin analyses on sex- and age-adjusted traits within SOLAR or generalized estimating equations. We also tested gene-by-sex interactions in marker-on-trait associations but did not find significant nonadditive interactions for C-824T or A-581G on either catecholamines or stress-induced BP increments. We also tested each sex separately for effects of TH promoter SNPs on these traits; significant effects were seen for female subject (n=267) alone but not for male subject alone (n=67), perhaps indicating a lack of power to detect male-specific effects in the relatively small number of male twins. Although California birth rates are approximately equal for males and females, females are far more likely to enroll in twin studies, especially younger females.25
Pleiotropy: Coordinate Effects of TH Promoter C-824T on Biochemical and Physiological Traits
C-824T: Biochemistry and Physiology
C-824T was associated with urine catecholamine excretions (epinephrine: P=0.0058; percent variation explained, 5.55%; and norepinephrine: P=0.0069; percent variation, 1.52%). The T-824C minor (T) allele was associated with higher urinary catecholamine excretion (Figure 3A) and greater changes in BP response to cold stress (Figure 3B; ΔSBP: P=0.01; percent variation, 1.54%; and ΔDBP: P=0.0069; percent variation, 3.98%).
C-824T: Pleiotropy (Norepinephrine and Stress BP)
C-824T exerted significant pleiotropic effects on the coupling between BP response to cold stress and urinary norepinephrine (Figure 3C). Increasing T-824C minor (T) allele copy number increased not only the change in DBP after cold stress but also urinary norepinephrine excretion (bivariate likelihood ratio analyses: ΔDBP, χ2=10.4, P=0.0013; ΔSBP, χ2=3.91, P=0.048). Similarly pleiotropic SBP results are shown in online Figure II.
However, promoter SNP C-824T alone did not exert a significant pleiotropic effect on the coupling between baroreceptor slope and norepinephrine secretion for either downward (bivariate χ2=3.77, P=0.0522) or upward (bivariate χ2=2.27, P=0.132) baroreceptor deflections (data not shown).
TH Promoter Haplotypes
From 10 SNPs in the TH proximal promoter, we inferred the presence of 17 haplotypes; the 3 most common haplotypes accounted for 84.8% of chromosomes examined (online Table V). The most common haplotype overall (No. 1, CCGGTAAACG) is rather frequent (>37.1%) in Asian, Hispanic, and white populations but relatively unusual (15.4%) in black samples. Haplotype 2 (CTGGTAGGCG) is the most common variant in blacks (at 35.6%) but is relatively rare in Asians (10%). Haplotype 3 (CCGCTAAGCG) is common in all (>11.6%) but blacks (7.7%). Similar haplotypes were derived by considering only the 4 most common promoter SNPs (C-824T, G-801C, A-581G, and G-494A).
Patterns of LD across the TH promoter are illustrated in online Figure III either across all ethnicities or in 4 discrete ethnic/ancestry groups: European, African, Asian, and Hispanic. LD was substantial in the total group and population subgroups, especially toward the 3′ end of the promoter. The Asian and Hispanic groups displayed very high LD (D’ >0.9) across virtually the entire promoter region. White subjects showed a falloff in LD toward the 5′ end of the region (D’ <0.4 at C-824T), whereas African ancestry subjects displayed especially low LD (D’ <0.1) in the 5′ region with preserved LD (D’ >0.9) at the 3′ end of the domain.
TH Promoter Haplotype Effects on Autonomic Traits
We formed haplotypes from the 4 most common promoter SNPs (C-824T, G-801C, A-581G, and G-494A) and then tested whether haplotype copy number (0, 1, or 2/genome) at the diploid locus influenced biochemical or physiological trait means in twins (online Table VI). This approach is feasible (ie, has sufficient power) for the most common haplotypes in a population (here, haplotype 1 [CGAA], 2 [TGGG], or 3 [CCAG]).
The second-most-frequent promoter haplotype (haplotype 2, TGGG) displayed copy number–dependent effects on stress change in BP (both ΔSBP: P=0.0154; percent variation, 1.90%; and ΔDBP: P=0.0004; percent variation, 3.73%; Figure 4A) and urinary epinephrine (P=0.0044; percent variation, 5.7%) and norepinephrine (P=0.0125; percent variation, 4.06%) excretion (online Table VI). Haplotype 3 (CCAG) displayed associations with both basal heart rate and change in heart rate during cold stress (P=0.0104; percent variation, 3.14%; online Table VI). However, there were no associations for trait means with copy number of the most frequent haplotype (haplotype 1, CGAA).
Diploid Haplotypes (“Diplotypes”)
Haplotype 1/2 (CGAA/TGGG) diploid haplotype pairs influenced change in DBP during cold stress (ΔDBP: P=0.0312; percent variation, 3.47%; online Table VIIA and Figure 4B). Haplotype 1/3 (CGAA/CCAG) diploid combinations displayed associations with basal HR (P=0.0335; percent variation, 4.13%) and upward baroreceptor slope (P=0.0476; percent variation, 4.49%; online Table VIIB). By contrast, individual promoter SNPs did not predict baroreceptor slope (all P>0.12).
TH Promoter Haplotypes and Pleiotropy: Joint Effects on Biochemical and Physiological Traits: Altering the Coupling Between Baroreceptor Function and Efferent Sympathetic Outflow
Baroreceptor slopes (downward and upward deflections) correlated highly with each other (ρ=0.690, P<0.0001; online Figure IVA). Urinary norepinephrine excretion inversely paralleled baroreceptor slopes, both upward (ρ=−0.303, P<0.001; Figure 5A) and downward (ρ=−0.278, P<0.001; online Figure IVB) slope.
Not only did haplotype 2 (TGGG) display copy number–dependent effects on urinary norepinephrine excretion (P=0.0125; percent variation, 4.06%), but bivariate likelihood ratio analyses also indicated that TH haplotype 2 altered the coupling of norepinephrine to baroreceptor activity for both upward (χ2=8.0, P=0.0047; Figure 5A) and downward (χ2=7.0, P=0.0082; online Figure IVB) deflection slope. There were no univariate effects of haplotype 2 on baroreceptor downward deflection slope (P=0.702) or upward deflection slope (P=0.497).
TH Promoter Haplotypes and Pleiotropy: Altering the Coupling Between Norepinephrine Release and Hemodynamic Responses to Environmental Stress
The effects of haplotype 2 (TGGG) displayed a pattern of pleiotropy (Figure 5B). Increasing copy number of haplotype 2 augmented changes in DBP (P=0.0004; 3.73% variation explained) during cold stress and urinary norepinephrine excretion (P=0.0125; 4.06% variation explained), and the likelihood ratio test indicated a coordinate effect on the 2 traits (bivariate χ2=14.2, P=0.0002).
TH Promoter Haplotype Lineage and Coalescence
Human phylogeny at the TH promoter was approached using haplotypes imputed from the 4 common (high minor allele frequency) SNPs in the proximal promoter (C-824T, G-801C, A-581G, and G-494A). Coalescent simulations (Figure 6A) provide a graphical representation of the most likely ancestry of the disease-associated haplotypes in recent human evolutionary history.16 The most common haplotype (haplotype 1, CGAA) seemed to arise by point mutation from its most likely ancestral haplotype, CGGG, corresponding to an A-581G purine/purine transition ≈224 000 years ago. The second most common haplotype (haplotype 2, TGGG) likely arose by point mutation from its most recent common ancestor, CGGG, corresponding to a C-824T pyrimidine/pyrimidine transition ≈381 000 years ago. Of note, haplotype 2, which is associated with greater BP and catecholamines (Figures 4 and 5⇑), is the most common haplotype in subjects of African ancestry (online Table V).
TH Promoter Variants: Functional Consequences Probed by Chromaffin Cell Transfection/Expression
We assayed haplotype-specific gene expression in PC12 chromaffin cells (Figure 6B) with TH promoter/luciferase reporters for the 2 most common promoter haplotypes. Haplotype 2 (TGGG) was substantially more active in chromaffin cells; Figure 6B illustrates the increased activities of haplotype 2 both in vitro (driving transcription) and in vivo (determining norepinephrine secretion).
Relatively low-expressing haplotype 1 (CGAA), is the most common variant in Asian (61.7%), white (46.8%), and Hispanic (37.1%) populations but is relatively unusual (15.4%) in blacks. The higher-expressing haplotype 2 is the most common variant in blacks (34.6%).
TH Promoter and Disease: Hypertension
Figure 7 illustrates a population-based case-control study in which >1200 subjects were drawn from the highest and lowest fifth percentiles of BP in a primary care practice of >53 000 adults. In a 2-way ANOVA, there was a significant sex-by-genotype (TH C-824T) interaction on DBP (P=0.044). As expected, sex also influenced DBP (P<0.001). When the ANOVA was run in the presence or absence of C-824T, the results indicated that C-824T variation accounted for 3.4% of population DBP variance. When analyses were conducted separately on the sexes, the C-824T genotype effect was found in male subjects (P=0.045) but not female subjects (P=0.985). Inspection of the bar graph indicates that increasing numbers of the minor (T) allele increased DBP in male but not female subjects.
To explore the findings in the same group in a model-free fashion, without relying on standard asymptotic assumptions, we also used more computationally intensive permutation tests. We dichotomized subjects by BP status (high versus low) and sex (male versus female). We then conducted 3×2 table permutation tests on the effect of C-824T diploid genotype (C/C, C/T, or T/T) on BP status; the effect remained significant in male (P=0.023) but not female (P=0.266) subject.
To replicate the TH effects on BP in an independent sample, we studied the C-824T polymorphism in 898 additional subjects, but this time not from population BP extremes: 352 with hypertension and 546 with normal BP. Once again, the C-824T affected DBP (F=7.73, P<0.001; Figure 7B), although without a gene-by-sex interaction in these subjects with less extreme BPs. If we adjusted DBP data for the effects of antihypertensive medications,26 the effect of C-824T on DBP persisted (F=7.88, P<0.001). Finally, C-824T also affected SBP (F=5.61, P=0.004).
Critical Role of the Enzyme TH in Catecholamine Metabolism
TH catalyzes the conversion of tyrosine to L-dihydroxyphenylalanine, the rate-limiting step in the biosynthesis of catecholamines.1,27 Profound TH deficiency, as occurs after unusual inactivating mutations (Leu205Pro or Gln381Lys) in homozygous individuals, results in widespread disturbance of neuropsychiatric function such as autosomal-recessive, L-dihydroxyphenylalanine–responsive dystonia.27 Complete homozygous ablation of the TH locus by homologous recombination-directed gene targeting in transgenic mice is lethal by the early postnatal period.28
Previous Work: (TCAT)n Intronic Polymorphism at TH
Differential allele frequencies for the intronic (TCAT)n microsatellite polymorphism have been associated with hypertension4 and BP regulation.8,29 However, the functional significance of the (TCAT)n polymorphism remains in doubt.8,9 Therefore, we systematically searched the TH locus for a functional polymorphism.
Comprehensive TH SNP Discovery and Role of Polymorphism in the TH Coding Region
Does common qualitative (amino acid changing) polymorphism in this gene contribute to variation in autonomic tone? In the coding region, we found 2 common biallelic variants, only one of which was nonsynonymous: Val81Met, at 37.4% (Figure 1A), a polymorphism of uncertain significance, lying outside the catalytic domain,30 which did not associate with autonomic traits. We also found 12 unusual nonsynonymous coding region variants (Figure 1A), but their allele frequencies were only 0.3% to 0.6%, not sufficient to account for population associations. Thus, we turned to potential regulatory (noncoding) variants.
Comprehensive Polymorphism Discovery in the TH Promoter
Visual inspection (Figure 1B) of GOLD plots of LD structure across TH reveals 2 blocks of particularly high LD: at the 5′ (promoter) and the mid to 3′ regions of the gene. Thus, we examined the promoter block for functional consequences. We found 10 SNPs in the proximal promoter (online Table I), 4 of which were common and resolved themselves into 4 common haplotypes (online Table V).
TH Promoter Variants and Autonomic Pathway Pleiotropy
Because the same TH promoter SNPs predicted both biochemical (Figures 2A and 3⇑A) and physiological (Figure 3A and 3B) traits, we undertook bivariate genetic analyses15 in search of pleiotropy or the coordinate effect of a single gene on multiple traits. Bivariate results indicate coupled genetic control of both catecholamine secretion and stress BP by C-824T (Figure 3C). Pleiotropy extended into haplotypic control of both catecholamine secretion and baroreceptor function (Figures 4 and 5⇑).
Multivariable Promoter Analyses
Because both C-824T and A-581G influenced both catecholamine secretion (Figure 2A) and BP response to stress (Figure 2B), we performed a multivariate analysis of all 4 common TH promoter polymorphisms to discern important variant(s). In this analysis, C-824T became the most significant predictor of change in DBP during cold stress (P=0.000164); G-494A also retained significance (P=0.0276).
Dating the Responsible TH Promoter Variants
We approached likely dates of origin of trait-associated TH promoter variants using the coalescent approach.16 A-581G likely arose ≈224 000 years ago; C-824T was even more ancient, arising ≈381 000 years ago (Figure 6A). What is the significance of such an ancient origin and modern persistence at high frequency (eg, −824T at ≈17% to 60%; online Table I) of such alleles? Of note, the −824T allele is associated with increased catecholamine production (Figure 3A), increased BP increments in response to stress (Figure 3B), and extreme BP values in the population (Figure 7). We speculate that the functional variation we observed in −824T carriers could be the outcome of environmental selective pressures acting on alleles augmenting catecholaminergic function.
Sex, TH Polymorphism, and BP
Twin data indicate that autonomic function differs between males and females in both BP and catecholamine secretion (online Table II). Sex differences in plasma epinephrine (female less than male) confirm our previous observations,14 whereas the sex differences in urinary epinephrine and norepinephrine (also female less than male) are novel. The most striking sex effect we noted was on TH polymorphism (C-824T) and population BP extremes (Figure 7); here, a gene-by-sex interaction was apparent (interaction F=3.14, P=0.044) whereby the polymorphism had affected BP in males (F=3.12, P=0.045) but not females (F=0.015, P=0.985). This gene-by-sex interaction may be a clue to genetic mechanisms underlying the well-known and substantial differences in adrenergic function between men and women.31 Of note, although a gene-by-sex interaction on BP was found in population BP extremes (Figure 7A), C-824T seemed to affect DBP in both male and female subjects in a sample of hypertensives and normotensives with less extreme BPs (F=7.73, P<0.001; Figure 7B).
Ethnicity and TH Polymorphism
TH allele (online Table I) and haplotype (online Table V) frequencies differed by ethnicity, with subjects of African ancestry displaying the most striking differences. In particular, the TH −824T allele and its associated haplotype 2 (TGGG) were especially frequent in blacks; −824T became the quantitatively major allele (at 57%; online Table I) and TGGG the most frequent haplotype (at 34.6%; online Table V). In white subjects, the −824T allele and haplotype 2 predicted greater catecholamine production (Figure 3A), stress BP increments (Figure 3B), and BP in the population (Figure 7). Because blacks have a greater population prevalence of hypertension32 and adrenergic reactivity to stress,33,34 we speculate that TH polymorphisms might be involved in such traits with disproportionate frequency across ethnic populations. However, explicit association tests of these TH variants with disease or physiology in subjects of African ancestry remain to be done.
Study Strengths: Coupling the Twin Method and Systematic Polymorphism Discovery With Adrenergic Phenotyping
Twin Phenotyping Protocol
We exploited the classic twin design.35,36 Twin data offer the advantage of heritability (h2) measurement, the fraction of phenotypic variance accounted for by genetic variance, a logical estimator of the tractability of any trait to genetic investigation. Because twins are randomly sampled from the population, genetic conclusions drawn from twin studies are likely to be generalizable to the population from which they were sampled.35 Multiple autonomic phenotypes in the twins, both biochemical and physiological, allowed construction of an integrated picture of the effects of particular genetic variants at TH (Figure 8).
Systematic Polymorphism Discovery
LD mapping is an increasingly powerful tool for exploring genetic determinants of disease.37 However, the LD approach requires fulfillment of many assumptions,38 including substantial LD between marker and trait alleles. Here, we took another approach: systematic polymorphism discovery at a candidate genetic locus. This approach enables direct testing of marker-on-trait allelic association rather than indirect testing relying on a hypothetical degree of LD between marker and trait alleles. Of note, at the TH locus, we discovered 2 blocks of LD (Figure 1B) and found that the SNP most commonly used in previous LD studies at TH, Val81Met (G2066A, exon 2), was in a different (downstream) “block” of LD from the promoter (upstream) block; of further note, Val81Met itself did not associate with autonomic traits, establishing the necessity of systematic SNP discovery as a prerequisite to effective exploration of the functional consequences of polymorphism at the TH locus.
SNPs and Haplotypes
Dense promoter genotyping (by systematic resequencing) in a large series of twins permitted both individual SNP (Figures 2 and 3⇑) and haplotype (online Table VI and Figure 4) approaches to trait associations. In the presence of already complete genomic information at the TH proximal promoter (online Table I and Figure 1), haplotypes might not provide any new associations by virtue of LD; indeed, we derived SNP genotypes in the TH promoter by resequencing that region in 172 twin pairs. Twin analyses in SOLAR allowed us to quantify the contribution of each polymorphism (single SNP or haplotype) to the adrenergic traits in the form of percent of trait variance explained. In general, trait predictions by individual SNPs (Figures 2 and 3⇑) and haplotypes (online Table VI and Figure 4) were comparable. However, several observations here suggest that >1 variant in the TH promoter may be important for trait determination: (1) Two strong univariate SNP-on-trait associations were found in the TH promoter at both C-824T and A-581G (Figures 2 and 3⇑); (2) a multivariable analysis of the TH promoter indicated that >1 SNP (both C-824T and G-494A) influenced stress BP responses; and (3) haplotypes, but not individual SNPs, predicted baroreceptor slope (Table IIIB) and its pleiotropy with catecholamine secretion (Figure 5A). In the population, however, only a single TH SNP (C-824T) predicted BP extremes (Figure 7). Diploid haplotype combinations (pairs) also predicted adrenergic traits (online Table VII and Figure 4); in general, pairs were not more predictive than individual haplotype copy numbers (0, 1, 2 copies).
Ultimate Disease State Association: Hypertension
To establish the pertinence of our observations for human disease, we also genotyped individuals with extreme of BP as a quantitative trait in a population-based cohort (Figure 7A). Indeed, common TH promoter variation at C-824T accounted for up to ≈3.4% of the population DBP variance, and the effect was replicated (Figure 7B). These results document the “intermediate phenotype”39,40 approach as a successful route to discovery of genetic variants underlying a complex disease trait.
Complex (Nonmendelian) Inheritance
“Intermediate” Phenotypes, Pathways, and Pleiotropy
Investigations of putative pathways toward disease (Figure 8) yielded coordinate or pleiotropic effects of TH promoter variants on both biochemical and physiological traits (Figures 3, 5, and 8⇑⇑). We documented such pleiotropy statistically using bivariate genetic analyses.15 Genetic pleiotropy is an integral component of the “intermediate phenotype” hypothesis (Figure 8), wherein 1 gene influences a series of traits over time.
In addition to TH effects on resting traits (Figure 2A), TH variation also predicted the BP response to cold stress (Figure 2B). This is a classic example of a gene-by-environment interaction,41,42 requiring both a specific genetic variant (here, C-824T) and an environmental perturbation (here, cold) for expression of the trait (in this case, ΔBP).
TH and Sex
Sex had a substantial effect on many of the adrenergic intermediate phenotypes we evaluated (online Table II). In addition, sex seemed to play a permissive role for the action of the TH C-824T genotype on individuals with the most extreme (highest and lowest) BP in the population (Figure 7A), but in people with less extreme BP values, the genotype affected DBP in both males and females (Figure 7B). We do not precisely understand the nature of this TH gene-by-sex interaction, although it may be rooted in the different hormonal milieu subserving autonomic and cardiovascular function in males and females.31 Fundamental molecular and cellular mechanisms of BP control may differ in males and females,31 but the ultimate implications of such differences for disease states such as hypertension are not clearly understood. Indeed, we have noted that aging-dependent changes in sympathetic activity differ between the sexes,43 as do vascular responses to adrenergic agonists.44 The C-824T genotype-by-sex interaction has potential implications. Since sex may modify the C-824T effect on BP, sex may contribute to the role of genotype in the diagnosis, pathogenesis, or treatment of hypertension.
Functional Documentation in Chromaffin Cells
What is the mechanism by which TH 5′ allelic variants (Figure 1A and online Table I) influence human autonomic traits? We tested differential regulation of TH transcription with 1155-bp promoter haplotype/reporter plasmids transfected and expressed in chromaffin cells (Figure 6B). Indeed, we confirmed functional differences between TH promoter variants in vitro (Figure 6B), and the differences paralleled associations of these same variants with catecholamine secretion in vivo (Figure 5B). Thus, the 4 TH variants under consideration clearly lie in a domain with transcriptional activity, altering TH promoter strength in vitro.
Study Limitations and Caveats
Complex Trait Genetics
Multiple alleles may yield multiple traits. Genetic analyses of a complex trait necessitate the consideration of multiple phenotypes and genotypes, raising the possibility of false-positive (type I) statistical errors. We approached this issue in several ways: haplotyping, pleiotropic/bivariate analyses (1 gene yields >1 trait), multivariable analyses (>1 SNP yields 1 trait), SNPSpD (determining the “effective” number of SNPs within a block of LD and thereby adjusting the required threshold for significance of a single phenotype), permutation (exact) tests, and finally replication. TH haplotypes, simultaneously considering each common variant within the promoter, predicted autonomic traits (Figures 4 and 5⇑). A multivariable analysis established the particular role of C-824T on BP. SNPSpD determined that the promoter SNP effects on single autonomic traits exceeded chance alone. Permutation established an empirical level of significance; the effects of C-824T on twin traits and BP in the population remained significant. Finally, replication established the effect of C-824T on hypertension.
Haplotype Assignment Uncertainty
Imputation of phase from diploid genotype data is inherently uncertain and occasionally prone to misclassification; the haplotype method we used assigns the 2 most likely haplotypes to each individual.45 Although emerging haplotype methods consider uncertainty in phasing,46,47 such methods have not yet been coupled with the computational needs of correlated twin pair statistics.
Although we conducted systematic variant discovery in both black and white subjects, the studies on autonomic physiology were analyzed only in white (European ancestry) subjects, initially to avoid the potentially spurious effects of population stratification on genetic trait associations.48 Only additional studies can determine whether the associations in white subjects are generalizable to other population groups.
We conclude that catecholamine secretory traits are heritable (online Table IV), displaying joint genetic determination (or pleiotropy) (Figure 5) with autonomic activity and finally with BP in the population (Figures 7 and 8⇑). Interindividual differences in catecholamine secretion are influenced by genetic variation in the adrenergic pathway encoding catecholamine synthesis, especially at the classically rate-limiting step, TH. These results document novel pathophysiological links between a key adrenergic locus, catecholamine metabolism, and BP (Figure 8) and suggest new strategies to approach the mechanism, diagnosis, and treatment of systemic hypertension.
A schematic formulating our results into a global hypothesis is presented in Figure 8 and outlines the role of intermediate phenotypes and the TH candidate gene (especially its C-824T promoter variant) in the determination of hypertension via the initial intermediary of catecholamine metabolism.
Implications for Pathophysiology/Mechanism, Prediction/Diagnosis, and Treatment
Our results suggest that the adrenergic pathway is centrally involved in the early pathogenesis of hypertension beginning in healthy individuals, perhaps initially by altering baroreceptor function (Figures 5 and 8⇑) or consequently the transient response BP response to environmental stress (Figures 2, 3, and 8⇑⇑). Adrenergic genetic determination of biochemical (Figures 2A and 3⇑A) and physiological (Figures 2B and 3⇑B) traits and the ultimate disease trait (Figure 7) suggests that treatments targeting the adrenergic pathway might be beneficial in preventing hypertension if administered to subjects at specific genetic risk. Our results also raise the possibility that adrenergic genetic profiling of patients with impaired autonomic activity or increased stress BPs might yield practical pharmacogenetic predictors of patients most likely to benefit from sympatholytic therapy.
Future Directions/Studies: Implications for Prevention: Heredity and Environment
Our results raise the possibility that profiling subjects for particular adrenergic and signaling polymorphisms would provide an index of risk for or susceptibility to hypertension. This prediction awaits testing in appropriate longitudinal cohorts.
The authors appreciate the assistance of the General Clinical Research Center (RR00827) and its core laboratory.
Sources of Funding
This work was supported by the Department of Veterans Affairs and the National Institutes of Health, Bethesda, Md.
Sharma P, Hingorani A, Jia H, Ashby M, Hopper R, Clayton D, Brown MJ. Positive association of tyrosine hydroxylase microsatellite marker to essential hypertension. Hypertension. 1998; 32: 676–682.
Thibaut F, Ribeyre JM, Dourmap N, Meloni R, Laurent C, Campion D, Menard JF, Dollfus S, Mallet J, Petit M. Association of DNA polymorphism in the first intron of the tyrosine hydroxylase gene with disturbances of the catecholaminergic system in schizophrenia. Schizophr Res. 1997; 23: 259–264.
Zhang L, Rao F, Wessel J, Kennedy BP, Rana BK, Taupenot L, Lillie EO, Cockburn M, Schork NJ, Ziegler MG, O’Connor DT. Functional allelic heterogeneity and pleiotropy of a repeat polymorphism in tyrosine hydroxylase: prediction of catecholamines and response to stress in twins. Physiol Genomics. 2004; 19: 277–291.
Albanese V, Biguet NF, Kiefer H, Bayard E, Mallet J, Meloni R. Quantitative effects on gene silencing by allelic variation at a tetranucleotide microsatellite. Hum Mol Genet. 2001; 10: 1785–1792.
Abecasis GR, Cookson WO. GOLD: graphical overview of linkage disequilibrium. Bioinformatics. 2000; 16: 182–183.
Falconer DS, Mackay TFC. Heritability. In: Introduction to Quantitative Genetics. Essex, England: Longman; 1996; 160–183.
Perneger TV. What’s wrong with Bonferroni adjustments. BMJ. 1998; 316: 1236–128.
Cui JS, Hopper JL, Harrap SB. Antihypertensive treatments obscure familial contributions to blood pressure variation. Hypertension. 2003; 41: 207–210.
Carson RP, Robertson D. Genetic manipulation of noradrenergic neurons. J Pharmacol Exp Ther. 2002; 301: 410–417.
Mendelsohn ME, Karas RH. Molecular and cellular basis of cardiovascular gender differences. Science. 2005; 308: 1583–1587.
Thomas KS, Nelesen RA, Ziegler MG, Bardwell WA, Dimsdale JE. Job strain, ethnicity, and sympathetic nervous system activity. Hypertension. 2004; 44: 891–896.
Stein CM, Lang CC, Singh I, He HB, Wood AJ. Increased vascular adrenergic vasoconstriction and decreased vasodilation in blacks: additive mechanisms leading to enhanced vascular reactivity. Hypertension. 2000; 36: 945–951.
Lillie EO, O’Connor DT. Early phenotypic changes in hypertension: a role for the autonomic nervous system and heredity. Hypertension. 2006; 47: 331–333.
Kraft P, Hunter D. Integrating epidemiology and genetic association: the challenge of gene-environment interaction. Philos Trans R Soc Lond B Biol Sci. 2005; 360: 1609–1616.
King D, Etzel JP, Chopra S, Smith J, Cadman PE, Rao F, Funk SD, Rana BK, Schork NJ, Insel PA, O’Connor DT. Human response to alpha2-adrenergic agonist stimulation studied in an isolated vascular bed in vivo: biphasic influence of dose, age, gender, and receptor genotype. Clin Pharmacol Ther. 2005; 77: 388–403.
Zaitlen NA, Kang HM, Feolo ML, Sherry ST, Halperin E, Eskin E. Inference and analysis of haplotypes from combined genotyping studies deposited in dbSNP. Genome Res. 2005; 15: 1594–1600.
Tyrosine hydroxylase (TH) is the rate-limiting enzyme in catecholamine biosynthesis. We undertook systematic polymorphism discovery at the TH locus and then tested variants for contributions to sympathetic function and blood pressure, initially in twin pairs to probe heritability and then in the population. We found that catecholamine secretory traits are heritable, displaying joint genetic determination with autonomic activity and finally with blood pressure in the population. Interindividual differences in catecholamine secretion are influenced by genetic variation in the adrenergic pathway encoding catecholamine synthesis, especially at the classically rate-limiting step, TH. These results document novel pathophysiological links between a key adrenergic locus, catecholamine metabolism, and blood pressure and suggest new strategies to approach the mechanism, diagnosis, and treatment of systemic hypertension.
↵*The first 2 authors contributed equally to the research.
The online Data Supplement, which includes Methods, tables, and figures, can be found with this article at http://circ.ahajournals.org/cgi/content/full/CIRCULATIONAHA.106.682302/DC1.