Pharmacogenomics of β‐Adrenergic Receptors and Their Accessory Signaling Proteins in Heart Failure

Abstract

β‐adrenergic receptors (βAR) are widely expressed on cardiovascular cells. Pharmacological stimulation or blockade of βAR signaling is the therapeutic mainstay in cardiogenic shock, hypertension, ischemia, arrhythmias, and heart failure. Interindividual variability in the response to βAR agonists and antagonists has prompted examination of variability in the genes encoding βAR signaling pathway members. Prominent among the genes that have been examined so far in heart failure are the β₁AR, β₂AR, and G‐protein‐coupled receptor kinase 5 (GRK5). Each has nonsynonymous polymorphisms that alter amino acid sequence and protein function and regulation in cell‐based systems, genetically altered mouse models, or human hearts. Here, we review these phenotypes and results from published clinical studies, with a focus on heart failure pharmacogenomics. Thus far, very few studies have utilized analogous protocols or drugs, and discrepancies in the clinical studies are apparent. A compelling approach is the use of multiple methods to understand the molecular, cellular, and organ phenotypes of a variant and couple these with clinical studies designed to specifically address the relevance of those phenotypes in humans. Undoubtedly, additional loci will be identified, and together, will provide for genetically driven, individualized treatments for heart failure.

Introduction

G‐protein‐coupled receptors (GPCR) are some of the most common targets of pharmacotherapeutics, accounting for approximately 30% of all prescribed medications. ¹ Drugs targeting the prototypical β‐adrenergic receptor (PAR) have been especially useful in cardiovascular disease. PAR agonists such as isoproterenol, dobutamine, and dopamine are given parenterally to acutely increase cardiac output via their direct inotropic activities. Antagonists of PAR (β‐blockers) are used to treat hypertension, to control heart rate and maintain normal rhythm, to control angina and myocardial infarction, and to improve cardiac function and prolong life in heart failure. ²

BAR‐mediated effects on the heart occur after an interaction between an adrenergic agonist and one of the cardiac BAR which initiates a signal amplifcation cascade involving a G‐protein coupled to adenylyl cyclase (AC), protein kinase A, and its phosphorylation substrates. Signaling is dependent upon the properties of the ligand‐receptor transduction system, as well as accessory proteins, such as regulators of G‐protein signaling (RGS) and G‐protein receptor kinases (GRK), that modulate the signal ( Figure 1 ). Identification of naturally occurring genetic variants of human pARs and associated proteins ( Table 1 ), and subsequent determination of their association with various cardiovascular diseases, has helped to generate new pathophysiological models and suggest refnements of current therapeutic approaches. Here, we review some of these findings.

Figure 1.

Schematic diagram of the principal components of beta‐adrenergic signaling and their cardiac effects. Norepi = norepinephrine, the major endogenous β₁AR agonist; GRK = G‐protein‐coupled receptor kinase.

Table 1.

Coding region variations of adrenergic signaling genes.

Gene name	Common name	Nucleotide variability*,†	Amino acid variability*	MAF(%)
				White	Black‡
ADRB1	β1AR	145 (A/G)	49 (Ser/Gly)	15	13
		1165 (C/G)	389 (Arg/Gly)	27	42
		1166 (G/T)	389 (Arg/Leu)	<0.1	0.9
ADRB2	β2AR	46 (G/A)	16 (Gly/Arg)	40	50
		79 (C/G)	27 (Gln/Glu)	43	27
		491 (C/T)	164 (Thr/lle)	<4	<4
GNAS1	Gβs	393(T/C)	131 (lle/lle)	42	80
GNB3	Gβ3	825 (C/T)	275 (leu/leu)	33	90
GRK4	GRK4	448 (G/T)	65 (Arg/Leu)	35	47
		679 (C/T)	142 (Ala/Val)	40	60
		1711 (C/T)	486 (Ala/Val)	40	19
GRK5	GRK5	122 (A/T)	41 (Gin/Leu)	1.3	23

*The most common allele in a general US population (composed of 15% African Americans) is the first provided.

†Nucleotide position is relative to the ATG initiator codon.

‡Allele frequencies, as reported in the literature and public databases, show some variance in the frequency in the African American population, likely due to differences in racial

admixture.

MAF = minor allele frequency.

β₁AR Polymorphisms

The β₁AR subtype is the most common PAR in the myocardium, which couples to the stimulatory G‐protein (G_s) that activates AC, and is felt to be responsible for many of the pathological effects of catecholamines in heart disease (reviewed in Ref. 3). The intronless human P₁AR gene exhibits two common variations in the form of single nucleotide polymorphisms (SNPs) that change the encoded amino acids (i.e., are nonsynonymous) SNPs ( Table 1 ).

The most common β₁AR SNP results in a substitution of Gly for Arg at amino acid 389, within the receptor's fourth intracellular loop. In this seven transmembrane‐spanning receptor, the fourth loop is formed by residues between the distal transmembrane domain and membrane‐anchoring palmitoylated cysteines of the carboxyl terminus, and is essential for G‐protein coupling. This region is highly conserved, and in β₁AR, Arg is found in the analogous position in every species from which sequence data are available, except for the human Gly variant. The functional consequences of the variation were first demonstrated by stably expressing either the Gly or the Arg389 receptor at equivalent levels in Chinese hamster flbroblasts, ⁴ and confirmed in Chinese hamster ovary cells. ⁵ β₁Arg389 exhibited greater agonist stimulation of AC than β₁Gly389 ( Figure 2A ), due to enhanced G_αs coupling. The binding afinities for multiple βAR antagonists were not different between the two receptors. ⁵

Figure 2.

Phenotypes of theβ₁Arg389‐Gly389 polymorphism in various model systems. (A) Results from transfected CHW cells expressing equal levels of the two receptors. (B and C) Results from transgenic mouse hearts studied in the ex vivo work‐performing mode. At 3 months, basal and maximal agonist‐stimulated contractility are enhanced in Arg389 hearts. At 6 months of age (C), basal contractility for Arg389 remained greater than Gly389, but the response to agonist was essentially absent. (D) Results from young transgenic mice treated with vehicle or propranolol in their drinking water for 1 month. Heart rates were determined by echocardiography and showed that only the Arg389 mice had a reduction in the heart rate from β‐blockade.

A goal of translational investigation is to integrate clinical findings from human studies with mechanistic findings from tissue culture and animal models. Accordingly, to assess the consequences of β₁Arg389 and β₁Gly389 on cardiovascular physiology, we generated transgenic mice (FVB/N strain) overexpressing the two human receptors ⁶ and examined their cardiac function over a 9‐month period. Consistent with enhanced receptor‐G‐protein coupling, cardiac contractility at baseline and after β₁AR stimulation with dobutamine was greater for Arg389 mice than for Gly389 mice at 3 months of age ( Figure 2B ). At 6 months of age, Arg389 mice continued to exhibit greater contractility in the resting state, but were desensitized, that is, had lost the ability to respond to dobutamine due to impaired receptor‐Gs coupling and decreased expression of G_αs and AC5/6 ( Figure 2C ). Likewise, there was a differential response to the β‐blocker propranolol, which substantially decreased contractility in Arg389 hearts, but did so minimally and only at the highest concentrations in Gly389 hearts. ⁶ These results suggested that Arg389, which exhibited a hyperactive acute inotropic response to β‐agonist, desensitization, and inhibition of cardiac function by β‐blockers, might be the more favorable allele for responsiveness to treatment in heart failure. Consistent with this hypothesis, an experiment designed to mimic daily β‐blocker treatment in human disease by giving propranolol in the drinking water of 4‐month‐old transgenic mice demonstrated that only the Arg389 transgenic mice showed a decrease in heart rate ( Figure 2D ).

Human clinical studies have largely confirmed the β₁Arg389 and Gly389 cardiac phenotypes seen in transgenic mice. Compared to Gly389 carriers (i.e., subjects with one or two Gly 389 alleles), homozygous Arg389 subjects demonstrate enhanced exercise capacity, ⁷ , ⁸ contractile response to dobutamine, ⁹ , ¹⁰ and short‐term improvement in left ventricular ejection fraction after β‐blockade. ⁶ , ¹¹ Subsequent studies of the effects of these two β₁AR variants compared the responses of right ventricular trabeculae isolated from nonfailing and failing hearts. ¹² In nonfailing hearts from individuals homozygous for Arg389, the maximal contractile force was greater than that of the hearts from Gly389 carriers ( Figure 3A ). In failing hearts, this phenotype was maintained, but the differences were not as great ( Figure 3B , note scale change). Likewise, the atypical β‐blocker bucindolol caused a dose‐dependent decrease in force generation in trabeculae from Arg389 homozygous hearts, but not from Gly389 carrier hearts ( Figure 3C ). In contrast, the β‐blocker carvedilol was a neutral antagonist for both receptor variants. ¹²

Figure 3.

Physiologic properties of human right ventricular trabeculae from nonfailing and failing hearts stratified by the β₁AR position 389 variants. Trabecular strips were studied in organ bath experiments by measuring contractile force in response to the indicated drugs. (A) Dose‐response to isoproterenol from human nonfailing hearts (normal hearts not used for transplantation). (B) Dose‐response to isoproterenol from failing hearts (end‐stage, explanted hearts from patients receiving cardiac transplants); note scale difference between (A) and (B). Regardless of whether the hearts were normal or failing, the Arg389 homozygous responses to agonist were greater than Gly389 heterozygotes. (C) Dose‐response to bucindolol in failing hearts. Trabeculae were co‐stimulated with forskolin, which allows for detection of partial and inverse agonist actions. Bucindolol had an inverse agonist effect only in Arg389 hearts.

To follow up the human trabeculae studies and bette examine the consequences of β₁AR 389 variation in heart failure the Liggett and Bristow laboratories genotyped archived DN from 1,040 participants in the placebo‐controlled Beta‐Blocke Evaluation of Survival Trial (BEST) trial that evaluated the efficac of bucindolol in severe heart failure. Clinical characteristics o the placebo and bucindolol groups, including age, sex, rac left ventricular ejection fraction, etiology of heart failure, the New York Heart Association clinical class, and baseline hear rate and blood pressure, were well matched. ¹² In the geneti substudy, the primary outcome was all‐cause mortality, hear failure hospitalization, and the combined outcome, comparin the efficacy of placebo and bucindolol between homozygou Arg389 subjects and Gly389 carriers. The most importan result, summarized in Figure 4 , is that homozygous β₁Arg38 patients who received bucindolol showed significantly improve survival, whereas Gly389 carriers had identical survival whethe they received placebo or bucindolol. By comparing the placeb groups after stratifcation by genotype, it is evident that surviv of patients with the Gly389 variant is not affected by treatmen with bucindolol: compared to placebo, Arg389 patients treate with bucindolol had a hazard ratio (HR) of 0.62 (95% confidence interval [CI]= 0.40–0.96, p= 0.03) for the endpoint of death indicating an improvement in survival with bucindolol treatmen in those with this genotype. The same comparison in Gly389 carriers revealed no bucindolol effect on survival (HR = 0.90, 95% CI = 0.62–1.30, p= 0.57). Tere was also an apparent influence of β₁AR genotype on heart failure exacerbation during bucindolol treatment, measured by time to heart failure hospitalization. For this outcome, bucindolol‐treated Arg389 patients had HR = 0.64 (95% CI = 0.46–0.88, p= 0.006), whereas bucindolol‐treated Gly389 carriers showed no benefit (HR = 0.86, 95% CI = 0.64–1.15, p= 0.30). For the combined outcome of time to hospitalization or death, a bucindolol‐associated favorable treatment effect was evident for Arg389 patients (HR = 0.66, 95% CI = 0.50–0.88, p= 0.004), but not Gly389 carriers (HR = 0.87, 95% CI = 0.67–1.11, p= 0.250). The less common β₁AR polymorphism at amino acid position 49 provided no additional predictive value.

Figure 4.

Kaplan‐Meier survival curves in heart failure patients stratified by bucindolol or placebo treatments and Arg or Gly389 β₁AR genotypes. The Arg389 patients were homozygous, and Gly389 patients were carriers of one or two Gly389 alleles. Comparisons were within genotype between bucindolol and placebo. For Arg389 patients treated with bucindolol compared to placebo (HR = 0.62, 95% CI = 0.40–0.96, p= 0.03), indicating an improvement in survival with bucindolol in those with this genotype. This same comparison in Gly carriers revealed no difference in survival (HR = 0.90, 95% CI = 0.62–1.30, p= 0.57), indicative of no treatment response to bucindolol.

These are intriguing results, suggesting a pharmacogenomic interaction between a specific β₁AR polymorphism and an experimental β‐blocker with unique pharmacological characteristics found only with Arg389 (enhanced sympatholysis ² , ³ and inverse agonist activity ¹² ). It is important to consider whether these findings might also apply to β‐blockers currently recommended for heart failure treatment, such as metoprolol and carvedilol. To our knowledge, there is only one study that assessed β₁AR polymorphism genotypes of metoprolol‐ and placebo‐treated patients, a substudy of Metoprolol Randomized Intervention Trial in Heart Failure (MERIT‐HF). ¹³ This analysis compared outcome by genotype in the combined cohort of placebo‐and metoprolol‐treated patients, and the major conclusion was that Gly389 patients did not have improved outcomes. It is not possible from this study to assess whether there was a pharmacogenetic effect since comparisons within the two treatment arms, by genotype, were not performed. Separately, Shin et al. ¹⁴ examined survival in 227 heart failure patients, of which 81% received an unspecified β‐blocker. No association with β₁AR polymorphisms was noted, but a 2‐locus haplotype of the β₂AR subtype (see below) was associated with poor survival. de Groote et al. ¹⁵ studied 444 Caucasian heart failure patients treated with either bisoprolol or carvedilol and found a different β₂AR 2‐locus haplotype associated with survival, but no association with a β₁AR polymorphism. Finally, Sehnert et al. ¹⁶ recently published a retrospective 2‐center catheterization‐laboratory registry study of 637 heart failure patients treated with either carvedilol or metoprolol and found no association between β₁AR or β₂AR polymorphisms and survival. This study was retrospective, patient enrollment was based on results from cardiac catheterization (a potential selection bias), a formal β‐blocker titration protocol was not in place, compliance was not monitored after 6 months, and there was no placebo group. In comparison to the prospective, placebo‐controlled, randomized BEST trial, these results may represent “real‐world” conditions. Thus, the applicability of the BEST substudy results to other β‐blockers, or to less formal treatment practices, is unknown.

Another polymorphism of β₁AR that has received less attention substitutes Gly for Ser at amino acid 49 of the receptor's extracellular amino terminus. When Ser49 and Gly49 β₁ARs were stably expressed in CHW as well as HEK‐293 cells, the Liggett group found no differences in agonist‐promoted coupling to AC. ¹⁷ However, another group has reported increased basal and agonist‐stimulated AC with the Gly49 receptor in transfected HEK‐293 cells, ¹⁸ and both groups observed increased agonist‐promoted downregulation of the Gly49 receptor, compared to the Ser49 receptor. Several small studies have described potentially important associations between the Ser49Gly polymorphism and cardiac disease: patients with Ser49 reportedly show increased myocardial ischemia ¹⁹ and decreased benefit from β‐blocker therapy. ²⁰ In contrast, subjects with Gly49 are described as having increased aerobic power on exercising testing ²¹ and a lower risk of heart failure. ²² Other clinical implications of this polymorphism in heart failure, if any, are uncertain at this time.

β₂AR Polymorphisms

There are two common and one rare polymorphism of the P₂AR coding block ( Table 1 ). In transfected cells and transgenic mice, P₂Ile164 displays depressed signaling to AC ²³ and contractility ²⁴

A homozygous subject has yet to be reported for this polymorphism, and the estimated allele frequency is <4%; all published studies that have examined the response to β‐blockers in heart failure (HF) are underpowered to detect an effect of the Ile164 polymorphism. The position 16 and 27 SNPs are essentially found as only 3 of the 4 possible haplotypes: Arg16/Gln27, Gly16/Glu27, and Gly16/Gln27. ²⁵ In transfected cells, the Gly16 receptor (regardless of the position 27 genotype) undergoes enhanced agonist‐promoted downregulation compared to the Arg16 receptor. ²⁶ This appears to be at a step after receptor internalization at the point where receptors are either recycled or degraded. ²⁶ The results from association studies of β₂AR polymorphisms and β‐blocker response in HF have not achieved consensus, ¹⁴ , ¹⁵ , ¹⁶ and supporting mechanistic studies, such as more HF‐relevant cell‐based experiments, mouse models, or ex vivo human heart studies, are lacking. The basis for the discrepancies in the clinical studies may be due to a small effect of these polymorphisms on differential drug response and multiple different study designs (reviewed in Refs. 27 and 28).

G_s Polymorphisms

The stimulatory effects of βAR are transduced by G_s‐mediated activation of AC. In the inactive state, G_s exists as a GDP‐bound heterotrimer (α, β, and γ subunits). Activation of βAR causes G_αs to bind GTP and dissociate from the G_αγ subunit, permitting each to transduce signals (reviewed in Ref. 29). Polymorphisms of G_αs are rare, ³⁰ and no nonsynonymous variants have yet been described that contribute to cardiovascular disease. However, a common synonymous polymorphism (ATT>ATC, Ile131) in exon 5 has been associated with hypertension and blood pressure response to β‐blockade. ³¹ It is notable that rare activating GNAS (G_αs gene) mutants resulting in constitutive cAMP production, such as Arg201Leu, cause hormonal oversecretion associated with spontaneous tumors of pituitary, thyroid, and adrenal glands, and café‐au‐lait skin lesions in McCune–Albright's syndrome, ³² whereas inactivating G_αs mutations cause Albright's hereditary osteodystrophy and resistance to parathormone and pituitary‐derived thyroid stimulating hormone (TSH) and growth hormone releasing hormone (GHRH). ³³ Amino acid conservation in G_αs a n d the striking pathological effects of nonsynonymous mutations in GNAS suggest that there is little tolerance for functional variability in this critical signaling factor.

The C825T variant of G_β3 is reported to reduce the risk of myocardial infarction and increase the effectiveness of cholesterol‐lowering statin therapy ³⁴ and has been associated with features of the metabolic syndrome, including hypertension, obesity, dyslipidemia, and insulin resistance ³⁵ , ³⁶ (although not every study has been positive). ³⁷ This polymorphism causes alternate splicing that produces a truncated protein which can increase G‐protein signaling. ³⁸

RGS Polymorphisms

RGS proteins terminate G_α‐mediated signaling by accelerating hydrolysis of GTP to GDP which promotes reassociation of G_α and G_β subunits into the inactive heterotrimer (see Figure 1 ). Two linked intronic insertion/deletion polymorphisms, 1891, 1892TC and 2138, 2139AA are associated with essential hypertension in African Americans. ³⁹ , ⁴⁰

GRK Polymorphisms

GPCR signaling is terminated by uncoupling receptors from their G‐protein effectors, a process called desensitization. As depicted in Figure 1 , the mechanism involves phosphorylation of the receptor by downstream kinases (protein kinase A for βAR, heterologous desensitization) or a family of dedicated kinases that are activated by ligand‐occupied receptors, the GRK (GRKs 2, 5, and 6 for βAR). ⁴¹ No nonsynonymous polymorphisms have been confirmed for GRK2. ⁴² However, deep resequencing identified several uncommon nonsynonymous SNPs of GRK5 and confirmed a common nonsynonymous SNP ( Table 1 ). The common variant, Glu41>Leu, is in a putative regulatory domain and confers enhanced agonist‐promoted desensitization, phosphorylation, and internalization of βAR responses in cell‐based recombinant expression systems. ⁴² , ⁴³ Glutamine at this position is completely conserved across human, bovine, mouse, rat, dog, and zebrafsh GRK5 analogs, and the leucine variant is disproportionately represented in African Americans. ⁴² βAR pathway gene variants that have increased frequency in African Americans are of particular interest in cardiac disease because this racial group is reported to benefit less from pharmacological β‐blockade in heart failure. ⁴⁴

Case–control analyses showed that the GRK5 Leu41 variant did not alter the risk of developing heart failure, but a pharmacogenomic interaction was detected between GRK5 Leu41 and β‐blockers for the endpoint of death. ⁴² In CHO cells co‐tranfected with βAR and equivalent levels of GRK5Q41 or L41, L41 cells revealed a more rapid and extensive desensitization of cAMP accumulation in response to agonist exposure ( Figure 5A ). Similarly, in transgenic mice with myocyte‐targeted expression of GRK5 variants infused with isoproterenol, desensitization of adenylyl cyclase activity by L41 was greater than by Q41 ( Figure 5B ). Compared to transgenic mice expressing GRK5 Glu41, mice expressing GRK5 Leu41 in the heart were protected from catecholamine‐induced left ventricular remodeling, affording similar protection as the β‐blocker propranolol ( Figure 5C, D ). The provocative results of a preliminary study of African Americans with heart failure are shown in Figure 5E, F . Approximately 40% of African Americans who carried one or two copies of the GRK5 Leu41 variant and were β‐blocker naïve were protected against death or cardiac transplantation to a degree that was similar to that afforded by pharmacological β‐blockade in subjects homozygous for the more common GRK5 Glu41 allele. ⁴² Taken together, these studies identified a gain‐of‐desensitization function for this GRK5 polymorphism that has similar effects as β‐blockade in cell and transgenic mouse models, and that appears to mimic some of the benefits of β‐blocker therapy in African American heart failure patients. Whether βAR desensitization is the only mechanism responsible for the observed GRK5 polymorphism effects, ⁴⁵ and whether this polymorphism also modifes other diseases in which βAR signaling plays a critical role and in diseases in which GRK5 interacts with other GPCRs, is not yet known.

Figure 5.

Functional and clinical characteristics of GRK5 Glu41 and Leu41. (A) Tissue culture studies showing GRK5 Leu41 (L41) desensitizing β₁AR‐stimulated cAMP accumulation more rapidly than GRK5 Glu41 (Q41). (B) Isolated perfused heart studies showing that GRK5 Leu41 (L41) desensitizes myocardial β₁AR‐stimulated adenylyl cyclase activity more than GRK5 Glu41 (Q41). (C) Treatment with the β‐blocker propanolol prevents left ventricular remodeling, measured as the increase in left ventricular end‐diastolic dimension (LVEDD), in chronic isoproterenol‐stimulated heart failure. (D) GRK5 L41 transgenic mice are resistant to ventricular remodeling induced by chronic isoproterenol (black bars), similar to propanolol treatment (gray bars). (E and F) Kaplan–Meier survival curves of African Americans stratified by β‐blocker treatment in homozygous GRK5 Q41 (E) and GRK5L41 carriers (F).

Genetic polymorphisms of GRK4 have also been associated with cardiovascular disease, but in an indirect way. GRK4 is expressed largely in tissues of the genitourinary tract. Three polymorphisms, Arg65Leu, Ala142Val, and Ala486Val, have been associated with essential hypertension. ⁴⁶ The proposed mechanism involves enhanced GRK‐mediated phosphorylation of renal dopamine D1 receptors that impairs the natriuretic and diuretic effects of dopamine. ⁴⁷

Conclusion

Genomics studies of GPCRs and their accessory proteins, especially of the βAR pathway in cardiac disease, are providing the foundation for new approaches to drug evaluation and disease management. The brief list of polymorphic receptors and signaling factors reviewed above illustrates the capacity of natural genetic variation to alter functional signaling in a manner that can either mimic, compliment, or oppose the effects of endogenous hormones and prescribed medications.

The time is approaching where it will be essential to account for genetic characteristics in clinical studies of drug efficacy to properly direct optimal therapeutics to the target patient population. This requires establishing a reasonable likelihood that a given genetic marker is associated with a particular disease or outcome. Various approaches, including cell and genetically modified mouse models and targeted physiological studies in humans, can be utilized to define the phenotype and mechanism and the effects on human end‐organ function. Ultimately, the relevance of a given genetic variant will require adequately powered clinical association studies with meaningful outcomes and replication. In addition, the effects will need to be placed into context with those of other variants in order to gain insight into the most useful pharmacogenetic loci for predicting responses.