ASSOCIATION ANALYSES OF ADRENERGIC RECEPTOR POLYMORPHISMS WITH OBESITY AND METABOLIC ALTERATIONS

Abstract

Genes involved in the regulation of catecholamine function may be important in obesity because of the role catecholamines play in energy expenditure and lipolysis. To determine if common single nucleotide polymorphisms (SNPs) in β₁ (ADRB1), β₂ (ADRB2), β₃ (ADRB3) and α₂a adrenergic receptor (ADRA2A) genes associate with obesity and metabolic alterations, we recruited 74 healthy African American and 161 Caucasian males and females (age: 18–49y) to participate in this case-control genetic association study. Genotypes were determined by PCR and RFLP. Associations between genotype and body mass index (BMI), percent body fat (by measuring skin fold thickness in 7 different sites), fasting (12-hour) plasma glucose, insulin, potassium concentrations, glycated hemoglobin (Hgb AIC) and insulin resistance (homeostasis model assessment score (HOMA-_IR)) were performed. Among Caucasians, the ADRB1 Arg389→Gly variant associated with insulin concentrations and HOMA_IR: mean ± SD values for insulin and HOMA_IR in Arg389 homozygotes and carriers of the Gly were 10 ± 7.0 and 12 ± 9.4 μIU/ml (p=0.02) and 2.1 ± 1.7 and 2.6 ± 2.2 (p=0.057), respectively. Systolic blood pressure was higher in Caucasian for carriers of the ADBR1 Ser49 compared to Gly49 homozygotes 124 ± 12.6 vs. 119 ± 11.3 mm Hg, respectively (p = 0.02). Subsequent analysis revealed that these associations were attributable to a higher BMI among obese participants. The ADRA2A G1780A SNP associated with BMI and percent body fat in African Americans (p=0.05). Interactions were detected between ADRA2A C-1291G and ADRB2 Gln27→Glu variants for obesity in African Americans; and between ADRA2A C-1291G SNP and ADBR1 haplotype for obesity in Caucasians. We conclude that common SNPs in adrenergic receptor genes may be important susceptibility loci for obesity and related alterations. Because of the limited size of our populations, our results should be interpreted with caution and should be replicated in larger populations.

INTRODUCTION

Obesity is a major health problem in the United States. Sixty-five percent of American adults are overweight, and over 30% are obese ¹. It is well accepted that obesity in some but not in all individuals leads to metabolic alterations including hyperinsulinemia and insulin resistance ², which can lead to the development of Type II diabetes and to cardiovascular diseases and some forms of cancer ^{3 4 –6}. Although rare, obesity syndromes can be caused by mutations in single genes. However, the greatest proportion of obesity involves variants in multiple genes interacting with environmental factors particularly diet ^7;8.

Genes that are involved in the regulation of catecholamine function may be important in obesity because of the role catecholamines play in energy expenditure and lipolysis. Fat stored in the body as triglycerides is hydrolyzed to free fatty acids and glycerol through the process of lipolysis ⁹. Activation of β adrenergic receptors (ADRB) expressed in adipocytes mediate lipolysis ^3;10;11 while stimulation of α₂ adrenergic receptors (ADRA2) inhibit lipolysis¹². Insulin is also an important inhibitor of catecholamine-stimulated lipolysis ¹³ by reducing the ADBR effects of epinephrine and by activating ADRA2 in adipocytes ¹⁴.

Additionally, catecholamine-stimulated whole body lipolysis and lipolysis in subcutaneous adipocytes are blunted in obesity ^10;15;16, thereby limiting lipid mobilization and favoring fat accumulation. The mechanisms underlying lipolytic resistance to catecholamines in obesity are not clear and may include desensitization of ADRB2 function ¹⁷, increased activity of ADRA and the hyperinsulinemia that accompanies obesity ^18;19.

Several mutations in genes encoding ADRA and ADRB have been identified that could alter receptor expression and function. Given the important role that adrenergic receptors play in regulating energy expenditure and lipolysis, it is possible that common genetic polymorphisms in these genes contribute to obesity and to the accompanying metabolic alterations. Numerous studies have reported on the relationship between obesity and genetic variants in adrenergic receptors in different populations with conflicting results ^20–22. Few however, have explored associations between adrenergic receptor polymorphisms and obesity and accompanying metabolic alterations in self-identified African Am ericans and Caucasians. In the present study we tested the hypothesis that obesity and metabolic alterations associate with common polymorphisms in adrenergic receptor genes.

METHODS

Study Participants, Outcome Measures

All participants were between the ages of 18 and 49 years old, healthy, non-smokers and were not taking any medications chronically. Body Mass Index (BMI) was calculated from height and weight using the formula: BMI=[Body weight (pounds)/height (inches)²]*704.5. Obesity was defined as a BMI ≥ 30 ¹. Percent body fat was determined by calculating the mean of 3 measurements of skin fold thickness in 7 different anatomical sites (triceps, biceps, sub scapula, abdominal, suprailiac, thigh and chest), using a Lange Skin fold Caliper (Beta Technology Inc., Cambridge, MD) and equations as previously described ²³. All participants were fasted overnight. Systolic and diastolic blood pressures were measured manually with a mercury sphygmomanometer by auscultation using a stethoscope, following a 5-minute period of supine rest. Fasting plasma concentrations of glucose were quantified by the glucose oxidase method with a Glucose Analyzer II (Beckman, Fullerton, CA). Anthropometric measurements were performed by one of two registered nurses. Fasting plasma concentrations of insulin were quantified by RIA (INSULIN RIA, Diagnostics Systems Laboratories, Inc., Webster, TX). Fasted glycated hemoglobin (Hgb AIC) and potassium concentrations were quantified by a Beckman Autoanalyzer Model LX20 (Beckman, Fullerton, CA). Insulin resistance was assessed using the homeostasis model assessment score (HOMA-_IR) ²⁴. Studies were performed in the Nemours Children’s Clinic and the University of Florida Shands Jacksonville/SJCHC, Jacksonville, FL, and were approved by the Nemours Children’s Clinic clinical research review committee, Nemours Children’s Clinic and University of Florida institutional review committees. All subjects gave informed written consent prior to entering the study.

Determination of Genotype, Haplotype

The genotype of common single nucleotide polymorphisms (SNP) for the adrenergic receptors was determined by PCR and RFLP. Genomic DNA was extracted from whole blood using a DNA isolation kit (Gentra Systems, Inc., Minneapolis, MN, USA). Primers were synthesized by Operon Technologies, Inc. (Alameda, CA, USA). The restriction enzymes were purchased from New England Biolabs, Inc. (Beverly, MD, USA). The following SNPs were typed for the ADRB1 (NCBI RefSeq: NM_000684): A145G (Ser49→Gly, rs1801252), C1165G (Arg389→Gly, rs1801253); ADRB2 (NCBI RefSeq: NM_000024): T-47C (Cys19→Arg, rs1042711), G46A (Gly16→Arg, rs1042713), C79G (Gln27→Glu, rs1042714); ADRB3 (NCBI RefSeq: NM_000025): T190C (Trp64→Arg, rs4994); ADRA2A (NCBI RefSeq: NM_000681): C-1291G (rs1800544); G1780A (3’UT, rs553668). The first nucleotide in the initiating codon of methionine for translation is referred as nucleotide position +1 in the reference sequences. Because of strong linkage disequilibrium between amino acids 49 and 389 ^25;26, we inferred the following ADRB1 haplotypes: Ser49/Arg389, Ser49/Gly389, Gly49/Arg389. ADRB2 haplotypes Cys19/Arg16/Gln27, Cys19/Gly16/Gln27 and Arg-19/Gly16/Glu27 were determined as previously described ²⁷. The following primers and restriction enzymes were used in PCR and RFLP for ADRB1 Ser49Gly: forward 5′-GTCCTGGGCGCCTCCGAG-3′/reverse 5′-GATGGCCACGATCACCAGCAC-3′ and Sau96 I; ADRB1 Arg389Gly: forward 5′-CGCTCTGCTGGCTGCCCTTCTTCC-3′/reverse 5′-GGTCTCCGTGGGTCGCGTGG-3′) and BstN I; ADRB3 Trp64Arg: forward 5′-CAATACCGCCAACACCAGTGG-3′/reverse 5′-GGTCATGGTCTGGAGTCTCG-3′ and Msp I; ADRA2A C-1291G: forward 5′-GGAGGTTACTTCCCTCG-3′/reverse 5′-GGTACCTTGAGCTAGAGAC-3′ and Msp I; A1780G: forward 5′-CAGAGCAGCACTGGACTAC-3′/reverse 5′-TGGAAGGCATCTCTCCCAAG-3′ and Dra I

Statistics

We estimated a sample size of 88 obese and 88 non-obese participants to address the hypothesis in this study. This estimate was based on the ADRB2 haplotype frequencies ²⁸, and assuming that the frequency of the Gly16/Glu27 haplotype is 20% higher in obesity compared to non-obese participants. Normal Q-Q plots ²⁹ with tests were used to assess normality of distribution. The data were log-transformed when characteristics departed from normality in African Americans and Caucasians. Hardy-Weinberg equilibria between expected and observed genotype distributions were calculated using Chi-square goodness-of-fit tests. Deviations between observed and expected values with probabilities larger than 10% was accepted as no statistical deviation from H-W equilibrium. Linkage disequilibrium (LD) ³⁰ was assessed by comparing expected and observed haplotype distributions, and analyzed by Haploview32. The statistical significance of the estimated LD was tested by converting LOD score into an equivalent chi-squared statistic using one degree of freedom.

The prevalence of obesity by gender and race were compared using Fisher’s exact test ³¹. Non-parametric Spearman correlation coefficients ³² between BMI and metabolic characteristics were calculated based on ranking the two variables. Metabolic characteristics were compared between non-obese and obese participants, of which Student’s unpaired t-test was used for continuous variables with normal distribution including body fat, Hgb A1C, systolic blood pressure, and diastolic blood pressure; nonparametric method, Mann-Whitney test, was used for data that are significantly skewed including BMI, potassium, glucose, insulin, and insulin resistance.

Frequency distributions of SNP genotypes and haplotypes between African Americans and Caucasians were compared by x² test. General linear models ³³ were used to assess qualitative and quantitative explanatory variables at the same time. Assuming a general linear model, logarithmic transformation was used to normalize the distributions of BMI, plasma potassium, glucose, insulin, and insulin resistance. The linear model was applied in all associations between the genetic variants with metabolic and clinical responses. Percentage of body fat and BMI were adjusted for age and sex; all other metabolic and clinical characteristics including potassium, glucose, insulin, Hgb A1c, insulin resistance were also adjusted for percentage of body fat.

To estimate the risk of obesity associated with candidate SNPs, odds ratios were calculated. Crude odds ratios were calculated by dividing the odds in the exposed group by the odds in the control group. Adjusted odds ratios for the risk of obesity were calculated by logistic regression model with adjustment of age and sex ³¹. For each odds ratio, two-tailed p-vale and 95% confidence interval (CI) were estimated. Statistical significance was made at a cutoff value of P-value<0.10 for screening risk factors ³⁴. Analysis of covariance was applied in associations between the numbers of haplotype copies and BMI adjusted for age and sex ³³. Chi-square test was used for testing obese prevalence differences between the numbers of haplotype copies. A general linear model with covariates of age and sex was used to assess interactions between ADRA and variants of ADRB1 and ADRB2. All analyses were performed in SPSS. P values < 0.1 were considered significant.

RESULTS

Participants

A total of 238 subjects participated in the study; 161 were self-identified Caucasians (7 Hispanic, 154 non-Hispanic), 74 were African Americans and the remaining participants were either Indian (2) or Asian (1). Because of the small numbers of Indians and Asians, analyses were performed on 235 participants; 161 self-identified Caucasians (72.7% female) and 74 African Americans (80% female). Mean ± SD age for African Americans and Caucasians were: 30 ± 8.7 vs. 30 ± 8.0 years, respectively (p = 0.63). Among African American, BMI and % body fat were higher compared to Caucasians: 33 ± 7.8 vs. 29 ± 7.5 kg/m² (p=0.001); and 36 ± 8.2 vs. 32 ± 10 %, respectively (p =0.008). Among Caucasians 36% were classified as obese, compared to 61% in African Americans.

Table 1 compares anthropometric, metabolic and other characteristics by ethnicity. For African Americans and Caucasians, BMI, percent body fat, insulin concentrations and values for HOMA_IR were higher among obese individuals compared to non-obese. For Caucasians, resting glucose concentrations, HgbAIC and systolic blood pressure were higher in obese participants compared to non-obese. Among African Americans, diastolic blood pressure was higher in obese compared to non-obese individuals. For African Americans, glucose concentrations and Hgb AIC were similar between obese and non-obese individuals.

Table 1.

Comparison of means ± SD for anthropometric measures and metabolic characteristics by ethnicity and body weight designation.

Measurement	Caucasians					African Americans
	obese		normal		p	Obese		normal		p
	n	x̄ ± sd	n	x̄ ± sd		n	x̄ ± sd	n	x̄ ± sd
BMI(Kg/m2)	58	37.02 ± 5.73	103	24.24 ± 3.26	<0.0001	45	37.26 ± 6.07	28	25.27 ± 3.33	<0.0001
Body Fat (%)	58	40.30 ± 6.65	101	27.59 ± 7.73	<0.0001	44	40.33 ± 5.26	28	28.52 ± 6.73	<0.0001
Potassium (meq/ml)	55	4. 38 ± 0.57	98	4. 37 ± 0.49	0. 959	45	4.17 ± 0.23	28	4.25 ± 0.53	0.522
Glucose (mg/dl)	55	87.27 ± 10.40	99	82.23 ± 9.14	0.015	45	86.13 ± 12.66	28	84.32 ± 11.11	0.364
Insulin (uIU/ml)	54	15.52 ± 10.16	98	9.48 ± 10.81	<0.0001	45	13.72 ± 7.94	28	8.44 ± 6.75	0.001
Hgb AIC (%)	51	5. 35 ± 0.46	91	5.12 ± 0.41	0.002	40	5.51 ± 0.49	26	5.42 ± 0.46	0.464
HOMA IR	54	3.46 ± 2.62	98	1. 91 ± 2.04	<0.0001	45	2.99 ± 1.88	28	1.80 ± 1.68	0.001
Systolic	58	123. 95 ± 10.58	103	118. 30 ± 11.99	0.003	44	123 ± 13.78	27	119.30 ± 12.23	0.215
Diastolic	58	76.50 ± 8.50	103	74.29 ± 9.08	0.132	44	78.00 ± 10.01	27	73.19 ± 8.21	0.039

Student t test was used for variables of body fat, Hgb A1C, systolic blood pressure, and diastolic blood pressure; Mann-Whitney U test was used for BMI, potassium, glucose, insulin, and HOMA _IR.

Relation between BMI and Metabolic Characteristics

Glucose, insulin, HOMA_IR and HA1c were related to BMI in both African Americans and Caucasians (Figure 1). The relationship between glucose concentrations and BMI was weak with an R² value of 0.018 (p = 0.045). Insulin concentrations and HOMA_IR values were more strongly related with obesity with R² values of 0.14 (p < 0.0001) and 0.18 (p<0.0001), respectively (Figure 1), demonstrating increased insulin resistance at normal glucose concentrations. Blood pressure was weakly associated with BMI: the R² values for systolic and diastolic blood pressures were 0.046 (p=0.001) and 0.018 (p = 0.037), respectively. Systolic and diastolic blood pressures were not associated with age (p = 0.21 and 0.36, respectively (data not shown)).

Figure 1.

Relationship between mean ± SD fasting glucose concentrations (A), insulin concentrations (B), insulin resistance (C) as determined by HOMA_IR and blood pressure (D) as a function of BMI in self-identified Caucasians and African Americans.

Genotype, Haplotype Frequencies

A high rate of genotyping (> 98%) was achieved for all SNPs, which were in Hardy-Weinberg equilibrium when assessed by ethnicity except for the ADBR2 Gly16→Arg variant in Caucasians (data not shown). The ADRB2 SNPs Cys19→Arg and Gln27→ Glu27 were in complete linkage disequilibrium (LD). ADRB1 SNPs Ser49→Gly and Arg389→Gly; ADRB2 SNPs amino acids Gln27→ Glu27; and Gly16→Arg and Gln27→Glu; and ADRA2A SNPs C–1291G and G→1780A were in tight LD (data not shown). Table 2 compares genotype frequencies of the 8 adrenergic receptor SNPs between 2 ethnic groups. Ethnic differences for genotype frequencies were found for ADRB1 (Ser49→Gly), ADRB2 (Cys19→Arg; Gln27→Glu) and the ADRA2A (C-1291G) SNPs. Ethnic differences were also noted for ADRB1 and ADRB2 haplotype frequencies (Table 3).

Table 2.

Adrenergic receptor polymorphism genotype frequencies in African American and Caucasian subjects.

SNPs	Genotype	Frequency (%)		Pearson Chi-Square	P-value
		African American	Caucasian
ADRB1 Codon 49	A/A	44 (59.5)	120(75.0)	7.440	0.02
	G/A	25 (33.8)	37(23.1)
	G/G	5 (6.8)	3 (1.9)
ADRB1 Codon 389	C/C	30 (40.5)	85 (53.1)	3.614	0.16
	C/G	35 (47.3)	63 (39.4)
	G/G	9 (12.2)	12(7.5)
ADRB2 5′ LC-19	CC	4 (5.4)	31(19.4)	15.91	<0.0001
	TC	22 (29.7)	67(41.9)
	TT	48 (64.9)	62(38.8)
ADRB2 Codon 16	CC	12 (16.2)	33(20.6)	4.808	0.09
	GC	39 (52.7)	60(37.5)
	GG	23 (31.1)	67(41.9)
ADRB2 Codon 27	CC	48 (64.9)	62(38.8)	15.905	<0.0001
	GC	22 (29.7)	67(41.9)
	GG	4 (5.4)	31(19.4)
ADRB3	CC	3 (4.2)	2 (1.3)	3.454	0.18
	TC	15 (20.8)	23(14.7)
	TT	54 (75)	131(84.0)
Alpha 2A C-1291G	CC	10 (13.7)	85(53.5)	39.827	<0.0001
	CG	37 (50.7)	58(36.5)
	GG	26 (35.6)	16(10.1)
Alpha 2A A2659G	AA	4 (5.5)	8(5.0)	5.470	0.07
	AG	29 (39.7)	40(25.0)
	GG	40 (54.8)	112(70.0)

Table 3.

Haplotype frequencies for ADRB1 and ADRB2 in African American and Caucasian subjects

Gene	Haplotype	Frequency		P value
		African American (%)	Caucasian (%)
ADRB1	Gly49-Arg389	35(23.6)	43(13.4)	<0.0001
	Ser49-Arg389	60(40.5)	194(60.6)
	Ser49-Gly389	53(35.8)	83(25.9)
ADRB2	5′LCCys19-Gly16-Gln27	55(37.2)	65(20.3)	<0.0001
	5′LCCys19-Arg16-Gln27 5′LCArg19-Gly16-Glu27	64(43.2) 29(19.6)	126(39.4) 129(40.3)

Genetic Association Analyses

Among Caucasians, significant associations were found for the ADRB1 Arg389→Gly variant and insulin concentrations and HOMA-_IR (Figure 2; table 4): mean ± SD values for insulin and HOMA-_IR in Arg389 homozygotes and carriers of the Gly were 10 ± 7.0 and 12 ± 9.4 μIU/ml (p=0.02) and 2.1 ± 1.7 and 2.6 ± 2.2 (p=0.057), respectively, at similar values for BMI, percent body fat and glucose concentrations. When we stratified by BMI (obese < 30 kg/m ² vs. non-obese ≥ 30 kg/m²), the associations between ADRB1 Arg389→Gly variant and insulin and HOMA_IR were lost (Table 4). However, among obese participants, BMI was significantly higher in carriers of the Gly allele compared to Arg389 homozygotes: 39 ± 6.5 vs. 35 ± 4.0; p = 0.018, suggesting that the association between the Arg389→Gly genotype and insulin and insulin resistance (Figure 2) was attributable to the higher BMI among carriers of the Gly16 allele. Systolic blood pressure was higher in Caucasians for carriers of the ADBR1 Gly49 compared to Ser49 homozygotes: 124 ± 12.6 vs. 119 ± 11.3 mm Hg, respectively, p= 0.023. For non-obese participants, systolic blood pressures in Ser49 homozygotes (n=74) and carriers of the Gly (n=28) were 117 ± 11 and 122 ± 14, respectively; p = 0.19. For obese participants, systolic blood pressures in Ser49 homozygotes (n=45) and carriers of the Gly (n=12) were 123 ± 11 and 129 ± 8.9, respectively, p= 0.088.

Figure 2.

Influence of the ADRB1 Arg389→Gly (G1165C) polymorphism on insulin resistance and insulin concentrations in Caucasians. Insulin concentrations and values for HOMA_IR were log transformed; means ± SE were determined and tested by t-test in Arg389 homozygotes and Gly389 carriers.

Table 4.

Comparison of mean ± SD BMI, insulin and insulin resistance (by HOMA_IR) by ADRB1389 genotype among obese and non-obese Caucasians.

Measurement (white)		ADRB1codon389(x̄ ± sd)/n
		CC	CG+GG	P
BMI(Kg/m2)		27.97 ± 6.04 N=85	29.63 ± 8.72 N=75	0.29
Insulin (Uiu/ml)		9.99 ± 6.96 N=81	12.28 ± 9.36 N=69	0.041
BMI (Kg/m2)	Non-obese	24.41 ± 3.11 N=56	24.04 ± 3.45 N=47	0.70
	obese	34.86 ± 3.99 N=29	39.02 ± 6.48 N=28	0.018
Insulin (uIU/ml)	Non-obese	7.77 ± 3.35 N=54	9.50 ± 7.24 N=43	0.102
	obese	14.43 ± 9.78 N=27	16.87 ± 10.71 N=26	0.33
HOMA		2.13 ± 1.72 N=81	2.63 ± 2.24 N=69	0.057
HOMA	Non-obese	1.59 ± 0.69 N=54	1.93 ± 1.47 N=43	0.15
	obese	3.20 ± 2.52 N=27	3.78 ± 2.78 N=26	0.34

Among African Americans, there was a significant gene dose-response relationship for the ADRA2A G1780A genotype with both BMI and percent body fat (Figure 3). Compared to carriers of the ADRB2 Glu27 (n=25), Gln27 homozygotes (n=48) had higher insulin concentrations: 12.3 ± 8.0 vs. 10.6 ± 7.7, p = 0.056; and were more insulin resistant as determined by HOMO_IR: 2.7 ± 2.0 vs. 2.2 ± 1.8, p = 0.06. There were no significant associations between the genotype of the ADRB3 Trp64→Arg SNP and anthropometric measures or metabolic characteristics in Caucasians or African Americans (data not shown).

Figure 3.

Mean ± SE of BMI and percentage of body fat by ADRA2A A1780G genotypes in African Americans

Gene-Gene Interactions

Among Caucasians there was a significant interaction for BMI between ADRB1 haplotype and ADRA2A C-1291G SNP (p=0.038): 5 participants with the Ser49-Arg389/Ser49-Gly389 diplotype and C1291G G homozygous genotype had the highest BMI: 39.4 ± 9.73 kg/m². Among African Americans, there was a significant interaction between the ADRB2 Gln27→Glu variant and the ADRA2A C-1291G SNP, p =0.010: 6 participants who were homozygous for Gln27 and homozygous for the C-1291 SNPs had the highest BMI compared to other combinations: 38.8 ± 5.98 kg/m².

Discussion

Obesity is a common complex disease that involves multiple genetic variants interacting with environmental and behavioral factors. Genes involved in the regulation of catecholamine function may be important in obesity because of the role catecholamines play in energy expenditure and lipolysis. Stimulation of ADRB increases lipolysis, which favors weight reduction, whereas ADRA2A inhibits lipolysis, which favors weight gain. In the present study we hypothesized that common SNPs in ADRB and ADRA2A associate with obesity and accompanying metabolic alterations. The results of our study demonstrate that common SNPs in ADRB and ADRA2A genes may contribute to obesity and to metabolic alterations in young healthy Caucasians and African Americans, although in a complex manner. Among Caucasians, genotype (and haplotype) of ADRB and ADRA2A SNPs contributed little or not at all to obesity, which is in keeping with previous studies in older adult women ²⁶ and children ³⁵, though in an earlier study, the Arg389 variant was weakly associated with obesity in older adult women (0.86 kd/m²/Arg allele) ³⁶. In the present study, Caucasian carriers of the ADRB1 Gly389 (heterozygotes and Gly homozygotes) had higher insulin concentrations and were more insulin resistant compared to Arg389 homozygotes (Figure 2) at similar BMI and percent body fat values. To determine what drove this association, we stratified our cohort by obesity status, and found that among obese participants (BMI ≥ 30 kg/m²), BMI was significantly higher among Gly389 carriers compared to Arg389 homozygotes (Table 4). These data suggest that the association we found between the Gly389 variant and insulin resistance may, at least in part, be attributed to the higher BMI in this genotype. We therefore conclude that the ADRB1 Arg389→Gly polymorphism contributes to obesity and to accompanying metabolic alterations in young healthy Caucasians.

The functional basis underlying the association between the ADRB1 Arg389→Gly variant with obesity and metabolic alterations may be related to the pivotal role the sympathetic nervous system (SNS) plays in obesity. Previous studies have reported that compared to the Gly389, the Arg389 form of the ADRB1 mediated a higher isoproterenol-stimulated adenylyl cyclase activity and greater coupling ³⁷, demonstrated a greater inotropic response to stimulation ³⁸, and was more responsive to beta blockade ²⁵, though several studies report no differences between the two variants for several phenotypes (see ref21). It has been hypothesized that SNS underactivity could lead to an inadequate thermogenic response to overeating, positive energy balance and weight gain ³⁹. Alternatively, SNS activation that occurs with chronic overeating can lead to enhanced sympathetic stimulation ^39;40. Our finding is consistent with the former hypothesis as the Gly389 variant is the less active form of the ADBR1. Whatever the mechanism, it apparently does not involve lipolysis since lipolytic response to ADRB1 stimulation in human adipocytes did not differ by amino acid 389 variants ⁴¹.

Among Caucasians, we found that carriers of the ADRB1 Gly49 had higher systolic blood pressures compared to Arg49 homozygotes. Although we did not find associations between BMI or percent body fat with Ser49→Gly variant in obese vs. non-obese participants, among obese individuals carrying the Gly49, systolic blood pressure was 6 mm Hg higher compared to obese Ser49 homozygotes (p = 0.088). These data suggest that the higher systolic blood pressure we observed in carriers of the Gly49 was driven by the higher BMI. These data are consistent with the study by Linne et al ⁴² who reported that the Gly49 was associated with long-term weight gain in women. In in-vitro studies the Gly49 variant demonstrated higher basal and agonist-stimulated adenylyl cyclase activity compared to the Ser49 ⁴³, and the Gly49 was down-regulated to a greater extent by continuous agonist exposure ^43;44. It is possible that increased sympathetic tone contributes to elevations in blood pressure in the obese state ⁴⁰, which are consistent with our findings.

Among African Americans, we found a weak association between ADRB2 Gln27→Glu and insulin: Gln27 homozygotes tended to have slightly higher insulin concentrations and were slightly more insulin resistant. The ADRB2 gene is highly polymorphic ⁴⁵. In studies of human adipocytes, the Glu27 variant has been associated with reduced lipolytic activity compared to the Gln27 ⁴⁶. Our results are not consistent with these studies. However, the results of several association studies of common SNPs (genotype and haplotype) with obesity, diabetes and other common complex diseases are conflicting ²¹. Taken together, and given the small numbers of African Americans in our study, the relevance of this finding is not clear.

Our data suggest that the ADRA2A gene may play an important role in obesity (Figure 3). The ADRA2A gene is located on chromosome 10q24-q26, is intronless and highly polymorphic ⁴⁷. The C -1291G SNP is located in the promoter or enhancer region with the G allele having frequencies of 0.27 and 0.67 in Caucasians and African Americans, respectively. The G1780A SNP is located in the 3’ UTR with the A allele having frequencies of 0.17 and 0.31, respectively ⁴⁷. In our study, the allele frequencies for the G-1291 allele in Caucasians and African Americans were 0.28 and 0.61, respectively; for the A1780 SNP the allele frequencies were 0.18 and 0.25, respectively. These ADRA2A variants were selected because associations with the G1780A SNP and hypertension in African Americans ⁴⁸, with obesity in Caucasians ^49;50, and with endurance in Caucasian athletes have been reported ⁵¹. The C -1291G SNP has been associated with changes in glucose and diastolic blood pressure in response to dexamethasone ⁵². In the present study, the G1780 allele was associated with obesity and percent body fat among African Americans. Moreover, we found evidence of interactions between ADRA2A and ADRB1 genes among Caucasians and, and ADRA2A and ADRB2 genes among African Americans with obesity. These data support a complex and important role for adrenergic receptor variants in obesity and the accompanying metabolic alterations, and agree with a study by Ukkola et al. ⁵³ that interactions among ADRA2A and ADRB genes contribute to obesity related phenotypes. In the future, large studies should explore interactions for genes that regulate catecholamine-mediated effects on energy expenditure.

The epidemic of obesity in this country and world-wide has led to a marked increase in the incidence of metabolic syndrome, which is variably defined as a constellation of metabolic abnormalities including glucose intolerance (type 2 diabetes), insulin resistance, central obesity (BMI ≥ 30 kg/m²) dyslipidemia and hypertension (blood pressure ≥ 130/85 mm Hg) ⁵⁴. The mechanism(s) underlying the development of the metabolic syndrome is not clear, but is thought to be largely attributable to insulin resistance with an excessive flux of fatty acids and the involvement of the proinflammatory state ⁵⁵. Although none of the obese individuals in our study could be characterized as having type 2 diabetes or metabolic syndrome, some but not all had higher concentrations of glucose and insulin, insulin resistance as assessed by HOMA_IR, and elevated Hgb AIC and blood pressures, compared to non-obese individuals (Table 1, Figure 1). It is likely that many of these participants, if left untreated (i.e., life-style modifications) would go on to develop type 2 diabetes, metabolic syndrome and cardiovascular and other diseases. A few studies support a genetic basis for metabolic syndrome including the ADRB2 Gly16→Arg and the Gln27→Glu variants ⁵⁶ and others ⁵⁷. It is in this context that our findings that common SNPs in the ADRB and ADRA genes may associate with hyperinsulinemia, insulin resistance and elevated systolic blood pressures portend significant clinical relevance.

We have uncovered several intriguing associations between common SNPs in ADRB2A and ADRB genes with obesity and related alterations. However, the present study has several limitations. Many gene variants contribute to obesity and related alterations and those in ADRA2A and ADRB may have modest effects thus requiring large sample sizes to detect associations ⁵⁸. Our sample size was small, and it is possible that the associations we observed between ADRA2A and ADRB SNPs with obesity could represent false positives. Because of the potential for population stratification ⁵⁹, we stratified our population by self-identified ethnicity, which further exacerbates our problems with small numbers. In addition, our findings are further confounded by the fact that 73% and 80% of our Caucasian and African American population was female even though we adjusted for sex in our analyses. We did not correct for multiple hypothesis testing, which could also contribute to false positive associations ⁵⁸. We chose not to adjust for multiple comparisons because given the small numbers of participants; we reasoned that it is important not to dismiss differences that could be real. For these reasons, the results of our study should be interpreted with caution and should regarded as exploratory, and underscore the need for replication in larger populations.