INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

PHYSIOLOGICAL VARIATION of blood pressure (BP) and body mass index (BMI) is important in relation to individual cardiovascular risk and the population prevalence of stroke and heart attack. The close relationship between BP and BMI (29) is one of the most consistently observed associations between cardiovascular risk factors seen within healthy populations as well as in certain disease states such as Cushing's syndrome. This association may indicate common underlying determinants of BP and BMI or a physiological cause-and-effect relationship between the two phenotypes.

Common genetic factors may determine both BP and BMI. The normal familial aggregation of BMI is consistent with genetic effects, and a high degree of heritability of BMI is observed in both young and adult subjects (5, 14). Familial aggregation of BP also follows a pattern consistent with significant but complex genetic determination (2). Moreover, multivariate biometric analyses indicate that a substantial proportion of the familial patterns of BP can be explained by covariation with body weight, suggesting a common genetic link between the two phenotypes (12).

Molecular approaches to BP have recently defined several genetic mechanisms causing rare Mendelian hypertensive diseases (23). However, these genetic mutations have not been proven to be relevant to normal physiological variation in BP or essential hypertension. Although a number of candidate genes have been proposed, the molecular genetics of normal variation in BMI has not been defined (6).

At a physiological level, the glucocortioids and the sympathetic nervous system are two important determinants of BP and BMI. Therefore, the relevant genes make good candidates for genetic analysis. Chromosome 5q31-5q32 is of particular interest in this regard because of the very close juxtaposition of two relevant genes, the GRL gene that encodes the glucocorticoid receptor (GR) and the gene ADRB2 that encodes the ₂-adrenergic receptor. These two genes are <1 cM apart, providing potential for interaction between glucocorticoids and the sympathetic nervous system at a molecular level as well as a physiological level.

Glucocorticoids play important roles in regulating carbohydrate, protein, and lipid metabolism and have critical actions on the cardiovascular and central nervous systems. Cortisol influences the responsiveness of the blood vessels to cardiovascular control systems, including catecholamines and angiotensin. High levels of cortisol (36) or increased sensitivity to cortisol (34, 35) have been associated with high blood pressure, and cortisol excess leads to hypertension (37). The effects of cortisol on lipid metabolism are complex but can lead to an increase in BMI through effects on appetite, the formation of adipocytes from precursor cells, and stimulation of lipogenesis (7, 13).

The actions of glucocorticoids depend on binding with the cytoplasmic GR. The resultant steroid hormone-receptor complex enters the nucleus where it binds to specific DNA sequences and acts as an important transciptional regulator of genes. The GR is one of the steroid hormone-receptor superfamily that includes receptors for mineralocorticoids, sex steroids, thyroxine, and vitamin D. The human GRL gene has been localized to chromosome 5q31 (32), and mutations in the gene have been shown to cause insensitivity to glucocorticoids (16, 25). Other studies have shown genetic variation in or around the GRL associated with less extreme effects on GR sensitivity (15, 27). The mechanism of insensitivity under these circumstances is unknown but may include the formation of alternative splicing of GRL pre-RNA (3). Some of the GRL DNA variants have been associated with high blood pressure (36), increased BMI (15), and increased abdominal visceral fat (9). These observations make GRL an important candidate for determination of both BP and BMI.

The sympathetic nervous system also exerts an important influence on cardiac and vascular function and metabolic processes including lipid metabolism. The sympathetic nervous system has been implicated in essential hypertension and energy expenditure (4, 11), and the response patterns of the sympathetic nervous system appear to be determined genetically (26). Therefore, genes that have some effects on the actions of catecholamines are candidate genes for BP and BMI.

The ₂-adrenergic receptor is a seven-transmembrane G protein-coupled receptor found in vascular and adipose tissues. Stimulation of this receptor results in vasodilation (10, 17) and promotes lipolysis in human adipose tissue (4, 24). The ADRB2 gene on chromosome 5q31 has been implicated in both BP and BMI. In subjects whose BP is sensitive to salt, ADRB2 expression is reduced in isolated cells (20), and impairment of isoproterenol-mediated vasodilation has been observed in black subjects predisposed to high blood pressure (21). Genetic variants of the ADRB2 gene have been associated with hypertension in Caucasian and African populations (19, 30, 31). ADRB2 has also been associated with the predisposition to hypertension in young normotensive subjects (33). In addition, genetic variants at the ADRB2 locus have been associated with obesity and increased receptor sensitivity in women (22).

The GRL and ADRB2 genes are in close proximity on human chromosome 5q31-5q32, and both are important candidates for the related phenotypes of BP and BMI. To date no genetic linkage study has been undertaken in relation to these loci. The aim of this study was to perform a linkage analysis of chromosome 5q31-5q32 in relation to the physiological variation in BP, BMI, and related phenotypes. Beyond fundamental physiological importance, evidence of linkage would be relevant to hypertension and obesity as well as the broad contribution of BP and BMI to cardiovascular disease.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Subjects and phenotype measurement. Subjects were drawn from the Victorian Family Heart Study (VFHS), which was established to examine familial patterns of cardiovascular risk factors in the general population. These studies were approved by the Ethics Review Committee of the Alfred Hospital, Melbourne, Australia, and informed consent was obtained from all participants.

Eligible families were Caucasian with both parents (ages 40 and 70 yr) and at least one natural offspring (ages 18 and 30 yr) available to participate. The lower age for offspring was set to minimize the potential confounding effect of growth on the phenotypes under study. Recruitment was limited to Caucasian families to reduce the possible confounding genetic differences that were determined racially. A family history of heart disease was not relevant to recruitment, the aim being to enroll a relatively representative sample of families exhibiting a broad cross section of cardiovascular risk factor levels.

Potential participating families were identified through a variety of community-based sources. These included the Australian NHRMC Twin Registry, the Melbourne Collaborative Cohort Study (Health 2000), general practitioners, and work sites (Common Scientific and Industrial Research Organisation). We asked a total of 8,060 individuals by letter or telephone to indicate whether their families would be willing to participate in the VFHS. We received 2,946 responses and excluded 1,928 families because they were ineligible (n = 1,108 responses: key family member unavailable, no natural children, ethnic origin) or declined (n = 820 responses). Recruitment was undertaken between the years 1991 and 1996.

The present study sample comprised a sample of 264 families from the VFHS, selected at random from families that had two participating siblings. Families that included monozygotic twins or same-sex twins of uncertain zygosity were excluded from these analyses.

At our research clinics, trained research nurses measured cardiovascular risk factors according to standardized measurement techniques. Height of subjects without shoes on was measured to the nearest 0.5 cm using a wall-mounted ruler. The weight of subjects without shoes on or items of heavy clothing worn was measured to the nearest 0.5 kg with scales that were calibrated regularly. Blood pressure was measured in the right arm using a standard mercury sphygmomanometer according to recommended guidelines (28). The cuff size was chosen according to the arm circumference. Systolic blood pressure (SBP) was taken as the return of arterial sounds (Korotkoff phase I), and diastolic blood pressure (DBP) was taken as the disappearance of sounds (Korotkoff phase V). Blood pressure measurements were made to the nearest 2 mmHg. Three measurements of SBP and DBP were taken with the subject lying down and then repeated after the subject had been standing for 2 min. The first blood pressure measurements were discarded, and the last two measurements of systolic and diastolic pressures in both the supine and standing positions were averaged to calculate the SBP and DBP, respectively, used in this analysis. Pulse pressure (PP) was calculated as SBP  DBP, and mean arterial pressure was calculated as DBP + PP/3. Pulse rate was the average of measurements made over 1 min in both the supine and standing positions. After phenotypic measurements, 10 ml of venous blood was collected into EDTA-coated tubes for DNA extraction and analysis.

Genotyping. Genomic DNA from parents and offspring was extracted by standard methods. According to available maps (http://cedar.genetics.soton.ac.uk/pub/chrom5/map.html), the GRL and the ADRB2 genes map to chromosomal region 5q31-5q32. We therefore selected the following polymorphic microsatellite markers from this region: D5S207 (5q31.3-5q33.3, 6 alleles detected, heterozygosity = 0.65), D5S210 (5q31.3-5q33.3, 13 alleles detected, heterozygosity = 0.79), D5S519 (5q32-5q33, 11 alleles detected, heterozygosity = 0.82), and D5S119 (5q31.3-5q33.3, 10 alleles detected, heterozygosity = 0.70). We designed PCR primers using primer sequence information in GDB (http://gdbwww.gdb.org/). PCR amplification of human genome was performed as follows. Each reaction mix contained 1-10 ng DNA, 0.9× GeneAmpPCR Buffer II (Perkin Elmer), 0.35 µM of each primer, 0.17 unit of AmpliTaq Gold DNA polymerase (Perkin Elmer), 0.25 mM of each deoxynucleotide triphosphate (dNTP), and 2.5 mM MgCl₂. The forward primer was 5'-fluorescently end-labeled with 6-carboxyfluorescein (6-FAM), tetrachlorinated analog of FAM (TET), or hexachlorinated analog of FAM (HEX). The reverse primer was modified with a GTTTCTT tail on the 5' end, which facilitated adenylation of the 3' end of the forward strand (8). The volume of each amplification was 5 µl. The cycling reaction was performed on an ABI 877 Integrated Thermal Cycler (Perkin Elmer). After being held at 95°C for 10 min (for activation of the AmpliTaq Gold enzyme), DNA underwent 10 repetitions of 95°C for 15 s, 55°C for 30 s, and 72°C for 60 s, followed by 20 repetitions of 89°C for 15 s, 55°C for 30 s, and 72°C for 60 s. A final extension was allowed for 10 min at 72°C. PCR products amplified by the five markers were pooled for each individual DNA sample. Alleles were visualized using an ABI 377 Sequencer (Perkin Elmer). Allele sizing was determined using GeneScan Analysis Version 2.1 and Genotyper Version 2.0 (Perkin Elmer). Not all markers amplified successfully in all members of all families. The number of complete families providing results is shown for each marker in RESULTS.

Linkage analysis. Summary data are given as median and interquartile range (IQR) unless stated otherwise. Statistical linkage analysis was performed using the Genetic Analysis System (GAS) package version 2.0 (Alan Young, Oxford University, 1993-1995). This linkage analysis makes use of quantitative measures of phenotypes in identifying important genetic loci. For a locus that influences a particular phenotype, the quantitative difference of that phenotype between siblings will be less when the siblings are genetically identical at that locus. For each genetic locus, GAS uses both parametric (Haseman-Elston algorithm) and nonparametric (Mann-Witney U test) statistical methods to test the relationship between sibling pair differences in phenotype versus similarity in genotype. Genetic similarity between siblings was measured by the number of alleles shared at a particular locus (0, 1, or 2). Allele sharing was assessed identity-by-descent (IBD), making use of the genotypes of the parents. Associations between BMI and BP were assessed with nonparametric Spearman correlation coefficients. It is not possible to specify accurately the statistical power of this study. However, with reference to power estimates for quantitative sibling pair analyses (1) based on the heritability estimates for mean arterial pressure and BMI derived from the entire VFHS sample (data not shown), our sample size is sufficient to achieve 90% power at a significance level of 5% for BMI and approximates a power of 80% for arterial pressure.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The characteristics of our study families are summarized in Table 1. These values were representative of the larger population sample from which these families were derived. Figure 1 shows the frequency distribution of BMI and mean arterial pressure in the siblings. These cover a broad range of values and display unimodal distributions. The distribution of BMI was skewed to the upper values, whereas that of mean arterial pressure was relatively symmetrical. Figure 2 shows the association between BMI and mean arterial pressure in the siblings. The Spearman correlation coefficient (r = 0.35) was significant statistically (P < 0.0001). A significant, but slightly less close correlation (r = 0.21, P < 0.0001) was observed in parents.

View larger version (20K): [in this window] [in a new window]   Fig. 1.   Frequency distribution (n = no. of individuals) of body mass index (kg/m2) and mean arterial pressure (diastolic + one-third of pulse pressure, mmHg) in siblings in this study.

View larger version (21K): [in this window] [in a new window]   Fig. 2.   Scatter diagram of body mass index (kg/m2) against mean arterial pressure (mmHg) in siblings in this study.

The IBD sibling pair analysis revealed no significant linkage among SBP, DBP, or BMI and any of the microsatellite markers. Regression analysis indicated no significant correlation between the number of allele-shared IBD and the difference in SBP, DBP, and BMI between sibling pairs (Table 2). The nonparametric analyses showed that the phenotypes were not related to genotypes at any of the four markers (Table 3).

Absence of genetic linkage between chromosome 5q31-5q33.3 was also observed for the related variables: mean arterial pressure, pulse pressure, pulse rate, body weight, and height (data not shown).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

This study represents the first reported linkage study of GRL and ADRB2 with BP and BMI. Previous molecular analyses have been simple association studies that do not test inheritance. Our study is also unique in that instead of focusing on extreme phenotypes, we have examined genetic linkage with physiological variation in BP and BMI. On the basis of linkage analysis with four highly polymorphic markers in a large number of families from a Caucasian population, we were unable to detect any influence of chromosome 5q31-5q33.3 on SBP, DBP, or BMI. We also could not find any effect on the physiological variation of the related phenotypes body weight, height, mean arterial pressure, pulse pressure, or pulse rate. The advantage of the linkage approach over association analyses is that it provides a comprehensive analysis of phenotypic effects of a chromosomal region and is not restricted to a few specific markers. Furthermore, quantitative linkage tests are more informative and more powerful because they make use of measured data rather than simple dimorphic categorization (e.g., normotension vs. hypertension or normal vs. obesity), and the IBD strategies make use of parental genotypes. From these results we can be reasonably confident of excluding any significant role of the GRL and ADRB2 genes on interindividual differences in BP and BMI in our population. However, linkage studies do not necessarily exclude the possibility that particular polymorphisms of the GRL and ADRB2 genes interact when in certain combinations to influence the phenotypes. To investigate this possibility, it would be necessary to evaluate the polymorphic spectrum of the two genes and associate components of this spectrum with phenotypic variation in BP and BMI.

As summarized in the introduction, there was good justification from both physiological and pathological perspectives to consider both GRL and ADRB2 as potentially important candidates for variation in BP and BMI. Both the GR and the ₂-adrenergic receptor play critical roles in the cardiovascular and metabolic actions of steroids and catecholamines. Furthermore, functional DNA mutations and variants have been described that have impact on BP and BMI. Some of these have been found in rare familial syndromes (16, 25) that are not likely to be relevant directly to physiological variation of the phenotypes in the general population. However, other studies have compared the genotypes and phenotypes in more representative samples (9, 15, 19, 22, 30, 31, 33). In relation to blood pressure, these studies have shown abnormalities in levels of cortisol, in the sensitivity to glucocortioids, and in the distribution of molecular variants of GRL (15, 27, 35, 36). However, not all results have been consistent. Some reports have not detected association between GRL polymorphism and differences in GR sensitivity (18). Different studies have reported association between GRL variants and abdominal visceral fat but not BMI (9) or association with BMI but not blood pressure (15).

There are a number of possible explanations for the discrepancy between studies. In the first instance, these might relate to sampling of populations. We cannot extrapolate our findings with any certainty to other geographical or racial groups. Other environmental circumstances or genetic background may expose the effects of GRL or ADRB2 variants that are not evident in our population. It might also be that different variants of the same genes might be informative in relation to particular physiological characteristics. However, our linkage analysis is a general test of the involvement of these genes and is not restricted to any specific DNA mutation or variant.

Perhaps differences in sampling within populations are important. It might be that the effects of certain GRL or ADRB2 variants are only obvious in those with extreme phenotypes, i.e., hypertension or obesity. Such an explanation assumes that other predisposing genetic factors or pathophysiological complications must exist before GRL or ADRB2 variants exert any actions. In such a case, the relevance of such variants is limited to a very small part of the population. It also would imply that the physiological consequences of such variants are not sufficiently robust to emerge alone. In essence this is what we have tested and we could find no evidence of such a robust effect. Our results do not negate the importance of glucocorticoids and catecholamines in the physiological determination of BP or BMI. They do, however, suggest that DNA variation in or around the GRL or ADRB2 genes does not play an important role in this process.