Linkage of Essential Hypertension to the Angiotensinogen Locus in Mexican Americans
Abstract Essential hypertension has been linked to a highly polymorphic marker at the angiotensinogen locus, and association with a polymorphism in this locus has been found in some populations. We tested the hypothesis that these same polymorphic markers are linked to essential hypertension in Mexican Americans. The data comprised all the affected relative pairs in 46 extended families chosen at random from a low-income barrio in San Antonio. Specifically, we searched for linkage by testing for excessive marker alleles shared identical by descent (IBD) among hypertensive relative pairs. When women taking oral contraceptives or hormones were excluded, the affected relative pairs shared a significant excess of alleles IBD for the highly heterozygous GT repeat polymorphism (P=.038) and were marginally significant for the M235T variant (P=.079), which has a much lower heterozygosity (0.43 versus 0.85 for the GT repeat). We also assayed plasma levels of angiotensinogen and, using likelihood methods, found no significant association (P=.43) between plasma levels of angiotensinogen and M235T genotypes. These results support the linkage of essential hypertension to the angiotensinogen locus but do not indicate a specific role for the M235T variant.
Essential hypertension is a multifactorial disease that results from the interaction of genes and environmental factors. Due to the central role of the renin-angiotensin system in the regulation of blood pressure, variations in the genes that encode components of this system are candidates for the development of essential hypertension. Previous studies on sibpairs have reported that DNA polymorphisms at the AGT locus are linked to essential hypertension. Using affected-pair methods, three independent studies reported that AGT is linked to essential hypertension in three geographically distinct groups of whites and one group of blacks. Jeunemaitre et al1 linked hypertension to a GT dinucleotide repeat at the AGT locus in groups of hypertensive sibling pairs in Salt Lake City, Utah, and Paris, France. Caulfield et al2 7 confirmed the linkage of hypertension to the GT dinucleotide repeat polymorphism by using data on hypertensive relative pairs from the United Kingdom and hypertensive sibling pairs from the African Caribbean.
On sequencing the locus, Jeunemaitre et al1 found two variants in exon 2, a threonine-to-methionine substitution at amino acid 174 (T174M) and a methionine-to-threonine substitution at amino acid 235 (M235T), that showed statistically significant associations with hypertension. Several case-control studies3 4 5 of M235T in the Japanese population also reported an association; however, case-control studies of this polymorphism in white,2 African American,6 African Caribbean,7 and Finnish8 populations did not. Finally, Jeunemaitre et al1 showed that there were significant differences in plasma levels of AGT in subjects with different AGT genotypes.
In this study we confirmed the linkage of hypertension to the GT dinucleotide repeat by using data on Mexican American hypertensive relative pairs from a low-income barrio in San Antonio. However, using maximum-likelihood methods, we found no association of genotypes for M235T with plasma levels of AGT in this Mexican American population.
Probands in the San Antonio Family Heart Study (SAFHS) were 40- to 60-year-old men and women selected at random from a low-income Mexican American barrio in San Antonio, Texas. The only eligibility criterion apart from age was that the proband have a living spouse willing to participate in the study and at least six first-degree relatives, excluding parents, 16 years old or older and living in the San Antonio area. Thus, an individual was eligible if he/she had a total of six siblings and/or offspring, in any combination, living in San Antonio. The proband and all first-, second-, and third-degree relatives willing to participate were subjected to an extensive data-gathering protocol in which information on demographics, morphometrics, cigarette and alcohol consumption, dietary behavior, and physical activity were obtained. In all individuals, blood pressure was measured three times after a 5-minute seated rest period with a random zero sphygmomanometer. The single measure of blood pressure we used in this study was calculated by dropping the first reading and averaging the latter two. Hypertension was defined as systolic blood pressure greater than or equal to 140 mm Hg, diastolic blood pressure greater than or equal to 90 mm Hg, or taking high blood pressure medication. This study consisted of approximately 1200 individuals in 46 extended families. All individuals gave informed consent, and all protocols were approved by the Institutional Review Board of the University of Texas Health Science Center. Demographic characteristics of all individuals broken down by hypertension status are shown in Table 1⇓.
Dinucleotide Repeat at the AGT Locus
The AGT 3′ dinucleotide repeat9 was analyzed by polyacrylamide electrophoresis of PCR amplification products produced using the “U” primer set of Jeunemaitre et al.1 These primers amplify an approximately 167–base pair (±dinucleotide insertions/deletions) fragment of the human AGT gene. Reaction mixtures of 25 μL total volume contained approximately 250 ng of sample DNA, 10 mmol/L Tris-HCl (pH 8.3), 50 mmol/L KCl, 1.5 mmol/L MgCl2, 10% DMSO, 0.001% gelatin, 125 μmol/L (each) dNTPs, 0.25 μmol/L (each) unlabeled primers, ≈0.05 pmol 32P end-labeled (≈0.75 μCi/pmol) upstream primer, and 0.5 U Taq polymerase (Perkin-Elmer Cetus). Touchdown PCR was performed in microtiter plates with a GeneAmp PCR System 9600 thermocycler (Perkin-Elmer), according to the following protocol: an initial denaturing step of 4 minutes at 94°C was conducted, followed by five cycles comprising a 15-second denaturation step at 94°C, a 15-second annealing step (starting at 72°C for the initial cycle and decreasing by 2°C in each subsequent cycle for a final temperature of 64°C in the fifth cycle), and a 30-second elongation step at 72°C. These cycles were followed by 31 cycles of 15 seconds at 94°C, 15 seconds at 62°C, 30 seconds at 72°C, and a final 5-minute elongation step at 72°C before cooling to 4°C. Subsequent to PCR amplification, 8 μL of denaturing/loading dye solution (98% deionized formamide, 10 mmol/L EDTA, 0.05% xylene cyanole, and 0.05% bromphenol blue, in water) was added to each reaction well. These mixtures were heated and held at 95°C for 5 minutes, then cooled to 4°C. The cooled reaction mixtures (6 μL of each) were applied to a well of denaturing polyacrylamide gel (8% acrylamide [1:19 bis-acrylamide], 0.09 mol/L Tris, 0.09 mol/L boric acid, 0.02 mol/L EDTA, 7 mol/L urea) and separated at 1250 V (constant) for 5 hours. After electrophoresis, gels were dried and exposed to Kodak XAR film (with one intensifying screen) for 2 days to produce autoradiographic records.
M235T and T174M Variants of the AGT Locus
The M235T polymorphism was typed by PCR amplification of lymphocyte DNA with a mismatched reverse primer and cleavage with Tth111I as described by Russ et al.10 The T174M polymorphism was typed by PCR amplification of AGT exon 2 sequences as previously described by Hixson and Powers11 and cleavage with Nco I. The Met allele contains an Nco I site (CC ATG G), and the Thr allele lacks the site (CC ACG G).
Genotyping of the M235T and T174M variants was performed earlier in the process of gathering individuals for the SAFHS than genotyping of the GT dinucleotide repeat; therefore, the total is smaller. Also, we stopped genotyping T174M early, when it became obvious that it was not informative. Heterozygosity and allele frequencies for each of the three polymorphisms are reported in Table 2⇓.
Assays of Plasma Levels of AGT
AGT levels in plasma were measured using a two-step procedure consisting of conversion of AGT to angiotensin I by adding primate renin in excess followed by measurement of angiotensin I by radioimmunoassay.12 The incubation buffer for the first step (0.6 mol/L sodium phosphate, pH 6.0, 36 mmol/L EDTA, 1 mg/mL gelatin) was boiled and then cooled to room temperature to eliminate protease activity. Baboon renin was prepared by method A as described by Haas et al.13 The conversion of AGT to angiotensin I was accomplished by incubating 0.25 mL of buffer with 20 μL of the renin preparation, 5 to 10 μL plasma, and water added to 1 mL for 3 hours at 37°C. Each batch of baboon renin routinely released all of the angiotensin I in 5 μL of human plasma within 1.5 hours under these incubation conditions. Aliquots were then assayed for angiotensin I generated by a commercially available radioimmunoassay kit (RIANEN, Du Pont). Blank controls containing renin but no added plasma, and a control plasma pool were included within each set of assays. This procedure was performed three times on 519 individuals in 46 extended families. If the coefficient of variation of the three readings was greater than 0.10 for any individual, then the entire assay was repeated for that individual. The plasma AGT level reported was the average of the three readings.
We performed nonparametric linkage analyses by using data on affected pairs of relatives only, thereby avoiding problems of incomplete penetrance and variable age of onset. The statistical test used, known as simIBD, was based on all affected relative pairs (for example, parent-offspring, uncle-niece) and was thus ideal for our large extended families. Linkage methods based on IBD sharing were recently shown14 to be more powerful than the methods based on identity by state used by Jeunemaitre et al1 and Caulfield et al.2 Using strictly sibling pair tests would have severely reduced our sample size, making the results unreliable.
The program simIBD computes a similarity statistic that is a weighted sum of the IBD probabilities. The significance of this statistic is determined empirically by using a null distribution that is calculated using simulation methods.14 15 16 17 This null distribution is simulated by first assuming that the affected individuals are untyped at the marker locus and then randomly generating genotypes at the marker locus for the affected individuals conditional on the genotypes of the unaffected individuals. The similarity statistic is then computed by using the generated random genotypes. This process is repeated 1000 times, and it is these 1000 simulated similarity statistics that form the empirical null distribution.
Jeunemaitre et al1 suggested on the basis of a case-control study that the M235T polymorphism is associated with essential hypertension. They also performed an analysis of variance, which showed that the different M235T genotypes were associated with different plasma levels of AGT, a quantitative trait. A case-control test of association between M235T genotypes and hypertension was invalid in our dataset owing to the inherent familial correlations; however, a test of association between plasma levels of AGT and M235T genotypes, using likelihood methods that incorporate the effects of residual familial correlation, was possible. We proceeded as follows. After excluding all individuals who were taking hypertensive medication, oral contraceptives, or other hormones (which reduced the sample size to 448), we initially used likelihood methods to estimate any significant effects of polygenes (ie, heritability) and environmental factors. These factors included BMI, physical activity, cigarette and alcohol consumption, and dietary consumption of protein, fat, carbohydrates, fiber, cholesterol, and sucrose. Incorporating heritability in the underlying model accounts for the nonindependence of individuals in the pedigrees. We then performed measured genotype analysis18 to determine whether genotypes of M235T were associated with different levels of AGT in plasma. Any significant covariates from the initial analysis were included in the measured genotype analysis. The GT dinucleotide repeat was not tested owing to the large number of genotypes. The T174M variant was considered uninformative owing to low heterozygosity. We computed the maximum likelihood of two models; the general model assumed that AGT levels were associated with M235T genotypes (ie, the three genotypic means differ) and the subset model assumed that AGT levels were not associated with M235T genotypes (ie, the three genotypic means were equal). Twice the difference in the two maximum likelihoods was assumed to be distributed as a χ2 distribution with 2 df. This procedure was carried out using the computer program package PAP.19
Results of the relative pairs tests for linkage are shown in Table 3⇓. These tests indicated a significant (P=.019) excess of marker alleles shared IBD for the GT dinucleotide repeat but not for the M235T variant (P=.162). Oral contraceptives and other hormones are known to affect blood pressure, so we repeated the analysis on a reduced dataset in which women using such medication were excluded. Tests on this reduced dataset showed that the GT dinucleotide repeat remained significant (P=.038) and the M235T variant became marginally significant (P=.079). The molecular variant T174M was also tested for linkage, but it had low power to detect linkage owing to low heterozygosity (see Table 2⇑). We also considered analyzing sibling pair tests, but the sample size was very small; in the reduced dataset there were 40 or fewer sibling pairs. Thus, results of sibling pair analyses are not reliable.
Results of the measured genotype test of association are shown in Table 4⇓. Of the factors initially tested for their effects on plasma levels of AGT, only BMI was significant (P<.0001). BMI was therefore included in the measured genotype analysis, along with sex, linear and quadratic age effects (which are known to affect blood pressure), and heritability. The measured genotype analysis showed that genotypes at M235T were not significantly associated with plasma levels of AGT (P=.43). A measured genotype analysis of the effect of M235T on blood pressure (both systolic and diastolic), with treated hypertensives excluded, was also not significant (results not shown).
Identification of genes that contribute to the development of hypertension would facilitate the identification of individuals at risk for this common disease. One approach by which to detect and locate genes that affect complex diseases such as hypertension is linkage analysis. This approach has become increasingly popular in the past decade given the development of methods that allow for the detection and subsequent rapid typing of highly polymorphic DNA markers and sophisticated statistical genetic methodologies. Recently, using affected sibpair and relative pair linkage analysis methods, two groups of investigators1 2 7 have reported linkage between a highly polymorphic GT dinucleotide repeat near the AGT locus and essential hypertension in three white populations and one black population.
In this study of extended Mexican American families, we confirmed the linkage between the highly polymorphic GT repeat and essential hypertension. At best, we found only marginal linkage between M235T and hypertension. Logically, it has to be linked if the GT repeat is linked. This seemingly paradoxical result can be explained by the low heterozygosity (0.43) of the M235T polymorphism and the smaller number of individuals that were typed for this polymorphism. These two factors would reduce the power to detect linkage to M235T. We note, however, that when individuals taking hormones were removed from the dataset, the test for linkage to M235T reaches marginal significance. This result would be expected if the individuals taking hormones represent hypertension “phenocopies” whose presence would reduce the power to detect true linkage.
It should also be emphasized that this study was conducted using a population-based random sample. It therefore had less power to detect linkage to hypertension than a sample of similar size ascertained on hypertension. Conversely, inferences to the general population based on this random sample would be valid, whereas general inferences from an ascertained sample would be more problematic.
Jeunemaitre et al1 also reported that the linkage with the AGT locus was stronger among individuals with more severe hypertension and among males. However, Caulfield et al2 did not find any significant differences between sexes or with increasing severity of hypertension after stratifying their families according to sex and severity of hypertension. We have large extended families with approximately five affected members per family and are therefore unable to stratify our families by sex or severity and can provide no information about their effect on linkage of hypertension to the AGT locus.
Our failure to find an association between M235T and plasma levels of AGT may be due to the exclusion of hypertensives on medication. These individuals could be expected to have a higher frequency of the T allele, and excluding them might lower the overall frequency of T. However, a comparison of the frequency of the T allele in Tables 2⇑ and 4⇑ shows that it did not decrease significantly, indicating that the exclusion does not bias the measured genotype analysis.
The lack of an association between M235T and AGT levels in this study and between M235T and essential hypertension in the report by Caulfield et al2 does not detract from the finding of linkage. Genetic association studies require that the polymorphism under study be either the genetic lesion that causes a defect or in strong linkage disequilibrium with the genetic lesion. In the latter case, the results of association studies would be population specific. However, the absence of an association between a specific polymorphism and a disease in different populations would indicate that this specific polymorphism is not the critical genetic lesion.
Linkage of hypertension to the AGT locus has now been shown in three white populations, one black population, and one Mexican American population. These results strongly reinforce the hypothesis that AGT, or another closely linked locus, affects essential hypertension. However, the exact location and nature of the genetic lesion that contributes to hypertension is still unknown.
Selected Abbreviations and Acronyms
|BMI||=||body mass index|
|IBD||=||identical by descent|
|PCR||=||polymerase chain reaction|
This work was supported by National Institutes of Health grants HL54707 and HL45522. The authors thank the Executive Committee of the San Antonio Family Heart Study (HL45522) for the use of stored plasma and the extensive database on participants in their study. The authors also thank Allen Ford, Desiree Villarreal, and Nicole Stowell for their technical assistance.
Reprint requests to Larry D. Atwood, Division of Epidemiology, School of Public Health, University of Minnesota, 1300 S Second St, Suite 300, Minneapolis, MN 55454-1015.
- Received September 18, 1996.
- Revision received October 11, 1996.
- Accepted February 18, 1997.
Iwai N, Shimoike H, Ohmichi N, Kinoshita M. Angiotensinogen gene and blood pressure in the Japanese population. Hypertension. 1995;25:688-693.
Rotimi C, Morrison L, Cooper R, Oyejide C, Effiong E, Ladipo M, Osotemihen B, Ward R. Angiotensinogen gene in human hypertension: lack of an association of the 235T among African Americans. Hypertension. 1994;24:591-594.
Caulfield M, Lavender P, Newell-Price J, Farrall M, Kamdar S, Daniel H, Lawson M, De Freitas P, Fogarty P, Clark AJL. Linkage of the angiotensinogen gene locus to human essential hypertension in African Caribbeans. J Clin Invest. 1995;95:687-692.
Kiema T, Kauma H, Rantala AO, Lilja M, Reunanen A, Kesäniemi YA, Savolainen MJ. Variation at the angiotensinogen-converting enzyme gene and angiotensinogen gene loci in relation to blood pressure. Hypertension. 1996;28:1070-1075.
Kotelevtsev YV, Clauser E, Corvol P, Soubrier F. Dinucleotide repeat polymorphism in the human angiotensinogen gene. Nucleic Acids Res. 1991;19:6978.
Russ AP, Maerz W, Ruzicka V, Stein U, Grob W. Rapid detection of the hypertension-associated Met235 → Thr allele of the human angiotensinogen gene. Hum Mol Genet. 1993;2:609-610.
Tewksbury DA, Dart RA. High molecular weight angiotensinogen levels in hypertensive pregnant women. Hypertension. 1982;4:729-734.
Haas E, Goldblatt H, Gipson EC, Lewis L. Extraction, purification, and assay of human renin free of angiotensinase. Circ Res. 1966;19:739-749.
Ott J. Computer-simulation methods in human linkage analysis. Proc Natl Acad Sci U S A. 1989;86:4175-4178.
Weeks DE, Ott J, Lathrop GM. SLINK: a general simulation program for linkage analysis. Am J Hum Genet. 1990;47:A204. Abstract.
Hasstedt SJ. Pedigree Analysis Package, Revision 3.0. Salt Lake City, Utah: Department of Human Genetics, University of Utah; 1989.