Nucleotide diversity and haplotype structure of the human angiotensinogen gene in two populations

Abstract

Variation in the angiotensinogen gene, AGT, has been associated with variation in plasma angiotensinogen levels. In addition, the T235M polymorphism in the AGT product is associated with an increased risk of essential hypertension in multiple populations, making AGT a good example of a quantitative-trait locus underlying susceptibility to a common disease. To better understand genetic variation in AGT, we sequenced a 14.4-kb genomic region spanning the entire AGT and identified 44 single-nucleotide polymorphisms (SNPs). Forty-two SNPs were observed both in 88 white and in 77 Japanese unselected subjects. Six major haplotypes accounted for most of the variation in this region, indicating less allelic complexity than in many other genomic regions. Although the two populations were found to share all of the major AGT haplotypes, there were substantial differences in haplotype frequencies. Pairwise linkage disequilibrium (LD), measured by the D', r(2), and d(2) statistics, demonstrated a general pattern of decline with increasing distance, but, as expected in a small genomic region, individual LD values were highly variable. LD between T235M and each of the other 39 SNPs was assessed in order to model the usefulness of LD to detect a disease-associated mutation. Among the Japanese subjects, 13 (33%) of the SNPs had r(2) values >0.1, whereas this statistic was substantially higher for the white subjects (occurring in 35/39 [90%]). LD between a hypertension-associated promoter mutation, A-6G, and 39 SNPs was also measured. Similar results were obtained, with 33% of the SNPs showing r(2)>0.1 in the Japanese subjects and 92% of the SNPs showing r(2)>0.1 in the white subjects. This difference, which occurs despite an overall similarity in LD patterns in the two populations, reflects a much higher frequency of the M235-associated haplotype in the white sample. These results have important implications for the usefulness of LD approaches in the mapping of genes underlying susceptibility to complex diseases.

Figure 1

Schematic diagram of AGT, showing locations of five exons (A), repeat elements (B), and 44 SNPs (C). The complete genome sequence containing the entire AGT, 14.4 kb (10.1% coding sequence), was determined. The exact sizes of intron 1, 2, 3, and 4 are 3,233, 3,794, 1,595, and 863 bp, respectively. Repetitive elements (SINE, LINE, and LTR) and simple repeat elements were analyzed by RepeatMasker (see the RepeatMasker Documentation web site). The location of the dinucleotide-repeat sequence is shown.

Figure 2

Pairwise LD between T235M and other SNPs in AGT, evaluated by either D′ (A) or r² (B), in whites and in Japanese. D′ is expressed as an absolute value.

Figure 3

Pairwise LD versus physical distance between all pairwise SNPs, based on the 861 marker pairs in Japanese, evaluated by either D′ (A) or r² (B). Average values of D′ and r² (B) at every 500 bp, in whites and in Japanese, show that LD declines with increasing physical distance between SNP pairs.

Figure 4

Pairwise LD in AGT, evaluated by r². LD between all pairs of SNPs (SNP_i and SNP_j, where i and j refer to the SNP number shown in table 2) was evaluated by the LD measure r². Pairwise LD was determined among the 861 marker pairs studied in whites (A) and Japanese (B), and pairs in LD (r²=0.5) are shown as blackened boxes. Several subgroups of SNPs in tight LD with each other are shown below. A dot in the center of a square indicates that no data are available, because SNP24 and SNP27 were not observed in whites.

Figure 5

AGT haplotypes in whites and Japanese. These haplotypes were constructed and the frequencies were estimated by the EM algorithm based on 21 SNPs in AGT. The number of chromosomes analyzed was 176 for whites and 154 for Japanese. Blackened boxes denote the minor allele in Japanese. The chimpanzee sequence is also shown.

Figure 6

Plot of DSS (Y-axis)—that is, the difference, in the sum of squares, between trees generated from two halves of a 1,500-bp sliding window of DNA sequence—versus the position of the center of each sliding window (X-axis). Gaps in the sequence represent those portions of the sequence in which no polymorphic variation was present.

Figure 7

Haplotype trees for AGT haplotype, based on 21 SNPs and the chimpanzee sequence. The sizes of the circles represent the frequencies of the haplotypes in whites (A) and Japanese (B).

Figure 8

Relationships between four major SNP haplotypes and the microsatellite marker. The distribution of the frequency of individual microsatellite alleles is shown for each of the common SNP haplotypes in AGT. Al though the distribution of the CA-repeat alleles (A) in whites is very different from that in Japanese, each SNP haplotype is associated with a specific CA-repeat allele in whites (B) and Japanese (C).