Apolipoprotein E variation at the sequence haplotype level: implications for the origin and maintenance of a major human polymorphism

Abstract

Three common protein isoforms of apolipoprotein E (apoE), encoded by the epsilon2, epsilon3, and epsilon4 alleles of the APOE gene, differ in their association with cardiovascular and Alzheimer's disease risk. To gain a better understanding of the genetic variation underlying this important polymorphism, we identified sequence haplotype variation in 5.5 kb of genomic DNA encompassing the whole of the APOE locus and adjoining flanking regions in 96 individuals from four populations: blacks from Jackson, MS (n=48 chromosomes), Mayans from Campeche, Mexico (n=48), Finns from North Karelia, Finland (n=48), and non-Hispanic whites from Rochester, MN (n=48). In the region sequenced, 23 sites varied (21 single nucleotide polymorphisms, or SNPs, 1 diallelic indel, and 1 multiallelic indel). The 22 diallelic sites defined 31 distinct haplotypes in the sample. The estimate of nucleotide diversity (site-specific heterozygosity) for the locus was 0.0005+/-0.0003. Sequence analysis of the chimpanzee APOE gene showed that it was most closely related to human epsilon4-type haplotypes, differing from the human consensus sequence at 67 synonymous (54 substitutions and 13 indels) and 9 nonsynonymous fixed positions. The evolutionary history of allelic divergence within humans was inferred from the pattern of haplotype relationships. This analysis suggests that haplotypes defining the epsilon3 and epsilon2 alleles are derived from the ancestral epsilon4s and that the epsilon3 group of haplotypes have increased in frequency, relative to epsilon4s, in the past 200,000 years. Substantial heterogeneity exists within all three classes of sequence haplotypes, and there are important interpopulation differences in the sequence variation underlying the protein isoforms that may be relevant to interpreting conflicting reports of phenotypic associations with variation in the common protein isoforms.

Figure 1

Locations of polymorphic variants in and around the APOE gene. The genomic location of the 23 DNA variants, identified by sequencing 5.5 kb in 96 individuals, is shown, below the exon-intron structure of the APOE gene. An asterisk (*) marks the new variant identified in the Mayan sample from Campeche at position 3701, and x’s show the population distribution of the observed polymorphisms (J = Jackson, C = Campeche, N = North Karelia, and R = Rochester). Variants that result in amino acid substitutions are boxed.

Figure 2

RM network of APOE sequence haplotypes in the pooled total sample. Mutational relationships are indicated by lines linking the 31 unique haplotypes, indicated in the network as circles (lighter lines indicate inferred mutational relationships that are less likely if homoplasy caused by recombination or recurrent mutation involving site 560 is assumed). The size of each circle is proportional to the relative frequency of the haplotype in the total sample. Mutational differences between haplotypes are indicated on the branches of the network (variants and haplotypes are numbered as in table 1; mutations that result in amino acid substitutions are boxed). Haplotype 12 (shaded) is the root haplotype in the network. The three large boxes indicate the groups of haplotypes that define the ε2, ε3, and ε4 APOE alleles.

Figure 3

RM networks of APOE haplotypes in the four population samples. Networks describing haplotype relationships, as they are found in each individual sample, are shown in black. Other branches, which are present in the total network but missing in a particular sample, are indicated in light gray. Numbers label haplotypes only (mutations are not indicated). Boxes indicate groups of haplotypes that define the ε2, ε3, and ε4 alleles, as presented in figure 2.

Figure 4

Scaled coalescent gene tree of IS-compatible APOE variation in the total (pooled) sample. The tree shows the inferred genealogical history of the 16 haplotypes compatible with the IS model (indicated by the numbered labels marking each terminal branch at the bottom of the tree; haplotype numbering as in table 1). The mean estimated ages of variants in the tree, indicated as numbered points within the genealogy, were derived by the program Genetree. For branches with multiple mutations, the order of mutations in time is arbitrary. Horizontal lines are drawn underneath the groups of haplotypes that define the ε2, ε3, and ε4 APOE alleles, respectively. Mutations that result in amino acid substitutions are boxed.

Figure 5

Sliding-window analysis of the F test statistic (Fu and Li 1993). Estimates of F were calculated for overlapping 750-bp-wide windows placed at 25-bp steps along the 5.5-kb sequenced region. The value for each window was plotted at the midpoint of the window and these points were linked by a continuous line for ease of inspection. Sliding window plots for the pooled sample and each of the separate population samples are shown (J = Jackson, C = Campeche, N = North Karelia, R = Rochester). The location of the APOE exons within the sequenced region are shown directly beneath the plot for the combined sample.

Figure 6

Sliding-window analysis of APOE polymorphism (π) and interspecific divergence (K). Estimates of average pairwise sequence difference, π, among the human haplotypes and net sequence divergence, K, between the haplotypes and the chimpanzee APOE sequence, were calculated for overlapping 750-bp-wide windows placed at 25-bp steps along the 5.5-kb sequenced region. π is plotted as a light line (and with respect to the scale on the left) and K is plotted in bold (and with respect to the scale on the right). The location of the APOE exons within the sequenced region are shown beneath the plot.