Nonrandom association of polymorphic restriction sites in the β-globin gene cluster

By using probes for ε-, ψβ₁-, and β-globin genes, we found four additional polymorphic restriction sites that have frequencies > 0.1 in persons of Mediterranean area origin, Asian Indians, and American Blacks. Three of these (HincII sites) and the two previously described polymorphic HindIII sites (one in intervening sequence (IVS) II of each γ-globin gene) are distributed over 32 kilobases (kb) of DNA located 5' to the δ-globin gene. This region of DNA comprises two-thirds of the β-globin gene cluster. Since each of these five polymorphic sites can be present (+) or absent (-), in theory there exist 32 possible combinations of sites (haplotypes). However, in Italians, Greeks, Indians, and Turks, 3 of the 32 haplotypes, (+----), (-+-++), and (-++-+), account for 92% of 89 β[A] chromosomes examined. The observed frequencies for these haplotypes are 0.64, 0.15, and 0.13 in the populations studied, in contrast to expected frequencies (based on the observed gene frequencies at each of the five sites) of 0.20, 0.006, and 0.005, respectively. In American Blacks, a fourth haplotype, (----+), which is rare in non-Black populations, has a frequency of 0.37 in contrast to its expected frequency of 0.05. These results suggest a nonrandom association of DNA sequences over 32 kb 5' to the δ-globin gene in all populations studied. Two other polymorphic sites 3' to the δ gene (the newly discovered Ava II site in IVS II of the β-globin gene and the Bam HI site 3' to it) are nonrandomly associated with each other but randomly distributed with respect to the above haplotypes. This suggests that randomization of sequences has occurred within 12 kb of DNA between these two nonrandomly associated sequence clusters. Nonrandom association of polymorphic restriction sites has practical consequences in that it limits the usefulness of these additional HincII sites for prenatal diagnosis of hemoglobinopathies by linkage analysis. These sites provide little additional information for detection of β-thalassemia, while the polymorphic Ava II site, which lies outside the nonrandomly associated sequences 5' to the δ gene, improves the test applicability from 52% to 70% of couples at risk.