High prevalence of the GSTM 3 * A / B polymorphism in sub-Sarahan African populations

A 3-bp insertion/deletion polymorphism in intron 6 of GSTM3 (rs1799735, GSTM3*A/*B) affects the activity of the phase 2 xenobiotic metabolizing enzyme GSTM3 and has been associated with increased cancer risk. The GSTM3*B allele is rare or absent in Southeast Asians, occurs in 5-20% of Europeans but was detected in 80% of Bantu from South Africa. The wide genetic diversity among Africans led us to investigate whether the high frequency of GSTM3*B prevailed in other sub-Saharan African populations. In 168 healthy individuals from Angola, Mozambique and the São Tomé e Príncipe islands, the GSTM3*B allele was three times more frequent (0.74-0.78) than the GSTM3*A allele (0.22-0.26), with no significant differences in allele frequency across the three groups. We combined these data with previously published results to carry out a multidimensional scaling analysis, which provided a visualization of the worldwide population affinities based on the GSTM3*A/*B polymorphism.


Introduction
Sistonen et al. (1,2) have recently described the worldwide patterns of distribution of functional polymorphisms in phase 1 drug-metabolizing enzymes of the cytochrome P450 (CYP) superfamily and emphasized the strong impact that population substructure may have on the variation seen in pharmacogenetic loci.The present communication extends these observations to GSTM3, a member of the glutathione S-transferase family of phase 2 enzymes, that catalyzes the conjugation of glutathione to a wide variety of xenobiotics.A 3-bp insertion/deletion polymorphism in intron 6 (rs1799735, GSTM3*A/*B) of the encoding GSTM3 gene, which may affect the regulation and ultimately the amount and activity of GSTM3 (3), provides an extreme example of inter-population differentiation.The homozygous GSTM3*B genotype is ca.20 times more frequent in an African Bantu group (frequency 0.64) than among Europeans (frequency <0.05) (4)(5)(6).The wide genetic diversity among African populations (7) led us to investigate whether the very high frequency of the GSTM3*B allele (0.80) in South African Bantu prevails in other sub-Saharan African populations.We screened population samples from Angola, Mozambique, São Tomé e Príncipe, and combined the new data with results from the literature to carry out a multidimensional scaling analysis, which provided a visualization of the worldwide population affinities based on the GSTM3*A/B polymorphism.

Material and Methods
A sample of 168 unrelated healthy, adult men and women from Cabinda (Angola, N = 73), Mozambique (Maputo, N = 66) and São Tomé e Príncipe (N = 29) was analyzed in the present study.The samples were collected by IPATIMUP collaborators according to research protocols approved by the ethic boards of all participating institutions in Africa.A single blood sample was drawn from each subject with informed consent.Genomic DNA was extracted using standard phenol-chloroform methodologies and genotyping for the GSTM3*A/B polymorphism was performed by restriction fragment length polymorphism (RFLP) as previously described (3).
The allelic frequencies were estimated by direct gene counting.Allele and genotype frequencies were compared using the χ 2 test, with the level of significance set at P < 0.05.The ARLEQUIN software, version 3.0 (http://cmpg.unibe.ch/software/arlequin3) was used to test for Hardy-Weinberg equilibrium and to produce pairwise fixation index (F ST ) genetic distances between populations.Multidimensional scaling (MDS) analysis based on F ST values was conducted by the method included in the STATISTICA data analysis software system, version 8.0.(http://www.statsoft.com).

Results and Discussion
The allele frequency and genotype distribution of the GSTM3*A/*B polymorphism in the three sub-Saharan population samples are shown in Table 1.The genotype frequencies in the three populations did not deviate from expected Hardy-Weinberg proportions.The GSTM3*B allele was three times more frequent (0.74-0.78) than the GSTM3*A allele (0.22-0.26) in each group.There were no significant differences in allele frequencies across the three populations.These results are in excellent agreement with previously published data for a Bantu group recruited at Durban, South Africa, where the frequency of GSTM3*B was 0.8 (5).Bantu people comprise the majority of the population from Mozambique and Angola, and were as well the main African contributors to the settlement of São Tomé e Príncipe in the 15th century.Thus, collectively our data indicate that the GSTM3*B variant is 3-4 times more frequent than the wild-type GSTM3*A in sub-Saharan populations of Bantu influence, spread out over a large geographical area.Within these populations, allele frequencies for the tested GSTM3 polymorphism are, however, very different from the ranges typically found either in populations of European descent, where the GSTM3*B allele usually does not add more than 20% (4-6), or in Asian populations, where the same vari-ant is virtually absent (5).Accordingly, when F ST distances were calculated between each of the African populations and different European or Asian populations, all values were found to be highly statistically significant (results not shown).The resulting F ST distances were submitted to MDS analysis and the corresponding plot (Figure 1) clearly demonstrates the ability of this unique polymorphism to discriminate among populations of African, Asian or European ancestry.Not surprisingly, therefore, populations such as the Creoles from Mauritius Island, U.S. African-Americans and Brazilians self-identified as White, Intermediate or Black (8) displaying variable degrees of genetic admixture of continental ancestral populations, all occupy intermediate positions in the MDS representation.Despite the limitation of our MDS analysis, based on F ST distances from a single gene, GSTM3, it does not seem that major distortions have been produced since the representation portraits reasonably well the known historical relationships between the populations under study.
The plot demonstrates how heterogeneous the populations referred to as African-Americans can be, as illustrated by the relative position of the two samples considered here (Af-Am1 and Af-Am2 in Figure 1).These two groups, which in fact are from U.S. states quite distant geographicallyone from California, on the West Coast ( 9) and the other from Pennsylvania and New York, on the East Coast (10) -show very distinct levels of admixture between European and African ancestry regarding GSTM3.As a matter of fact, the demographic history varied enough among different African-American populations to explain the diverse genetic structure of contemporary African-Americans (11).
Concerning the highly heterogeneous Brazilian population, the MDS representation is consistent with our previous demonstration that the frequency of the GSTM3*B allele increases from self-reported White, to Intermediate and to Black individuals, in parallel to the increase in the average proportion of African ancestry across these groups, although at the individual level, the frequency of the GSTM3*B allele increases continuously as the individual proportion of African ancestry increases, irrespective of self-reported "color/race" (8).

Table 1 .
Allele frequencies and observed genotype distributions of GSTM3*A/B in African populations.
Data are reported as means and 95%CI.N = number of individuals.a Probability values for tests of departure from Hardy-Weinberg expectations (HWE).b χ 2 tests for comparison across the three population groups.