Impaired collagen chaperone results in preterm PROM

It is well known that risk for many polygenic diseases, such as coronary heart disease, hypertension, prostate cancer, and asthma, varies according to race, ethnicity, or population. This phenomenon has led to the suggestion that mapping by admixture linkage disequilibrium (MALD) may hold the power to elucidate the mutations that underlie these common and complex chronic diseases (1). However, the concept of race is ill-defined because self-reported racial descriptors are often not accurate indicators of ancestry and are frequently confounded with sociocultural factors that underlie the risk of disease. Consequently, there has been significant debate concerning the scientific value of incorporating race or ethnicity descriptors into biomedical research (2–4). Indeed, a recent metaanalysis of 43 validated gene–disease associations across 697 populations found no race-associated variation in risk of disease in 86% of cases (5). However, this remarkable level of homogeneity of genotypic associations across populations does not imply that the risk of disease is homogeneous across populations. In fact, 58% of the markers demonstrated allelic heterogeneity among the control populations, which suggests that these diseases have population-specific differences in risk. Therefore, over one-half of the studied gene–disease associations should theoretically be detectable in recently admixed populations by association analysis. MALD has yet to successfully identify genetic variants responsible for polygenic disease (6) primarily because of the lack, until recently, of dense maps of validated markers with large frequency differences between populations (7). In this issue of PNAS, Wang et al. (8) report the identification of the first ancestry-informative mutation responsible for an increased risk of preterm, premature rupture of membranes (PPROM) in African Americans. PPROM occurs in ≈3% of all pregnancies and accounts for one-third of all preterm births (9). PPROM is also the leading identifiable cause of preterm birth and is more prevalent in African Americans than in European Americans. The findings of Wang et al. are important because the elucidation of a specific mechanism that underlies PPROM may lead to opportunities for developing management interventions. Furthermore, their discovery affirms the relevance of MALD for the elucidation of disease associations for diseases that differ significantly in risk among racially defined populations.

Establishing causality for a mutation underlying polygenic disease in non-model species is an enormous challenge that requires several independent lines of circumstantial and corroborating evidence in support of causality that are considered beyond reasonable doubt by the scientific community (10). Wang et al. (8) identified SERPINH1, which encodes Heat-shock protein 47 (Hsp47) as a candidate gene for PPROM because of Hsp47's role in collagen synthesis, because knockout mice die before birth with ruptured blood vessels and reduced type I collagen, because type I procollagen increases with Hsp47, and because PPROM fetal membranes have reduced collagen content. Through a series of functional and association analyses, Wang et al. provide compelling evidence for the causality of a previously reported C→T SNP within the SERPINH1 promoter (11) for an elevated risk of PPROM in African Americans. Two particularly elegant attributes of their statistical analyses concern the control of false-positive results due to population stratification and the potential for linkage of the SERPINH1 promoter SNP to other disease-associated mutations. African Americans are a recent admixture of African and European ancestries, and the extent of the European contribution varies by population geographical location (12.7–22.5%) (12) but much more significantly between individuals within populations (13). Because of the difference in risk of PPROM between African Americans and European Americans, it is possible that selection of the case and control cohorts based on the presence or absence of PPROM could have produced samples that were stratified according to European ancestry. By using 29 ancestry-informative markers, no difference in admixture was found between the case and controls. After statistical adjustment for ancestry (14), the association between SERPINH1 genotypes and risk of PPROM was unchanged. Strauss and coworkers (15, 16) have previously reported associations between SNPs within the promoters of MMP1 and MMP8 and PPROM, which are located ≈27 megabases from SERPINH1 on chromosome 11q (Fig. 1A). Although linkage disequilibrium (LD) is unlikely to extend over this range in human populations, the authors demonstrate that there is no LD between either the MMP1 or MMP8 promoter SNPs and the SERPINH1 promoter SNP. This finding eliminates the possibility that the detected SERPINH1–PPROM association is an indirect effect caused by disequilibrium with a disease-associated mutation in the vicinity of MMP1 and MMP8. The apparent clustering on chromosome 11q of genes involved in collagen synthesis or degradation or that have antiinflammatory activity (17) and that are associated with risk of PPROM is fascinating, and we wonder whether all four genes have independent effects on risk.

Fig. 1.

Cluster of PPROM susceptibility loci on HSA11q. (A) Polymorphisms in genes encoding an immunomodulatory protein (SCGB1A1), heat shock protein (SERPINH1), and two matrix metalloproteinases (MMP8 and MMP1) contribute to PPROM risk (8, 15–17). (B) Asterisks indicate polymorphism positions. SCGB1A1 +38 A→G increases the concentration of CC16, an immunomodulatory protein, in amniotic fluid. SERPINH1 −656 C→T decreases promoter activity in amnion fibroblasts but increases activity in dermal fibroblasts and uterine smooth muscle cells. The MMP8 haplotype of SNPs −799 C→T, −381 A→G, and +17 C→G increases promoter activity in trophoblasts and the MMP1 insertion polymorphism −1607 G increases promoter activity in amnion mesenchymal cells. SERPINH1 −656 C→T may also contribute to keloid and fibroid tumor risk.

Several questions remain unanswered by this study. First, is the SERPINH1–PPROM association detected and consistent in its effect on PPROM risk in European Americans? The limited evidence to date suggests that genotype–disease associations that have attained significance in a metaanalysis in at least one population behave homogeneously across racial populations (5); however, the relatively low frequency of the minor T allele in Caucasians (4.1%) will make testing this hypothesis difficult. Second, why is the frequency of the disease-associated T allele so high in Africans and African Americans? By sequence alignments to the chimpanzee and macaque genomes, we established that the SERPINH1 −656 C allele is ancestral and that the T allele arose after the divergence of the primates. Therefore, the T allele must have reached its present frequency in African populations through either drift or positive selection. However, the T allele is clearly selectively disadvantageous in contemporary African Americans because, until recently, neonates born 7.2 weeks prematurely would not have survived. In fact, Wang et al. (8) estimate the SERPINH1 T allele population-attributable risk of PPROM to be 12.3%. The authors also demonstrate that the SERPINH1 promoter alleles have different transcriptional activities within amnion fibroblasts, dermal fibroblasts, and uterine smooth muscle cells and speculate that the T allele may be deleteriously associated with the risk of keloids and uterine fibroid tumors, which are more prevalent in African Americans (Fig. 1B). Although these associations may not further reduce genotype fitness, the deleterious effects of the T allele are clear and argue against either drift or positive selection and in favor of negative selection on the allele. Therefore, as opposed to the current high frequency, we would have anticipated that historic selection would have reduced the frequency of the T allele to a low level in contemporary African and African American populations. Several alleles responsible for an increased risk of complex diseases in contemporary human populations have been shown to be ancestral, suggesting that the coevolution of genetically independent physiological traits and recent environmental changes have been responsible for altering genotype fitness from selective neutrality to disadvantage (18). It is therefore intriguing to speculate that the reduced fitness of the SERPINH1 −656 CT and TT genotypes is a modern phenomenon possibly caused by relatively recent increases in human adult size and birth weight and that the T allele was selectively neutral for much of the evolution of the modern human lineage. If so, it is indeed likely that drift is responsible for the remarkably high current frequencies of the T allele.