the presence of the UTR-3 haplotype.42
For the HLA-G 3’UTR, Castelli et al. characterized the variability and its haplotype structure
in one of the most heterogeneous populations of the world.8 In their population only less
than 1% of haplotypes was undefined, which is in line with our data, and most likely not clinically relevant. The allelic and haplotype frequencies of our control group were similar
to those described for European populations.43
Among the eight haplotypes, UTR-4 was less frequently seen in recurrent miscarriage, and more frequently in uneventful pregnancy, suggesting a protective effect of this haplotype in pregnancy. In contrast to pregnancy where a high HLA-G expression seems to be beneficial, HLA-G expression in tumour cells contributes to the escape of tumour
cells from a tumour specific immune response.44 In line with our data, in prostate cancer
UTR-4 was associated with an increased risk in disease development and supports the
idea that the presence of UTR-4 haplotype influences HLA-G expression.45 Conversely,
UTR-4 was associated with a lower risk of colorectal carcinoma, although after adjustment
for possible confounders no significance was reached.46
The UTR-4 haplotype is the only haplotype bearing the +3003C allele, which was also increased in women with uneventful pregnancy. The +3003C allele is in linkage
disequilibrium with the allele -725G at the promotor region.43 Although this allele was
previously associated with increased risk for miscarriage in couples,47 it was also found
to be related with a higher HLA-G expression, as compared to the -725C or T allele.48
The affinity of microRNAs binding to this region is probably dependent on the type
of nucleotide present at this +3003 position.49 In our study, women with recurrent
miscarriage had more often the genotypic variants +3010CC and +3003TT. The presence of +3003T, in combination with +3010C, shows low affinity for miR-513a compared to
the +3003C/+3010G haplotype.49 The combination of the polymorphisms +3010C and
+3003T are part of the UTR-2, UTR-3, UTR-5, and UTR-7 haplotypes, of which only the UTR-3 haplotype contains the 14-bp insertion. The UTR-2, UTR-5 and UTR-7 were
found to be associated with lower expression of sHLA-G.13 Interestingly, in the subgroup
analysis in the current study the frequency of UTR-2 was significantly higher in women with secondary recurrent miscarriage compared to control subjects. In addition, UTR-4 was only significantly associated with uneventful pregnancy compared to women with secondary recurrent miscarriage, and not with women with primary recurrent miscarriage. Although the size of this group was limited, this finding may support the idea that secondary recurrent miscarriage has different underlying pathology than primary recurrent
3
live birth increases immunity to fetal antigens in the next pregnancy by for example the
presence of HLA antibodies or H-Y antibodies.50,51 Hypothetically, women with secondary
recurrent miscarriage and an UTR-2 haplotype could have less immunomodulatory capacity during implantation and therefore possibly an increased risk of miscarriage due to increased immunity by the presence of antibodies from a previous live birth. Although the presence of HLA antibodies increases after 28 weeks of pregnancy and antibodies can
still be present at time of a new conception,52,53 no significant differences were observed
in paternal-specific HLA antibodies between women with primary recurrent miscarriage
and women with secondary recurrent miscarriage in a recent study.54 These differences
between primary recurrent miscarriage and secondary miscarriage should be subject for further studies.
Furthermore, the UTR-2, UTR-5, and UTR-7 haplotypes share three genotypic variants which have previously been associated with decreased HLA-G levels: 1.) the 14-bp
insertion influencing mRNA stability,10 resulting in lower levels or even absence of sHLA-G
in plasma11-13; 2.) the +3142G allele, which increases the affinity of several microRNAs,
including miR-148a, miR148b, and miR152, leading to downregulation of HLA-G
expression36; and 3.) the +3187A allele, which is related to decreased HLA-G expression
due to increased number of adenines in AU-rich motif mediating mRNA degradation.35 In
line with these data we observed a decreased frequency of the 14-bp deletion/deletion in women with recurrent miscarriage (27%) compared to the control group (43.8%), the latter being comparable to the Dutch genome reference group (37.1%) (data not shown). The current study shows that the 3’UTR HLA-G haplotype, rather than single polymorphisms in the 3’UTR region of HLA-G, is most significantly associated with recurrent miscarriage. On the basis of these data we hypothesize that a combination of polymorphisms in the 3’UTR region of the HLA-G gene affects the degree of microRNA binding to these regions, which influences HLA-G expression. If this is the case, microRNAs may be used in the future as biomarkers and possible targets for therapeutic strategies, aiming at the induction of a higher HLA-G expression and improved pregnancy outcome.
Acknowledgements
The authors would like to thank Clara Kolster, Marjolein Bourgonje-Verhart, and Marise Wagner for their help obtaining blood samples from all the cases. The authors would like to thank Geert Haasnoot for his help to assess the Hardy-Weinberg equilibrium for genotypes and haplotypes of the HLA-G studied.