Population and study design
Our study population was recruited from Maternidade Escola Januário Cicco, a tertiary center for women’s health, located in Natal, Rio Grande do Norte state, Brazil. A total of 1693 women were recruited from 2002 to 2010, as part of a broader study aiming to investigate clinical, epidemiological and genetic aspects of hypertensive disorders of pregnancy. Clinical data as well as blood samples were collected at the time of enrollment. For the current study, we retrospectively selected 1282 women based on their pregnancy outcome: 693 normotensive women (control), 342 preeclampsia (PE), 61 superimposed preeclampsia (PEsuper), 74 eclampsia, and 112 HELLP syndrome cases. All Methods were performed in accordance with the Declaration of Helsink and followed the Brazilian ethical standards of scientific research. The research protocol was reviewed and approved by the Federal University of Rio Grande do Norte (CEP-UFRN 88) and Brazilian National Ethical Committee (CONEP 5059). All research participants or their legal guardian provided informed consent.
The diagnostic criteria followed the recommendations from the American College of Obstetrician and Gynecologists 19. Preeclampsia was defined as the new onset hypertension (SBP ≥ 140 mmHg or DBP ≥ 90 mmHg) and proteinuria (≥ +1 on dipstick) after 20 weeks of gestation. Superimposed preeclampsia occurred when the woman had a previous diagnosis of chronic hypertension and developed proteinuria after 20 weeks of gestation. Eclampsia was defined by the presence of seizure, while HELLP syndrome diagnosis was based on Mississippi Class III system (AST > 40 IU/L and LDH > 600 IU/L and platelets < 150,000/μL) 20. Controls were healthy pregnant women with no history of hypertension. Women with multiple pregnancies, diabetes or other chronic diseases were excluded from study.
The variants in ERAP2 (rs2549796, rs2927609, rs11135484) and LNPEP (rs27300, rs38034, rs2303138) were all tag-variants, identified through a pairwise selection strategy with an r2 threshold ≥ 0.8 in Haploview 4.2 21 using the HapMap CEU population genotype data (HapMap Rel 27 phase II+III). Variants rs30187 and rs27044, in ERAP1, were chosen based on their effect on protein function 22,23 as well as their implication in other diseases 17.
DNA extraction was carried out as previously described24. Samples were genotyped by SNaPshot® technique and the capillary electrophoresis performed on ABI PRISM® 3100 Avant Genetic Analyzer (Applied Biosystems). Technique standardization was carried out according to Lins and colleagues 25. GeneMapper® software (Applied Biosystems, CA, USA) was used for the genotype calling.
Population stratification assessment
To avoid confounding by ethnicity we used a panel with 27 ancestry informative markers (AIMs) particularly designed for the Brazilian population 26. A sub-sample of 756 women randomly selected was used to assess the genetic ancestry of our study population (n=1282) using principal component analysis in SNPRelate R package 27. Samples from The 1000 Genomes Project 28 of European (IBS), African (ASW, MSL, YRI) and American (CLM) origins were used as reference populations.
Variant effects on mRNA and protein levels were assessed from GTEx (dbGaP Accession phs000424.v8.p2) 29 and single nucleotide polymorphisms annotator (SNiPA) 30 databases. Aiming to functionally validate the ERAP1 genetic finding, we recruited an additional cohort of 65 pregnant women, including 29 normotensive controls and 36 severe preeclampsia cases, that had their Ang II plasma concentration measured by ELISA commercial kit (MyBioSource, San Diego, CA, USA, Cat.Num. MBS453098). Briefly, blood samples were systematically collected between 7 and 9 am in EDTA tubes and immediately centrifuged. The obtained plasma was stored at -80 °C until assay.
Clinical and demographic data were analyzed through chi-squared and t-test for categorical and quantitative variables, respectively. With regard to the genetic data, allele frequencies were compared by Fisher exact test, whereas genotype and haplotype association tests were performed through logistic regression models including maternal age and parity (primigesta vs others) as covariates. Haplotype frequencies were estimated by Expectation-Maximization algorithm with a minor haplotype frequency threshold of 0.03. The p-values were corrected for family-wise error rate by permutation procedures (10,000x) implemented in PLINK 31. All analyses were performed by comparing each case phenotype (i.e. PE, PEsuper, eclampsia and HELLP) against the normotensive control group.