Mutations in Complement Regulatory Proteins Predispose to Preeclampsia: A Genetic Analysis of the PROMISSE Cohort

Participants

The first group of patients studied were enrolled in The PROMISSE Study (Predictors of pRegnancy Outcome: bioMarkers In antiphospholipid antibody Syndrome and Systemic lupus Erythematosus), a prospective multi-center observational study to identify predictors of pregnancy outcome initiated in 2003. Patients with SLE (defined as ≥4 ACR criteria) and/or APL Ab (defined as at least one of following documented twice between 6 wk and 5 y apart: ACL [IgG or IgM ≥40 units], lupus anticoagulant, or anti-β2GPI [IgG or IgM ≥40 units]) and disease controls were recruited by 12 wk gestation. At screening, patients with renal disease (proteinuria >1,000 mg/24 h, RBC casts, or serum creatinine >1.2 mg/dl), taking prednisone >20 mg/d, diabetes mellitus (Type I and Type II antedating pregnancy), hypertension (blood pressure ≥140/90 mmHg), or multiple gestations were excluded. Detailed medical and obstetrical information and serial blood specimens were obtained monthly during the course of pregnancy. Preeclampsia was defined as new onset of elevated systolic blood pressure (≥140 mmHg) and/or elevated diastolic blood pressure (≥90 mmHg) after 20 wk gestation on two occasions at least 4 h apart and proteinuria of 300 mg or greater in a 24 h urine specimen or ≥1+ on dipstick at least 4 h apart in the absence of pyelonephritis or hematuria. Severe preeclampsia and HELLP are defined by standard criteria [16]. In women with proteinuria present in early gestation, preeclampsia was defined as an increase in proteinuria to double their baseline level. For patients with a history of preeclampsia in previous pregnancies, medical charts were reviewed to document that patients met the above criteria above.

Of 250 enrolled patients with SLE and/or APL Ab who completed their pregnancies before 31 May 2008, 30 patients developed preeclampsia during the study and 10 additional patients had documented preeclampsia in a previous pregnancy (Table 1). Three of the 30 patients with preeclampsia during the PROMISSE Study pregnancy had documented preeclampsia in a previous pregnancy. None of the disease-free controls developed preeclampsia. The racial/ethnic distribution of the 40 patients with preeclampsia was 67% white, 17% black, 7% Asian, 5% Hispanic, and 4% other.

The Institutional Review Board at each of the PROMISSE Study sites (Hospital for Special Surgery, NY, NY, US; NYU School of Medicine, NY, NY, US; Johns Hopkins University, Department of Rheumatology, Johns Hopkins Medical Institutions, Baltimore, MD, US; Mount Sinai Hospital, Toronto, Canada; University of Utah, Intermountain Health Care, Salt Lake City, UT, US; Oklahoma Medical Research Foundation, University of Oklahoma Health Sciences Center, Oklahoma City, OK, US) gave approval for human participation for patients. Written informed consent that included the use of material for genetic research was obtained from all PROMISSE participants who contributed samples toward this study. All study samples were deidentified.

The second group of patients studied was from the Universal Samples Database at University of Utah, which stores maternal blood and clinical data from women who present to labor and delivery. Charts were reviewed to confirm the diagnosis of severe preeclampsia and HELLP defined according to established criteria, and 59 patients meeting criteria and with available DNA were identified [16]. The racial/ethnic distribution of patients with severe preeclampsia and/or HELLP was 74% white (non-Hispanic), 25% Hispanic, and 1% other. Control pregnancies defined as uncomplicated term deliveries in women with no more than one pregnancy loss and at least one live birth were identified in the Universal Samples Database at University of Utah. The Institutional Review Board at University of Utah, Salt Lake City, UT, gave approval for human participation for the participants whose samples are stored at the Universal Samples Database at University of Utah. Samples from 143 ethnically matched controls were available.

The allele frequency of MCP A304V and CFI IVS12+5 G>T variants in white individuals were determined using DNA from The UK Blood Services collection of Common Controls (UKBS collection) which forms part of the Wellcome Trust Case Control Consortium (WTCCC) (http://www.wtccc.org.uk) [17],[18].

Genetic and Statistical Analyses

Genomic DNA was extracted from peripheral blood leukocytes, and coding sequence and intronic flanking regions of MCP, CFH, and CFI were screened as described [19],[20]. Mutations considered for our analysis were either missense or nonsense coding mutations and mutations in conserved positions of splice sites and promoters. To further restrict our study to mutations that are likely deleterious, we ignored mutations present in dbSNP (http://www.ncbi.nlm.nih.gov/projects/SNP/), except for those previously observed in aHUS. This left novel variants and known aHUS alleles with the described properties. Mutations that occur in less than 5% of the population are termed low-frequency variants, and those in <0.5% are rare variants [21].

The rare variants could not be tested for Hardy-Weinberg equilibrium (HWE) because they do not appear in controls. The A304V variant was tested and found to be in HWE in 40 PROMISSE and 59 Utah cases (p = 1.0) and in the WTCCC controls (p = 0.15). The 34 PROMISSE controls could not be assessed for HWE because no alleles were present, and the 143 Utah controls were found not to be in HWE, largely because of a single homozygote that was sampled (p = 0.02).

To test the hypothesis that A304V is associated with preeclampsia, we combined cases and controls from the PROMISSE and Utah studies. These two populations were combined because the frequencies of A304V were similar, and Fisher's exact test was used to test for enrichment of a deleterious, dominant allele.

Expression and Functional Assessment of MCP

Substitution mutagenesis was performed using the QuikChange site-directed mutagenesis kit (Stratagene). The MCP template was the MCP isoform BC1 (GenBank accession no. X59405) cloned into the EcoRI site of plasmid pCDNA3.1+ (Stratagene). Transient transfections were performed with Fugene-6 (Roche Molecular Biochemicals) into Chinese hamster ovary (CHO) K1 cells. After cell lysis, MCP was quantified in an ELISA as described [22]. Briefly, MCP mAb GB24 was coated at 2.5 µg/ml on microtiter wells followed by blocking in PBS containing 4% BSA and 0.05% Tween 20. Cell lysates and standards were incubated in wells followed by reaction with a rabbit anti-MCP antiserum, horseradish- peroxidase–labeled donkey anti-rabbit IgG (Jackson ImmunoResearch), and ImmunoPure TMB Substrate kit (Pierce).

Ligand binding and cofactor assays for C4b and C3b have been described [22]. Briefly, an ELISA format was used for characterizing ligand binding. C3b or C4b (Complement Technologies) was coated onto wells of a microtiter plate. Next, dilutions of MCP wild-type or mutant CHO lysates were added. Cofactor assays employed biotinylated ligands in low- or physiologic salt buffer (i.e., 10 mM Tris [pH 7.4], with 25 or 150 mM NaCl), CFI (100 ng), and cell lysates (25 pg of wild-type or mutant MCP). After 20 and 90 min, cleavage fragments were analyzed by a 10% size fractionation on reducing SDS/PAGE followed by transfer, Western blotting, and densitometric scanning [22].

Patients with Preeclampsia

The PROMISSE Study is a prospective multi-center observational study in which patients with SLE and/or APL Ab were enrolled at 12 wk gestation and followed through pregnancy. Patients with renal insufficiency, diabetes mellitus (antedating pregnancy), hypertension, or multiple gestations (risk factors predisposing to preeclampsia) were excluded. Between 2003 and 2008, 250 patients with SLE and/or APL Ab completed their pregnancies, 40 of whom developed preeclampsia during the study or had preeclampsia in a previous pregnancy (Table 1). All patients with preeclampsia were screened for mutations in the genes encoding the complement regulatory proteins MCP, CFI, and CFH. Mutations that occur in less than 5% of the population (low-frequency alleles) are termed uncommon variants [21].

We identified heterozygous mutations in complement system pathway genes in seven of 40 patients with preeclampsia (18%)—four patients with mutations in MCP, two with CFI mutations, and one with a CFH mutation. Complement C3 levels were normal at enrollment in these patients. We did not find any mutations in 34 patients without preeclampsia (15 matched for disease [SLE and/or APL Ab], age, and ethnicity, and 19 healthy control patients matched for age and ethnicity) (7/40 versus 0/34, p<0.01; also see results of mutation screening described for each gene in the sections to follow). We reject the null hypothesis of equal or fewer rare, deleterious alleles in controls at the 0.02 level (p = 0.0105 for the one-tailed test). This approach represents a variation of existing collapsing methods for the analysis of low-frequency variants [23]. The frequency of these mutations and variants in other healthy populations is described below. The clinical characteristics of patients with mutations are presented in Table 2.

To replicate the relationship of complement system variants with preeclampsia and to determine whether mutations in complement regulatory proteins are present in non-autoimmune patients with severe preeclampsia and/or HELLP, we screened 59 patients and 143 ethnically matched healthy controls with normal pregnancies from the Universal Samples Database at the University of Utah for mutations in MCP, CFI, and CFH. We identified heterozygous mutations in the genes of five patients—four in MCP and one in CFI. No significant sequence variants were detected in CFH. The clinical characteristics of the patients from this cohort are presented in Table 3. Three individuals among the healthy controls had the A304V variant (5/59 versus 3/143, p<0.05). Details of the mutations and variants and their distribution are described below.

Mutations and Variants of MCP

MCP is a widely expressed transmembrane protein that binds C3b and C4b deposited on cell surfaces and serves as a cofactor for their cleavage by CFI. The resulting attached fragments, iC3b and C4d, are not capable of forming convertases. By cleaving C3b, MCP is a key regulator on host cells of the amplification loop of the alternative pathway. Four patients with preeclampsia (10%) had mutations in MCP. Three patients were heterozygous for MCP A304V, a low-frequency hypomorphic variant previously shown to be deficient in control of alternative pathway activation on the cell surface [14]. The allele occurred at a frequency of 3.7% in the PROMISSE Study patients with preeclampsia. All three patients with A304V were white, and one had SLE only, one APL Ab only, and one both SLE and APL Ab. Two patients had severe preeclampsia and two patients reported prior pregnancy complications, including preeclampsia (Table 2). This variant has been identified in four of 238 patients from the French multi-center aHUS cohort and in two of 181 healthy French controls, and in none of 120 white participants from another study [12]. Because the frequency of MCP A304V was not reported in publically available databases, including HapMap, we genotyped a large population of whites. In 1,529 individuals from the WTCCC, the allele frequency was 2.1% (61 heterozygotes and two homozygotes). Of note, no pregnancy histories or other medical information was available for these individuals.

We confirmed the association of MCP A304V with preeclampsia in a second cohort identified from the Universal Samples Database at the University of Utah. In 59 non-autoimmune patients with severe preeclampsia and/or HELLP, we identified four individuals (6.8% of patients) heterozygous for MCP A304V (allele frequency 3.4%). The details of their pregnancies are presented on Table 3. In 144 ethnically and geographically matched controls with uncomplicated pregnancies, two individuals were heterozygous and one homozygous for MCP A304V (allele frequency 1.4%), none of whom had histories of thrombotic microangiopathy or preeclampsia. The frequency of MCP A304V was consistent with our findings in the WTCCC. Combining our autoimmune and non-autoimmune cohorts with preeclampsia (n = 99), we identified seven individuals heterozygous for MCP A304. In both groups separately and when combined (99 cases versus 178 controls), A304V confers risk of preeclampsia. The variant acts in a dominant fashion with an odds ratio of 4.4 (95% confidence interval 1.1–17.6; p = 0.027).

We have previously studied in detail the complement inhibitory profile of A304V employing permanent CHO cell lines expressing an equal copy number of wild-type or A304V. While there was no detectable difference in ELISA type assays of solubilized protein, in situ the A304V was ∼50% as effective at controlling C3b deposition by the alternative pathway [14]. In addition, a 25%–50% reduction in levels of wild-type complement regulatory proteins expressed by CHO cells impairs regulation of the alternative pathway, showing that haploinsufficiency has functional consequences [14],[24].

The potential role of complement dysregulation in preeclampsia is underscored by our discovery of a novel mutation in exon 2 of the gene encoding MCP (K32N) in a patient from the PROMISSE Study who had SLE. Her pregnancy ended with preeclampsia and fetal death at 25 wk. She had a history of two early pregnancy losses. MCP K32N is a missense mutation that alters a charged solvent-exposed amino acid in complement control protein repeat (CCP) 1 [25]. To examine its functional consequences, the K32N mutant construct was transiently transfected in CHO cells. The mutant protein was quantified in cellular lysates and binding to human C4b and C3b was assessed (Figure 2A). While binding to C3b was similar to wild-type MCP, the mutant showed an approximately 50% decrease in C4b binding (p<0.001) compared to wild-type MCP. Given the lower binding to C4b, we suspected that the K32N mutant form of MCP would have a decreased capacity to serve as a cofactor for CFI-mediated cleavage of C4b. To examine this possibility, we performed cofactor assays comparing the ability of wild-type MCP and rare mutant K32N to serve as cofactors for cleavage of C3b and C4b by monitoring loss of the α′ chain and generation of cleavage fragments (C4d in the case of C4b and α1 in the case of C3b) (Figure 2B) [24]. As predicted by the binding studies, cofactor activity for C3b was similar for the K32N mutant and wild-type MCP, whereas the K32N mutant showed decreased CFI-mediated cleavage activity for C4b. At 20 min in the C4b cofactor activity assay, no C4d was detected in the K32N lane, but an easily discernable band in wild-type MCP. We performed densitometric scanning of the C4d after 90 min and found that cleavage of C4b to C4d in wild-type was four times greater than in the K32N mutant (Figure 2C). Our results define the functional defect in K32N and identify the initial disease-related mutation in MCP that results exclusively in impaired ability to bind C4b and mediate C4b cofactor activity. Because SLE is characterized by autoantibodies that trigger the classical pathway, defective regulation of C4b is likely to influence the severity of tissue injury and risk for preeclampsia (Figure 1). Indeed, the presence of C4 degradation products in placentas from patients with APL Ab and in kidneys from lupus patients underscores the importance of C4 in disease pathogenesis [10],[26].

Mutation in CFI

CFI, a plasma glycoprotein, is a serine protease that cleaves the α′ chain of C3b and C4b in the presence of a cofactor protein, such as CFH or MCP. Inactivation of C3b and C4b prevents the formation of C3 and C5 convertases and thereby down-modulates the classical, lectin, and alternative pathways. We found identical heterozygous mutations in the gene encoding CFI in two unrelated white women with preeclampsia in the PROMISSE Study (Table 2). Similar to three of the four patients with MCP variants, those with CFI mutations had a history of previous pregnancy complications, specifically preterm deliveries at 24 and 28 wk, respectively. The mutation identified in CFI, an IVS12+5 G>T change, has been reported in patients with aHUS and was shown to affect the donor site of intron 12 reducing the splice score from 93 to 86 [27] but did not alter CFI serum levels [12]. The CFI IVS12+5 G>T variant has been identified in six of 325 French control individuals (0.9% allele frequency) and in eight of 238 patients from the French multi-center aHUS cohort (1.7% allele frequency) (alone in four patients and in association with a second mutation in four cases; V. Fremeaux-Bacchi, unpublished results). The frequency of this variant is not reported in publically available databases. Our findings suggest that CFI IVS12+5 G>T is a low- frequency variant that may be enriched in aHUS. Its clinical impact cannot be assessed in the control patients because pregnancy histories and other medical information were not available for these individuals.

The two patients presented here and other patients with the IVS12+5 G>T variant have had normal CFI levels (66 and 63 mg/l, nl 42-78 mg/l), but CFI levels within the normal range does not exclude alterations in local production [28],[29]. In the 12 Eutherian mammals that have a similar CFI gene architecture to human CFI (Ensemble release 52), conservation is perfect at the +5 intron position. The lack of variation at this position relative to the surrounding bases argues that a G at this base is functionally important.

We screened CFI in 57 non-autoimmune patients with severe preeclampsia and/or HELLP in our second cohort from the Universal Samples Database at University of Utah. In one patient, we identified a heterozygous nonsynonymous sequence variant in exon 11 (c.1246A>G, p.Ile398Leu). We did not find this change in 117 white British controls. Bienaime et al. and Sellier-Leclerc et al. have previously demonstrated this variant in three aHUS patients (two heterozygous, one homozygous) but not in 100 controls [29],[30]. CFI was undetectable in the individual with the homozygous change [29]. Recombinant expression of the I398L mutant confirmed nearly complete absence of production compared to wild-type CFI, which was associated with altered subcellular localization detected by immunofluorescence and altered endoglycosidase H digestion [29]. We identified two other nonsynonymous changes in this cohort: c. 1217G>A, p.Arg388His [19],[28] and c.1322A>G, p.Lys423Arg [rs41278047]. Both changes have previously been described in control populations and are classed as SNPs in our study. Identification of CFI I398L mutation, a complete loss of function leading to haploinsufficiency, supports a role for defects in complement regulation by CFI in the pathogenesis of preeclampsia. This association also suggests that enrichment of IVS12+5G>T variants in patients with preeclampsia and aHUS [12] is functionally important, although its mechanism is as yet undefined.

Mutations in Factor H

The single factor H mutation (S40A) in the PROMISSE cohort was extensively investigated and not shown to have a functional defect (normal secretion in a transfection system, normal plasma levels, and normal to increased C3b binding and cofactor activity) (unpublished data). No factor H mutations were found in the Utah cohort.