Background: Several findings link systemic lupus erythematosus (SLE) with C1q, the first molecule of the classical complement pathway. Polymorphisms of the C1qA gene are associated with low serum C1q levels in patients with cutaneous LE, but C1q polymorphisms have not been studied in patients with systemic lupus.
Objective: To determine whether polymorphisms of the C1q genes are associated with SLE, disease phenotypes, serum C1q and CH50 levels.
Methods: DNA for genetic analysis was obtained from 103 Caucasian patients with SLE and their family members. Five tag single nucleotide polymorphisms (tag SNPs) served as unique markers for underlying SNPs in the genes of the C1q protein. The pedigree disequilibrium test (PDT) was applied to trios to determine association of markers with SLE, SLE phenotypes, low serum C1q and low CH50. Single SNP association and haplotype analysis was also performed.
Results: The PDT revealed a significant association of the tag SNP rs631090 (covering the C1qB gene) with SLE (p = 0.02). Rs631090 was moderately associated with low serum C1q levels (p = 0.06). In addition, the tag SNPs rs292001 and rs294183 were associated with more severe SLE (Systemic Lupus Erythematosus International Collaborating Clinics (SLICC) damage index score>0; p = 0.007 and p = 0.02, respectively). Haplotype analysis and single SNP association analysis showed no significant associations, but additional analyses revealed that marker rs587585 is associated with low serum C1q and CH50 levels.
Conclusions: C1q polymorphisms are associated with SLE, serum C1q and CH50 levels in a stable founder population of patients with SLE. Although the studied population was small and allele frequencies were low, this is the first study to suggest an association of C1q polymorphisms with SLE.
Statistics from Altmetric.com
Systemic lupus erythematosus (SLE) is a systemic autoimmune disease characterised by the production of autoantibodies against self-antigens. These autoantibodies can affect several organs or organ systems, resulting in a broad spectrum of clinical and immunological manifestations. The exact pathogenesis of SLE is unknown, but genetic predisposition is supposed to play an important role.1 2
Several findings link SLE with C1q, the first molecule of the classical complement pathway. First, most individuals with hereditary C1q deficiency develop a clinical syndrome that closely resembles SLE (>90%). Genetic analysis has revealed mutations in the C1q gene in individuals with hereditary C1q deficiency, leading either to the absence of C1q or to the production of a non-functional, low molecular weight C1q.3 4 Secondly, antibodies directed against C1q (anti-C1q) are present in 20% to 50% of patients with SLE. The presence of anti-C1q is strongly associated with hypocomplementaemia, disease activity and the appearance of renal involvement in patients with SLE.5 6 Finally, active disease in SLE is often accompanied by low levels of C1q and other complement factors of the classical pathway, reflecting consumption of complement by active inflammation. As C1q plays a critical role in the clearance of immune complexes and apoptotic cells,7–9 this clearance may be impaired by low C1q levels.
Although various, predominantly immunological, studies link SLE with C1q and C1q deficiency, there is only little data on an association of SLE with polymorphisms of the genes encoding for C1q. The coding region for C1q is localised on chromosome 1p34–36 and consists of three genes, C1qA, C1qB and C1qC. Besides the few cases of lupus with hereditary C1q deficiency, only one other genetic study of the C1q gene has been performed in patients with lupus. In this study, a single nucleotide polymorphism of the C1qA gene was associated with decreased levels of serum C1q.10 However, this study was performed in patients with only a cutaneous form of lupus, and not in patients with systemic lupus.
In this study, we analysed whether C1q polymorphisms are associated with SLE and, therefore, are susceptibility loci for SLE. In addition, we determined whether C1q polymorphisms are accompanied by low levels of C1q or decreased activity of the classical complement pathway (CH50) in a population of SLE cases and first-degree relatives.
In the period from November 2000 to November 2001, all patients from our SLE cohort were invited to participate in a study on the genetic predisposition of SLE. All patients met the criteria for SLE according to the American College of Rheumatology (ACR).11 In total, 107 out of 164 patients (65%) gave written informed consent and were included. To avoid influence of ethnicity, 4 patients who were not of Caucasian descent were excluded, yielding 103 patients that could be studied for genetic associations with SLE (table 1). The Systemic Lupus International Collaborating Clinics (SLICC)12 damage index score was obtained for each patient as a measure of chronic damage caused by SLE. We classified disease manifestations using cumulative ACR criteria (table 1).
Participants invited their family members, ideally both parents or spouse and a child, to join the study. None of the family members (n = 195) had SLE. Either the non-transmitted chromosomes of the parents or the chromosomes of the spouse served as controls. These could be either determined directly (n = 60) or determined through other family members (eg, children or sibs of the patient) (n = 13). In the case that neither both parents nor spouse and child were participating (for n = 23 patients), other family members were used for linkage phase determination, but in most cases contributed to only one control haplotype. For seven patients, no family members were willing to participate. The study was approved by the local medical ethics committee.
C1q single nucleotide polymorphism (SNP) selection and genotyping
Five haplotype tagging SNPs from the C1q gene cluster (comprising C1qA, C1qC and C1qB) were selected to be genotyped. Using SNP browser software, V. 3.0 (Applied Biosystems, Forster City, California, USA) the HapMap build 35 database was loaded and subsequently the genotype data of the SNPs from a 30 kb region containing the C1qC gene, with flanking sequences including C1qA upstream and C1qB downstream, for which a TaqMan assay was available, were exported as PED format. Subsequently, the exported data for the HapMap Caucasian sample panel were loaded into Haploview (http://www.broad.mit.edu/mpg/haploview/) using the Tagger SNP selection algorithm with default settings, but with forced inclusion of three intragenic SNPs for each of C1qA, C1qC and C1qB, respectively. This resulted in the selection of five SNPs to cover the region: rs587585, rs292001, rs294183, rs294179 and rs631090 (see fig 1 and table 2). The five corresponding TaqMan assays, hCV3176797, hCV3176787, hCV3176770, hCV992987 and hCV992968, respectively, were ordered (Applied Biosystems) and genotyping was performed on an ABI7900HT apparatus.
Complement C1q in serum was measured by nephelometry. Activation of the classical complement pathway (CH50) was assessed by measuring haemolysis of antibody-coated sheep erythrocytes. Normal values for C1q are 0.10–0.25 g/litre, for CH50 >65%. Samples were obtained during quiescent disease (Systemic Lupus Erythematosus Disease Activity Index (SLEDAI)⩽4). C1q and CH50 values were available for 102 and 94 of the 103 patients, respectively.
The total number of available family members was 195. For the haplotype analysis, the non-transmitted haplotypes of the parents or the haplotypes of the spouse served as controls, either determined directly or through reconstruction using other family members yielding 146 haplotypes.
The pedigree disequilibrium test (PDT)13 14 was applied to the trios consisting of the patient and both parents (n = 42) and the families consisting of the patient and unaffected sib(s) (n = 27). This test assesses whether a specific allele is more often transmitted from a parent to the affected child or more often present in an affected sib than in an unaffected sib than expected. Classical association analysis was performed by comparing the genotype frequencies of each tag SNP between cases and controls using a χ2 test or Fisher exact test, as appropriate. Haplotype frequencies were estimated using the expectation–maximisation (EM) algorithm and compared between cases and controls by means of a likelihood ratio test (in-house software).
In addition, patients were analysed for serum C1q, CH50 levels and SLICC score. The PDT was applied separately to subgroups of patients with values above and below the median. Haplotype analyses were performed comparing patients with values above the median to those with values below. In a quartile analysis of serum C1q levels and CH50, haplotype groups were compared using a χ2 test. The two major haplotypes were named 1.1 and 1.2. The minor haplotype A (MHA) was defined as the haplotypes containing the minor allele of the tag SNP rs586585. Minor haplotype B (MHB) represented all other minor haplotypes. Correlation between C1q and CH50 levels was determined according to the Pearson correlation test.
Before association of C1q gene polymorphisms was SLE analysed, linkage disequilibrium was determined (table 3). This showed a weak disequilibrium between the markers rs292001, rs294183 and rs294179.
To determine whether polymorphisms of the C1q genes are associated with SLE, we performed a PDT. An association between SLE and the tag SNP rs631090 could be established (number transmitted 7, non-transmitted 1; p = 0.02) (table 4). To see whether polymorphisms have functional consequences (eg, leading to lower or less functional C1q), we analysed the association of C1q polymorphisms with serum C1q and CH50 levels. Although such an association of tag SNP rs631090 with low serum C1q levels is suggested, a significant relation could not be established (p = 0.06). To determine whether polymorphisms are involved in severity of disease, association of C1q polymorphisms with the SLICC score was analysed. Transmission of rs631090 was not associated with the SLICC score, but an association of the tag SNPs rs292001 and rs294183 with SLICC score was found (p = 0.007 and p = 0.02, respectively). There was no association between the other tag SNPs with either SLE, C1q, CH50 levels.
When single SNP association analysis was performed, none of the tag SNPs were associated with SLE, serum C1q, CH50 levels or SLICC score (table 5). However, a tendency towards an association of tag SNP rs587585 with low serum C1q (minor allele frequencies C1q below median: 16.4%, above median: 8.7%; p = 0.11) and low CH50 levels (allele frequencies CH50 below median: 17%, above median: 8.7%; p = 0.09) with the tag SNP rs587585 was suggested.
To further investigate the influence of this C1q polymorphism on the levels of serum C1q and CH50, an additional quartile analysis was performed. In this analysis serum C1q levels and the CH50 activity of different haplotype groups were compared. The MHA (containing the rs587585 minor allele) was more associated with C1q levels and CH50 in the lower quartile then were the major haplotypes 1.1 and 2.1, although not reaching significance (p = 0.051 and p = 0.077, respectively; fig 2A). When the MHA was compared with all the other haplotypes, there were significantly more patients with C1q and CH50 values in the lower quartile in the MHA group (p = 0.049, fig 2B). This difference confirms the trends already found for rs587585 in the single SNP association analysis.
No difference in haplotype frequencies was observed between SLE subjects and controls (table 6). To investigate the association of different SLE phenotypes (disease manifestations according to ACR criteria) with polymorphisms of the C1q region, haplotype analysis was performed, but no significant association was found (data not shown). Of note, a stepwise increase of percentage of patients positive for ACR skin criteria (1, 2 and 3) was observed after stratification by rs292001 genotypes: 65% (11/17) AA, 70% (35/50) AG and 81% (29/36) GG. However, this also did not reach statistical significance.
This study shows that polymorphisms of the C1q gene are associated with SLE in a sample of 103 Caucasian trios from the northern part of The Netherlands. In addition, another polymorphism in the same region is associated with low serum C1q and CH50 levels.
A significant association of SLE with the tag SNP rs631090, which covers part of the C1qB gene, was found in the PDT analysis. Our results suggest an association with low serum C1q levels for this tag SNP. Patients with an absolute C1q deficiency were not present in our cohort. This might be expected, as hereditary C1q deficiency is a very rare condition and is the cause of SLE only in a few cases. The presence of anti-C1q antibodies might also contribute to lower C1q and CH50 levels, but these antibodies were not measured in our population. A possible explanation for the association between polymorphisms in the C1q region and low CH50 levels is that patients having these polymorphisms might be the producers of less or less functional C1q.
Several studies suggest that lower or less functional C1q might contribute to the development of autoimmunity. In C1q deficient mice, the development of autoimmunity is related to impaired clearance of apoptotic cells.15 An in vitro study demonstrated reduced uptake of apoptotic cells by macrophages in the presence of sera of patients with SLE with low complement levels.16 In addition, low C1q and CH50 levels may also reflect complement consumption by disease activity. However, all samples in this study were obtained during quiescent disease (SLEDAI⩽4). Furthermore, the C1q and CH50 levels (see table 1) are fully comparable with those found in an other study where patients with SLE with low disease activity were analysed.17 In addition, the median complement C3 and complement C4 levels were within the normal range. Thus, although the influence of complement consumption on C1q and CH50 levels cannot be excluded, this effect is probably negligible.
Our results from the single tag SNP analysis were not as strong as the results from the PDT analysis. A trend for low serum C1q and CH50 levels was found in association with tag SNP rs587585, which covers the C1qA promoter region. The association of tag SNP rs587585 with lower serum C1q and CH50 levels was further analysed by comparing C1q levels and CH50 for different haplotype groups. Lower C1q levels and CH50 were found for the minor haplotype A group when compared with the other groups. This finding confirms the association of the rs587585 minor allele with low serum C1q and CH50, found in the single-SNP analysis. The finding that polymorphisms of the C1qA promoter region are associated with low C1q and CH50 levels is supported by a study by Miura-Shimura et al.18 In this study, a polymorphism of the C1qA upstream region was associated with low C1q and the development of nephritis in New Zealand Black mice.
None of the other tag SNPs in the PDT and single tag SNP analyses were associated with either SLE, serum C1q or CH50 levels. However, an association of rs292001 and rs294183 with a SLICC score >0 was found. In the haplotype analysis, C1q polymorphisms were not associated with SLE. The differences in results between PDT, the single tag SNP analysis and the haplotypes analysis can be explained by different kinds of information obtained from the data. The PDT focuses specifically on inheritance information whereas tag SNP analysis only compares allele frequencies between patients and controls. The haplotype analysis analyses the evolutionary history of mutations.
Because SLE is a very heterogeneous disease with a large variety of phenotypes, we also performed a haplotype analysis for the different SLE phenotypes according to the (cumulative) ACR criteria. This analysis revealed no significant association between polymorphisms of the C1q region and the different phenotypes, probably because groups with certain disease manifestations were either very small (eg, neurological manifestations) or contained almost all patients, as in the case of anti-nuclear antibodies and antibodies to anti-double-stranded DNA (dsDNA).
The finding that low C1q levels are associated with a polymorphism in the C1qA region,10 was not confirmed in this study. In our study, the marker rs587585 covered part of the C1qA promoter region and not the C1qA gene itself. We did not find significant results for our marker rs292001, covering part of the C1qA gene itself. This difference in findings can be explained by the fact that the study by Racila et al contained a relatively small number of patients. Furthermore, these patients had exclusively subacute cutaneous LE. Our study is the first that analysed the involvement of polymorphisms of the C1q region in systemic LE.
Even though the number of patients in our cohort is relatively low and allele frequencies were low, some conclusions might be drawn. Our study had 80% power to discover associations for alleles. Allele frequencies of 5% and 40% have a relative risk for susceptibility for SLE of 3.8 and 2.2, respectively; for difference in subgroups of SLE cases a relative risk of 4.1 and 2.3, respectively.19 These are moderate to high risks, but the crude odds ratios in this study (∼11 for rs631090 for SLE susceptibility, ∼6 for rs587585 for high CH50 level, and ∼3.0 and ∼2.4 for rs292001 and rs294183, respectively, for positive SLICC scores) are higher. Considering the size of our study population, we expect that the effects of C1q polymorphisms will be more pronounced in a larger population. Nevertheless, the results need to be confirmed in future study, preferably in al larger population of patients with SLE. Furthermore, this study was performed in a 100% Caucasian, stable founder population, in contrast to a lot of other genetic studies concerning SLE which have been performed in multiethnic populations.
In conclusion, this is the first study that suggests an association of C1q polymorphisms with SLE, serum C1q levels, CH50 and disease severity. Further studies should confirm this association and reveal the functional genetic variation that underlies our observations.
Competing interests: None declared.
Ethics approval: The study was approved by the local medical ethics committee.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.