Complement C4 Copy Number Variation is Linked to SSA/Ro and SSB/La Autoantibodies in Systemic Inflammatory Autoimmune Diseases.

OBJECTIVE: Copy number variation of the C4 complement components, C4A and C4B, has been associated with systemic inflammatory autoimmune diseases. This study was undertaken to investigate whether C4 copy number variation is connected to the autoimmune repertoire in systemic lupus erythematosus (SLE), primary Sjögren's syndrome (SS), or myositis.
METHODS: Using targeted DNA sequencing, we determined the copy number and genetic variants of C4 in 2,290 well-characterized Scandinavian patients with SLE, primary SS, or myositis and 1,251 healthy controls.
RESULTS: A prominent relationship was observed between C4A copy number and the presence of SSA/SSB autoantibodies, which was shared between the 3 diseases. The strongest association was detected in patients with autoantibodies against both SSA and SSB and 0 C4A copies when compared to healthy controls (odds ratio [OR] 18.0 [95% confidence interval (95% CI) 10.2-33.3]), whereas a weaker association was seen in patients without SSA/SSB autoantibodies (OR 3.1 [95% CI 1.7-5.5]). The copy number of C4 correlated positively with C4 plasma levels. Further, a common loss-of-function variant in C4A leading to reduced plasma C4 was more prevalent in SLE patients with a low copy number of C4A. Functionally, we showed that absence of C4A reduced the individuals' capacity to deposit C4b on immune complexes.
CONCLUSION: We show that a low C4A copy number is more strongly associated with the autoantibody repertoire than with the clinically defined disease entities. These findings may have implications for understanding the etiopathogenetic mechanisms of systemic inflammatory autoimmune diseases and for patient stratification when taking the genetic profile into account.

INTRODUCTION

Systemic inflammatory autoimmune diseases are a group of diseases characterized by inflammation in multiple organs and the presence of antibodies targeting different ubiquitously expressed cytoplasmic or nuclear proteins. Systemic lupus erythematosus (SLE), primary Sjögren's syndrome (SS), and idiopathic inflammatory myopathies (myositis) are all categorized as systemic inflammatory autoimmune diseases. In SLE, multiple tissues and organs such as the skin, joints, cardiovascular system, and kidneys are commonly affected, whereas a more specific involvement is seen with lacrimal and salivary glands affected in primary SS, and muscle as well as connective tissues in myositis. Despite different patterns of tissue and organ involvement, the 3 diseases share important features, including clinical manifestations, presence of autoantibodies to nuclear antigens, and several genetic loci (1, 2, 3, 4, 5, 6).

The complement system plays an important role for clearance of immune complexes and apoptotic cells. Impaired removal of cellular debris may lead to exposure of cellular self‐antigens and loss of tolerance (7, 8). Although mainly studied in SLE, the complement system has also been shown to play a role in the pathogenesis of primary SS and myositis (9, 10). Particularly relevant are the genes in the early classical complement pathway, for which deficiency in any of the genes C1Q, C1R, C1S, C2, or C4 may lead to SLE or lupus‐like disease (11). Further, low plasma levels of C3 and C4 are routinely used in the clinical setting as biomarkers for complement activation associated with disease activity and flares in SLE (12, 13).

C4 is coded by the paralogous genes C4A and C4B located between the HLA class I and class II regions on chromosome 6. A high level of copy number variation exists for the 2 C4 genes, and while most individuals carry 2 C4A copies and 2 C4B copies, the number of genes may range from 0 to 5 copies for C4A and 0–4 copies for C4B (14). By definition, C4A and C4B differ by 4 amino acids (PCPVLD and LSPVIH, respectively) in exon 26 that also affect the biochemical reactivity toward either amino groups or hydroxyl groups, respectively (15, 16, 17).

Previous studies have shown an association between low copy number of C4A and systemic inflammatory autoimmune diseases (14, 18, 19, 20). However, due to the strong linkage disequilibrium (LD) between C4A copy number and the extended C*07:01–B*08:01–DRB1*03:01–DQB1*02:01 risk haplotype, which is associated with SLE, primary SS, and myositis in populations of European origin, it has been difficult to define whether the association is with HLA or with the complement system (21). In a recent study, C4 copy number rather than HLA was suggested to be the main risk factor for SLE (22). This conclusion was based on a parallel analysis of C4 copy number and HLA alleles in patients of European and African American origin, for which the latter population shows low LD between C4A copy number and DRB1*03:01.

Due to a high homology between the genomic reference sequences of C4A and C4B (99.91% identical), the 2 genes are generally excluded from variant calling analysis due to ambiguous mapping of sequencing reads. In addition, a diploid state is assumed in variant calling of human autosomes, which is incompatible with the high level of variation in the C4 copy number ranging from 2 to 8 C4 copies. Therefore, variation in C4 genes at nucleotide level and the association to disease remain relatively unexplored.

The purpose of the current study was to perform a focused analysis of C4 copy number and C4 nucleotide variation in patients with 3 systemic inflammatory autoimmune diseases and in healthy controls. In parallel, we aimed to evaluate the relation between HLA alleles and C4 copy number variation, for which the combined analysis of 3 cohorts of patients with SLE, primary SS, or myositis allowed the comparison of shared and distinct patterns within and between the diseases.

PATIENTS AND METHODS

Study participants

Patients diagnosed as having SLE, primary SS, or myositis at Scandinavian rheumatology clinics, as well as healthy blood donors/population controls have been previously described in detail (23, 24, 25), and basic characteristics are presented in Supplementary Table 1, available on the Arthritis & Rheumatology website at https://onlinelibrary.wiley.com/doi/10.1002/art.42122. For validation of C4 copy number calls and HLA alleles, we included whole‐genome sequencing data from 75 parent/offspring trios, in which offspring were diagnosed as having SLE (26). The individual studies were approved by local ethics committees, and all study participants provided written informed consent.

Targeted sequencing, genotyping, and quality control

The capturing array and the targeted DNA sequencing of patients and healthy controls has previously been described (23, 24, 25, 27). A detailed description of the sequencing workflow, alignment, genotype calling, and quality control at variant‐ and individual‐based levels can be found in the Supplementary Information, available on the Arthritis & Rheumatology website at https://onlinelibrary.wiley.com/doi/10.1002/art.42122.

DNA sequencing and genotyping was performed at the SNP&SEQ Technology Platform in Uppsala, part of the National Genomics Infrastructure Sweden and Science for Life Laboratory. The computations were enabled by resources in projects sens2017142 and sens2020577, provided by the Swedish National Infrastructure for Computing at Uppsala Multidisciplinary Center for Advanced Computational Science.

Estimation of C4 copy number and calling of HLA alleles from targeted sequencing data

The total copy number of C4 was estimated based on read depth using the GermlineCNVCaller (GATK), and the relative read depths of 5 C4A/C4B‐specific single‐nucleotide variants were used for ascertainment of total number of C4 copies into copy number of C4A and C4B. Genetic variants in the C4A/C4B genes were called at bp resolution using the HaplotypeCaller (GATK). The human endogenous retrovirus element present in some C4 genes (28) was not analyzed (Supplementary Information, https://onlinelibrary.wiley.com/doi/10.1002/art.42122).

HLA alleles of 6 HLA genes (HLA–A, HLA–B, HLA–C, HLA–DPB1, HLA–DQB1, and HLA–DRB1) were called at 4‐digit resolution from sequencing data using xHLA (29). A detailed workflow for the analysis of C4 and HLA is available in the Supplementary Information together with method validation using results from previous polymerase chain reaction–based C4 analyses (30, 31, 32).

Plasma C4 concentration and autoantibody status of patients

Measurement of plasma C4 in SLE and primary SS patients was performed at local centers as part of clinical routine (33, 34). Information about autoantibodies was extracted from medical records, and healthy controls were included as a reference cohort under the assumption that autoantibodies are specific to patients. Nevertheless, a small percentage of the individuals in the general population may be positive for autoantibodies (34), but the impact on this study is considered negligible.

C4b deposition on heat‐aggregated human IgG

Serum from healthy individuals carrying only C4A genes or only C4B genes were incubated with heat‐aggregated IgG, and deposition of C4b was detected by enzyme‐linked immunosorbent assay, as described in the Supplementary Information (https://onlinelibrary.wiley.com/doi/10.1002/art.42122).

Statistical analysis

Statistical analyses were performed in R version 4.0.4 (35). Two‐tailed P values less than 0.05 were considered significant. The statistical tests and covariates that were included (e.g., principal components for genetic population structure) are described in the text and figure legends. Odds ratios (ORs) and 95% confidence intervals (95% CIs) were calculated. For analysis of associations in the HLA region, a Bonferroni correction for 5,000 tests was used, corresponding to a statistical significance threshold of 1 × 10−5.

RESULTS

Association of C4A copy number with systemic inflammatory autoimmune diseases

Using targeted sequencing data, we estimated the C4 copy number in 2,290 patients diagnosed as having SLE, primary SS, or myositis, and in 1,251 healthy controls (Figure 1A). The total C4 copy number calls ranged from 2 to 8 C4 copies, but since only 2 individuals had more than 6 C4 copies (Supplementary Information, https://onlinelibrary.wiley.com/doi/10.1002/art.42122), we focused on subjects with 2–6 C4 copies.

Figure 1

Association between complement C4 copy number and 3 systemic inflammatory autoimmune diseases (SIADs). A Workflow for analysis. Three patient groups with systemic inflammatory autoimmune diseases and 1 reference cohort were analyzed for HLA alleles and copy number of the paralogous C4 genes C4A and C4B, using targeted sequencing data. Association analysis of C4 was performed for clinical subsets of the diseases. Additionally, common and rare variants in the C4 genes were analyzed from the sequencing data. B, Total C4 copy number in healthy controls and patients with systemic lupus erythematous (SLE), primary Sjögren's syndrome (SS), or myositis (n = 3,541). Logistic regression was performed to analyze associations in the combined patient group compared to healthy controls, with adjustment for sex and principal components 1–4 (PC1–PC4). Odds ratios (ORs) represent disease risk in association with each decrease in C4 copy number. C, Copy number of C4A and C4B in each patient group and healthy controls (n = 3,520). Analysis is based on logistic regression with both C4A and C4B included as additive variables and with adjustment for sex and PC1–PC4. ORs represent disease risk in association with each decrease in C4A or C4B copy number. MHC = major histocompatibility complex; CI95% = 95% confidence interval.

Notably, the pattern in C4 copy number was relatively similar among the 3 systemic inflammatory autoimmune diseases, and therefore, we performed both joint and separate analyses of the diseases. As previously observed, a low C4 copy number was associated with an increased risk of all 3 diseases compared to healthy controls (P for C4 = 2 × 10−38) (Figure 1B and Supplementary Figure 1A, https://onlinelibrary.wiley.com/doi/10.1002/art.42122), and this association was almost exclusively explained by the copy number of the C4A gene (P for C4A = 5 × 10−45) (Figure 1C). We noted a negative correlation between C4A copy number and C4B copy number (rs = −0.50) (Supplementary Figure 1B), and although the copy number of C4B was slightly higher in patients compared to controls, each decrease in C4B copy number was modestly associated with an increased risk of systemic inflammatory autoimmune disease when adjusting for the effect of C4A (P for C4B = 6 × 10−4) (Figure 1C and Supplementary Figure 1A). Based on these results, we concluded that low C4A copy number is a strong risk factor for systemic inflammatory autoimmune disease, whereas the effect of C4B is limited.

Overlap of association with autoantibodies against SSA and SSB between systemic inflammatory autoimmune diseases

Autoantibodies against nuclear antigens are a common feature and part of the clinical evaluation of SLE, primary SS, and myositis, and therefore, we analyzed whether C4A copy number was associated with the autoantibody repertoire in patients. A strong and consistent association was detected between low C4A copy number and presence of anti‐SSA and anti‐SSB autoantibodies in all 3 diseases (Supplementary Table 2, https://onlinelibrary.wiley.com/doi/10.1002/art.42122). None of the other autoantibodies investigated for all diseases were consistently associated with C4A copy number (Supplementary Table 2).

The prevalence of anti‐SSA/SSB autoantibodies differed highly among the diseases and was highest in primary SS, with 73% of patients having autoantibodies against SSA and/or SSB (Figure 2A). For myositis, 32% of patients had anti‐SSA/SSB autoantibodies with most being against SSA and only 2.4% of patients being positive for both anti‐SSA and anti‐SSB. In SLE, 41% of patients had autoantibodies against SSA and/or SSB. A few patients (2.4%) had autoantibodies against SSB only, which is in line with the notion that anti‐SSA antibodies appear first and may be followed by anti‐SSB antibodies due to epitope spreading (36).

Figure 2

Association between C4 copy number and anti‐SSA/SSB autoantibodies. A, Prevalence of anti‐SSA/anti‐SSB autoantibodies in patients with SLE (n = 919), primary SS (n = 902), or myositis (n = 364). B, Logistic regression analysis of association between each decrease in C4A copy number and anti‐SSA/anti‐SSB autoantibody status among each patient group compared to healthy controls. Dotted lines indicate ORs for association in the combined group of 3 diseases. C, Logistic regression analysis of association between C4A copy number and anti‐SAA/anti‐SSB autoantibody status in the combined patient group compared to healthy controls. In B and C, bars represent 95% confidence intervals, and models have been adjusted for presence of C4B, sex, and PC1–PC4. D, Plasma C4 levels in patients with primary SS. Groups were compared by analysis of variance with square root–transformed values for the C4 concentration, adjusted for sex and cohort (n = 470). C4A and C4B copy number was included in the model; the x‐axis shows total C4 copy number for simplicity. E, Deposition of the complement activation product C4b on heat‐aggregated human IgG, analyzed with varying concentrations of serum from healthy individuals carrying C4A genes only (n = 5) or C4B genes only (n = 5). The samples were analyzed ≥3 times, and the mean absorbance for each sample was evaluated using the Mann‐Whitney U test. In D and E, Data are shown as box plots. Each box represents the 25th to 75th percentiles. Lines inside the boxes represent the median, and whiskers extend to 1.5 times the interquartile range. See Figure 1 for other definitions. Color figure can be viewed in the online issue, which is available at http://onlinelibrary.wiley.com/doi/10.1002/art.42122/abstract.

Remarkably, we observed a dose–response relationship between C4A copy number and the prevalence of autoantibodies against SSA and SSB in all 3 diseases. Each decrease in C4A copy number was associated with an increased risk of autoantibodies against both SSA and SSB (OR 5.89 [95% CI 4.83–7.23]), whereas the association with the systemic inflammatory autoimmune diseases without any autoantibodies against SSA/SSB was more modest (OR 1.53 [95% CI 1.36–1.73]) (Figure 2B). Patients with autoantibodies against either SSA or SSB showed a similar intermediate association, and therefore, we combined anti‐SSA+SSB− and anti‐SSA−SSB+ patients into 1 group (OR 2.37 [95% CI 2.02–2.77]).

Despite slight differences between the diseases, the associations of C4A copy number were more specific to the presence of anti‐SSA/SSB autoantibodies than to the individual diseases, and thus, we evaluated the prevalence of autoantibodies against number of C4A copies collectively for the 3 diseases. This revealed a strong effect of C4A copy number on the association with autoantibodies against SSA/SSB, in which the risk of disease with anti‐SSA/SSB autoantibodies for individuals with a C4A copy number of 0 was ~80 times higher than for individuals with a C4A copy number of 3 (Figure 2C). Interestingly, each change in C4A copy number was associated with a consistent change in disease risk.

In addition to a genetic association between C4 copy number and autoantibodies against SSA and SSB, we also identified a functional association linking the two parameters. As previously shown (22, 37), plasma C4 levels in patients with primary SS showed higher concentrations with an increasing copy number of C4A and C4B (Figure 2D). Additionally, we detected lower plasma C4 levels in patients with primary SS with autoantibodies against SSA/SSB when compared to patients with primary SS without anti‐SSA/SSB autoantibodies (Figure 2D), suggesting a direct connection between anti‐SSA/SSB autoantibodies and plasma C4. However, a similar association between plasma C4 and anti‐SSA/SSB was not found for SLE patients (P = 0.41; n = 411) (Supplementary Figure 2A, https://onlinelibrary.wiley.com/doi/10.1002/art.42122).

By analyzing serum from healthy individuals carrying C4A genes only or C4B genes only, we detected a higher deposition of the C4 activation product C4b on aggregated human IgG for C4A carriers (Figure 2E). These results are consistent with the proposed role of C4A being more efficient than C4B in clearance of immune complexes and suggest a connection between low C4A copy number and impaired removal of immune complexes in systemic inflammatory autoimmune disease.

In summary, we showed a strong association between C4A copy number and autoantibodies against SSA and SSB, which to a greater extent represented an association with anti‐SSA/SSB autoantibodies rather than with the systemic inflammatory autoimmune diseases themselves.

Higher proportion of SLE patients carrying the C4A loss‐of‐function (LoF) variant rs760602547

We continued by evaluating a common LoF variant (rs760602547) mainly found in C4A. The LoF variant results in a CT insertion in exon 29, which introduces a premature stop codon in C4A (38).

We called the LoF variant rs760602547 for patients and controls and detected the CT insertion in 7.7% of all individuals. Among the LoF carriers, 98% had 1 LoF variant, and the variant was not found among any of those carrying only C4B genes, thus supporting the notion that the CT insertion mainly is found in C4A and rarely in C4B (39). When analyzing the number of LoF variants in patients with SLE, primary SS, or myositis compared to healthy controls, no enrichment of the variant was seen in patients (P > 0.10 by logistic regression). However, logistic regression analysis allowing for interaction between C4A copy number and the LoF variant rs760602547 showed that SLE patients with a low C4A copy number carried the LoF variant to a greater extent than controls (P for LoF = 0.01; P for interaction = 0.03 [n = 2,136]) (Figure 3A), indicating an additional mechanism of impaired C4A function in SLE. The increased frequency of the LoF variant among SLE patients with a low C4A copy number was also seen in a focused analysis of SLE patients and healthy controls with 1–2 C4A copies (P = 0.01 by Fisher's exact test; n = 1,559). For patients with primary SS and those with myositis, no association was found for the LoF variant (P for interaction > 0.10) (Figure 3A). Grouping patients based on anti‐SSA/SSB autoantibody status showed a tendency toward an increased frequency of the LoF variant in SSA/SSB‐negative patients when compared to healthy controls, whereas no association was seen for the patients with SSA/SSB autoantibodies (Figure 3B).

Figure 3

Loss‐of‐function (LoF) variant in C4A and rare variants in C4 genes. A, Proportion of patients and healthy controls carrying the C4A LoF variant rs760602547. B, Proportion of the LoF variant rs760602547 among SSA/SSB autoantibody subgroups combined across the 3 systemic inflammatory autoimmune diseases and healthy controls. Patients (or SSA/SSB subgroups) and controls were grouped based on C4A copy number, and the size of points indicate the total number of individuals in each group with the specific C4A copy number. P values are based on logistic regression with interaction between C4A copy number and rs760602547. The LoF variant was only present among individuals with 1–4 C4A genes. C, Number of individuals carrying rare (present among <0.5% of all individuals) coding variants in ≥1 C4 gene (synonymous, n = 44; missense, n = 100). Variants present among 10–17 individuals have been combined. D, Number of C4 genes carrying a rare coding variant in each disease cohort. The number of variants in each disease group has been adjusted for total C4 copy number in order to account for lower copy number of C4 among patients with systemic inflammatory autoimmune diseases. See Figure 1 for other definitions. Color figure can be viewed in the online issue, which is available at http://onlinelibrary.wiley.com/doi/10.1002/art.42122/abstract.

Analysis of plasma C4 by linear regression showed lower C4 concentrations for carriers of the LoF variant rs760602547 (SLE: P = 8 × 10−5 [n = 407]; primary SS: P = 5 × 10−4 [n = 471]) (Supplementary Figure 3A, https://onlinelibrary.wiley.com/doi/10.1002/art.42122) after accounting for the copy number of C4A and C4B, demonstrating that the LoF variant directly affects plasma C4 concentration.

To ensure that the enrichment of the LoF variant rs760602547 in SLE patients with a low C4A copy number was not caused by an indirect linkage to DRB1*03:01 or other SLE‐associated HLA alleles, we analyzed the LD between the LoF variant and common variants in C4, HLA alleles, and single‐nucleotide polymorphisms (SNPs) in the HLA region. We detected multiple SNPs and C4 variants in high LD with the LoF variant (Supplementary Figure 3B, Supplementary Data). For HLA alleles, the strongest LD was seen with DQB1*06:04 (R2 = 0.56), whereas C4 copy number or DRB1*03:01 was not in LD with the LoF variant rs760602547 (R2 < 0.10). Therefore, the association of the LoF variant was independent of SLE‐associated HLA alleles.

Overall, we detected lower plasma C4 levels in carriers of the LoF variant rs760602547, and the LoF variant was enriched in SLE patients with a low C4A copy number, thereby adding another level of complexity in the genetic variation of the complement system.

No enrichment of rare C4 variants in systemic inflammatory autoimmune diseases

Due to the common variation in C4 copy number, together with the high sequence homology between C4A and C4B (99.91% identical), nucleotide variants in the C4 genes are generally omitted from sequencing‐based analyses. By using information about the C4 copy number for each individual while analyzing C4 nucleotide variants (Supplementary Information, https://onlinelibrary.wiley.com/doi/10.1002/art.42122), we called common and rare variants for all study participants, focusing on coding variants. As variants could not be assigned unambiguously to C4A or C4B, variants were analyzed relative to the total C4 copy number.

Overall, we detected 144 rare coding variants (present among <0.5% of all study participants). Of these variants, 65% were found in only 1 individual, and 69% of the variants were missense variants (Figure 3C and Supplementary Information). Analysis of rare coding variants in each disease group did not indicate an enrichment of rare synonymous or missense variants in any of the systemic inflammatory autoimmune diseases (Figure 3D). Further, prediction of the effect of missense variants did not show an enrichment of putative benign or deleterious variants in systemic inflammatory autoimmune diseases (Supplementary Figure 3C, https://onlinelibrary.wiley.com/doi/10.1002/art.42122), and no difference was detected for rare C4 variants in patients grouped based on anti‐SSA/SSB autoantibody status (Supplementary Figure 3D).

In summary, we detected 144 rare variants in the coding sequence of C4, but no enrichment was observed in C4 genes of patients with systemic inflammatory autoimmune disease.

Anti‐SSA/SSB autoantibody subgroups and association with HLA and C4

Although the association of C4A copy number was more specific to anti‐SSA/SSB autoantibody status than to the individual disease entity (Figure 2B), we noted some differences between the 3 systemic inflammatory autoimmune diseases. The most striking difference was observed for patients without anti‐SSA/SSB autoantibodies, in which no association was seen between C4A copy number and primary SS (OR 1.13, P = 0.23), while a strong association was seen for a decrease in C4A copy number among myositis patients (OR 1.91, P = 4 × 10−11) and to a lesser extent among SLE patients (OR 1.55, P = 2 × 10−8) (Figure 2B).

In order to evaluate whether other variants in the HLA region could explain these differences, we analyzed the association of SNPs, HLA alleles, C4 copy number, and common variants in the C4 genes in relation to the autoantibody status for each of the individual diseases. Global analysis of the HLA region generally showed an association with the extended C*07:01–B*08:01–DRB1*03:01–DQB1*02:01 haplotype together with C4A copy number in both SLE and myositis patients without anti‐SSA/SSB autoantibodies (Figure 4A). No additional associations were present after conditioning for the HLA allele with the strongest association (Supplementary Figure 4, https://onlinelibrary.wiley.com/doi/10.1002/art.42122). For patients with primary SS without autoantibodies to SSA/SSB, no association was seen with the HLA region (Figure 4A), suggesting at least partially different autoimmune processes among the 3 systemic inflammatory autoimmune diseases for the anti‐SSA/SSB–negative subset of patients.

Figure 4

Association of variants in the HLA region with anti‐SSA/SSB–negative and anti‐SSA/SSB–positive patients. A, Association of variants in HLA region in patients with SLE (n = 544), primary SS (n = 241), or myositis (n = 247) who were negative for autoantibodies against SSA/SSB, as compared to healthy controls (n = 1,251). B, Regional association plot of HLA variants in patients with SLE (n = 168) or primary SS (n = 368) who were positive for autoantibodies against both SSA and SSB, as compared to healthy controls (n = 1,251). Few myositis patients (n = 11) had autoantibodies to both SSA and SSB, and therefore the data for these patients were not plotted. HLA alleles for 6 genes (HLA–A, HLA–C, HLA–B, HLA–DRB1, HLA–DQB1, and HLA–DPB1) and variants in C4 present in >1% of individuals were included in the analysis. Groups were analyzed for associations using logistic regression with adjustment for sex and PC1–PC4. Dashed lines represent the Bonferroni‐corrected significance threshold (P = 1 × 10−5). Color figure can be viewed in the online issue, which is available at http://onlinelibrary.wiley.com/doi/10.1002/art.42122/abstract.

We continued analyzing genetic variation in the HLA region for patients with autoantibodies against both SSA and SSB. Due to the low number of anti‐SSA/SSB–positive myositis patients (n = 11), we focused on SLE and primary SS patients. Again, the extended C*07:01–B*08:01–DRB1*03:01–DQB1*02:01 haplotype together with C4A copy number generally showed a strong association with SLE and primary SS patients with autoantibodies against both SSA and SSB (Figure 4B). After conditioning for DQB1*02:01 and C4A copy number, no additional association with HLA alleles was found for SLE (Supplementary Figure 5A, https://onlinelibrary.wiley.com/doi/10.1002/art.42122), whereas conditional analysis showed an association with DRB1*03:01, DRB1*15:01, DQB1*04:02, and B*08:01 in anti‐SSA/SSB–positive patients with primary SS patients (Supplementary Figure 5B).

The genomic reference sequence for C4A and C4B differs at 18 nucleotide positions of which 5 variants in exon 26 are used for defining copy number of C4A and C4B (Supplementary Information). In addition to the variants differing between C4A/C4B reference sequences, we detected 78 variants found in ≥1 C4A/C4B gene among more than 1% of the individuals. We included all 96 common variants in the analysis of the HLA region, using the number of C4 genes, with alternative alleles for each variant as variables in the analysis. However, the copy number of C4A generally explained the largest part of the associations with minimal effect of individual variants in the C4 genes (Figure 4 and Supplementary Figure 5).

In addition to the analysis of anti‐SSA/SSB–negative and anti‐SSA/SSB–positive patients against healthy controls, we also performed a pairwise case–case analysis of autoantibody subsets within the individual diseases and for the 3 systemic inflammatory autoimmune diseases combined. Again, the extended C*07:01–B*08:01–DRB1*03:01–DQB1*02:01 haplotype together with C4A copy number explained the main association (Supplementary Figure 6, https://onlinelibrary.wiley.com/doi/10.1002/art.42122).

In summary, no association was seen between the HLA region and primary SS patients without any autoantibodies against SSA and SSB, in contrast with anti‐SSA/SSB–negative SLE and myositis patients. Further, the extended C*07:01–B*08:01–DRB1*03:01–DQB1*02:01 haplotype together with C4A copy number was associated with both anti‐SSA/SSB–positive systemic inflammatory autoimmune diseases as well as SLE and myositis without anti‐SSA/SSB autoantibodies. In conditional analyses, C4A copy number generally followed the extended HLA haplotype, and the copy number of C4A largely explained the association with C4 with a minor effect of individual variants in the C4 genes.

DISCUSSION

In the current study, we demonstrated a strong association between low C4A copy number and the presence of anti‐SSA/SSB autoantibodies in the 3 systemic inflammatory autoimmune diseases: SLE, primary SS, and myositis. The similarities between the disease associations indicated that a low C4A copy number to a higher extent is associated with anti‐SSA/SSB autoantibodies rather than to the individual disease entities.

Autoantibodies, such as anti‐SSA and anti‐SSB, generally appear several years before clinical onset of both SLE (40, 41) and primary SS (42, 43), indicating a slow progression from asymptomatic autoimmunity to clinical manifestations. Here, we demonstrated that a genetic susceptibility with a low C4A copy number is a risk factor for development of anti‐SSA/SSB autoantibodies, and when present, the autoantibodies may contribute to systemic inflammatory autoimmune disease in a subset of individuals. Moreover, these observations may explain the partial overlap in clinical manifestations between diseases for a subgroup of patients (e.g., in SLE, in which 25% of patients are affected by secondary SS [44]). However, the low number of myositis patients with autoantibodies against both SSA and SSB limits the conclusions that can be drawn from this subset of patients.

Although the C4A copy number was associated with anti‐SSA/SSB autoantibodies in a dose‐dependent manner, the association between C4A copy number and systemic inflammatory autoimmune diseases without anti‐SSA/SSB autoantibodies was limited. As described previously, no association with the HLA region was found for anti‐SSA/SSB–negative patients with primary SS (24, 45). In contrast, C4A copy number together with the extended C*07:01–B*08:01–DRB1*03:01–DQB1*02:01 haplotype showed a residual association with anti‐SSA/SSB–negative patients with SLE and myositis, which is consistent with earlier reports indicating DRB1*03:01 as a risk factor in SLE patients without anti‐SSA/SSB autoantibodies (46, 47). This suggests partly different etiopathogenetic mechanisms for anti‐SSA/SSB–negative patients with primary SS compared to patients with SLE or myositis.

We note that the Scandinavian study population in the current analysis is relatively homogenous, which may limit the generalizability of the study. Still, the variation in C4 copy numbers generally follows the pattern reported in other studies of individuals of European ancestry (14, 18, 22).

While it is difficult to distinguish the association signal from C4A copy number and DRB1*03:01 in populations of European descent, several lines of evidence support the notion that C4A copy number plays a central role in systemic inflammatory autoimmune diseases. Analysis of C4 copy number in SLE patients of East Asian (48) and African American origin (22)—populations in which the LD between C4A copy number and DRB1*03:01 is low—showed a strong risk with a low C4 copy number. The importance of the complement system is further supported by the strong risk of SLE and lupus‐like symptoms often including anti‐SSA autoantibodies in individuals with deficiencies in any of the early classical complement pathway genes C1Q, C1R, C1S, C4, and C2 (10, 11).

The higher frequency of the common deleterious C4A variant among SLE patients with a low C4A copy number adds an additional modulating factor to the variation in C4A copy number and plasma levels of C4 among SLE patients. Previous analyses of the LoF variant did not detect any enrichment in SLE patients (21, 49). However, a larger cohort, along with a slightly increased frequency of the LoF variant, may explain the enrichment detected in the current study. Further, the enrichment was only seen when taking the C4A copy number into account, which was not done in the previous studies. Nevertheless, no enrichment of the LoF variant was seen in patients with primary SS or myositis, suggesting a higher vulnerability among SLE patients.

Despite a high similarity between the C4A and C4B proteins, a low copy number of C4A explained the major risk for systemic inflammatory autoimmune disease with only a minor effect for the C4B copy number. The 4 C4A/C4B‐defining amino acids in exon 26 of C4 alter the reactivity of C4A and C4B toward amino groups and hydroxyl groups, respectively. This is thought to increase the efficiency of C4A in the clearance of immune complexes and apoptotic cells, whereas C4B is more efficient in targeting microbes (9). We found more extensive depositions of the C4 activation product C4b on aggregated human IgG when adding serum from individuals carrying only the C4A gene as compared to serum from individuals carrying only the C4B gene, thus indicating that C4A has enhanced capability to remove immune complexes. The functional differences between C4A and C4B were also investigated in a recent study by Simoni et al (50), in which lupus‐prone mice were genetically modified to express C4 with the human C4A/C4B‐defining amino acids. When compared to mice coding for human C4B, mice coding for C4A showed enhanced clearance of apoptotic cells, less autoreactive B cells and lower titers of anti‐SSA autoantibodies, overall supporting the notion that C4A has a role in prevention of autoantibody generation and autoimmunity.

In conclusion, we demonstrated a strong association between low C4A copy numbers and the presence of anti‐SSA/SSB autoantibodies in SLE, primary SS, and myositis. Similar relationships between C4A copy numbers and autoantibody status were observed in all 3 diseases. Our findings suggest that anti‐SSA/SSB autoantibodies are largely dependent on genetic predisposition, and this subset of autoimmune patients may be considered a specific diagnostic entity.

AUTHOR CONTRIBUTIONS

All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Lundtoft had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study conception and design

Lundtoft, Pucholt, Martin, Eloranta, Almlöf, Syvänen, Lindblad‐Toh, Nilsson, Blom, Diaz‐Gallo, Svenungsson, Rönnblom.

Acquisition of data

Lundtoft, Pucholt, Martin, Bianchi, Lundström, Eloranta, Sandling, Sjöwall, Jönsen, Gunnarsson, Rantapää‐Dahlqvist, Bengtsson, Leonard, Baecklund, Jonsson, Hammenfors, Forsblad‐d'Elia, Eriksson, Mandl, Bucher, Norheim, Johnsen, Omdal, Kvarnström, Wahren‐Herlenius, Notarnicola, Andersson, Molberg, Diederichsen, Almlöf, Syvänen, Kozyrev, Lindblad‐Toh, Blom, Lundberg, Nordmark, Diaz‐Gallo, Svenungsson, Rönnblom.

Analysis and interpretation of data

Lundtoft, Pucholt, Martin, Bianchi, Lundström, Blom, Diaz‐Gallo, Svenungsson, Rönnblom.

ADDITIONAL DISCLOSURES

Author Pucholt is currently an employee of Olink Proteomics, but was employed by Uppsala University during the time the study was conducted. Author Mandl is an employee of Novartis.

Disclosure Form

Click here for additional data file.

Appendix S1: Supporting Information

Click here for additional data file.

Appendix S2: Supporting Information

Click here for additional data file.

Supplementary Appendix file.

Click here for additional data file.