Article Text
PDF
Abstract
Objectives The goals of these studies were to elucidate the inter-relationships of specific anti-nuclear antibody (ANA), complement, and the interferon gene signature (IGS) in the pathogenesis of systemic lupus erythematosus (SLE).
Methods Data from the Illuminate trials were analysed for antibodies to dsDNA as well as RNA-binding proteins (RBP), levels of C3, C4 and various IGS. Statistical hypothesis testing, linear regression analyses and classification and regression trees analysis were employed to assess relationships between the laboratory features of SLE.
Results Inter-relationships of ANAs, complement and the IGS differed between patients of African Ancestry (AA) and European Ancestry (EA); anti-RNP and multiple autoantibodies were more common in AA patients and, although both related to the presence of the IGS, relationships between autoantibodies and complement differed. Whereas, anti-dsDNA had an inverse relationship to C3 and C4, levels of anti-RNP were not related to these markers. The IGS was only correlated with anti-dsDNA in EA SLE and complement was more correlated to the IGS in AA SLE. Finally, autoantibodies occurred in the presence and absence of the IGS, whereas the IGS was infrequent in anti-dsDNA/anti-RBP-negative SLE patients.
Conclusion There is a complex relationship between autoantibodies and the IGS, with anti-RNP associated in AA and both anti-dsDNA and RNP associated in EA. Moreover, there was a difference in the relationship between anti-dsDNA, but not anti-RBP, with complement levels. The lack of a relationship of anti-RNP with C3 and C4 suggests that anti-RNP immune complexes (ICs) may drive the IGS without complement fixation, whereas anti-dsDNA ICs involve complement consumption.
- autoantibodies
- lupus erythematosus
- systemic
- cytokines
Data availability statement
Data are available in a public, open access repository. Data are available in a public, open access repository. Data were downloaded from Gene Expression Omnibus (GEO) under accession GSE88884.
Statistics from Altmetric.com
Key messages
What is already known about this subject?
Anti-RNP and anti-dsDNA autoantibodies form immune complexes (ICs) that induce interferon (IFN).
Anti-dsDNA ICs deposit in lupus kidneys and contribute to renal pathology by fixation and activation of complement.
What does this study add?
Anti-RNP autoantibody ICs have a stronger capacity to induce IFN than anti-dsDNA autoantibodies but are not related to depression of complement.
Autoantibodies are likely required for the induction of IFN, but the IFN gene signature (IGS) is not required for the production of autoantibodies.
How might this impact on clinical practice or future developments?
Development of a more precise way to detect the presence of pathogenic ICs and identify the mechanisms underlying the IGS.
Introduction
Systemic lupus erythematosus (SLE) is a prototypic autoimmune disease that primarily affects young women and is especially severe in patients of African Ancestry (AA).1–4 Among the most prominent immunological features of SLE is the production of anti-nuclear antibodies (ANAs); these antibodies target nucleic acids, proteins and protein-nucleic acid complexes.5 Among ANAs, anti-dsDNA antibodies are unique markers for both classification and disease activity.6 In SLE, anti-dsDNA antibodies form immune complexes (ICs) that deposit in the kidney to activate complement and provoke inflammation; anti-dsDNA can also stimulate the production of type I interferon (IFN) by cells of the innate immune system.7 8
Like anti-dsDNA, ANAs directed to RNA-binding proteins (RBPs) can form ICs inducing IFN.9–15 As a group, anti-RBPs target RNA–protein complexes, although antibodies bind to the protein and not to the nucleic acid. Although anti-RNP antibodies are frequent in SLE, they are not disease specific and their levels do not obviously change with disease activity.5 16 Therefore, their role in SLE pathogenesis has been less well studied.
The IFN response is usually assessed by analysis of gene expression of peripheral blood cells, which can show an IFN gene signature (IGS).17–20 The stimulation of IFN results from the interaction of DNA or RNA with internal nucleic acid sensors, both toll-like receptor (TLR) and non-TLR, following uptake into cells. These internal receptors are elements of an internal host defence system that can recognise foreign DNA or cytoplasmic DNA from cell stress or damage.21 Recent studies have demonstrated a strong relationship between the IGS and anti-RNP antibodies, suggesting a pathogenic role for these autoantibodies.15 22 23
Anti-dsDNA and anti-RBPs, while both targeting nuclear macromolecules, nevertheless differ in their pattern of expression. Thus, anti-dsDNA levels can vary markedly during the course of disease especially during nephritis, and can decrease and even disappear with therapy.24 During flare, levels of C3 and C4 can decrease, consistent with activation of the complement system by ICs.25–29 In contrast, anti-RBP levels tend to remain relatively static over time, making the relationship of anti-RBPs and disease flares or complement unclear. With the approval of anifrolumab, a monoclonal antibody to the type I IFN receptor, to treat SLE, it is important to understand the drivers of the IGS and the interplay of biomarkers related to IC formation.30–32
In these studies, we investigated the relationship between ANA levels, the IGS and levels of C3 and C4 in patient samples from two clinical trials of tabalumab in SLE, testing the associations of these biomarkers in patients of AA and European Ancestry (EA). As our results indicate, anti-dsDNA levels have an inverse relationship with C3 and C4, whereas anti-RNP levels were not related to depression of complement despite a strong association of anti-RNP with the IGS in both ancestral groups. These findings suggest differences in the properties of ICs formed by anti-dsDNA and anti-RBP antibodies in SLE and the impact of ancestry on serological disease manifestations.
Methods
Patient involvement
Patients were not directly recruited or involved in this study. Rather, patient’s data from previously completed clinical trials were obtained and analysed.33 SLE patients enrolled in the trials had a clinical diagnosis of SLE defined as having ≥4 of the American College of Rheumatology 1997 criteria, positive ANA, and active disease defined as SELENA-SLEDAI ≥6. Exclusion criteria included having active nephritis, active CNS or peripheral neurologic disease, having previously received rituximab, having received IVIg within 180 days of randomisation, and other criteria relating to previous infections and details relating to treatment regimens.34
Clinical information and microarray data from whole blood of patients from two clinical trials of tabalumab, a monoclonal antibody to BAFF, in SLE were downloaded from Gene Expression Omnibus under accession GSE88884.33–36 Additional clinical metadata was provided by Matthew D. Linnik of Eli Lilly & Co. Anti-dsDNA was determined by IgG INOVA QUANTA Lite SC ELISA (INOVA Diagnostics, San Diego, California, USA). Anti-RNP was determined by QUANTA Lite RNP ELISA (INOVA Diagnostics, San Diego, California, USA). Anti-Sm, anti-SSA, anti-SSB and C3 and C4 levels were also determined by ELISA. Gene expression and autoantibody levels, complement levels and all clinical data were analysed using baseline values, before initiation of study drug. Patients with missing data were excluded from the appropriate analyses.
Gene set variation analysis (GSVA)
The GSVA R package was used as a non-parametric, unsupervised gene set enrichment method.37 The inputs for the GSVA algorithm were log2 microarray expression values (Affymetrix Human Transcriptome Array V.2.0) and predefined gene sets describing IFN stimulated gene signatures,15 38–41 TNF38 and interleukin 1 (IL-1) cytokine signatures.15 Probes were filtered out if their IQR was equal to zero. GSVA was conducted on the remaining network. Enrichment scores were calculated using a Kolmogorov-Smirnov (KS)-like random walk statistic to estimate variation of predefined gene sets. The enrichment scores take on values between −1 and 1, where 1 represents enrichment of every gene in a given gene set among the samples analysed compared with every other gene not included in the specified gene set, whereas −1 represents a relative lack of enrichment. Each gene in a gene set is given a rank based on expression values and the KS-like random walk statistic is calculated.
Classification and regression trees (CART)
CART analysis was performed in R V.4.0.4 using the rpart, rpart.plot, and ggplot2 packages.42–44 Regression trees were initially visualised in R and reimaged in GraphPad Prism (V.9.1.0.221). Categorical variables were used as input to the CART algorithm, including autoantibody status of anti-dsDNA, anti-RNP, anti-Sm, anti-SSA and anti-SSB (positive, negative or borderline) and complement C3 and C4 status. C3 levels were considered low if <0.9 g/L and normal if ≥0.9 g/L. C4 levels were considered low if <0.1 g/L and normal if ≥0.1 g/L. A numeric range for C3 and C4 levels considered high was not available. Antibody levels were considered positive if >20 IU/mL, borderline if ≥11 IU/mL and ≤20 IU/mL and negative if <11 IU/mL. GSVA enrichment scores of the core IGS were used as the dependent variable of the estimated regression trees. In every CART analysis, regression trees were pruned once, and then two times but with no observed reduction in cross-validated error, and, therefore the original unpruned trees were decided on as the best estimators of the IGS.
Statistical analyses
Statistical analyses were conducted in GraphPad Prism (V.9.1.0.221) including linear regression. All violin plots, scatterplots and pie charts resulting from these analyses were also rendered in Prism. To compare autoantibody positivity between AA and EA patients, an online calculator (https://wwwsocscistatisticscom/defaultaspx) was used to carry out the χ2 test with Yates correction.
To determine statistical significance between GSVA enrichment scores of two groups, the Mann-Whitney U test was carried out in GraphPad Prism. To determine statistical significance between enrichment scores of three or more groups, the Kruskal-Wallis and Dunn’s multiple comparisons test were performed in GraphPad Prism. Dunn’s multiple comparisons test accounts for the number of comparisons made and adjusted p values were reported.
Results
Female SLE patients from GSE88884 (n=1620) were stratified by the presence of five autoantibodies (online supplemental table 1) to determine the association with gene sets representing various IGS. Among autoantibody-stratified patients, GSVA enrichment scores showed an increase in the IGS in anti-dsDNA positive (anti-dsDNA+) patients compared with patients negative for the five measured autoantibodies (anti-dsDNA− anti-RBP−) (figure 1A). Notably, a significant increase in the IGS was also present in patients positive for anti-RNP (anti-RNP+) antibodies only when compared with the anti-dsDNA-anti-RBP− group and when compared with the anti-dsDNA+ group. This relationship was observed for type I, type II and core IFN signatures shared by both types I and II (online supplemental figure 1). To investigate whether these relationships could have been influenced by antibody titre, we correlated antibody levels with IGS enrichment scores in anti-dsDNA+ anti-RNP− patients and anti-dsDNA− anti-RNP+ patients, respectively, (online supplemental figure 2A). We observed a minimal but significant relationship between anti-dsDNA titre and the IGS (R2=0.05, p<0.0001), whereas any amount of RNP appeared sufficient to be associated with the IGS (p=0.485). Furthermore, anti-dsDNA, anti-SSA and anti-SSB antibody titers were comparable among EA and AA groups, whereas anti-RNP and anti-Sm antibody levels were elevated in AA SLE (online supplemental figure 2B).
Supplemental material
Overall, 1620 active, female SLE patients were stratified by the presence of five autoantibodies. GSVA was carried out on microarray data of these patients using various IGS (A) and cytokine signatures (B). Dunn’s multiple comparisons test was performed to determine significant differences in IGS enrichment among the groups. Numbers of patients (n) in each of the comparator groups are annotated on the x-axis. Violin plots display median values (solid lines) and upper and lower quartiles (dashed lines). *p<0.05; **p<0.01; ***p<0.001; ****p<0.0001.
GSVA enrichment scores of other cytokine-induced signatures were similarly compared among the autoantibody-stratified groups (figure 1B). Whereas, anti-dsDNA+ and anti-RNP+ patients showed similar increases in the TNF signature compared with anti-dsDNA− anti-RBP− patients, a significant increase in the IL-1 signature was only present in anti-dsDNA+ patients.
The relationships between the IGS and complement levels were next evaluated, with linear regression identifying significant inverse relationships between C3 and C4 levels and IGS GSVA scores (figure 2A). For anti-dsDNA+ only patients (n=251), a significant relationship between C3/C4 and IGS GSVA scores was present (figure 2B). In contrast, no relationship between complement levels and the IGS was identified in the anti-RNP+ only patients (figure 2C). Furthermore, linear regression demonstrated significant relationships between increasing levels of anti-dsDNA antibody and decreasing C3 and C4 levels (figure 3A), but no such relationship between complement levels and anti-RNP levels (figure 3B).
Linear regression analyses show relationships between complement levels and enrichment of interferon gene signature (IGS) in systemic lupus erythematosus (SLE) whole blood, where each dot on the scatterplots represents one patient sample. (A) Represents 1620 active, female SLE patients. (B) Represents 251 active, female SLE patients positive for anti-dsDNA and none of the other five autoantibodies measured. (C) Represents 102 active, female SLE patients positive for anti-RNP only and none of the other five autoantibodies measured. Dotted lines represent 95% confidence bands of the best-fit line.
Linear regression analyses show relationships between autoantibody levels and complement levels in systemic lupus erythematosus (SLE) whole blood, where each dot on the scatterplots represents one patient sample. (A) Represents 480 active, female SLE patients positive for anti-dsDNA but negative for anti-RNP antibodies. (B) Represents 220 active, female SLE patients positive for anti-RNP but negative for anti-dsDNA antibodies. Dotted lines represent 95% confidence bands of the best-fit line.
We next stratified the anti-dsDNA+ only patients (figure 4A) and the anti-RNP+ only patients (figure 4B) in terms of low or high/normal C3 levels and by low or high/normal C4 levels; we then examined GSVA scores of the IGS in these groups. Again, significant increases in IGS expression were present in the low complement groups in the anti-dsDNA+ patients; whereas, in the anti-RNP+ patients, there were no significant differences in the IGS.
Gene set variation analysis (GSVA) was carried out on microarray data of systemic lupus erythematosus (SLE) patient whole blood, using various interferon (IFN) gene signatures (IGS). Subjects positive for anti-dsDNA only (A) or anti-RNP only (B) were stratified by the presence of low or normal/high complement C3 and C4 levels as shown. The Mann-Whitney U test was performed to determine significant differences in IGS enrichment between groups. Numbers of patients (n) in each of the comparator groups are annotated on the x-axis. Violin plots display median values (solid lines) and upper and lower quartiles (dashed lines). ****p<0.0001; n/s=not significant.
Since hydroxychloroquine (HCQ) is reported to inhibit TLR7 and TLR9 activity and thus block downstream induction of IFN,45 we also analysed the IGS among patients taking HCQ (or an equivalent antimalarial drug) and those not taking HCQ. For both anti-dDNA+ anti-RNP− and anti-dsDNA− anti-RNP+ patients, no significant differences were observed in GSVA scores of the type I or type II IGS related to HCQ administration (online supplemental figure 3).
To assess the influence of ancestry on these inter-relationships, we then analysed autoantibody stratification among EA and AA patients (online supplemental table 2). As these data indicate, over one-third of EA patients were anti-dsDNA− anti-RBP− and an almost equivalent proportion was positive for anti-dsDNA alone. In contrast, 18.2% of AA patients were anti-dsDNA− anti-RBP−; 14% were positive for only anti-dsDNA and 15.7% were anti-RNP+ only, compared with just 10.2% of EA patients. Overall, 18.2% of AA patients were both anti-dsDNA+ and anti-RNP+, while negative for anti-Sm, anti-SSA and anti-SSB, compared with just 6.19% of EA patients with the same autoantibody profile. Overall, 65.9% of EA patients expressed at least one of the five autoantibodies, whereas 81.8% of AA patients expressed one or more autoantibodies (p=8.08E−4, χ2 with Yates correction). Notably, 34.6% of EA and 67.8% of AA SLE patients were anti-RNP+ (p<1.00E−5, χ2 with Yates correction), whereas 52.2% of EA and 56.2% of AA SLE patients were anti-dsDNA+ (p=0.482, χ2 with Yates correction).
In AA patients, IGS GSVA scores were significantly elevated in those that were anti-RNP+ or anti-RNP+ in combination with other autoantibodies, but not in anti-dsDNA+ only patients (figure 5). In contrast, in EA patients, GSVA scores were elevated in all autoantibody-positive groups compared with anti-dsDNA− anti-RBP− patients.
Overall, 121 African Ancestry (AA) and 630 European Ancestry (EA) active, female systemic lupus erythematosus (SLE) patients were stratified by the presence of five autoantibodies. Gene set variation analysis (GSVA) was carried out on microarray data of these patients using various interferon gene signatures (IGS). Dunn’s multiple comparisons test was performed to determine significant differences in IGS enrichment among the groups. Numbers of patients (n) in each of the comparator groups are annotated on the x-axis. Violin plots display median values (solid lines) and upper and lower quartiles (dashed lines). *p<0.05; **p<0.01; ***p<0.001; ****p<0.0001.
A unique IGS has previously been reported to be related to ancestry rather than disease activity.40 We, therefore, assessed this relationship by GSVA to determine its association with autoantibodies. First, we confirmed its relationship to ancestry, where GSVA enrichment scores were higher among AA SLE patients compared with EA SLE (online supplemental figure 4A). Next, we examined this unique signature in SLE patients stratified by ancestry and autoantibody status (online supplemental figure 4B). Consistent with our previous results using different IGS, the ancestry-specific IGS was significantly enriched in autoantibody-positive subjects compared with anti-dsDNA− anti-RBP− patients, regardless of ancestry. Notably, GSVA enrichment scores of the ancestry-specific IGS were significantly higher in anti-dsDNA+ and anti-RNP+ EA patients and significantly more enriched in the anti-RBP+ group, whereas this signature was significantly enriched only in the anti-RBP+ group but not in anti-dsDNA+ AA SLE. These results indicate that enrichment of this ancestry-specific subset of the IGS follows the pattern of expression of the more global IGS, tracking more with the presence of anti-RNP antibodies, which are more frequent in AA SLE.
In both AA and EA anti-dsDNA+ anti-RNP− patients, we identified significant relationships between increasing anti-dsDNA levels and depression of C3 and C4 by linear regression (figure 6A). Interestingly, the regression coefficients were much stronger in the AA than EA cohort. In contrast, in anti-dsDNA− anti-RNP+ atients, anti-RNP levels did not exhibit relationships with C3 or C4 levels in either ancestral group (figure 6B).
Linear regression analyses show relationships between autoantibody levels and complement levels in African Ancestry (AA) and European Ancestry (EA) systemic lupus erythematosus (SLE) patient whole blood, where each dot on the scatterplots represents one patient sample. (A) Represents 37 AA and 376 EA active, female SLE patients positive for anti-dsDNA but negative for anti-RNP antibodies. (B) Represents 44 AA and 132 EA active, female SLE patients positive for anti-RNP but negative for anti-dsDNA antibodies. Dotted lines represent 95% confidence bands of the best-fit line.
Next, we used CART analysis to determine the most important predictors of the dependent variable, IGS enrichment. Status (positive, negative or borderline) of the five autoantibodies measured and C3 and C4 status (high, normal, low) were input as independent variables. Regression trees were visualised to display the hierarchy of importance of each variable on IGS GSVA Score.
In this analysis, anti-RNP was identified as the strongest predictor of IGS expression in both AA and EA cohorts (figure 7). Anti-dsDNA and anti-SSA/Ro antibodies were also identified as predictors of the IGS, but not anti-Sm or anti-SSB/La. Of note, whereas anti-dsDNA contributed to the IGS in global and EA SLE, it did not contribute to IGS enrichment in AA SLE. C3 was important after anti-RNP positivity in all SLE and in AA cohorts but appeared later in the hierarchy in the EA cohort. Only AA patients depended on C4 status for IGS enrichment secondary to having high or normal C3 levels.
Classification and regression tree (CART) analysis was employed to determine the most important contributors to the core interferon gene signature (IGS) (shared by type I and II interferon (IFN)) as measured by gene set variation analysis (GSVA) enrichment scores. Resultant regression trees are visualised in (A) for all 1589 systemic lupus erythematosus (SLE) patients who had measurements of C3, C4 and all five autoantibodies, (B) 208/1589 SLE patients of African Ancestry (AA) and (C) 1100/1589 SLE patients of European Ancestry (EA). Autoantibody status was determined as follows: positive (<20 IU/mL), borderline (≥11 IU/mL and ≤20 IU/mL) or negative (<11 IU/mL). C3 levels were considered low if <0.9 g/L and normal if ≥0.9 g/L. C4 levels were considered low if <0.1 g/L and normal if ≥0.1 g/L.
Whereas anti-RNP and anti-dsDNA antibodies were associated with the IGS, we were not able to establish directional relationships. We, therefore, assessed autoantibody expression in IGS+ (defined as having an IGS GSVA enrichment score>0) and IGS− (defined as having as IGS GSVA enrichment score<0) patients. Autoantibodies were detected in IGS+ and IGS− samples. In the IGS+ group, 82.41% of patients were anti-dsDNA+ only, anti-RNP+ only or positive for both (online supplemental figure 5A), whereas 54.48% of IGS− patients were also positive for either anti-dsDNA only, anti-RNP only or both.
In contrast, the IGS was uncommon in subjects lacking autoantibodies, with 86.3% of anti-dsDNA− anti-RBP− SLE patients lacking the IGS. We did not detect an association with elevated serum IFN and anticardiolipin antibodies (online supplemental figure 6). There appeared to be a stepwise decrease in the frequency of the IGS related to the number of autoantibodies expressed (online supplemental figure 5B). This finding suggests that patients can produce autoantibodies without the IGS whereas it is less likely that patients with the IGS are negative for autoantibodies.
Finally, we analysed relationships between complement levels, IGS expression and anti-SSA/Ro levels in patients positive for anti-SSA but negative for the other measured autoantibodies (anti-SSA+). Notably, a significant relationship between depressed C4 levels and IGS enrichment was identified (online supplemental figure 7A). Moreover, IGS GSVA scores were increased in anti-SSA+ patients stratified by low C3 and low C4 (online supplemental figure 7B). Despite these significant relationships with the IGS, neither C3 nor C4 levels were related to anti-SSA titre by linear regression (online supplemental figure 7C). Since only three patients were positive for anti-SSB antibodies alone, we were unable to similarly probe these relationships in anti-SSB+ patients further.
Discussion
We employed a large clinical trial database to assess the relationship between autoantibody levels, complement consumption and the IGS. As previously reported, the presence of various autoantibodies differed among ancestral groups with anti-RNP especially prevalent in AA SLE and anti-RNP levels higher in AA SLE. Notably, the IGS was most prominently associated with the presence of anti-RNP antibodies, whereas it was associated with anti-dsDNA antibodies only in EA SLE. The apparent differences between anti-RNP and anti-dsDNA to associate with the IGS might relate to different signalling potentials of the intracellular TLRs engaged by their respective cargos in ICs.21 Indeed, the anti-RNP assays employed used native RNP/Sm as antigens, and, therefore, would be likely to detect autoantibodies that incorporated RNA species into ICs that could engage endosomal TLRs. Alternatively, since anti-RNP can be present in larger amounts,46 it is possible that this translates to a greater mass effect of anti-RNP versus anti-dsDNA ICs. This appears to be less likely, since the association of the IGS and anti-RNP antibodies did not appear to depend on titre, since no correlation was found between the IGS and anti-RNP levels. However, a significant but only a very modest correlation was detected between the anti-dsDNA titre and the IGS. The association of autoantibodies and complement levels was also complex, with anti-dsDNA levels inversely related to levels of C3 and C4, but not anti-RNP levels. Since ICs with anti-dsDNA or anti-RNP can induce IFN production,12 47 these findings suggest that properties of ICs formed by various ANAs are functionally different, with ICs with anti-RNP likely unable to activate complement despite a capacity to induce IFN. In this regard, linear regression identified significant relationships between levels of anti-dsDNA and complement in both ancestral groups; however, regression coefficients were stronger in the AA cohort.
Few prior studies have addressed depression of complement by anti-RBPs because of the focus on anti-dsDNA in pathogenesis and the utility of anti-dsDNA antibodies as markers of disease activity especially in conjunction with complement.48 In contrast to variable expression of anti-dsDNA, levels of antibodies to RBPs tend to remain relatively constant over time and they are therefore not routinely assessed especially with quantitative assays.16 21 In addition, whereas anti-dsDNA antibodies are disease-specific, anti-RNP, anti-SSA/Ro and anti-SSB/La are disease-related, perhaps suggesting a lesser role in lupus pathology.5
Whereas simultaneous expression of multiple ANAs can limit assessment of the relationship between autoantibodies and complement, our study had a sufficiently large number of patients to allow analysis of samples that contained either anti-dsDNA only or anti-RNP only. In this way, we could show that anti-RNP, in contrast with anti-dsDNA, was not associated with depressed levels of either C3 or C4. Furthermore, we could show that, among patients with just anti-RNP, complement levels were not related to the IGS. These findings suggest that complexes formed by anti-RBPs, while associated with the IGS, appear to lack the requisite size or antigen distribution to activate complement. In this regard, previous studies indicated blood in patients with SLE contains ICs comprised of anti-RBPs.49 Furthermore, studies on renal eluates have demonstrated the presence of anti-RBP antibodies.50 It is possible, therefore, that anti-RBPs may form ICs that localise in the kidney, but they may not induce renal inflammation because, unlike anti-dsDNA, they do not activate complement.
While our results provide evidence that anti-dsDNA and anti-RBP antibodies differ in their ability to form ICs that likely activate complement, the basis of this difference is unclear. Previous studies have indicated that anti-RNP and other anti-RBP antibodies can activate complement when tested in in vitro assays such as a complement-fixing immunofluorescence assay (CFANA)51–53; the CFANA is similar to a classical ANA assay except for the use of immunofluorescence reagents to detect bound C3, C4 or properdin. In ANA assays of this kind, chemical fixation may alter the structure or distribution of the target antigens to facilitate antibody binding and complement fixation. Pending more information on the in vivo structure and location of nuclear antigens during disease, we can only speculate that the nature of DNA and RNP (and other RBPs) in patients differs with respect to antigen charge, density or size in ways that affect the binding of antibodies and complement engagement. Finally, the Ig heavy chain isotype of the autoantibodies may contribute to their biologic activity. In this regard, one study indicated that anti-RNP antibodies are primarily of the IgG2 subclass (which activate complement ineffectively),54 whereas other studies have indicated that antibodies to RBPs are similar to anti-dsDNA and are predominantly IgG1, which engage complement efficiently.55
In characterising the impact of ancestry on serology, we showed that a much larger proportion of EA patients were anti-dsDNA− anti-RBP− or anti-dsDNA+ only, whereas AA SLE patients were more likely to be anti-RNP+, anti-dsDNA+ and anti-RNP+, or anti-dsDNA+, anti-RNP+ and anti-Sm+. Indeed, a significantly greater proportion of AA patients were positive for at least one autoantibody than EA patients. A larger proportion of AA patients were also anti-RNP+ compared to EA patients, whereas similar proportions of patients of each ancestry were anti-dsDNA+.
As has been previously reported,15 22 23 we found a significantly stronger association between anti-RNP and the IGS in both AA and EA SLE. In AA, we noted no significant association between anti-dsDNA and the IGS, but a significant association with anti-RNP, whereas in EA SLE the association between the IGS and anti-RNP was significantly greater than that of anti-dsDNA and the IGS. This is consistent with a recent report noting an AA-specific IGS, characterised by greater enrichment in AA SLE patients than EA patients and enrichment occurring with lower frequency and magnitude in autoantibody-stratified EA than AA patients.41 It is notable that we found that this subset of the IGS was more frequent in AA SLE, but it appeared to be more related to the presence of anti-RNP antibodies, which were more common in AA. Of note, CART also predicted anti-RNP antibodies to be the most important contributors of the IGS in AA SLE, with C3 and C4 the next most important contributors, in contrast with anti-dsDNA antibodies as the next most important contributors in EA SLE. These results suggest that there may be ancestral differences in the biologic functions of autoantibodies. Whereas anti-RNP was associated with the IGS in both ancestries, anti-dsDNA was only associated with the IGS in EA SLE.
We found that the measured autoantibodies can largely explain the presence of the IGS, since 86.3% of patients negative for all five autoantibodies lacked the IGS. We also observed a stepwise decrease in IGS enrichment with progression towards autoantibody negativity. Nevertheless, over half of patients expressing the IGS were positive for either anti-dsDNA or anti-RNP antibodies. These data suggest that, whereas the IGS is not required for autoantibody production, the presence of autoantibodies may be necessary for IGS induction. In this regard, we did not find differences in the IGS in patients related to HCQ use in anti-dsDNA+ anti-RNP− or anti-dsDNA− anti-RNP+ populations. These findings suggest either significant non-adherence among patients with antimalarial treatment at baseline or a lack of effect of these drugs on TLR signalling and IFN induction in treated patients. Still, 13.7% of patients lacking the five autoantibodies were IGS+, which may be explained by other mechanisms not involving ICs such as spontaneous MAVS oligomerisation as described by Buskiewicz et al or associations with other autoantibodies that were not measured in this study.56
While our findings involve large patient numbers, they may be limited by study of a clinical trial population. For entry, patients had to have active SLE but not active nephritis and no active CNS involvement; it is possible that other populations would show different biomarker relationships, thus these findings need not necessarily be generalised to the entire SLE population. Nevertheless, our findings indicate heterogeneity of ICs in SLE, thereby affecting the use of complement to infer the presence of ICs that induce IFN.
Data availability statement
Data are available in a public, open access repository. Data are available in a public, open access repository. Data were downloaded from Gene Expression Omnibus (GEO) under accession GSE88884.
Ethics statements
Patient consent for publication
Ethics approval
This study does not involve human participants.
Acknowledgments
The authors would like to acknowledge and thank Mary Artley for her assistance with the preparation of the manuscript. This work was funded by the RILITE Foundation. An abstract (https://acrabstracts.org/abstract/association-between-anti-rnp-antibodies-and-interferon-gene-expression-but-not-complement-consumption-in-sle/) describing this work was accepted for a poster presentation at the 2021 Annual Meeting of the American College of Rheumatology and is available online.
References
Supplementary materials
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Footnotes
Handling editor Josef S Smolen
Contributors ELH analysed the data and wrote the manuscript. DSP and PEL contributed to the conception, design and oversaw conduct of the research, edited the manuscript and made the decision to publish; thus, DSP and PEL were the guarantors.
Funding This work was supported by a grant from the RILITE Foundation.
Competing interests The authors have no competing interests to disclose. ELH and PEL are salaried employees of AMPEL BioSolutions, LLC.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.