The research was performed in accordance with the Helsinki Declaration and approved by the Oklahoma Medical Research Foundation Institutional Review Board. Banked serum samples and clinical data from SLE patients seen between May 2002 and October 2014 were obtained from the Oklahoma Rheumatic Disease Research Core Center (ORDRCC). The patients who met ≥4 of the 1997 modified American College of Rheumatology Classification Criteria for SLE (17, 18) were evaluated for disease activity and serum autoantibody profiles. The demographics of the patients (n=303) in this study are shown in Supplementary Table 1 . Serum autoantibodies were measured using multiplex fluorescent bead-based assays. The antigens studied were dsDNA, chromatin, Ro/SSA, La/SSB, Sm, smRNP, RNP, RNP-A, RNP-68, Centromere B, Scl-70, and Ribosomal P. The antibody levels were quantified based on the fluorescence intensity for each specificity. The positive cut-off for the anti-dsDNA was set at 10 IU/mL (range 0- >300) and for all other specificities was 1.2 IU/mL (range 0- >8) per manufacturer’s recommendations. Clinical assessments of SLE were performed using the hybrid SELENA- SLE Disease Activity Index (SLEDAI) (19) and the British Isles Lupus Assessment Group (BILAG-2004) Index (20). Serum samples from de-identified healthy volunteers (n=66) were studied for antibodies to E. gallinarum, E. faecalis, and human RNA.

Enterococcus gallinarum (ATCC#BAA-748) and Enterococcus faecalis (ATCC#19433, Type strain) were obtained from the American Type Culture Collection (ATCC, Manassas, VA). These strains are utilized extensively as control/reference strains, and their use will allow for comparisons with studies performed by other investigators in future. The bacteria were cultured in Brain Heart Infusion broth, harvested, washed extensively with PBS, and stored as pellets in single-use aliquots at -80°C. An ELISA-based assay was used to measure antibodies to formalin-fixed whole bacteria as previously described (21). All sera from SLE patients and healthy controls were tested at a 1:500 dilution for anti-bacterial IgG and 1:100 dilution for anti-E. gallinarum IgA antibody titers. Serial dilutions from a pooled serum sample were included in each assay as a calibrator. A standard curve was constructed, and the titers of anti-bacterial antibody were calculated for each sample and expressed as units/mL.

Human RNA was purified from THP1 (ATCC#TIB-202), a human monocytic cell line, propagated in RPMI-1640 with 10% bovine calf serum. RNA was extracted from THP1 cells using the RNeasy Mini Kit (Qiagen, Germantown, MD). Genomic DNA contaminants in the human RNA were eliminated by RNase-free DNase1 digestion using manufacturer’s protocols (Qiagen, Germantown, MD), followed by purification using RNeasy Mini columns (Qiagen, Germantown, MD).

Synthetic double-stranded RNA (poly I:C) HMW was purchased from Invivogen (San Diego, CA).

IgG antibodies to RNA [human RNA, and poly (I:C)] were measured using an ELISA. RNA (5μg/mL) dissolved in PBS with 1 mM EDTA was coated on DNA-BIND ELISA plates (Corning, Glendale, AZ) overnight at 4°C. After blocking, the plates were incubated with serum samples (1:100 dilution) for 2 hours. Bound antibodies were detected with HRP-conjugated goat anti-human IgG (Southern Biotechnology, Birmingham, AL) and enzyme activity determined by tetramethylbenzidine substrate (Bio-Rad Laboratories, Hercules, CA). The reaction was stopped with 2.5N sulfuric acid, and the absorbance was read at 450nm.

Graph Pad Prism 9.0 software (GraphPad Software, San Diego, CA) was used for statistical analyses. Anti-bacterial antibody titers were log₁₀ transformed. Normality tests were performed on each dataset, and non-parametric tests were used for non-Gaussian distributions. Antibody titers between two groups were compared using a t-test for normal distributions or Mann-Whitney test for non-Gaussian distributions. Antibody titers between multiple groups were compared using a one-way ANOVA test, and Sidak’s multiple comparisons post-test determined adjusted p values. For non-Gaussian distributions, antibody reactivity in multiple groups was compared by the Kruskal-Wallis test, followed by Dunn’s multiple comparison post-test. Correlations were determined by Pearson’s method for normal distributions and Spearman’s method for non-Gaussian distributions. Proportions were compared by the Chi-square test. A p-value of <0.05 was considered significant. Post-hoc power calculations were performed using https://epitools.ausvet.com.au/.

IgG antibody titers to formalin-fixed whole E. gallinarum bacteria were measured in sera from lupus patients (n=303) and healthy donors (n=66). Anti-Eg IgG were detected in all the sera tested, and the titers were not significantly different between the two groups ( Figure 1A ). The anti-Eg IgG titers between SLE patients based on self-reported race/ethnicities were also not different ( Supplementary Figure 1A ). Since E. gallinarum is associated with the gut mucosa, serum IgA antibody titers were also measured. No significant differences were seen in anti-Eg IgA titers between SLE patients and healthy donors or between patients in different racial/ethnic groups ( Figure 1B and Supplementary Figure 1B ). Anti-Eg IgG or anti-Eg IgA titers were not different between male and female patients (data not shown). No correlation was noted between age and anti-Eg IgG titers. However, anti-Eg IgA titers showed a statistically significant inverse correlation with age (Spearman r= -0.1941; p=0.0013). The anti-Eg IgG and IgA titers in the SLE patients showed a statistically significant, albeit modest, correlation ( Figure 1C ). The finding that anti-Eg IgG and IgA titers are not different suggests a comparable exposure to E. gallinarum in all groups.

Patients were stratified into groups based on the presence or absence of autoantibodies to different lupus-associated antigens. The anti-Eg titers between each autoantibody-positive and -negative group were compared ( Table 1 ). As shown in Figure 2 , higher anti-Eg IgG titers were associated with antibodies to Ribosomal P (p=0.0059), dsDNA (p=0.0093), and Sm (p=0.0315).

A comparison between patients positive for anti-Ribosomal P, anti-dsDNA, or anti-Sm antibodies with healthy controls showed that anti-Ribosomal P reactivity in patients was consistently associated with higher anti-Eg IgG titers (adjusted p value = 0.0178; Figure 2B ). Compared to healthy controls, higher anti-Eg IgG was also seen in patients with anti-dsDNA or anti-Sm following pair-wise analyses ( Supplementary Table 2 ).

Statistical significance was not reached in comparisons of anti-Eg IgG titers between the other autoantibody-positive and -negative groups ( Supplementary Figure 2 ) or between autoantibody positive patients and healthy controls (data not shown).

SLE patient categorization based on disease activity measures, including SLEDAI scores or BILAG indices, or clinical subsets failed to correlate with anti-Eg IgG titers. Similarly, the anti-Eg IgA titers failed to show association with the presence or absence of autoantibody specificity ( Supplementary Figure 3 ), disease activity measures or clinical subsets (data not shown).

To determine whether exposure to other Enterococci also shows associations with lupus autoantibodies, we measured IgG antibodies to E. faecalis, a commensal bacterium represented in the gut microbiome. Anti-E. faecalis IgG titers were not significantly different between healthy donors and SLE patients ( Supplementary Figure 4A ). Further, anti-Eg, and anti-E. faecalis IgG titers in patients showed a significant correlation ( Supplementary Figure 4B ), suggesting comparable exposure to the immune system and the possibility of cross-reactive antibodies.

Further analysis showed that anti-E. faecalis IgG titers were significantly higher in patients positive for antibodies to dsDNA, Sm, chromatin, and RNP autoantigens. However, the anti-E. faecalis titers between anti-Ribosomal P positive and negative patients failed to reach statistical significance ( Table 1 ). A post hoc analysis showed that in this experiment, sample sizes gave >80% power to detect a significant difference in a two-tailed statistical test with a confidence level of 0.95. Thus, the negative result was likely not due to insufficient power, suggesting that the anti-Ribosomal P positivity and anti-Eg association is specific compared to E. faecalis-directed immune responses.

We have previously reported that in SLE patients, high titer antibodies to pathogenic periodontal but not commensal bacteria are associated with increased disease activity indices (13). Streptococcus gordonii is a gram-positive commensal bacterium present in dental plaque and also found in the gut mucosa. Anti-S. gordonii IgG titers failed to show significant associations with any of the lupus autoantibody specificities ( Table 1 ).

A close association was reported between anti-Ribosomal P and anti-dsDNA in SLE patients (22, 23) and is replicated in our SLE patients (65% of anti-Ribosomal P positive patients are also anti-dsDNA positive). However, since ribosomes are closely bound to RNA, we postulated that the lack of immunoregulation in SLE patients would favor the presence of antibodies to human RNA in anti-Ribosomal P positive patients. To test this hypothesis, we purified RNA from a human monocytic cell line as a substrate to measure anti-human RNA in SLE patients who were Ribosomal P antibody positive (n=26) or randomly selected Ribosomal P negative (n=33). Patients positive for anti-Ribosomal P had higher anti-human RNA titers than anti-Ribosomal P negative patients ( Figure 3A ). Further, anti-human RNA titers in anti-Ribosomal P positive patients showed a modest but significant correlation with anti-Eg IgG antibody (Spearman r= 0.422, p=0.0319) ( Figure 3B ).

Anti-RNA antibodies in SLE patients also react with viral dsRNA and synthetic dsRNA (24). To investigate whether anti-RNA reactivity was skewed by RNA binding protein contaminants co-purified in the human RNA preparation, the same sera were screened for antibodies to synthetic dsRNA (poly I:C) coated on an ELISA plate. Anti-dsRNA reactivity was higher in Ribosomal P antibody-positive patients ( Supplementary Figure 5A ). The anti-human RNA and anti-dsRNA titers showed a strong correlation, Spearman r=0.782, p=2.57x10^-13, n=59 ( Supplementary Figure 5B ), thereby confirming the reactivity to the nucleotide backbone.

Higher anti-RNA antibody titers are associated with higher disease activity ( Supplementary Table 3 ) and a diversified autoantibody repertoire. Therefore, the association of anti-human RNA with anti-Eg titers might not be unique to Ribosomal P positivity. To investigate whether other autoantibody specificities also showed a similar relationship, patients were stratified into autoantibody-positive and -negative groups, and the correlation between anti-Eg and anti-human RNA titers in each group was studied ( Supplementary Table 4 ). In addition to anti-Ribosomal P, anti-human RNA titers were also higher in patients positive for anti-dsDNA ( Figure 3C ). Further, anti-Eg IgG also showed modest but significant correlations with anti-human RNA titers in anti-dsDNA positive patients (r=0.492, p=0.0146). ( Figure 3D and Supplementary Table 4 ).