Multifaceted immune dysregulation characterizes individuals at-risk for rheumatoid arthritis

Abstract

Molecular markers of autoimmunity, such as antibodies to citrullinated protein antigens (ACPA), are detectable prior to inflammatory arthritis (IA) in rheumatoid arthritis (RA) and may define a state that is 'at-risk' for future RA. Here we present a cross-sectional comparative analysis among three groups that include ACPA positive individuals without IA (At-Risk), ACPA negative individuals and individuals with early, ACPA positive clinical RA (Early RA). Differential methylation analysis among the groups identifies non-specific dysregulation in peripheral B, memory and naïve T cells in At-Risk participants, with more specific immunological pathway abnormalities in Early RA. Tetramer studies show increased abundance of T cells recognizing citrullinated (cit) epitopes in At-Risk participants, including expansion of T cells reactive to citrullinated cartilage intermediate layer protein I (cit-CILP); these T cells have Th1, Th17, and T stem cell memory-like phenotypes. Antibody-antigen array analyses show that antibodies targeting cit-clusterin, cit-fibrinogen and cit-histone H4 are elevated in At-Risk and Early RA participants, with the highest levels of antibodies detected in those with Early RA. These findings indicate that an ACPA positive at-risk state is associated with multifaceted immune dysregulation that may represent a potential opportunity for targeted intervention.

Fig. 1. Methylation signatures of anti-CCP3(−), At-Risk and Early RA in peripheral blood B cells, memory T cells and naive T cells.

A Dimension reduction analysis of samples on all filtered loci confirmed variances of methylation from cell lineage specific methylation in B cell, memory T cell and naїve T cells. B DML and DMG identified among anti-CCP3(−), At-Risk and Early RA in each cell type (Unadjusted p < 0.05, based on two-sided Student’s t-test and differences of average β > 0.1 were used as cutoffs for DML). C PCA of cohort 2 samples and DML on B cell, memory T cell and naїve T cells. D Selected differentially modified pathways identified in each cell lineage among anti-CCP3(−), At-Risk and Early RA samples. E Combined results of one-vs-one predictive models separates anti-CCP3(−) control, At-Risk and Early RA samples. DML differentially methylated locus, DMG differentially methylated genes, PCA principal component analysis. Source data are provided as a Source Data file.

Fig. 2. The frequency of cit-reactive T cells is increased in At-Risk subjects.

Samples from the subset of Anti-CCP3(−), At-Risk, and Early RA participants who were DRB1*04:01 positive (30, 24, and 17 subjects, respectively) were stained with HLA class II tetramers to enumerate CD4 + T cells specific for citrullinated (cit) aggrecan, cartilage intermediate layer protein (CILP), vimentin and fibrinogen, or α-enolase. A The combined frequency of cit-reactive T cells was significantly higher in At-Risk participants compared to Anti-CCP3(−) controls (p = 0.035), but was not significantly different between At-Risk and Early RA participants. B Considering individual antigen specificities, the frequency of CILP reactive T cells was significantly higher for At-Risk subjects compared to Anti-CCP3(−) Controls and Early RA participants (p = 0.042). Only modest differences were seen for other antigens, indicating that the increase in cit-specific T cells in At-Risk subjects was driven by CILP reactive T cells. C The frequency of influenza (flu) reactive T cells did not differ between Anti-CCP3(−) Controls, At-Risk, and Early RA participants. Error bars indicate standard deviation. P-values indicate unpaired comparisons using Wilcoxon’s nonparametric two-tailed test. Source data are provided as a Source Data file.

Fig. 3. Examining the phenotypes of CILP specific CD4 + T cells.

A Aligned clusters within the CD4 + T cell landscape were classified into T cell surface phenotype groups based on six surface markers (CD45RA, CD38, CCR7, CXCR3, CCR4, and CCD6) using the DISCOV-R approach and assigned identities as elaborated in Supplementary Fig. 3. B For At-Risk participants, a significantly higher proportion of cartilage intermediate layer protein (CILP) reactive T cells resided within AC4 as compared with Anti-CCP3(−) Controls (p = 0.020). Early RA participants also tended to have a lower proportion of CILP reactive T cells within AC4 (p = 0.072). C For Early RA participants a significantly lower proportion of aggrecan reactive T cells resided within AC7 as compared with Anti-CCP3(−) Controls (p = 0.008). Early RA participants also tended to have a higher proportion of aggrecan reactive T cells within AC7 (p = 0.06). Those with at least 8 total antigen specific events are shown. For CILP graph anti-CCP3(−) N = 15, at-risk N = 16, and early-RA N = 12. For aggrecan graph anti-CCP3(−) N = 22, at-risk N = 19, early-RA N = 11. Error bars indicate standard deviation. P-values indicate unpaired comparisons using Wilcoxon’s nonparametric two-tailed test. Source data are provided as a Source Data file.