Fig 2: CD8+ T cells in ACPA+ RA blood proliferate in response to citrullinated antigens.a, b PBMCs (5 × 105) of ACPA+ RA patients were incubated with anti-CD3/28 antibodies (n = 14), NP (influenza A)/ pp65 (CMV) protein (50 μM of each, n = 14), native vimentin protein (100 μM, n = 5), or citrullinated vimentin (cit-vimentin) protein (100 μM, n = 14) with or without anti-CD8 and/or HLA class I-blocking antibody (anti-CD8 Ab n = 6, anti-HLA class I Ab n = 6 or anti-CD8/ HLA class I Abs n = 10), or no stimulation (n = 15) for 3 days. The percentage of Ki-67-expressing CD8+ (a) or CD4+ (b) T cells in ACPA+ RA PBMCs was measured by flow cytometry. For a, ***P < 0.0001, **P = 0.026, **P = 0.0021, #P = 0.0471, #P = 0.026, ##P = 0.0056 or ##P = 0.0021. For b, ***P < 0.0001, or *P = 0.0383. c, d Native vimentin or cit–vimentin pulsed monocyte-derived dendritic cells (MoDCs) were cocultured with Cell Proliferation Dye eF450-labeled CD3+ T cells isolated from ACPA+ RA PBMCs (n = 9) with or without anti-CD8/HLA class I-blocking antibody for 10 days, followed by flow cytometry analysis. Representative flow cytometry results (c, left) and quantification of the percentage of dyelow proliferating CD8+ T cells (c, right) or CD4+ T cells (d). For c, ***P < 0.0001, **P = 0.0054, **P = 0.0015, #P = 0.0494 or #P = 0.0251. For d, ***P < 0.0001, *P = 0.0171, or #P = 0.0484. e Proliferation capacity of ACPA+ RA CD8+ T cells based on proliferation dye eF450-labeled CD8+ T cells co-cultured with cit-vimentin or native vimentin-loaded MoDCs in the absence or presence of anti-CD8/HLA class I-blocking antibody for 10 days (n = 4). **P = 0.0011, **P = 0.0027, #P = 0.0486, or ###P = 0.0007. Representative histograms (left) and quantification of the proliferating CD8+ T cells (right). Bars represent means ± SEM. *P < 0.05, **P < 0.01, or ***P < 0.001 versus no treatment by unpaired t-test with two-tailed test; and #P < 0.05, ##P < 0.01, or ###P < 0.001 versus cit-vimentin treatment by unpaired t-test with two-tailed test. Source data are provided as a Source Data file. HC healthy control, ACPA anti-citrullinated protein antibodies, RA rheumatoid arthritis, Cit-Vim citrullinated vimentin.

Fig 3: ACPA+ RA blood CD8+ T cells upregulate activation and cytotoxic markers in response to citrullinated antigens.a Stimulation of fresh blood from ACPA+ RA (n = 8, red bars) or HCs (n = 7, blue bars) with anti-CD3/28 Abs, NP/pp65, cit- or native vimentin for 6 h with addition of Golgi inhibitor for the last 5 h. Quantification of CD69+ and GzmB+IFNγ+ CD8+ T cells, and IFNγ+ CD4+ T cells. For CD69+CD8+, *P = 0.049, +P = 0.0397, **P = 0.001, ##P = 0.009, or ###P = 0.0008. For GzmB+IFNγ+CD8+, +P = 0.0175, **P = 0.0071, ##P = 0.0071, or ##P = 0.0037. For IFNγ+CD4+, ++P = 0.0065, +P = 0.0489, *P = 0.0155, #P = 0.0351, or ##P = 0.0038. b PBMCs of ACPA+ RA patients (n = 10) were incubated with anti-CD3/28 Abs, NP/pp65 (50 μM of each), individual citrullinated proteins (100 μM), all citrullinated proteins (20 μM each), or all native proteins (20 μM each) for 16 h. Percentage of IFNγ or Granzyme B (GzmB) expressing CD8+ T cells measured by intracellular staining. c Quantification of IFNγ+ or GzmB+IFNγ+ expressing CD8+ T cells in ACPA+ RA PBMCs treated with cit- or native vimentin in a dose-dependent manner (1, 10 or 100 μM) (n = 7). d Quantitative analysis of IFNγ+ or GzmB+IFNγ+ expressing CD8+ T cells in ACPA+ RA PBMCs stimulated with cit-vimentin in the presence or absence of anti-CD8 and/or HLA class I-blocking antibody (n = 10). Data are presented as means ± SEM. For a, two-tailed unpaired t-test: *P < 0.05 or **P < 0.01 versus no treatment in ACPA+ RA; +P < 0.05, or ++P < 0.01 versus no treatment in HC; and #P < 0.05, ##P < 0.01, or ###P < 0.001 versus cit-vimentin in ACPA+ RA. For b, two-tailed unpaired t-test: *P < 0.05, **P < 0.01, or ***P < 0.001 versus no treatment; #P < 0.05, or ##P < 0.01 versus all native proteins. For c, two-tailed unpaired t-test: *P < 0.05, or **P < 0.01 versus no treatment; #P < 0.05 for cit-vimentin 1 μM versus vimentin 1 μM; $P < 0.05 for cit-vimentin 10 μM versus native vimentin 10 μM by two-tailed unpaired t-test; +P < 0.05 for cit-vimentin 100 μM versus native vimentin 100 μM. For d, by two-tailed unpaired t-test: *P < 0.05, or **P < 0.01 versus no treatment; #P < 0.05, or ##P < 0.01 versus cit-vimentin 100 μM. Source data are provided as a Source Data file. HC healthy control, ACPA anti-citrullinated protein antibodies, RA rheumatoid arthritis, Cit-Vim citrullinated vimentin.

Fig 4: The Toll-like receptor 4 serves as a ligand anchor for eVim and CitVim on the surface of human neutrophils. a Images of human neutrophils showing eVim and CitVim colocalization with TLR4. In control (CT) and treated (+ 1 µg/mL for 1 h) groups, colocalization areas are highlighted in orange. The scale bar, ~ 10 μm. b Quantitative analysis of eVim/CitVim colocalization with TLR4 at the neutrophil membrane. n = 40, 44, 31, and 49 cells for CT (eVim), + eVim at 1 µg/mL, CT (CitVim), and + CitVim (1 µg/mL) groups, respectively. Each symbol represents one cell (n = 3). c Dot blot assay assessed eVim/CitVim interactions with TLR4, quantified by a bar plot correlating interaction intensities with immobilized protein amounts (n = 3). d Functionalization of AFM tips with eVim and CitVim, for interaction studies. e A box plot demonstrates the specific binding probabilities between eVim/CitVim and TLR4-immobilized mica. Each data point on a force map reflects the percentage of adhesive force curves (n = 43). f Rupture forces between eVim or CitVim and TLR4 on the mica surface. Each data point corresponds to the rupture force detected for Vim (n = 590) and CitVim (n = 663) derived from 43 force-distance maps. g Reactive oxygen species (ROS) release measured after exposing human neutrophils to varying concentrations of eVim, CitVim, or LPS with and without TLR4 inhibition (TLR-4 antibody). ROS levels were normalized to untreated conditions (n = 3). h The presumed mechanism of action of eVim/CitVim on TLR4. i Mean fluorescence intensity of GFP-NF-kB reporter in HEK cells expressing TLR4 (TLR4+) and expressing (MD2+) or not MD2 (MD2-). Cells were untreated or exposed for 24 h to eVim, (10 µg/mL), CitVim (10 µg/mL), and LPS (10 ng/mL). Data are presented as the mean ± standard deviation of the mean. *, #, P ≤ 0.05; **, ##, P < 0.01; ***, ##, P < 0.001. A hash sign (#) indicates a statistical significance comparison between corresponding eVim/CitVim and LPS compared to conditions with anti-TLR4 antibodies (g). Significance was determined by one-way ANOVA with Tukey’s test or Student’s t-test (e)