Single cell sequencing identifies clonally expanded synovial CD4+ TPH cells expressing GPR56 in rheumatoid arthritis

Abstract

Rheumatoid arthritis (RA) is an autoimmune disease affecting synovial joints where different CD4⁺ T cell subsets may contribute to pathology. Here, we perform single cell sequencing on synovial CD4⁺ T cells from anti-citrullinated protein antibodies (ACPA)+ and ACPA- RA patients and identify two peripheral helper T cell (T_PH) states and a cytotoxic CD4⁺ T cell subset. We show that the adhesion G-protein coupled receptor 56 (GPR56) delineates synovial CXCL13^high T_PH CD4⁺ T cells expressing LAG-3 and the tissue-resident memory receptors CXCR6 and CD69. In ACPA- SF, T_PH cells display lower levels of GPR56 and LAG-3. Further, most expanded T cell clones in the joint are within CXCL13^high T_PH CD4⁺ T cells. Finally, RNA-velocity analyses suggest a common differentiation pathway between the two T_PH clusters and effector CD4⁺ T cells. Our study provides comprehensive immunoprofiling of the synovial CD4⁺ T cell subsets in ACPA+ and ACPA- RA.

Fig. 1. Cytotoxic CD4⁺ T-cell frequency in synovial fluid of ACPA− and ACPA+ RA patients.

a Representative flow cytometry dot plot staining of effector molecules, receptors and transcription factors associated with cytotoxic functions in CD4⁺ T cells from synovial fluid (SF) from ACPA− (upper panel) and ACPA+ (lower panel) RA patients, quantified in b, (ACPA−, n = 7–9) (ACPA+ , n = 10–12). Line represents median, two-tailed Mann–Whitney U test, P = 0.0072 (%GZMB⁺PRF1⁺), ns (%GZMA), P = 0.0018 (%Hobit), ns (%Eomes), P = 0.0036 (%NKG7-high), P = 0.0007 (%GPR56), ns: not significant. c Correlation between the frequency of GZMB⁺ PRF1⁺ in CD4⁺ T cells in SF and the level of serum anti-CCP (cyclic citrullinated peptide) antibodies, n = 21, Spearman two-tailed test, P = 0.0070. d Correlation between the frequency of GPR56 in CD4⁺ T cells in SF and the level of anti-CCP (cyclic citrullinated peptide) antibodies, n = 21, Spearman two-tailed test, P < 0.0001. a–d Data are from a pool of nine independent experiments where a circle is a single replicate. Blue dots indicate ACPA− RA SF and red dots indicate ACPA+ RA SF.

Fig. 2. Single-cell RNA sequencing of CD4⁺ T cells from SF and PB of RA patients.

a Technical workflow including CD4⁺ T-cell enrichment and 10X 5′ single-cell sequencing coupled to TCRαβ sequencing on 15 paired PB and SF from RA patients (11 ACPA+, 4 ACPA−). b UMAP displaying 12 CD4⁺ T-cell clusters in combined SF and PB (upper panel, n = 15 RA (11 ACPA+, 4 ACPA−), n = 166,944 cells), SF (middle panel, n = 8 RA (4 ACPA+, 4 ACPA−), n = 31,089 cells) and PB (lower panel, n = 8 RA (4 ACPA+, 4 ACPA−), n = 48,167 cells). c Heatmap showing selected differentially-expressed genes (DEGs) in the different CD4⁺ T-cell clusters in combined PB and SF, P < 0.01, model-based analysis of single-cell transcriptomics (MAST). d Frequencies of CD4⁺ T-cell clusters in PB and SF from n = 8 RA patients (4 ACPA+, 4 ACPA−). Circle indicates ACPA+ RA patients, triangle indicates ACPA− patients. Line represents median, two-tailed Mann–Whitney U test.

Fig. 3. GPR56 expression on peripheral helper CD4⁺ T cells in ACPA+ and ACPA− RA SF.

a 2-D dot plots showing the expression of selected genes in the different CD4⁺ T-cell clusters in ACPA+ and ACPA− SF (circle size indicates the percentage of cells expressing, color intensity indicates average expression) (n = 4 ACPA+, n = 4 ACPA− RA SF). b Representative flow cytometry dot plot of GPR56 expression within PD-1^high (T_PH) and PD-1⁻ (non-T_PH) CD4⁺ T cells in ACPA+ (upper panel) and ACPA− (lower panel) RA SF, quantified in n = 9 ACPA− SF and n = 12 ACPA+ SF, P = 0.0006 (%PD-1^high), P < 0.0001 (%GPR56 in PD-1⁻ versus PD-1^high in ACPA+), P < 0.0001 (%GPR56 in PD-1⁻ versus PD-1^high in ACPA−), P = 0.0007 (%GPR56 in PD-1^high in ACPA− versus ACPA+). c Left panel, representative flow cytometry dot plot of the two T_PH states (PD-1^highGPR56^low and PD-1^highGPR56⁺) and quantification of their frequency in SF, ns (%PD-1^highGPR56^low), P = 0.0005 (%PD-1^highGPR56⁺) in ACPA+ versus ACPA−. Middle panel, MHC-II MFI: P < 0.0001 (PD-1⁻GPR56⁻ versus PD-1^highGPR56^low), P < 0.0001 (PD-1⁻GPR56⁻ versus PD-1^highGPR56⁺) in ACPA− SF; P = 0.0001 (PD-1⁻GPR56⁻ versus PD-1^highGPR56^low), P < 0.0001 (PD-1⁻GPR56⁻ versus PD-1^highGPR56⁺), P = 0.0192 (PD-1^highGPR56^low versus PD-1^highGPR56⁺) in ACPA+ SF; P = 0.0381 (PD-1^highGPR56^low in ACPA− versus ACPA+), ns (PD-1^highGPR56⁺ in ACPA− versus ACPA+). Right panel, PD-1 MFI: P < 0.0001 (PD-1⁻GPR56⁻ versus PD-1^highGPR56^low), P < 0.0001 (PD-1⁻GPR56⁻ versus PD-1^highGPR56⁺) in ACPA− SF; P < 0.0001 (PD-1⁻GPR56⁻ versus PD-1^highGPR56^low), P < 0.0001 (PD-1⁻GPR56⁻ versus PD-1^highGPR56⁺), P = 0.0059 (PD-1^highGPR56^low versus PD-1^highGPR56⁺) in ACPA+ SF; P = 0.0491 (PD-1^highGPR56^low in ACPA− versus ACPA+), P = 0.0056 (PD-1^highGPR56⁺ in ACPA− versus ACPA+). (n = 9 ACPA−, n = 11 ACPA+, MHC-II) (n = 9 ACPA−, n = 12 ACPA+, PD-1). d Representative flow cytometry dot plots of CXCL13 and BLIMP-1 expression within PD-1⁻GPR56⁻ (non-T_PH) and PD-1^highGPR56^low and PD-1^highGPR56⁺ (2 T_PH states) in CD4⁺ T cells in ACPA+ and ACPA− RA SF, quantified in e in n = 5 ACPA− SF and n = 5 ACPA+ SF. Upper panel, %CXCL13: P = 0.0079 (PD-1⁻GPR56⁻ versus PD-1^highGPR56^low), P = 0.0079 (PD-1⁻GPR56⁻ versus PD-1^highGPR56⁺), P = 0.0079 (PD-1^highGPR56^low versus PD-1^highGPR56⁺) in ACPA− SF; P = 0.0079 (PD-1⁻GPR56⁻ versus PD-1^highGPR56^low), P = 0.0079 (PD-1⁻GPR56⁻ versus PD-1^highGPR56⁺), P = 0.0317 (PD-1^highGPR56^low versus PD-1^highGPR56⁺) in ACPA+ SF; ns (PD-1^highGPR56⁺ in ACPA− versus ACPA+). Lower panel, %BLIMP-1: P = 0.0159 (PD-1⁻GPR56⁻ versus PD-1^highGPR56^low), P = 0.0079 (PD-1⁻GPR56⁻ versus PD-1^highGPR56⁺), ns (PD-1^highGPR56^low versus PD-1^highGPR56⁺) in ACPA− SF; P = 0.0079 (PD-1⁻GPR56⁻ versus PD-1^highGPR56^low), P = 0.0079 (PD-1⁻GPR56⁻ versus PD-1^highGPR56⁺), P = 0.0079 (PD-1^highGPR56^low versus PD-1^highGPR56⁺) in ACPA+ SF; ns (PD-1^highGPR56⁺ in ACPA− versus ACPA+). b, c Data are from a pool of nine independent experiments where a circle is a single replicate. e Data are from a pool of five independent experiments where a circle is a single replicate. b, c, e Line represents median, two-tailed Mann–Whitney U test. Blue dots indicate ACPA− RA SF and red dots indicate ACPA+ RA SF. ns not significant, MFI mean fluorescence intensity.

Fig. 4. Tissue-resident memory receptors on CD4+ T cells in ACPA+ and ACPA− RA SF.

a 2-D dot plots showing the expression of selected genes in the different CD4⁺ T-cell clusters in ACPA+ and ACPA− SF (circle size indicates the percentage of cells expressing, color intensity indicates average expression) (n = 4 ACPA+, n = 4 ACPA− RA SF). b Representative flow cytometry dot plots showing the expression of LAG-3, CXCR6, CD69, CD49a, and CX3CR1 on GPR56⁺ and GPR56⁻ CD4⁺ T cells in ACPA+ RA, quantified in ACPA+ and ACPA− SF in c) P < 0.0001 (%LAG-3), P = 0.0052 (%CXCR6), P < 0.0001 (%CD69), P < 0.0001 (%CD49a), P = 0.0379 (%CX3CR1) in GPR56⁻ versus GPR56⁺ in ACPA+ SF; P = 0.0002 (%LAG-3), P = 0.0006 (%CXCR6), P = 0.0078 (%CD69), ns (%CD49a), P = 0.0262 (%CX3CR1) in GPR56⁻ versus GPR56⁺ in ACPA− SF; P < 0.0001 (%LAG-3) in GPR56⁺ in ACPA− versus ACPA+; P = 0.0052 (MFI LAG-3), P = 0.0007 (MFI CXCR6), P = 0.0002 (MFI CD69), P = 0.0196 (MFI CD49a) in GPR56⁻ versus GPR56⁺ in ACPA+ SF; ns (MFI LAG-3), P = 0.0002 (MFI CXCR6), P = 0.0002 (MFI CD69), ns (MFI CD49a) in GPR56⁻ versus GPR56⁺ in ACPA − SF; (n = 10 ACPA+, n = 8 ACPA−, LAG-3, CXCR6, CD69, CD49a) (n = 8 ACPA+, n = 7 ACPA−, CX3CR1). White dots indicate GPR56⁻ CD4⁺ T cells and black dots indicate GPR56⁺ CD4⁺ T cells. d Expression of LAG-3, CXCR6, CD69, CD49a and CX3CR1 on CD4⁺ T cells in SF; P = 0.0005 (%LAG-3 in ACPA− versus ACPA+); (n = 8 ACPA−, n = 10 ACPA+, LAG-3, CXCR6, CD69, CD49a) and (n = 7 ACPA−, n = 8 ACPA+, CXC3CR1). Blue dots indicate ACPA− RA SF and red dots indicate ACPA+ RA SF. c, d Line represents median, two-tailed Mann–Whitney U test. Data are from a pool of eight independent experiments where a circle is a single replicate. e Correlation between the frequency of LAG-3 in CD4⁺ T cells in RA SF and the levels of anti-CCP (cyclic citrullinated peptide) antibodies (n = 18), P < 0.0001, Spearman two-tailed test. ns not significant.

Fig. 5. Clonally expanded CD4⁺ T cells in RA.

a UMAP plots showing expanded clones in ACPA+ (n = 4) and ACPA− (n = 4) RA PB and SF, colored based on sample size. Orange dots indicate small clones (1e-4 < X ≤0.001), purple dots indicate medium size clones (0.001 < X ≤0.01), green dots indicate large size clones (0.01 < X ≤0.1). Each dot displays a cell. b Stacked barplots displaying the phenotype of the SF (left) and PB (right) expanded clones (n ≥ 2 cells) in each RA patient, quadrant represents an individual clone, total number of clones per patient are indicated above each column. c Frequency of expanded clones within each CD4⁺ T-cell cluster in SF (left panel) and PB (right panel) for each RA patient (n = 4 ACPA+, 3 ACPA−). d Frequency of FOXP3⁺ cells among CD4⁺ T cells in ACPA− (n = 9) and ACPA⁺ (n = 12) RA SF, P = 0.027. Data are from a pool of nine independent experiments where a circle is a single replicate. Blue dots indicate ACPA− RA SF and red dots indicate ACPA+ RA SF. Line represents median, two-tailed Mann–Whitney U test.

Fig. 6. CD4⁺ T-cell subsets interconnectivity.

a Chord diagrams showing the interconnectivity between CD4⁺ T cells sharing the same CDR3 (paired TCRα, and TCRβ chains) amino acid sequences within CD4⁺ T-cell clusters in RA SF (left panel) and between compartments in selected clusters (right panel) (n = 4 ACPA−, n = 4 ACPA+). b Jaccard overlap quantifications for clonal overlap between cell clusters across compartments (n = 4 ACPA−, n = 4 ACPA+). c Percentage of shared expanded clones (n ≥ 2 cells) shared with selected subsets among total cell clones, n = 7 RA patients (n = 4 ACPA+, 3 ACPA−). d Representative flow cytometry dot plots showing the expression of PD-1, GPR56, CXCL13, BLIMP-1, and PRF1 in control (upper panel) or after 48 h CD3/CD28-beads activation of T_PH cell states (lower panel) (PD-1^highGPR56^low and PD-1^highGPR56⁺ from ACPA+ SF), quantified in n = 3 ACPA+ RA patients. d Data are from a pool of three independent experiments where a circle is a single replicate. Line indicates median, two-tailed Wilcoxon paired test, P = 0.0005 (MFI PD-1 on PD-1^highGPR56^low after activation), P = 0.0740 (MFI PD-1 on PD-1^highGPR56⁺ after activation), P = 0.0869 (MFI GPR56 on PD-1^highGPR56^low after activation), P = 0.0276 (MFI GPR56 on PD-1^highGPR56⁺ after activation), P = 0.0563 (% BLIMP-1 on PD-1^highGPR56^low after activation), P = 0.0775 (% BLIMP-1 on PD-1^highGPR56⁺ after activation), P = 0.019 (% PRF1 on PD-1^highGPR56^low after activation). e Velocity plots showing the phenotype directionality between the different CD4⁺ T-cell clusters in RA split in ACPA− (left panel) and ACPA+ (right panel) (n = 4 ACPA−, n = 11 ACPA+, pool PB and SF). f Graphical summary showing the proposed developmental link between the two T_PH subsets, effector and cytotoxic CD4⁺ T cells in SF as well as the identified receptors.