Antigen-presenting autoreactive B cells activate regulatory T cells and suppress autoimmune arthritis in mice

Abstract

B cells undergo several rounds of selection to eliminate potentially pathogenic autoreactive clones, but in contrast to T cells, evidence of positive selection of autoreactive B cells remains moot. Using unique tetramers, we traced natural autoreactive B cells (C1-B) specific for a defined triple-helical epitope on collagen type-II (COL2), constituting a sizeable fraction of the physiological B cell repertoire in mice, rats, and humans. Adoptive transfer of C1-B suppressed arthritis independently of IL10, separating them from IL10-secreting regulatory B cells. Single-cell sequencing revealed an antigen processing and presentation signature, including induced expression of CD72 and CCR7 as surface markers. C1-B presented COL2 to T cells and induced the expansion of regulatory T cells in a contact-dependent manner. CD72 blockade impeded this effect suggesting a new downstream suppressor mechanism that regulates antigen-specific T cell tolerization. Thus, our results indicate that autoreactive antigen-specific naïve B cells tolerize infiltrating T cells against self-antigens to impede the development of tissue-specific autoimmune inflammation.

Figure 1.

Collagen-reactive C1-B are naturally detected in mice, rats, and humans. (A and B) Schematic design of the synthesized triple helical C1 tetramer and GPO control tetramer. (C) Representative flow cytometry plots depicting the frequency of C1-B from pooled spleens before enrichment in mice and following post-enrichment protocol in mice, rats, and PBMCs of HD and RA samples. (D) Ex vivo visualization of the C1-epitope using CB20 antibodies on cryofixed joint tissues of neonatal (upper panel) and adult mice (lower panel), revealing the exposure of the C1 epitope in the BM trabeculae. (E) Binding of five unique anti-C1 clones (MAK1, 4, 5, and 6) assessed by ELISA. (F) Binding of MAK antibodies to other defined COL2 epitope libraries assessed by bead-based multiplex assay. Error bars represent mean ± SEM. Scale bars are 50 μm.

Figure S1.

C1-B frequency is reduced in RA patients. (A) Representative flow cytometry staining of spleen and BM of BQ mice before and after enrichment in the absence of tetramers or in the presence of different control tetramers. (B) Representative flow cytometry staining of rat spleen and BM after C1-tetramer enrichment. (C) Representative flow cytometry staining of HD and RA PBMC before enrichment using C1- and GPO- tetramers. (D) The absolute number of C1-B in spleen and BM of mouse and rats. Each square represents one mouse or rat. (E) Frequency of C1-B and polyreactive (GPO) B cells in HD (n = 7) and RA patients (n = 6); each dot represents one donor. The ratio between C1-B and polyreactive B cells in HD and RA patients. (F) Luminex analysis of triple helical C1-specific antibody response in RA patients (n = 1,125) and age-matched HD (n = 398). (G) Schematic of the experimental design adopted to clone and express MAK antibodies. Error bars represent mean ± SEM. Statistical significance in E (% of enriched) was determined using a paired t test, while the ratio in F was determined using two-tailed Mann–Whitney U test. *P < 0.05, **P < 0.01, and ***P < 0.001.

Figure 2.

C1-B bypass mechanisms of negative selection. (A) Frequency of persistent C1-B in the BM and spleen of wild-type mice (n = 5 pooled). (B) Mass spectrometry–based proteomic analysis of sorted C1-B from heavy chain CB20 knock-in (green) ACB mice. Light chain Igkv 3–7 was the most significantly upregulated protein in C1-B compared to the rest of B cells (n = 5 mice/group). (C) IL2 titers from the supernatants were assessed by ELISA after co-culture of COL2-reactive T cell hybridoma together with purified B cells from LN of ACB mice with or without COL2. (D) Like C, but cells were cultured in the presence of MHCII-restricted COL2 peptide (gal-COL2) recognized by the T cell hybridoma. (E) In vivo COL2-pulsed B cells with either ovalbumin or COL2, purified, and co-cultured with T cell hybridoma without the addition of exogenous antigens. IL2 titers were assessed by ELISA 24 h later. (F) Like in C, but B cells were either fixed with formaldehyde or maintained live. (G) ACB.Col2^R360Q mouse expresses a mutated COL2 protein at the C1-epitope, point mutation Q–R, leading to the abrogation of CB20 binding to the mutated epitope. (H and I) Frequency and absolute number of C1-B in ACB, heterozygous, and homozygous ACB.Col2^R360Q LN and spleens (n = 5–8 mice/group). Error bars represent mean ± SEM. Statistical significance was determined using two-way ANOVA followed by Sidak’s test. *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001.

Figure S2.

Col2^R360Q mutation induces somatic hypermutation and alters the usage of the light chain by C1-B. (A) Enumeration of C1-B at different stages of B cell development in BQ mice. Each dot represents one mouse. (B) Schematic representation of the CB20 VDJ knock-in ACB mouse. (C) Measuring median affinity of C1-B. Cells from the spleen and BM of ACB mice were split equally in sever wells and incubated with noted concentration (M) of C1 or GPO THP or ovalbumin protein before C1-tetramer staining and flow cytometric analysis. Representative flow cytometry plots showing the retrieval of C1-B after incubations. (D) Percent of recovered C1-B (C1-tetramer⁺ B cells) is shown compared with the number of C1-B detected in the absence of competitor antigen. (E) Heatmap of significantly downregulated and upregulated proteins between C1⁺ and C1⁻ B cells. (F) Sera from naïve ACB at increasing dilutions tested for specificity against different THPs. (G) Representative flow cytometry plots exhibiting the frequency of C1-B detected in the spleen and inguinal LN (iLN) of ACB and ACB.Col2^R360Q mice. FMO controls were included for the C1-tetramer staining. (H) Frequency and absolute number of C1-B detected after C1 enrichment in the spleens of BQ and BQ.Col2^R360Q mice. Each dot represents one mouse. (I) Frequency and absolute number of C1-B detected in the BM of ACB and ACB.Col2^R360Q (het and homo). (J) Frequency of mutations in the CDR3 and FR4 light chain of enriched C1-B from BM of ACB and ACB.Col2^R360Q mice (n = 3/group). (K) V_k gene usage detected from light chain next-generation sequencing of enriched C1-B from ACB and ACB.Col2^R360Q (n = 3/group). Error bars represent mean ± SEM. Statistical significance in H–K was determined using two-tailed Mann–Whitney U test. *P < 0.05, **P < 0.01, and ***P < 0.001.

Figure 3.

C1-B suppress tissue-specific inflammation independent of IL10. (A–C) Adoptive transfer of naïve B cells from ACB and BQ mice into syngeneic autoimmune-prone BQ.Ncf1^m1j recipient. Arthritis clinical score, representative histological H&E staining of tibia (Ti) and talus (Ta), and histopathological assessment in recipient mice. (D) EAE clinical score in ACB and BQ mice after MOG protein and MOG peptide immunization. (E–G) Adoptive transfer of naïve B cells from ACB.IL10^−/−, BQ.IL10^−/− mice or no transfer (PBS) into syngeneic autoimmune-prone BQ.Ncf1^m1j recipient. Arthritis clinical score, representative histological H&E staining of tibia (Ti) and talus (Ta), and histopathological assessment in recipient mice. (H) Arthritis score of syngeneic autoimmune-prone BQ.Ncf1^m1j recipient mice after adoptive transfer of 0.5 × 10⁶ C1-B or B cells. (I and J) Serum titers of anti-COL2 and anti-C1 IgG2b antibody in recipient mice of H 25 d after immunization was assessed by ELISA. (K) Frequency of CD44⁺ and CD62L⁺ Tregs (Foxp3) in the LNs of recipient mice of H 90 d after immunization. (L and M) In vitro recall assay of LNs derived from H depicting Tregs proliferation and activation after 48 h of culture with different antigens. Error bars represent mean ± SEM. Statistics in A, C, E, and G–K were determined by two-tailed Mann–Whitney U test. Significance in L and M was determined by two-way ANOVA followed by Sidak’s test. *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001. Scale bar for 2.5× is 500 μm and for 10× is 100 μm.

Figure S3.

Antibody response and IL10 production by C1-B. (A) Adoptive transfer of naïve B cells from ACB and BQ mice into syngeneic autoimmune-prone BQ.Ncf1^m1j recipients. PBS group included where no adoptive transfer was carried out. Arthritis clinical score in recipient mice is plotted. (B) Titer of anti-COL2 antibody of IgG2b isotype measured by ELISA in the sera of recipient mice 50 d after CIA. (C and D) Comparative flow cytometric plots of C1⁺ and C1⁻ B cells depicting the frequency of IL10 production after in vitro stimulation using IL10-GFP reporter mice. (E) Titer of anti-COL2 IgG antibody measured by ELISA in the sera of recipient mice 23 and 87 d after CIA. (F) Frequency of C1-B in the spleen of ACB mice using dual tetramer staining. (G) Representative plot exhibiting the purity of magnetically enriched C1-B used for downstream procedures such as adoptive transfers. (H) Adoptive transfer of B cells from ACB (C1-B enriched or depleted) and BQ mice into syngeneic autoimmune-prone QD recipients. PBS group included where no adoptive transfer was carried out. Arthritis clinical score in recipient mice is plotted. (I) Adoptive transfer of B cells from ACB and BQ mice 10 d after CIA induction in syngeneic autoimmune-prone BQ.Ncf1^m1j mice. Arthritis clinical score in recipient mice is plotted. (J) Titer of anti-COL2 IgG antibody measured by ELISA in the sera of recipient mice 25 d after CIA. Error bars represent mean ± SEM. Statistical significance in A, B, E, and H–J was determined using two-tailed Mann–Whitney U test while in D was determined using one-way ANOVA followed by Sidak’s correction test. *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001.

Figure 4.

C1-B are potent inducers of Tregs. (A and B) Frequency, activation, and proliferation of FOXP3⁺ Tregs in the LNs of 10 d immunized ACB and BQ mice (n = 5 mice/group). (C–E) Frequency and activation of LAG3⁺CD49b⁺ Tr1 cells in the LNs of 10 d immunized ACB and BQ. (F and G) Flow cytometry plots depicting the frequency of endogenous COL2-reactive Vβ8.3 proliferating Tregs (FOXP3⁺Ki67⁺) after in vitro culture of LN cells derived from ACB, heterozygous and homozygous ACB.Col2^R360Q mice (n = 5 mice/group). (H) Frequency of CD44⁺ Tregs gated on F. (I) Experimental setup for contact-dependent and independent Treg induction experiment. (J and K) Flow cytometry plots depicting the frequency of proliferating Tregs in direct contact (pink), contactless (gray) culture alone or with natural C1-B and B cells purified from BQ mice in the presence or absence of COL2 (n = 18 mice, each symbol represents three mice pooled). Error bars represent mean ± SEM. Statistics in A, B, D, and E were determined by two-tailed Mann–Whitney U test. Significance in G, H, and K was determined by two-way ANOVA followed by Sidak’s test. ****P < 0.0001.

Figure S4.

C1-B controls the expansion of Tregs in vitro. (A) Representative flow cytometry plot of the Foxp3⁺Ki67⁺ gating 48 h after coculture of sorted QC T cells with sorted B cells from ACB, ACB.Col2^360Q, or BQ WT mice in the presence or absence of COL2. (B) Frequency of proliferating Tregs within the Vβ8.3⁺CD4⁺ T cell population. (C) Representative histograms exhibiting the expression of CD44 on proliferating Tregs. (D) Quantification of CD44 MFI on proliferating Tregs from A. (E) Gating strategy and experimental design of the sorted 10,000 C1-B, C1-GPO-specific, and naïve B cells cocultured with sorted QC T cells for 5 d in the presence of COL2. (F) Frequency of Foxp3⁺ T cells within the Vβ8.3⁺CD4⁺ population. (G) Quantification of CD44 MFI on Tregs from F. (H) Frequency of activated Tregs within the Vβ8.3⁺CD4⁺ T cell population following coculture with C1-B (C1⁺), C1-B depleted (C1⁻), or without (total QC LN or isolated QC T cells alone) in the presence or absence of COL2. Error bars represent mean ± SEM. Statistical significance in B and D was determined using two-tailed Mann–Whitney U test while in F–H was determined using one-way ANOVA followed by Sidak’s correction test. **P < 0.01 and ***P < 0.001.

Figure 5.

Antigen-presenting C1-B upregulates CCR7 and CD72 upon activation. (A) RNA sequencing of single sorted C1-B, GPO-specific, and B cells from HD (n = 3). Data presented as t-SNE. (B) The distribution of different groups within the clusters. (C) The expression of B cell subtype markers where a dot represents each feature (row) in each group of cells (column). The proportion of detected expression values and the average expression for each feature in each group of cells is visualized using the size and color, respectively, of each dot. (D) Heatmap of selected immune-related genes that are significantly differentially expressed (in red) and belong to the top 50 differentially regulated genes but not statistically significant (in bold black). (E) The summary of top enriched terms between C1-B and GPO-specific B cells depicting the enrichment score, gene ratio, and gene count. (F and G) Representative flow cytometry plots and mean fluorescent intensity (MFI) quantification depicting surface MHCII expression on BM- and spleen-derived B cells from ACB and BQ mice (n = 4/group). (H and I) Frequency and MFI of CCR7 on circulating C1-B days after ACB immunization. (J) CCR7 MFI on natural C1-B derived from LN of 10 d immunized BQ mice. (K and L) LN from day 30 immunized BQ mice were stained ex vivo and analysis of CD72 and CCR7 was assessed on C1-B (red), GPO-specific (blue), B cells (black), and non-B cells (gray). Error bars represent mean ± SEM. Statistics in J were determined by two-tailed Mann–Whitney U test. Significance in I and L was determined by two-way ANOVA followed by Sidak’s test. **P < 0.01 and ****P < 0.0001.

Figure S5.

CCR7/CD72 expression is restricted to autoreactive C1-B. (A) Heatmap of the genes that are significantly expressed grouped according to the three donors (A, B, and C), the three B cell population groups (C1-specific, GPO-specific, and double negative B cells), and the three clusters (cluster 1 = memory cells, cluster 2 = naïve cells, cluster 3 = B1-like). Normalized log counts are plotted after scaling by row for the top 50 markers (Wilcoxon rank sum test, Scran). (B) Enriched PE-specific B cells and C1-B from spleens and iLN of BQ mice 10 d after immunization with PE and Col2, respectively. Naïve mice were also included. Representative flow cytometry plots exhibiting the surface expression of CD72 and CCR7 on different gated (G) populations. (C) Dot plot quantifying the frequency of CCR7/CD72 double-positive cells within the different gated populations. (D) Representative flow cytometry plots exhibiting IgM and IgD surface expression on PE- and C1- B cells enriched from naïve and 10 d immunized BQ mice. (E) Bar plot quantifying the frequency of naïve (IgD^high/IgM^high) and switched (IgD^low/IgM^low) C1- and PE-specific B cells enriched from naïve (n = 1–2/group) and 10 d immunized (n = 3/group) BQ mice. (F) MFI of surface CD72 on enriched C1-B, IgG⁻, and IgG⁺ B cells isolated from HD PBMC (n = 4). Error bars represent mean ± SEM. Statistical significance in F was determined using one-way ANOVA followed by Sidak’s correction test. **P < 0.01, ***P < 0.001, and ****P < 0.0001.

Figure 6.

Treg induction by C1-B is abrogated by blocking CD72. (A and B) Flow cytometry plots depicting the proliferation of Tregs after culture of naïve LN in vitro from ACB.QC and BQ.QC mice in the presence of COL2 and mouse anti-mouse CD72 blocking antibody or mouse IgG1 isotype control. (C) Schematic design of CD72 blocking in vivo. (D and E) Flow cytometry plots depicting the frequency of circulating proliferating (Ki67⁺) and non-proliferating (Ki67⁻) Foxp3⁺ T cells in the blood (day 14 after COL2 immunization) of ACB mice injected anti-CD72 mAb or PBS (n = 8 mice/group). (F) Frequency of circulating B cells (CD19⁺B220⁺) cells as in D. (G and H) Serum titers of anti-COL2 and anti-C1 antibodies with different isotypes 14 d after immunization in anti-CD72 or PBS-treated groups. (I and J) Arthritis incidence and clinical score of CIA in ACB mice injected or not with anti-CD72 antibody. Only sick mice were plotted in J. (K) Representative histological H&E staining of tibia (Ti) and talus (Ta) in recipient mice. Error bars represent mean ± SEM. Statistics in E–H and J were determined by two-tailed Mann–Whitney U test. Significance in B was determined by two-way ANOVA followed by Sidak’s test. *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001. Scale bar for 2.5× is 500 μm and for 10× is 100 μm.

Figure 7.

CD72 controls Treg expansion and activation by C1-B. (A) Frequency of C1-B and quantification of CD72 MFI on C1-B 24 h after the addition of IL4 to the culture. (B) Representative histograms exhibiting the expression of CD72 on C1-B from A. (C) Representative flow cytometry plot of the Foxp3⁺Ki67⁺ and Foxp3⁺CD44^high gating 48 h after coculture of naïve iLN from ACB.QC and BQ.QC mice in the presence or absence of IL4. (D) Frequency of proliferating and activated Tregs from C. (E) Absolute number of proliferating Tregs and frequency of activated Tregs within the Vβ8.3⁺CD4⁺ population after blocking CD86. (F) Representative flow cytometry plot of the Foxp3⁺CD44^high gating 48 h after coculture of naïve iLN from ACB.QC and BQ.QC mice in the presence or absence of different blocking antibodies. (G) Representative histogram validating the blockade of CD72. (H) MFI quantification of antigen presentation-related proteins (CD86, MHCII), activation markers (CD44, CD69), and chemokine receptors (CXCR5 and CXCR4) on 48 h cultured B cells after anti-CD72 or isotype control treatment in the presence of different stimuli. (I) Isolated C1-B were treated ex vivo with blocking CD72 antibody (anti-CD72) or isotype control following adoptive transfer into syngeneic BQ.Ncf1^m1j recipient mice. Arthritis incidence and clinical score in BQ.Ncf1^m1j recipient mice are plotted. (J) Titer of anti-COL2 and anti-C1 IgG antibody measured by ELISA in the sera of recipient mice 25 d after CIA. (K) Flow cytometry plots and histograms exhibiting the surface expression of certain markers on transferred C1-B in I. Error bars represent mean ± SEM. Statistical significance in A was determined using paired t test, while in D, E, I, and J, it was determined using one way ANOVA followed by Sidak’s correction test. *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001.