Defensive tolerance drives the reprogramming and dysfunction of infiltrating pathogenic B cells assuring the maintenance of tolerance

Abstract

We previously showed that infiltrating cytotoxic immune cells are reprogrammed to regulatory-like/exhausted cells within accepted kidney allografts through a 'defensive tolerance' mechanism. We observed a regulatory B cell (Breg) signature within the accepted kidney. Here we show that despite a Breg phenotype, neither B cell depletion nor the use of μMT recipients which lack B cells, resulted in kidney allograft rejection. Negative regulators of B cell function, Siglecg and Fcgr2b, show increased expression in both accepted kidney and lung allografts. Kidney allografts transplanted in B6.Fcgr2b KO recipients underwent antibody mediated rejection. Hypothesizing that similar mechanisms in a tumor microenvironment may attenuate anti-tumor immunity, we observed that expression of SIGLEC10, the human homolog of Siglecg, was associated with resistance to anti-PD1 therapy in human melanomas. In conclusion, B cell expression of FcγRIIB and Siglec-G appear to play an essential role in maintaining transplant tolerance and in tumor evasion of anti-tumor immunity.

Fig. 1.. B cells are not required for the induction and maintenance of kidney allograft acceptance.

(A) Schematic showing the experimental design for CD19⁺ cell depletion using B6. [CD19^Cre/+iDTR^fl/+] mice and intraperitoneal (IP) injection of diphtheria toxin (DT). (B) Serial flow cytometric analysis of CD19+ and CD20+ cells in peripheral blood mononuclear cells (PBMC) following DT administration. (C) Kidney allograft survival following DT injection at post-operative day (POD) 21 (n = 2) or POD 180 (n = 3), compared to vehicle-treated controls (n = 2). (D) Schematic of the experimental design for CD20⁺ cells depletion using IP injection of anti-CD20 antibody (CD20Ab) into B6 wildtype (WT) recipients of DBA/2 kidneys. (E) Serial flow cytometric analysis of CD19+ and CD20+ cells in PBMC over time following CD20Ab administration. (F) Kidney allograft survival following CD20Ab injection at POD 28 (n = 3) or POD 180 (n = 2), compared to untreated controls (n = 3). (G) DBA/2 kidney allograft survival in B6.μMT recipients (n=4), which lack mature B cells, compared to B6 WT recipients (n = 4).

Fig. 2.. Temporal expression analysis of Fcgr2b and Siglecg in accepted and rejecting kidney allografts.

(A) Density plots of Siglecg⁺, Fcgr2b⁺, and Siglecg⁺Fcgr2b⁺ cells within B cell clusters in accepted kidney allografts at 1-week post-transplantation. (B) Density plots of Siglecg⁺, Fcgr2b⁺, Siglecg⁺Fcgr2b⁺, and Ighd⁺Ighm⁺ cells in B cell clusters at 3 weeks post-transplantation. (C) Density plots show expression of Siglecg⁺, Fcgr2b⁺, Siglecg⁺Fcgr2b⁺, Ighd⁺Ighm⁺, and Breg markers, including Havcr1, Cd1d1, Cd5, and IL10 at 3 weeks post-transplantation. (D, E) Pseudotime analysis of B cell clusters at 3weeks (D) and 6 months (E) post-transplantation. The red arrow marks the origin point representing naïve B cells, and the red circle highlights a cluster expressing Siglecg or Fcgr2b. Pseudotime graphic is overlaid on the UMAP plot on a gradient color scale. (F) Dot plots show progressive upregulation of Siglecg and Fcgr2b over time (1 week, 3 weeks, and 6 months post-transplantation) within B cell clusters. (G) Dot plot analysis showing higher expression of Siglecg and Fcgr2b in the spontaneous acceptance model (accept 1, 3, 24 weeks) and in an induced kidney acceptance model (where recipients are infused with donor splenocytes treated with the chemical cross-linker ethylenecarbodiimide (ECDI-SP)), when compared to a kidney allograft rejection model (B6 to DBA/2).

Fig. 3.. Temporal expression of Fcgr2b, Siglecg, and Breg markers in accepted lung allografts.

(A) Schematic showing transplantation of Balb/c lungs into B6 WT recipients (CD45.2), treated with peri-operative co-stimulatory blockade, followed by re-transplantation into non-immunosuppressed B6 WT (CD45.1) recipients ≥30 days later. Graphic created with BioRender. (B) Dot plots of Fcgr2b and Siglecg expression in the CD45.1⁺ and CD45.2⁺ B cell clusters. (C) Dot plot analyses of Breg markers - Havcr1, Cr2, Cd5, Cd24a, Cd38 - within the CD45.1+ and CD45.2⁺ B cell clusters and Siglecg expression in the CD45.1⁺ and CD45.2⁺ B cell clusters. (D) Pseudotime analysis of B cell clusters at seven days after re-transplantation. The red arrow marks the origin point representing Siglecg⁺Fcgr2b⁺ B cells, and the red circle highlights clusters expressing the Breg marker Havcr1. Pseudotime graphic is overlaid on the UMAP plot on a gradient color scale. (E) Dot plots show expression of Fgl2 within the T cell cluster.

Fig. 4. Antibody-mediated rejection of kidney allografts in FcγRIIB^−/− recipients.

(A) DBA/2 kidney allograft survival in FcγRIIB^−/− recipients (n=13), compared to wild-type (WT) recipients (n = 13; P<0.001). (B-D) Histopathological analysis of rejected kidney allografts obtained from FcγRIIB −/− recipients. (B) H&E staining shows thrombotic microangiopathy. Arrows indicate mononuclear cell infiltration. Day 35. Scale bars: 100μm. (C) PAS staining shows glomerulitis and capillaritis. Day 35. Scale bars: 20μm. (D) Immunohistchemistory shows increased C4d deposition in FcγRIIB −/− kidney allograft. Day 39. Scale bars: 50μm. (E) Flow cytometric analysis of serum antibodies reactive to mouse spleen cells. Sera from FcγRIIB −/− recipients exhibited higher antibody reactivity against DBA/2 (donor, 64.3%) and C3H (third-party, 40.0%) cells, compared to B6 (self, 8.2%) and negative control (0.93%). (F-H) Flow cytometric analysis of SiglecG⁺ FcγRIIB⁺ cells within the CD19⁺ B cell population isolated from spleens and kidney allografts at 1 week (F) and 3 weeks (G) post-transplantation in the tolerance model and 6 weeks post-transplantation in the FcγRIIB −/− rejection model (DBA/2 to B6 FcγRIIB^−/−) (H). The data in (F-H) are representative of triplicate experiments. (I, J) Histopathological analysis of kidney allografts obtained from FcγRIIB −/− B cells injected recipients 4 weeks post-transplantation (n=3). (I) PAS staining shows glomerulitis. Scale bars: 20μm. (J) Immunohistchemistory shows increased C4d deposition in FcγRIIB −/− B cells injected kidney allograft. Scale bars: 100μm.

Fig. 5.. Expression of Fcgr2b and Siglecg within the B cell cluster in murine KPC tumors.

(A) UMAP plot of all cells isolated from murine KPC tumors. The red arrow indicates the B cell cluster. (B) Violin and dot plots analysis of Fcgr2b and Siglecg expression in total cells. The B cell cluster is boxed in red. (C) Violin, dot and density plots of Fcgr2b and Siglecg expression in the B cell cluster. (D) Pseudotime analysis of the B cell cluster in murine KPC tumors. The red arrow indicates the origin point representing naïve B cells, and the red circle highlights a subpopulation expressing Fcgr2b and Siglecg. Pseudotime graphic is overlaid on the UMAP plot on a gradient color scale.

Fig. 6.. Expression of FCGR2B and SIGLEC10 within the B cell cluster in human melanomas.

(A) UMAP plot of tumor-infiltrating immune cells isolated from human melanomas. The red arrow indicates the B cell cluster. (B) Violin, dot and density plots show FCGR2B and SIGLEC10 expression within the B cell cluster among all tumor-infiltrating immune cells. The B cell cluster is boxed in red, and the red arrow indicates its location. (C) Boxplot comparing the B cell normalized enrichment score estimated via ssGSEA in patients treated with anti-PD-1 therapy(aPD1), stratified by clinical response (progressive disease [PD] vs. non-progressive disease [notPD]). The B cell signature includes the following genes: BLK, CD19, MS4A1, TNFRSF17, FCRL2, FAM30A, PNOC, SPIB, and TCL1A. (D) Forest plot showing odds ratios (OR) and 95% confidence intervals (CI) for the association between B cell abundance and SIGLEC10 expression with response to anti-PD1 therapy. ORs are shown on a log scale, with a dotted vertical line indicating the null value (OR = 1). Each square represents the estimated OR for one predictor, and horizontal lines denote the 95% CI. Values were standardized (z-score) prior to model fitting to allow interpretation per 1 standard deviation increase (n=124). Statistically significant associations (p < 0.05) are indicated with asterisks.