Fig 2: P. anaerobius activates MDSCs activation via lytC_22/Slamf4 interaction.a. The 50–100 kDa fraction from Pa-CM was separated using sodium dodecyl sulfate-polyacrylamide gel electrophoresis. b. Representative images of MDSCs migrated to the lower chamber and quantification of MDSC migration towards different concentration of lytC_22 (n = 4 biologically independent samples). Scale bar: 100 μm. c. tSNE plots showing expression levels of Cd244a in MDSCs clusters from tumours treated with P. anaerobius or not (Ctrl, control). d. Percentage of CD244+ MDSCs in MDSCs after LytC_22 treatment was determined by FACS analysis. e. MDSCs were pretreated with lytC_22 and lytC_22 mutant protein. CD8+ T cell proliferation after co-culture with MDSCs. c-d n = 3 biologically independent sample. Data are presented as mean ± SEM. P values were calculated by one-way ANOVA followed by Tukey’s post-hoc test (b, e) and unpaired two-tailed Student’s t-test (d). Source data

Fig 3: The integrin α2β1–CXCL1 and lytC_22–Slamf4 axis function independently.(a-c) 5×105 MC38 cells were subcutaneously injected into C57BL/6 J mice (n = 6). After 6 days, mice were injected with P. anaerobius or lytC_22 intratumorally and followed by PD1 mAb treatment. 100ug anti-PD1 intraperitoneally and 50ug anti-Slamf4 antibody intravenously injected every 3 days. a. Schematic diagram of experimental design (left) and representative images of tumours (right). b. Gzmb+CD8+T cells in tumours from different groups. 6 mice were used in each group. c. CXCL1 level in tumour tissue were detected using ELISA. 4-5 mice were used in each group including n = 4 mice in Ctrl, lytC_22, Ctrl+a-Slamf4, lytC_22+a-Slamf4 and n = 5 mice in PA, PA + a-Slamf4 group. d. Schematic diagram of experimental design. CRC cell line (Caco-2, HCT116) pretreated with 20ug/ml anti-Slamf4 antibody, followed by P. anaerobius infection. Then culture medium was collected. e. Quantification of secreted CXCL1 from human CRC cells Caco-2, HCT116 treated with vehicle or P. anaerobius after blocking Slamf4 receptor. PBS (n = 3), PA (n = 4), PBS+a-Slamf4 (n = 3), PA + a-Slamf4 (n = 3). f. Schematic diagram of experimental design. Human CRC cell line (Caco-2, HCT116) and mouse CRC cell line (MC38) pretreated with 1 mM RGDS peptide, followed by P. anaerobius infection. Then culture medium was collected and condensed using 50 kDa protein concentration column. g. Quantification of secreted lytC_22 from P. anaerobius was detected using western blot. Data are presented as mean ± SEM. P values were calculated by one-way ANOVA followed by Fisher’s LSD test (b) and one-way ANOVA followed by Tukey’s post-hoc test (c, e). Source data

Fig 4: Schematic model of P. anaerobius-driven anti-PD1 resistance in CRC and potential therapeutic approaches.P. anaerobius engaged with CRC cells via integrin α2/β1 receptor to promote nuclear translocation of NF-κB p65, which in turn stimulates release of CXCL1 to recruit CXCR2+ MDSCs to tumours. In addition, lytC_22 protein secreted from P. anaerobius directly activates Slamf4 receptor on MDSCs, leading to elevated arginase 1 (Arg1) and iNOS expression. P. anaerobius infection thus increased MDSCs accumulation and activation, leading to suppression of CD8+ T cells and non-response to anti-PD1 therapy. Blockade of integrin α2/β1 or Slamf4 could interrupt this process to impair MDSCs in tumour microenvironment and reactivate IFNγ expressing CD8+ T cells to overcome P. anaerobius-driven anti-PD1 resistance.

Fig 5: Blocking the lytC_22–Slamf4 interaction can reverse the effect of lytC_22 and P. anaerobius on MDSC modulation.a, Schematic of the lytC_22–Slamf4 binding sites and their corresponding prediction result (pink, lytC_22; cyan, Slamf4 and yellow dash, hydrogen bond). LytC_22 binds to Slamf4 protein with a binding energy of −416.43 kcal mol−1. LytC_22 is attached to Slamf4 with seven key residues: Glu570, Asp653, Asp725, Tyr661, Asp712, Arg660 and Lys669. b, LytC_22 mutant protein was purified following substitution of key binding residues with alanine. GST-tag pulldown of the GST–lytC_22 mutant with recombinant hSLAMF4. c, MDSCs were treated with lytC_22 or its mutant. Flow cytometry was used to determine the MFI of Arg1 and iNOS in MDSCs; n = 3 biologically independent samples. d, Following MDSC pre-treatment with lytC_22 or lytC_22 mutant protein, the expression of IFN-γ and GzmB in CD4+ and CD8+ T cells was assessed after co-culture with MDSCs; n = 4 biologically independent samples. e–g, Anti-Slamf4 reversed lytC_22-induced anti-PD1 resistance in the MC38 allograft mouse model; n = 6 mice per group. e, Tumour growth (left) and tumour weight (left). f, Percentage of MDSCs (left) as well as Arg1+ (middle) and iNOS+(right) MDSCs in tumours from mice in the different treatment groups. g, Percentage of IFN-γ+CD8+ T cells in the tumours. c–g, Data are the mean ± s.e.m. P values were calculated using a one-way ANOVA, followed by Tukey’s post-hoc test (c,d) or a one-way ANOVA, followed by Fisher’s least significant difference test (e–g).Source data