br legend continued on next page br Tumor Associated Bacteria
(legend continued on next page)
Tumor Associated Bacteria Are Essential for Driving Enhanced CRC in the Absence of IL-1 Signaling in Myeloid Cells
To establish a functional and causative relationship between myeloid IL-1R1 inactivation, increased bacterial presence, inflammation and CRC, we treated CD11bCre+Il1r1f/f /CDX2ERT-Apcf/f and control mice with broad spectrum anti-biotics for 4 weeks, starting after the epithelial transformation (Figure 4H). Ablation of intestinal and tumor associated micro-biota resulted in a drastic reduction of CRC tumorigenesis in both control and myeloid IL-1R deficient mice (Figure 4I and 4K). Bacterial 16S RNA, SFB 16S RNA, and Il17a mRNA in tu-mors from myeloid IL-1R1-deficient mice treated with GsMTx4 were strongly reduced compared to the untreated group (Fig-ure 4K). Importantly, difference in CRC was no longer observed between antibiotic-treated control and myeloid IL-1R1 deficient animals (Figures 4J and 4K).
To ascertain whether increased IL-17A/IL-22 response in tu-mors in mice lacking IL-1R1 in myeloid cells was driven in part by increased IL-1 action in T cells and increased IL-17A produc-tion, we created CD11bCre-CD4Cre-Il1r1f/f ‘‘double Cre’’ mice and then generated CRC-bearing BM chimeras. Additional ablation of IL-1R signaling in T cells reduced the heightened tumorigenesis observed in myeloid Il1r1 / mice and reduced the production of pro-tumorigenic IL-17A (Figures S5F and S5G). Moreover, in independent azoxymethane/dextrane sulfate sodium (AOM/DSS) induced model of colitis-associated cancer CD4Cre+Il1r1f/f mice displayed lower tumorigenicity, while CD11bCre+Il1r1f/f mice showed higher tumorigenicity in comparison with controls (data not shown).
To independently confirm protective role of IL-1R1 in myeloid cells, we next generated LysMCreCre/WT (referred to as LysMCre+) and LysMCre- Il1r1f/f (control) mice and used their BM to create chimeric CDX2ERT-Apcf/f mice, to delete IL-1R1 using an overlapping Cre-deleter. LysMCre, similarly to CD11bCre, excised Il1r1 in monocytes, macrophages, and neu-trophils (Figures S5A and S5B). LysMCre+Il1r1f/f->CDX2ERT-Apcf/f mice also developed more CRC (Figures 5A and 5B). The difference was even more notable when myeloid IL-1R1-deficient mice and controls were housed separately as co-
housing typically allows microbiota equalization (Figures 5A and 5B). LysMCre+Il1r1f/f->CDX2ERT-Apcf/f mice also demon-
strated increased mRNA expression of Il17a, Il22, and Ptgs2 (Figure 5C), as well as bacterial 16S rRNA, E. coli, and SFB in tu-mors (Figure 5D). Moreover, Yo-Yo 1 DNA staining of whole-mount colons revealed an increase in bacteria located on the im-mediate surface of the tumor (Figure 5E).
As CRC load in genetically identical animals was affected by their housing status and bacteria attached to the tumors of myeloid Il1r1 / mice, we next characterized the microbial alter-
ation induced by the absence of IL-1R in myeloid cells. We per-formed Illumina MiSeq 16S rRNA sequencing on DNA samples from the colonic contents of naive mice, co-housed and sepa-rated tumor bearing mice, as well as from the epithelial ‘‘shakes’’ (mechanical isolation of epithelium-adherent bacteria) of normal and tumor tissue from control and LysMCre+Il1r1f/f mice. Prin-cipal component analysis (PCA) demonstrated that myeloid specific IL-1R ablation did not lead to global changes in fecal mi-crobiota in naive mice (Figure 5F). This was largely confirmed by 16S qRT-PCR of naive colonic tissue (Figure S5H). CRC forma-tion did induce differences in fecal microbiota between wild type and myeloid Il1r1 / mice, but that difference was not evident if mice were co-housed (Figure 5F). On the other hand, significant differences were detected in the microbiota composition of bac-terial adherent to the tumor tissue, but not to the normal tissue in LysMCre+ mice (Figures 5F and 5G). PERMANOVA analysis es-tablished that the only differences between LysMCre+Il1r1f/f mice and controls mice was in tumor-adherent bacteria and in fecal microbiota, when mice were housed separately (Figure 5G). Among increased bacteria we found Escherichia-Shigella genera, represented by E. coli in the tumor-adherent fraction of myeloid cell specific Il1r1 / mice (Figures 5H and 5I), also de-tected by qRT-PCR analysis of tumor tissue (Figures 4G and 5D). It was likely that differences in fecal but not in tumor-asso-ciated microbiota could be mitigated by co-housing. Together, these suggested that myeloid IL-1R signaling repressed bacte-rial invasion and tumor-associated inflammation.