Immune checkpoint blockade induces gut microbiota translocation that augments extraintestinal antitumor immunity

Abstract

Gut microbiota, specifically gut bacteria, are critical for effective immune checkpoint blockade therapy (ICT) for cancer. The mechanisms by which gut microbiota augment extraintestinal anticancer immune responses, however, are largely unknown. Here, we find that ICT induces the translocation of specific endogenous gut bacteria into secondary lymphoid organs and subcutaneous melanoma tumors. Mechanistically, ICT induces lymph node remodeling and dendritic cell (DC) activation, which facilitates the translocation of a selective subset of gut bacteria to extraintestinal tissues to promote optimal antitumor T cell responses in both the tumor-draining lymph nodes (TDLNs) and the primary tumor. Antibiotic treatment results in decreased gut microbiota translocation into mesenteric lymph nodes (MLNs) and TDLNs, diminished DC and effector CD8⁺ T cell responses, and attenuated responses to ICT. Our findings illuminate a key mechanism by which gut microbiota promote extraintestinal anticancer immunity.

Fig. 1.. Immune Checkpoint Inhibitor Therapy (ICT) Induces Gut Microbiota Translocation into Secondary Lymphoid Organs and Tumor

(A) Schematic diagram of the strategy used to assess the temporal dynamics of bacterial translocation into secondary lymphoid organs and tumor in C57BL/6 mice (female, 6–8 wks, Jackson) bearing B16-F10 melanoma tumors and receiving ICT (anti-PD-1 and anti-CTLA-4 mAb, 200μg). (B) Cultured bacterial levels in MLN, Spleen, TDLN, and Tumor. Tissue homogenates were serially diluted, plated on BHI/Blood agar, and cultured at 37°C under anaerobic conditions for 24–72 hours. n=3–4 per time point per experiment. Two experiments were performed for a final sample size of n=6–8 per group. Points represent values from individual mice. Bars represent the mean ± SEM. Green dotted lines represent the limit of detection. (C) Relative abundance of cultured bacteria from secondary lymphoid organs and tumor, as determined by full length (V1-V9) 16S rRNA amplicon sequencing (Sanger). Sequences were entered into the NCBI standard nucleotide Basic Local Alignment Search (BLAST) tool utilizing the rRNA/ITS databases. Bacterial species identification was ascertained from BLASTN results with the highest Total Score, with percent identity score >95% and E value <0.01. (D-F) 16sRNA gene sequencing (V4 region) of tissue gDNA isolated from mice as detailed in fig. 1A. (D) Relative abundance of microbiota in MLN, spleen, TDLN, as determined by 16S rRNA sequencing. (E) Relative abundance of microbiota in the gut (feces), as determined by 16S rRNA sequencing. (F) Principal coordinate analysis of tissue and gut 16S rRNA sequencing data, weighted and normalized by Bray-Curtis distances. The proportion of variance accounted by each principal component is indicated

Figure 2.. Highly abundant microbiota translocators into secondary lymphoid organs activate DCs and induce anti-tumor effector T cell responses

(A) Schematic overview of the protocol used to assess the dendritic cell (DC)-activating potential of various gut bacteria. CD11c+ DCs were isolated from the spleen of C57BL6/J mice (female, 6–8 wks, Jackson) bearing B16-FLT3L tumors. Isolated DCs were stimulated with vehicle (PBS), Escherichia coli LPS, Enterococcus faecalis (Ef), Lactobacillus johnsonii (Lj), Escherichia coli (Ec), or Lactobacillus acidophilus (La) lysates for 6 hours. DCs were then analyzed by flow cytometry. Proportion of (B) MHC-II high, CD40+ cells and (C) MHC-II high, CD80+ cells among CD11c+ DCs. (D) Schematic overview of the protocol used to assess T cell activating and priming potential of DCs stimulated with different gut bacteria. CD11c+ DCs were pulsed with OVA 257–264 peptide and bacterial lysates for 6h. Stimulated DCs were then co-cultured with naïve CD8+ T cells isolated from age- and sex-matched OT-I mice for 7d. Surface expression of T cell activation marker CD69 and intracellular interferon-γ (IFN-γ) were quantified by flow cytometry. Proportion of (E) CD69+ and (F) IFN-γ+ cells among CD8+ T cells. Bars represent the mean ± SEM. All assays were performed in triplicate. Statistical analysis by Mann-Whitney test. *P<0.05, **P<0.01, ***P<0.001, ****P<0.0001.

Figure 3.. Mesenteric lymph nodes modulate gut microbiota-dependent anti-tumor priming responses in the tumor-draining lymph node and tumor

(A) Schematic overview of the protocol used to assess the impact of surgical resection of secondary lymphoid organs or control (sham surgery involving longitudinal abdominal incision only) on ICT efficacy. C57BL/6 mice (female, 6–8 wks, Jackson) were inoculated with 1 × 10⁵ B16-F10 cells subcutaneously in the right flank. Mice with comparable tumor volumes were randomized to receive surgery. 200μg anti-PD-1 and 200μg anti-CTLA-4 mAb (ICT) were injected intraperitoneally on days 4, 8, and 12 post-surgery. n=8–9 per group. (B) Tumor volume and (C) survival of tumor-bearing mice as detailed in fig 3A. Total bacterial load in tumor-draining lymph node (TDLN) (D) and tumor (E) of mice ± MLN ± ICT, as determined by bacterial group (Eubacteria, all bacteria) quantitative-PCR (qPCR) of tissue gDNA collected from mice as detailed in fig. 3A. (D-E) Dotted lines represent the limit of detection. n=5–7 per group. (F) Schematic overview of the protocol used to assess the impact of MLN removal on the CD4+ and CD8+ T cell immune responses in TDLN and tumor of B16-F10 tumor-bearing mice receiving ICT. n=4–6 per group (G-L) Quantification of CD4+ and CD8+ T cell immune response in the TDLN of mice ± MLN by flow cytometry. The proportion of (G, I) activated (CD69+) and (H, J) effector (CD62L-) T cells among CD4+ T cells and CD8+ T cells respectively. The proportion of (K) IFN-γ and (L) granzyme B (GzmB) producing cells among CD8+ T cells. (M-N) Quantification of CD8+ T cell immune response in the tumor of mice ± MLN by flow cytometry. The proportion of (M) IFN-γ and (N) granzyme B (GzmB) producing cells among CD8+ T cells. For all experiments, points represent results from individual animals. n=4–6 per group. Bars represent the mean ± SEM. Statistical analysis by Mann-Whitney test. *P<0.05, **P<0.01, ***P<0.001.

Figure 4.. DCs facilitate ICT-induced gut microbiota translocation into secondary lymphoid organs

(A) Schematic overview of the protocol used to assess the impact of dendritic cell (DC) depletion on ICT-induced microbial translocation into MLN. CD11c-dtr mice (female, 6–8 wks, Jackson) were injected with 100 ng diphtheria toxin (DT) intraperitoneally on day 3 post tumor implantation to deplete CD11c+ DCs. Wild-type C57BL/6 and DT-treated CD11c-dtr mice were injected with 200μg anti-PD-1 and 200μg anti-CTLA-4 mAb intraperitoneally on days 4 and 8 post tumor implantation. (B) Representative flow cytometry plot of CD11c+ subsets. (C) Bacterial load of MLN in wild-type or CD11c-dtr DC-depleted mice ± ICT, as determined by bacterial group (Eubacteria, all bacteria) quantitative-PCR of MLN gDNA collected from mice as detailed in fig. 4A. n=4 per group. Bacterial load in (D) MLN and (E) TDLN recovered from wild-type (C57BL/6J) and Ccr7^−/− mice ± ICT. n=10 per group. (C-E) Dotted lines represent the limit of detection. (F-H) Bacterial load in DCs. CD11c+ DCs were isolated from MLN of C57BL/6 mice (female, 6–8 wks, Jackson) bearing B16-F10 melanoma tumors ± ICT. gDNA was isolated from DCs (from 5 mice pooled into one sample). Bacteria load was determined by bacterial group (Eubacteria, all bacteria) quantitative-PCR of DC gDNA. (F) Number of CD11c+ DCs isolated from MLN of wild-type mice ± ICT. (G) Quantification of bacterial load within DCs isolated from MLN of mice ± ICT. Dotted line represents the limit of detection. (H) Quantification of bacterial load within DCs normalized to the total number of DCs. (I) Relative abundance of microbiota in dendritic cells, as determined by 16S rRNA sequencing. For (D, E), points represent results from individual animals. For (F-H), each point represents a single biological replicate with DCs pooled from 5 mice. Total of n=20 per group. Bars represent the mean ± SEM. Statistical analysis by Mann-Whitney test. *P<0.05, **P<0.01, ***P<0.001.

Figure 5.. ICT induces MLN remodeling

(A-D) C57BL/6 mice (female, 6–8 wks, Jackson) were inoculated with 1 × 10⁵ B16-F10 cells subcutaneously in the right flank. 200μg anti-PD-1 and 200μg anti-CTLA-4 mAb (ICT) or isotype antibodies were injected intraperitoneally × 3. MLN tissue was fixed and processed for high endothelial venule marker MECA-79 immunohistochemistry staining (A, B) and hematoxylin and eosin (H&E) staining (C, D). (A) Representative images of MECA-79 staining. Scale bar = 200μm. (B) Quantification of high endothelial venules (HEVs) diameter in MLN from mice bearing B16-F10 tumors and treated with or without ICT. n=5 mice per group. Statistical analysis by t-test. ***P<0.001 (C) Representative image of H&E staining. Scale bar = 600μm (Left panels), 300μm (Right panels). Yellow arrows indicate blood vessels in the medullary space (ms) (D) Quantification of total number of blood vessels in the MLN medullary space of mice bearing B16-F10 tumors treated with or without ICT. n=5 mice per group. Statistical analysis by t-test. ***P<0.001 (E-H) C57BL/6 mice (female, 6–8 wks, Jackson) were inoculated with 1 × 10⁵ B16-F10 cells subcutaneously in the right flank. Mice with comparable tumor volumes were randomized before the ICT. 3 doses of ICT or isotype controls were injected intraperitoneally (n=6–7 per group). 1 × 10⁷ GFP+ E. coli were injected directly into the MLN. Tumor, TDLN and blood was collected 24 hours post E. coli injection. Tissue homogenates and blood were spread on TSA-Kanamycin Agar plates. GFP+ colonies were enumerated after 24 hours of incubation at 37°C. (E) Representative image of GFP+ colonies. Quantification of GFP+ CFUs in (F) tumor, (G) TDLN and (H) Blood. Bars represent the mean ± SEM. Statistical analysis by t-test. *P<0.05, **P<0.01, ***P<0.001.

Figure 6.. Antibiotic treatment results in decreased gut microbiota translocation into MLN, decreased polyfunctional CD8+ T cell effector responses, and diminished ICT efficacy

Schematic overview of the protocol used to assess the impact of antibiotic-induced gut microbiota depletion on ICT efficacy, gut microbiota translocation and effector CD8+ T cell response. C57BL/6 mice (female, 6–8 wks, Jackson) were treated ± antibiotics (ABX, 2 mg/ml streptomycin and 1500 U/ml penicillin G in drinking water) for 7d before B16-F10 tumor inoculation. Mice were treated with 200μg anti-PD-1 and 200μg anti-CTLA-4 mAb intraperitoneally on days 4, 8, and 12 after tumor implantation (A) Tumor volume of mice ± ABX and ICT. n=5–7 mice per group. (B) Bacterial load of MLN in mice + ABX and ICT, as determined by bacterial group (Eubacteria, all bacteria) quantitative-PCR of MLN gDNA. n=8 per group. Dotted line represents the limit of detection. Three-dimensional t-distributed stochastic neighbor embedding (t-SNE) plot of secretory cytokine profiles of CD8+ T cells isolated from (C) MLN (n=3 per group) and (D) TDLN (n=10 per group) of mice ± ABX + ICT, as determined by using Isoplexis IsoSpark; 28-plex mouse adaptive immune panel. Only CD8+ T cells secreting at least one cytokine are represented. tSNE plot displaying immune functional categories of CD8+ T cells from (E) MLN and (F) TDLN, defined as Effector: Granzyme B, IFN-γ, MIP-1α, TNF-α Stimulatory: GM-CSF, IL-12p70, IL-15, IL-18, IL-2, IL-21, IL-5, IL-7 Chemoattractive: BCA-1, CCL-11, IP-10, RANTES, CXCL1, CXCL13 Regulatory: FAS, IL-10, IL-13, IL-27, IL-4, sCD137 Inflammatory: IL-17A, IL-1β, IL-6, MCP-1 Absolute quantification of IFN-γ and Granzyme B secreting CD8+ T cells isolated from (G) MLN and (H) TDLN of mice ± ABX + ICT. For (B), each point represents individual animal. For (C, D, E, F), each point represents individual CD8+ T cells isolated from MLN and TDLN of mice ± ABX + ICT. Bars represent the mean ± SEM. Statistical analysis by Mann-Whitney test. *P<0.05, **P<0.01, ***P<0.001, ****P<0.0001.