The gut microbiome enhances breast cancer immunotherapy following bariatric surgery

Abstract

Bariatric surgery is associated with improved breast cancer (BC) outcomes, including greater immunotherapy effectiveness in a preclinical BC model. A potential mechanism of bariatric surgery-associated protection is the gut microbiota. Here, we demonstrate the dependency of improved immunotherapy response on the post-bariatric surgery gut microbiome via fecal microbiota transplantation (FMT). Response to αPD-1 immunotherapy was significantly improved following FMT from formerly obese bariatric surgery-treated mice. When stool from post-bariatric surgery patients was transplanted into recipient mice and compared to the patients' presurgery transplants, postsurgery microbes significantly reduced tumor burden and doubled immunotherapy effectiveness. Microbes impact tumor burden through microbially derived metabolites, including branched-chain amino acids (BCAAs). Circulating BCAAs correlated significantly with natural killer T (NKT) cell content in the tumor microenvironment in donor mice after bariatric surgery and FMT recipients of donor cecal content after bariatric surgery compared with obese controls. BCAA supplementation replicated improved αPD-1 effectiveness in 2 BC models, supporting the role of BCAAs in increased immunotherapy effectiveness after bariatric surgery. Ex vivo exposure increased primary NKT cell expression of antitumor cytokines, demonstrating direct activation of NKT cells by BCAAs. Together, the findings suggest that reinvigorating antitumor immunity may depend on bariatric surgery-associated microbially derived metabolites, namely BCAAs.

Keywords: Breast cancer; Cancer immunotherapy; Immunology; Microbiology; Obesity; Oncology.

Figure 1. Microbes transplanted from VSG donors improved response to αPD-1 immune checkpoint blockade therapy.

(A) Study schema: Female C57BL/6J mice were purchased from The Jackson Laboratory at 4 weeks of age. On day –13, a broad-spectrum antibiotic cocktail (1 g/L neomycin, 1 g/L ampicillin, 1 g/L cefoperazone, and 0.5 g/L vancomycin) was administered for 8 days via drinking water to ablate the commensal gut microbiome prior to fecal microbiota transplantation (FMT). Following 2 FMT gavages (FMT A and B), E0771 breast cancer cells were injected into the fourth right mammary fat pad and monitored for 3 weeks. Two additional FMT gavages occurred following cell injection (FMT C and D). αPD-1 immune checkpoint blockade (ICB) therapy or IgG2a isotype control was administered at 200 μg/mouse intraperitoneally every 3 days from tumor injection to endpoint. Additional controls not depicted in the cartoon are reported in Supplemental Figure 1. (B) Tumor progression was recorded by digital caliper and quantified until endpoint. n = 8–10 per group. (C) Tumor volume at endpoint presented as mean ± SEM with comparisons determined by 2-way ANOVA. n = 8–10 per group. (D) Semisupervised heatmap of RNA-seq transcriptomic analysis of tumors from mice receiving Obese or VSG FMT and αPD-1 ICB or IgG2a isotype control. Top 50% of differentially expressed genes are shown. n = 3–6 per group. (E) Dot plots of gene set enrichment analysis (GSEA) results representing upregulated and downregulated HALLMARK pathways in VSG FMT recipients plotted as αPD-1 ICB versus IgG2a isotype control, with significant nominal P values denoted by dot color and gene count signified by dot size. Pathways are ranked by normalized enrichment score (NES). (F and G) Volcano plots of significant differentially expressed genes in tumors from Obese FMT (F) and VSG FMT (G) recipients plotted as αPD-1 ICB versus IgG2a isotype control with log₂ fold change (FC) of greater than 1.5 and adjusted P value of less than 0.05. n = 3–6 per group. *P < 0.05.

Figure 2. Microbes from patients after bariatric surgery reduced tumor burden and improved response to immunotherapy in recipient mice.

(A) FMT gavages were derived from stool collected from patients before bariatric surgery or 6 months after bariatric surgery transplanted into separate recipient groups. (B) Study schema is identical to Figure 1A with the exception that FMT samples were from patient stool. (C) Tumor progression following FMT from Donor A was recorded by digital caliper. n = 8–10 recipients per donor group. (D) Tumor volume at endpoint following FMT from Donor A presented as mean ± SEM with comparisons determined by 2-way ANOVA. n = 8–10 recipients per donor group. (E) Tumor regression following FMT from Donor A presented as percentage change in tumor volume is shown as rank-sorted treatment groups. The number of mice with regressed tumors is shown out of the total. (F) Tumor progression following FMT from Donor B was recorded by digital caliper. n = 10 recipients per donor group. (G) Tumor volume at endpoint following FMT from Donor B presented as mean ± SEM with comparisons determined by 2-way ANOVA. n = 10 recipients per donor group. (H) Tumor regression following FMT from Donor B presented as percentage change in tumor volume is shown as rank-sorted treatment groups. The number of mice with regressed tumors is shown out of the total. *P < 0.05; ***P < 0.001; ****P < 0.0001.

Figure 3. Microbes transplanted from VSG donors suggest elevated order Clostridiales in response to αPD-1 immune checkpoint blockade therapy.

(A) Relative abundance of microbial families is reported from donor cecal contents harvested from obese sham or VSG donors at endpoint. Relative abundance was calculated using the top 1000 taxa identified using phyloseq (https://www.bioconductor.org/packages/release/bioc/html/phyloseq.html). n = 6–7 per group. (B) The α diversity in recipients across the timeline of the study included baseline stool collected on day –13, postantibiotic stool collected on day –5, and cecal contents taken at endpoint. Shannon and Simpson indices were calculated using phyloseq. n = 8–10 per group for VSG FMT recipients. Boxes represent the interquartile range (IQR) between the first and third quartiles, and the horizontal line inside the box defines the median. Whiskers represent the range of values from highest to lowest. Black dots inside the box represent the mean. (C) Relative abundance of microbial families is reported in recipient cecal contents harvested at FMT study endpoint. n = 8–10 per group. (D and E) Differential abundance analysis of endpoint cecal contents for family Peptostreptococcaceae (D) and order Clostridiales (E). Significant comparisons between relative abundance values were determined by 2-way ANOVA. n = 9–10 per group. *P < 0.05. (F) Abundance of order Clostridiales members in the cecal contents at endpoint were correlated with endpoint tumor volume via MaAsLin2 analysis. R² = 0.12, P = 0.037, n = 38.

Figure 4. Elevated circulating branched chain amino acids were detected after FMT of VSG donor microbes, which correlated with improved antitumor immune response and NKT cells.

(A) Relative concentration of circulating branched chain amino acids (BCAAs, sum of valine, leucine, and isoleucine) quantified by GC-MS analysis of plasma. Data are presented as mean ± SEM with 2-way ANOVA comparisons for n = 9–10 per group. Black bars represent groups treated with IgG2a isotype control, while green bars represent groups treated with α-PD1 ICB. (B–D) Circulating individual BCAAs correlated with tumor volume at endpoint via linear regression analysis for valine (B; R² = 0.12, P = 0.037, n = 36), leucine (C; R² = 0.14, P = 0.022, n = 36), and isoleucine (D; R² = 0.13, P = 0.034, n = 36). (E) Abundance of order Clostridiales members in cecal contents at endpoint were correlated with relative concentration of circulating BCAAs via MaAsLin2 analysis. R² = 0.28, P = 0.029, n = 17. (F–J) Flow cytometric analysis of the tumor immune microenvironment (TIME) is shown as M1-like MHC II^hi macrophages (F; CD11b⁺Ly6C^–Ly6G^–F480⁺MHC II^hi), classical dendritic cells type 1 (G; cDC1s, CD11c⁺MHCII⁺CD11b^loCD10^hi), CD4⁺ memory T cells (H; CD4⁺CD44⁺CD69^–), CD8⁺ memory T cells (I; CD8⁺CD44⁺CD69^–), or NK/NKT cells (J, NK1.1⁺) out of total CD45⁺ immune cells. Data are presented as mean ± SEM with 2-way ANOVA comparisons for n = 6–9 per group. *P < 0.05; ***P < 0.001; ****P < 0.0001. (K) Percentage of NK/NKT cells out of total immune cells within the TIME were correlated with tumor volume at endpoint via linear regression analysis. R² = 0.27, P = 0.002, n = 33. (L) Percentage of NK/NK T cells out of total immune cells within the TIME were correlated with circulating BCAA levels via linear regression analysis. R² = 0.16, P = 0.034, n = 29.

Figure 5. Branched chain amino acid supplementation mimics the effects of the post–bariatric surgery microbiome.

(A) Study schema: Female C57BL/6J mice were purchased from The Jackson Laboratory at 4 weeks of age. On day –7, a branched chain amino acid (BCAA) cocktail (15 g/L leucine, 15 g/L isoleucine, and 15 g/L valine) was administered via drinking water for the duration of the study. Following 7 days of BCAA supplementation, E0771 BC cells were injected into the fourth right mammary fat pad and monitored for 3 weeks. αPD-1 ICB therapy or IgG2a isotype control was administered at 200 μg/mouse intraperitoneally every 3 days from tumor injection to endpoint. (B) E0771 tumor progression was recorded by digital caliper. n = 9–10 per group. (C) E0771 tumor volume at endpoint is presented as mean ± SEM with comparisons determined by 2-way ANOVA; n = 9–10 per group. (D and E) Flow cytometric analysis of E0771 tumor immune microenvironment (TIME) shown as NK cells out of total CD45⁺ immune cells (D, NK1.1⁺CD3^–) and invariant NKT (iNKT) cells (E; CD3⁺, PBS57-loaded mouse CD1d tetramer⁺) out of total CD3⁺ cells. Mean ± SEM with 2-way ANOVA for n = 5–9 per group. (F) Percentage of iNKT cells out of total CD3⁺ cells within E0771 TIME were correlated with tumor volume at endpoint via linear regression analysis. R² = 0.13, P = 0.046, n = 30. (G) 4T1 tumor progression in BALB/c mice was recorded by digital caliper. n = 5 per group. (H–K) Flow cytometric analysis of 4T1 TIME shown as iNKT cells (H, PBS57-loaded mouse CD1d tetramer⁺) out of total CD45⁺ cells, and CD69⁺ iNKT cells (I), IFN-γ⁺ iNKT cells (J), and TNF-α⁺ iNKT cells (K) out of total iNKT cells. Data presented as mean ± SEM with 2-way ANOVA for n = 5 per group. *P < 0.05; ***P < 0.001; ****P < 0.0001.

Figure 6. Branched chain amino acid treatment increases iNKT cell production of proinflammatory cytokines.

(A) Study schema: Liver lymphocytes were isolated from female C57BL/6J mice and FACS was used to isolate iNKT cells via PBS57-loaded mouse tetramer staining. Cells were expanded in culture for 20 days. On day 20, cells were treated with a branched chain amino acid (BCAA) cocktail (10 mM leucine, 10 mM isoleucine, and 10 mM valine) for 24 hours. On day 21, cells were stained for flow cytometric analysis. (B–D) Flow cytometric analysis of cultured cells at endpoint shown as iNKT cells (B, PBS57-loaded mouse tetramer⁺) out of total live cells, IFN-γ⁺ cells (C) and TNF-α⁺ cells (D) out of iNKT cells. Data presented as mean ± SEM with 2-tailed Student’s t test for n = 4 wells per group. *P < 0.05, **P < 0.01. (E) Proposed model: NKT cells are induced to mature by dendritic cells. Mature NKT cells activate antitumor immune cells including T cells and NK cells, which induce cancer cell death. Findings presented suggest that BCAAs are elevated in mice after FMT from VSG donors and are essential drivers of the NKT cell antitumor immune response. Solid lines indicate known interactions, while dashed lines represent proposed interactions.