Breast tumor microbiome regulates anti-tumor immunity and T cell-associated metabolites

Abstract

Background: Breast cancer, the most common cancer type among women, was recently found to contain a specific tumor microbiome, but its impact on host biology remains unclear. CD8⁺ tumor-infiltrating lymphocytes (TILs) are pivotal effectors of anti-tumor immunity that influence cancer prognosis and response to therapy. This study aims to elucidate interactions between CD8⁺ TILs and the breast tumor microbiome and metabolites, as well as how the breast tumor microbiome may affect the tumor metabolome.

Methods: We investigated the interplay among CD8⁺ TILs, the tumor microbiome, and the metabolome in a cohort of 46 breast cancer patients with mixed subtypes (Cohort A). We characterized the tumor metabolome by mass spectrometry and CD8⁺ TILs by immunohistochemistry. Microbiome composition and T cell gene transcript levels were obtained from data from our previous study, which utilized 16S rRNA gene sequencing and a targeted mRNA expression panel. To examine interactions between intratumoral Staphylococcus and specific breast cancer subtypes, we analyzed RNA sequencing data from an independent cohort of 370 breast cancer patients (Cohort B). We explored the functions of the tumor microbiome using mouse models of triple-negative breast cancer (TNBC).

Results: In tumors from Cohort A, the relative abundance of Staphylococcus positively correlated with the expression of T cell activation genes. The abundances of multiple metabolites exhibited significant correlations with CD8⁺ TILs, of which NADH, γ-glutamyltryptophan, and γ-glutamylglutamate displayed differential abundance in Staphylococcus-positive versus Staphylococcus-negative breast tumors. In a larger breast cancer cohort (Cohort B), we observed positive correlations between tumoral Staphylococcus and CD8⁺ TIL activity exclusively in TNBC. Preclinical experiments demonstrated that intratumoral administration of S. aureus, the predominant species of Staphylococcus in human breast tumors, resulted in a depletion of total NAD metabolites, and reduced the growth of TNBC tumors by activating CD8⁺ TILs.

Conclusions: We identified specific metabolites and microbial taxa associated with CD8⁺ TILs, delineated interactions between the breast tumor microbiome and metabolome, and demonstrated that intratumoral Staphylococcus influences anti-tumor immunity and TIL-associated metabolites. These findings highlight the role of low-biomass microbes in tumor tissues and provide potential biomarkers and therapeutic agents for breast cancer immunotherapy that merit further investigation.

Keywords: Microbiome; S. aureus; Staphylococcus; T cells; anti-tumor immunity; breast cancer; metabolism; metabolome; tumor microenvironment; tumor-infiltrating lymphocytes.

Figure 1 |. Presence of tumoral bacteria and their correlations with TIL abundance and activation genes.

a RNAscope-fluorescence in situ hybridization (FISH) images of human breast tumors using a probe targeting the bacterial 16S gene (yellow). Nuclei were stained with DAPI (blue). Scale bars are shown in white at 2000 and 25 μm in the images of low and high magnification, respectively. b, c Heatmaps displaying correlations between the abundance of bacterial genera and CD8⁺ and FoxP3⁺ cells (b) as well as the genes indicative of T cell activation, including GZMA/B/K and IFNG (c). Spearman’s correlation (b, c). *P < 0.05.

Figure 2 |. Identification of tumoral metabolites associated with CD8⁺ TILs.

a Categorization of hot and cold tumors based on CD8⁺ cell density quantified using immunohistochemistry. b Principal component analysis (PCA) of metabolites in hot and cold tumors (n=11 and 35, respectively). c The left panel shows the two statistical approaches employed to identify hot and cold metabolites. The right figure depicts hot and cold metabolites that exhibit significant p-values in both analyses in red. d Heatmap showing the identified hot and cold metabolites. The color gradient represents the log₂ fold change (FC) of each metabolite between hot and cold tumors (left column) and the correlation between each metabolite and CD8⁺ cell density (right column). e Volcano plot showing hot and cold metabolites (both highlighted in red) derived from Mann-Whitney U test (left) and Spearman’s rank correlation (right). f The abundance of NAD⁺ (left panel) and NADH (right panel) in healthy breast tissues (Healthy, n=25), cold tumors (n=35), and hot tumors (n=11). g Heatmap representing correlations between transcript levels of T cell-related genes and the abundance of NAD⁺, NADH, taurolithocholate 3-sulfate, and 3-formylindole. One-way analysis of variance (ANOVA) with multiple comparisons (f). Spearman’s correlation (g). *P < 0.05; **P < 0.01; ***P < 0.001.

Figure 3 |. Staphylococcus-associated tumoral metabolites.

a Heatmap displaying differentially abundant metabolites in Staphylococcus-positive compared to Staphylococcus-negative (Staph+ vs −) human breast tumors. The color gradient represents the log₂ fold change (FC) relative to the Staphylococcus-negative group. Types of metabolites and their associated breast cancer features (detailed in Fig. 2, supplementary Fig. 3, and supplementary Table 1) are indicated on the left side of the heatmap. Unpaired two-tailed Student’s t-test.

Figure 4 |. Associations between Staphylococcus, clinicopathological features, and CD8⁺ TILs.

a, b Percentages of Cohort B human breast tumor samples identified as positive for Staphylococcus (Staph+ tumors) across cancer subtypes (a) and stages (b). c Kaplan–Meier survival curves illustrating overall survival among patients of all breast cancer subtypes (left panel), TNBC (middle panel), and ER+/PR+ subtype (right panel) stratified by the presence or absence of intratumoral Staphylococcus (Staph+ and Staph-, shown as red and black lines, respectively). All p-values > 0.05 by log-likelihood test. d, e Comparisons of CD8+ T cell abundance (d) and activity (e) in TNBC and ER+/PR+ tumors with and without the presence of Staphylococcus (shown as red and black dots, respectively). f, g Comparisons of T cells-related genes in TNBC (f) and ER+/PR+ tumors (g) with and without the presence of Staphylococcus. Z-score transformed signature scores, and log-transformed transcripts per kilobase-million (TPKM) counts were compared by t-test. ns, not significant.

Figure 5 |. Intratumoral S. aureus modulates CD8⁺ TILs and total NAD level in TNBC models.

a, b Growth of 4T1 (a) and EO771 (b) tumors after intratumoral injection of various bacterial species compared to PBS-treated control (n=3–5). c Growth of EO771 tumors after intratumoral injection of S. aureus with and without antibody-based depletion of CD8⁺ or CD4⁺ T cells (n=4). d Representative immunohistochemistry (IHC) images showing GzmB⁺ and CD8⁺ cells in EO771 tumors after intratumoral injection of S. aureus (SA) or S. mitis (SM), with PBS treatment as a control. Scale bars are shown in black at 50 μm. e Quantification of densities of GzmB⁺ and CD8⁺ cells based on IHC staining shown in d. f Representative contour plots of flow cytometric analysis showing the percentage of GzmB-expressing CD8+ T cells in EO771 tumors after intratumoral injection of S. aureus or S. mitis, with PBS treatment as a control. g Quantification of the percentage of GzmB+ cells among CD8⁺ T cells based on the flow cytometric plots shown in f. h Total levels of NAD+ and NADH in the EO771 tumors after intratumoral injection of S. aureus or S. mitis, with PBS treatment as a control. Two-way analysis of variance (ANOVA) with multiple comparisons (a-c). One-way ANOVA with multiple comparisons (e, g, h). NS, not significant.