Studies have identified a link between specific microbiome-derived bacteria and immune checkpoint blockade (ICB) efficacy. However, these species lack consistency across studies, and their immunomodulatory mechanisms remain elusive. To understand the influence of the microbiome on ICB response, we studied its functional capacity. Using pan-cancer metagenomics data from ICB-treated patients, we showed that community-level metabolic pathways are stable across individuals, making them suitable for predicting ICB response. We identified several microbial metabolic processes significantly associated with response, including the methylerythritol 4-phosphate (MEP) pathway, which was associated with response and induced Vδ2 T cell-mediated antitumor responses in patient-derived tumor organoids. In contrast, riboflavin synthesis was associated with ICB resistance, and its intermediates induced mucosal-associated invariant T (MAIT) cell-mediated immune suppression. Moreover, gut metabolomics revealed that high riboflavin levels were linked to worse survival in patients with abundant intratumoral MAIT cells. Collectively, our results highlight the relevance of metabolite-mediated microbiome-immune cell cross-talk. SIGNIFICANCE: Microbial metabolic pathways are highly conserved across individuals and therefore offer an opportunity to link the microbiome to immunotherapy efficacy. We identified specific microbial metabolic pathways associated with response to ICB and provided mechanistic insights into the immunomodulatory influence of these pathways on antitumor immunity.

Bacteriophages (phages) are viruses that specifically attack bacteria. Their use as therapeutics, which constitutes a promising alternative to antibiotics, heavily relies on selecting effective lytic phages against the pathogen of interest. Current selection techniques are laborious and do not allow for direct visualization of phage infection dynamics. Here, we present a method that circumvents these limitations. It can be scaled for high-throughput and permits monitoring of the phage infection in real time via a fluorescence signal readout. This is achieved through the use of a membrane-impermeant nucleic acid dye that stains the DNA of damaged or lysed bacteria and new phage progeny. We have tested the method on Pseudomonas aeruginosa and Klebsiella pneumoniae and show that an increase in fluorescence reflects phage-mediated killing. This is confirmed by other techniques including spot tests, colony plating, flow cytometry and metabolic activity measurements. Furthermore, we illustrate how our method may be used to compare the activity of different phages and to screen the susceptibility of clinical isolates to phage. Altogether, we present a fast, reliable way of selecting phages against Gram-negative bacteria, which may be valuable in optimizing the process of selecting phages for therapeutic use.