Metabolization of microbial postbiotic pentanoate drives anti-cancer CAR T cells

Abstract

The microbiome is a complex host factor and key determinant of the outcome of antibody-based and cellular immunotherapy. Its postbiotics are a blend of soluble commensal byproducts that are released into the host environment and have been associated with the regulation of immune homeostasis, particularly through impacts on epigenetics and cell signaling. In this study, we show that the postbiotic pentanoate is metabolized to citrate within the TCA cycle via both the acetyl- and succinyl-CoA entry points, a feature uniquely enabled by the chemical structure of the C5 aliphatic chain. We identified ATP-citrate lyase as the crucial factor that redirects pentanoate-derived citrate from the succinyl-CoA route to the nucleus, thereby linking metabolic output and histone acetylation. This epigenetic-metabolic crosstalk mitigated T cell exhaustion and promoted naive-like differentiation in pentanoate-programmed chimeric antigen receptor (CAR) T cells. The predictive and therapeutic potential of pentanoate was corroborated in two independent patient cohorts and three syngeneic models of CAR T adoptive therapy. Our data demonstrate that postbiotics are integrated into mitochondrial metabolism and subsequently incorporated as epigenetic imprints. This bridge between microbial and mammalian interspecies communication can ultimately impact T cell differentiation and efficacy.

Figure 1:. Pentanoate abundance is a predictor of clinical CAR T cell response.

a-d, Kaplan-Meier curves showing PFS (upper panel) and OS (lower panel) depending on the abundance of pentanoate (a), butyrate (b), propionate (c) and acetate (d). e and f, Probability of PFS (e) and overall survival (f) in a cohort of patients receiving CAR T cell therapy depending on the pentanoate concentration. g, h, Probability of PFS (g) and overall survival (h) in a cohort of patients receiving CAR T cells depending on different antibiotic-(ABX)-treatment regimen. PIM (piperacillin/tazobactam, imipenem, meropenem); PFS, progression-free survival; OS, overall survival. i, Levels of multiple short-chain fatty acids following antibiotic exposure.

Figure 2:. Microbial metabolites improve engineering and boost CTL-phenotype of murine CAR T cells.

a, Schematic illustration of experimental setup for murine CAR T cell generation with pretreatment and functional analysis. b, Representative contour plots of the transduction marker and bar plots show frequencies of CD8⁺ BCMA- CAR T cells on day 6. Mean ± SEM from n = 3. c, Mean fluorescence intensity (MFI) of the transduction marker of generated CAR⁺ T cells. Mean ± SEM from n = 3. d, Total CAR⁺ T cell yield and viability on day 7. Mean ± SEM from n = 3. e, Representative contour plots and bar graphs show transduction efficiency of CD8 CD19- CAR T cells on day 6. Mean ± SEM from n = 3. f, Mean fluorescence intensity (MFI) for generated CAR⁺ T cells. Mean ± SEM from n = 3. g, Yield and viability of CAR⁺ T cells on day. Mean ± SEM from n = 3. h, Representative contour plots of transduction marker and bar graph show frequency of CD8⁺ ROR1- CAR T cells on day 6. Mean ± SEM from n = 3. i, Mean fluorescence intensity (MFI) of the transduction marker for generated CAR⁺ T cells. Mean ± SEM from n = 3. j, Yield and viability of ROR1- CAR⁺ T cells on day. Mean ± SEM from n = 3. k, Frequency of IFN-γ and TNF-α producing BCMA- CAR⁺ T cells on day 4 following antigen-independent restimulation for 5 hours. Mean ± SEM from n = 3. l, Flow cytometric analysis of Granzyme B and IL-2 production by CD8⁺ BCMA- CAR T cells after restimulation on day 4. Mean ± SEM from n = 4. m, Specific cytolytic activity of BCMA- CAR T cells at different E:T ratios at 4 and 8h timepoint. Mean ± SEM from n = 3. n, Cytokine secretion of IFN-γ, TNF-α and IL-2 after 24-hour co-culture of BCMA- specific CAR⁺ T cells with target cells. Mean ± SEM from n = 3. (b-n) Data represent pooled data from independent experiments. (m and n) n=3 biological replicates; pooled data from n=3 independent experiments; mean ± SEM was calculated for n=3 independent experiments. Statistical analysis was performed using unpaired two-tailed Student’s t test (b-l) and two-way analysis of variance (ANOVA) with Tukey’s multiple-comparison test (m and n).

Figure 3:. Anti-tumor reactivity in hematological and solid malignancies.

a, Schematic representation of the experimental timeline of BCMA- CAR T cells in a MM5080 tumor model. b, Kaplan-Meier analysis of overall survival for all tested mice (n = 5 mice/group). c, Scheme illustrating the analysis of ROR1-CAR T cells in a MC38ROR1 tumor model. d, Survival of MC38ROR1 tumor bearing mice. Mean ± SEM from n = 8 mice for control group, n = 9 for CAR or CAR_Penta groups. e, Tumor weight at the endpoint. Mean ± SEM from n = 6 for CAR and n = 8 for CAR_Penta groups. f, Single tumor weight over the course of the experiment. g, Scheme illustrating the analysis of ROR1- CAR T cells in a PancROR1 tumor model. h, PancROR1 tumor growth after treatment with CD8⁺ CAR T cells. i, Tumor volume at endpoint day 21. Mean ± SEM from n = 3 mice/group for mock, n = 6 mice/group for CAR treated groups. j, Frequency of CD8⁺ ROR1- CAR T cells (left panel) and endogenous CD8⁺ T cells in the tumor on day 21. Mean ± SEM from n = 5 for CAR, n = 6 for CAR_Penta. k, Frequency of CD8⁺ CAR T cells in the draining lymph node (dLN), spleen, bone marrow and blood on day 21. Mean ± SEM from spleen, BM and blood; Mean ± SEM from n = 4 for CAR, n = 5 for CAR_Penta for dLN. l – n, Flow cytometric analysis of IFN-γ⁺ TNF-α⁺ and GranzymeB⁺IL-2⁺-double positive ROR1-CAR T cells from the spleen (l), dLN (m) and tumor (n) following restimulation with PMA/Ionomycin for 5 h. Mean ± SEM from n = 5 mice/group and n= 4 mice/group. Survival curves were compared by the log-rank Mantel-Cox test (b and d). Statistical analysis was performed using unpaired two-tailed Student’s t test (e,i,j - n).

Figure 4:. HDAC class I inhibition-mediated hyperacetylation improves CAR T cell function.

a, Schematic pathway illustration detailing the HDAC-mediated T cell regulation. b and c, Shown are SCFAs (blue sticks), i.e. acetate, propionate, butyrate, pentanoate and hexanoate docked into AlphaFold 3-predicted structures of zinc cofactor-bound (spheres) HDAC1 (gray cartoons). d, Influence of bacterial SCFAs on the activity of recombinant class I and class II HDAC enzymes. e, Mocetinostat (blue sticks, HDAC class I inhibitor) docked into an AlphaFold 3-predicted model of zinc cofactor-bound (spheres) HDAC1 (gray cartoons). f and g, Histone acetylation status of T cells are measured by the expression of H3/K9–14 (f) and H3K27 (g) via flow cytometry treated with indicated substances. 1 out of 3 representative experiments is shown. h, Specific cytolytic activity of CD8⁺ CAR T cells generated in the presence of indicated substances against ROR1-expressing tumor cell lines at different E:T ratios after 6 h. Mean ± SEM from n = 3. i Cytokine secretion of CD8⁺ CAR T cells after 24 h co-culture with target antigen expressing cell lines, measured for IFN-γ, TNF-α and IL-2 by ELISA. Mean ± SEM from n = 3. h and i, biological replicates; pooled data from independent experiments. Statistical analysis was performed using one-way ANOVA (h) or two-way analysis of variance (ANOVA) with Tukey’s multiple-comparison test (i).

Figure 5:. Pentanoate synergizes epigenetic and metabolic modulation to boost effector function.

a, Schematic pathway representation. b, Flow cytometry analysis and quantification of PGC-1α in CD8⁺ CAR T cells either untreated or treated with pentanoate. Mean ± SEM from n = 3. c, Flow cytometric analysis of MitoFM in CD8⁺ CAR T cells treated with indicated substances. Mean ± SEM from n = 3. d, Representation of workflow with combinatorial or single treatment for CD8⁺ T cells. e and f, Cytolytic activity of ROR1- CAR T cells against PancROR1 (e) and MC38ROR1 (f) at variable E:T ratios after 4 h and 8 h. Mean ± SEM from n = 3. g-j, Secretion of IFN-y (g), IL-2 (h) and TNF-α (i) after 24 hours by CD8⁺ CAR T cells upon co-incubation with ROR1-expressing tumor cell lines. Mean ± SEM from n = 3. j, Scheme illustrating the different pretreatments of CAR⁺ T cells generated from wildtype or HDAC knockout mice. k, Cytolytic activity of CD8⁺ CAR T cells against target cells with different E:T ratios after 4 h. CAR T cells were generated as shown in j. Mean ± SEM from n = 3. l, Experimental setup of pretreatment with pentanoate or DCA/mocetinostat combination during the CAR⁺ T cell manufacturing process followed by injection in PancROR1 tumor bearing mice. m, Tumor growth (left) over time as well as tumor volume (middle) and tumor weight on day 14 (right) after tumor inoculation are shown. Mean ± SEM from n = 10 mice/group until day 7, then n = 5 mice/group until day 14. b, c, e-l and k, independent experiments. e-l, k, biological replicates; pooled data from independent experiments. Statistical analysis was performed using one-way ANOVA (b, c, m) or two-way analysis of variance (ANOVA) with Tukey’s multiple-comparison test (e-I and k).

Figure 6:. Pentanoate hijacks the cellular metabolism and becomes part of the epigenetic imprint.

a, Scheme illustrating pentanoate modulation and metabolism hijacking of T cells. b, Schematic for the generation of CAR⁺ T cells in combination with pretreatment prior to metabolite tracing c, GC-MS isotope tracing of ¹³C-glucose-derived metabolites in CAR T cells engineered in the presence of specified substances. Histograms show fractional enrichment of ¹³C-glucose-derived TCA metabolites. Mean ± SEM from n = 3 biological replicates. d, GC-MS isotope tracing of ¹³C-labeled glutamine in untreated or pentanoate-treated CAR T cells regarding citrate, α-ketoglutarate and malate. Mean ± SEM from n = 3 biological replicates. e, GC-MS tracing of ¹³C-pentanoate in CD8⁺ CAR T cells pretreated with indicated substrates. Histograms show the frequency of pentanoate-derived carbons in the downstream TCA cycle intermediates. n=2 independent T cell pools. f and g, PancROR1 (f) or MC38ROR1 (g) tumor killing of CAR⁺ T at different E:T ratios after 6 h. Mean ± SEM from n = 3. Statistical analysis was performed using unpaired two-tailed Student’s t test (d), one-way ANOVA (c) or two-way analysis of variance (ANOVA) with Tukey’s multiple-comparison test (f and g).

Figure 7:. Epigenetic- metabolic reprogramming induces a naïve-like phenotype in vivo.

a, Schematic representation of PancROR1 tumor model for the collection of T cells for scRNA seq. b, Uniform Manifold Approximation and Projection (UMAP) showing the distribution of immune cells collected from mice using scRNAseq data in the LSI space. Each point represents one cell. The cells are marked by color code based on cell annotations. Red indicates T cells, blue non-T cells. c, UMAP showing the distribution of T cells based on their cell states. The cells are marked by color code based on the different cluster they belong to. d and e, Barplots indicating the changes in percentage of the different CD4 (d) and CD8 (e) T cell subsets noticed following the different treatments and time points (day 0, 7 and 14).