Formate Supplementation Enhances Antitumor CD8+ T-cell Fitness and Efficacy of PD-1 Blockade

Abstract

The tumor microenvironment (TME) restricts antitumor CD8+ T-cell function and immunotherapy responses. Cancer cells compromise the metabolic fitness of CD8+ T cells within the TME, but the mechanisms are largely unknown. Here we demonstrate that one-carbon (1C) metabolism is enhanced in T cells in an antigen-specific manner. Therapeutic supplementation of 1C metabolism using formate enhances CD8+ T-cell fitness and antitumor efficacy of PD-1 blockade in B16-OVA tumors. Formate supplementation drives transcriptional alterations in CD8+ T-cell metabolism and increases gene signatures for cellular proliferation and activation. Combined formate and anti-PD-1 therapy increases tumor-infiltrating CD8+ T cells, which are essential for enhanced tumor control. Our data demonstrate that formate provides metabolic support to CD8+ T cells reinvigorated by anti-PD-1 to overcome a metabolic vulnerability in 1C metabolism in the TME to further improve T-cell function.

Significance: This study identifies that deficiencies in 1C metabolism limit the efficacy of PD-1 blockade in B16-OVA tumors. Supplementing 1C metabolism with formate during anti-PD-1 therapy enhances CD8+ T-cell fitness in the TME and CD8+ T-cell-mediated tumor clearance. These findings demonstrate that formate supplementation can enhance exhausted CD8+ T-cell function. See related commentary by Lin et al., p. 2507. This article is featured in Selected Articles from This Issue, p. 2489.

Figure 1:. Co-culture of tumor cells with T cells reveals TCR dependent activation of serine biosynthesis.

A) Schematic representation of culture experiments using activated OT-1 T cells in mono- or co-culture with tumor cells expressing cognate peptide (B16-OVA) or not (B16) for metabolite profiling. B) Carbon fates of ¹³C-glucose tracing. Percent glucose derived C) serine M+3, D) glycine M+2, E) pyruvate M+3, and F) lactate M+3 from OT-1 cells in mono- (light gray), B16-OVA co- (red), or B16 co- (dark gray) cultures. G) Carbon fates of ¹³C-serine and ¹³C-glycine tracing. H) ¹³C-serine derived serine M+3 and I) ¹³C-glycine derived serine M+2 from OT-1 cells in mono- (light gray) or B16-OVA co- (red) cultures. J) Schematic representation of OT-1 activation, proliferation, and tumor cell killing in vitro. Representative K) cell trace violet (CTV) and L) CD44 expression flow cytometry plots, and quantified M) percent (%) divided and N) CD44 expression following in vitro CD8⁺ T cell activation in control (RPMI) or serine and glycine (Ser/Gly) deficient media. O) Percentage of B16-OVA cell killing in vitro by activated OT-1 cells from control (RPMI) or serine and glycine (Ser/Gly) deficient media. Data are representative of 3 independent experiments. Significance determined using Student’s unpaired t test *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001.

Figure 2:. Activation of CD8⁺ T cell serine metabolism is TCR dependent.

A) Schematic of serine biosynthesis and the folate cycle. B) Schematic representation of in vitro T cell activation. C) Western blot from bulk CD8⁺ T cells at indicated time points for PHGDH, SHMT1, SHMT2, and β-actin. D) % glucose-derived serine (¹³C-glucose tracing) in naïve and anti-CD3/CD28 activated CD8⁺ T cells. E) Schematic representation of naïve OT-1 cell activation with varying concentrations of SIINFEKYL peptide in vitro. F) Western blot from CD8⁺ T cells at indicated concentrations of SIINFEKYL for PHGDH, SHMT1, SHMT2, and β-actin. G) % glucose-derived serine (¹³C-glucose tracing) in naïve CD8⁺ T cells activated with the indicated concentrations of SIINFEKYL. H) Schematic representation of activation of naïve CD8⁺ T cells with anti-CD3/anti-CD28 plus IL-2 in vitro in the presence of DMSO (control) or the SHMT1/2 inhibitor (SHMTi) +/− formate rescue for three days. I) Representative (left) and quantified (right) CTV dilution and J) CD44 expression in the naive CD8⁺ T cells activated with anti-CD3/anti-CD28 plus IL-2 and treated with vehicle (DMSO), SHMTi, or SHMTi with formate rescue. Data are representative of 2 (western blots) or 3 independent experiments. Significance determined using Student’s unpaired t test *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001.

Figure 3:. Serine depletion in the TME drives CD8⁺ T cell serine biosynthesis.

A) Metabolite analysis of B16-OVA tumor cell conditioned RPMI media (for 24 hours) normalized to RPMI (log₂ fold change). B) % glucose-derived serine (¹³C-glucose tracing) in B16-OVA or B16 cells cultured in control (RPMI) or tumor conditioned media (CM) alone or with OT-1 CD8⁺ T cells. C) Identification of metabolites in the tumor interstitial fluid (TIF) in vivo. Metabolite analysis of plasma (gray) or TIF (green) for D) serine and E) glucose from mice with B16-OVA tumors (days 14–21 from implantation). F) Schematic representation of in vivo competition assay of CRISPR modified CD8⁺ T cells. G) Paired analysis (normalized to the input ratio) of Control (Cntrl)-gRNA (gray) or Phgdh-gRNA (Phgdh-KO, red) OT-1 cells isolated from B16-OVA tumors of mice (14 days post implantation). Data are representative of 3 independent experiments. Significance determined using Student’s unpaired t test **p<0.01, ****p<0.0001.

Figure 4:. Formate synergizes with anti-PD-1 to enhance tumor clearance in B16-OVA tumors.

A) Schematic representation of formate and anti-PD-1 treatment. B) Tumor growth in mice given control (gray) or formate (red) drinking water (5 mg/mL) and rat IgG2a isotype control antibody (100 μg/dose). C) Tumor growth in mice given control (blue) or formate (purple) drinking water and anti-PD-1 (αPD-1) antibody (100 μg/dose). D) Survival of mice from B/C. E) Tumor growth in mice following CD8β cell depletion in control (gray) or formate (red) drinking water (5 mg/ml) given isotype control antibody (100 μg/dose). F) Tumor growth in mice following CD8β cell depletion in control (blue) or formate (purple) drinking water (5 mg/ml) given anti-PD-1 (αPD-1) antibody. G) Survival of mice from E/F. H) Schematic representation of formate and anti-PD-1 therapeutic model. I) Tumor growth in mice given control (gray) or formate (red) drinking water (5 mg/ml) and treated isotype control antibody (100 μg/dose). J) Tumor growth in mice given control (blue) or formate (purple) drinking water (5 mg/ml) and treated with anti-PD-1 (αPD-1) antibody (100 μg/dose). K) Survival of tumor implanted mice from I/J. Data are representative of 3 independent experiments. Significance values determined using Students unpaired t test **p<0.01, ***p<0.001, ****p<0.0001. Survival log-rank Mantel-Cox test.

Figure 5:. Single cell RNA-seq reveals formate therapy metabolically rewires CD8⁺ T cells in anti-PD-1 therapy.

A) Schematic representation of isolation of tumor-infiltrating CD8⁺ T cell populations for single cell RNA-seq (scRNA-seq) analysis from mice given control or formate (5 mg/mL) drinking water (starting day 9 post tumor implantation) and treated with rat IgG2a isotype control or anti-PD-1 mAb (100 μg/dose on day 10 post tumor implantation). B) Uniform manifold approximation and projection (UMAP) of single cell RNA-seq profiles. C) Barplots comparing percentages of individual CD8⁺ T cell populations between treatment groups: Control (Cntrl), Formate (Form.), Isotype control antibody (isotype), and anti-PD-1 (αPD-1). D) Comparison of KEGG metabolic gene signature scores between Control αPD-1 (blue) and Formate αPD-1 (purple). E) Comparison of Control αPD-1 and Formate αPD-1 KEGG metabolic gene signature scores by individual cell cluster, with differences between cell cluster gene expression FDR significance (p<0.05) indicated by red circles and non-significance (p≥0.05) indicated by gray circles (for panels F, G, I, J, and K). Comparison of Control plus αPD-1 versus Formate αPD-1 of F) Glycosphingolipid 3 and G) Glycolysis/Gluconeogenesis KEGG pathway scores by cell cluster. H) Heatmap of gene expression for enzymes of the one carbon metabolic pathway between treatment groups, and by cluster comparison of Control αPD-1 versus Formate αPD-1 of I) one carbon (1C) pathway enzymes, J) Phgdh, and K) Shmt2. Each group represents 2–3 replicates of purified CD8⁺ T cells pooled from 3 mice. Significance values determined using binomial test **p<0.01, ***p<0.001, ****p<0.0001.

Figure 6:. Formate therapy increases CX3CR1⁺CD8⁺ T cells in B16-OVA tumors.

A) Schematic representation of isolation of tumor infiltrating lymphocytes (TIL) for flow cytometric analysis on day 14 post B16-OVA tumor cell implantation and therapeutic formate and anti-PD-1 administration. Representative plots (left) and numbers (right) of OVA_257–264 specific CD8⁺ T cells per mg of tumor tissue from mice given no formate (control) or formate (5 mg/mL in drinking water) and receiving either B) rat IgG2a isotype or C) anti-PD-1 (αPD-1) antibodies each at 100 μg/dose. Representative plots (left) and numbers (right) of BrdU⁺ OVA-specific CD8⁺ T cells per mg tumor tissue in control versus formate treated animals receiving either D) isotype or E) anti-PD-1 (αPD-1) antibodies. Representative histograms (left) of tumor-infiltrating Granzyme B⁺ OVA-specific CD8⁺ T cells compared with dLN CD8⁺CD44⁺ cells from the draining lymph node (dark gray) within the same mouse and total number (right) per mg of tumor tissue animals receiving treatment with either F) control isotype (light gray) and formate isotype (red fill) or G) control anti-PD-1 (αPD-1) (blue fill) and formate anti-PD-1 (αPD-1) (purple fill). Representative plots (left) and numbers (right) of CX3CR1⁺TIM-3⁺ OVA-specific CD8⁺ T cells per mg tumor tissue in formate versus control animals receiving either H) isotype or I) anti-PD-1 (αPD-1) antibodies. Data are representative of 3 independent experiments. Significance values determined using Student’s unpaired t test *p<0.05, **p<0.01.