Although obesity is associated with cancer risk, the effects of obesity on the tumor microenvironment (TME) and antitumor immune responses are not fully known. Ringel and Drijvers et al. investigated how a high-fat diet shapes the TME using metabolic, proteomic, and genomic techniques. Their results, recently published in Cell, suggest that diet affects the metabolic profile of tumor cells, which limits CD8+ tumor-infiltrating lymphocyte (TIL) responses.
Ringel and Drijvers et al. used a mouse model fed a control diet (CD) or a high-fat diet (HFD). HFD mice gained more weight and displayed obesity-associated metabolic changes. The mice were inoculated with tumor lines with varying immunogenicity. MC38, E07771 (breast), B16 and B16-OVA (melanoma), and CT26 (colon) tumors grew faster in HFD mice, which was positively correlated with immunogenic status. To evaluate whether this growth rate difference was due to effects on T cells, the MC38 tumors were established in mice with a T cell receptor alpha knockout, as well as CD8-depleted mice. No difference in tumor growth was found between the two diets, suggesting that a high-fat diet may inhibit CD8+ T cell-specific antitumor responses.
The researchers then profiled the immune landscape of MC38 tumors in the HFD and CD mice using flow cytometry. The tumors in HFD mice had a smaller proportion of CD8+ T cells within the CD45+ compartment, while this difference was not seen in the spleen or draining lymph node. There were no differences in the number of NK cells, CD4+ T cells, or CD11c+ dendritic cells (or expression levels of MHC-I, MHC-II, or CD40 on CD11c+ cells); diet also had no effect on the expression of MHC-I or PD-L1 on the tumor cells. However, HFD tumors had a higher percentage of immunosuppressive GR1+CD11b+ myeloid-derived suppressor cells and F4/80+GR1-CD11b+ tumor-associated macrophages.
As the most dramatic effects were seen in the CD8+ compartment, Ringel and Drijvers et al. then studied markers of T cell function. CD8+ TIL in HFD mice had a lower expression of Ki67, ICOS, PD-1, and granzyme B, suggesting a decreased activation phenotype. Spleen CD8+ T cells were similar in both diets, indicating that the differences found in T cells were specific to the TME.
The researchers then moved to single-cell RNAseq of TIL. Like the flow cytometry data, fewer lymphocytes were found in HFD mice, but no difference in myeloid populations was observed. A curated set of 61 metabolic gene signatures was mapped on the cells to assess metabolic changes in each cell compartment. Leukocytes from the HFD mice were enriched for pathways associated with fat and cholesterol metabolism, folate biosynthesis, and pentose and glucuronate interconversion. Monocyte, T cell, and M2 macrophage clusters were also affected by HFD and had metabolic signatures related to glycolysis and the TCA cycle.
A TIL localization study using immunofluorescence showed that TIL were not restricted to the tumor margins, a hallmark of immune exclusion. Interestingly, GLUT1-high regions were associated with fewer T cells, particularly in HFD mice, suggesting that T cell infiltration was influenced by the local metabolic niche.
In the HFD TIL, the researchers found reductions in chemokine and T cell receptor signaling. Metabolic signatures in these TIL were associated with naive or unstimulated T cells. There were also transcriptional differences in genes involved in fat synthesis and cholesterol metabolism. These changes did not overlap with those found in tumor cells, suggesting TIL and tumor cells adapted to HFD differently. Focusing on potential tumor-specific changes, differential gene expression analysis revealed that in HFD tumor cells, there was a significant decrease in the expression of Phd3, a prolyl hydroxylase involved in regulating the response to hypoxia and fatty acid oxidation.
Neutral lipid storage profiles in CD8+ TIL and tumor cells were similar, but ex vivo measurement of palmitate influx revealed that tumor cells from HFD mice took up more fatty acid, and the CD8+ TIL from HFD mice took up less of the fatty acid palmitate compared to those in CD mice. Therefore, the increased fatty acid uptake of tumor cells may deprive CD8+ T cells of fatty acids in the TME.
The proteome of tumor cells was then assessed, revealing that fatty acid metabolism and oxidative phosphorylation were enriched in HFD tumor cells. Since fat oxidation signatures were enriched in tumor cells, targeted lipidomics was used to measure the effect of HFD on lipid levels in plasma, tumor cells, and tumor interstitial fluid (TIF). Out of these, the largest impact of diet was on the lipid availability in plasma and TIF. Triglyceride and diglyceride levels were significantly enriched in the TIF of HFD mice, but not in plasma. Both diets had similar local lipoprotein lipase activity, suggesting that tumors from HFD mice are lipid-rich and have local lipase activity that facilitates cellular uptake.
The researchers then tested whether prevention of metabolic changes in HFD tumor cells could restore the CD8+ TIL responses and prevent the increased tumor growth seen in this model. To do so, PHD3, the primary metabolic regulator changed in HFD, was overexpressed in MC38 cells. This resulted in increased free fatty acids in the TIF and recovery of palmitate availability in the TME. Importantly, this change resulted in increased CD8+ T cell infiltration and reduced tumor growth.
Finally, the researchers assessed obesity effects in TCGA data. Patients with severe obesity had fewer infiltrating CD8+ T cells, and significant positive correlations were found between MHC-I and GZMB with PHD3 expression across multiple cancer types. Five TCGA datasets were then assessed for immunological and metabolomic subtypes. Low-PHD3 samples were associated with ‘immune cold’ signatures in five out of six cancer types, confirming that the results in the mouse model might represent the situation in human tumors.
Overall, this study showed that a high-fat diet advantageously modifies tumor cell metabolism, resulting in a reduction in CD8+ T cell responses, suggesting that metabolic parameters in tumors may provide useful new biomarkers. Further mechanistic studies to confirm and extend these data and to define potential therapeutic targets could have important implications for novel interventions impacting immunotherapy.
Write-up by Maartje Wouters and image by Lauren Hitchings