Interrogation of T Cell–enriched Tumors Reveals Prognostic and Immunotherapeutic Implications of Polyamine Metabolism

Metabolic features of the tumor microenvironment (TME) antagonize antitumor immunity. We hypothesized that T cell–infiltrated (Thi) tumors with a known antigen should exhibit superior clinical outcomes, though some fare worse given unfavorable metabolic features leveraging T cell–infiltrated (Thi), human papillomavirus–related (HPV+) head and neck squamous cell carcinomas (HNSC) to test this hypothesis. Expression of 2,520 metabolic genes was analyzed among Thi HPV+ HNSCs stratified by high-risk molecular subtype. RNA sequencing (RNA-seq) data from The Cancer Genome Atlas (TCGA; 10 cancer types), single-cell RNA-seq data, and an immunotherapy-treated melanoma cohort were used to test the association between metabolic gene expression and clinical outcomes and contribution of tumor versus stromal cells to metabolic gene expression. Polyamine (PA) metabolism genes were overexpressed in high-risk, Thi HPV+ HNSCs. Genes involved in PA biosynthesis and transport were associated with T-cell infiltration, recurrent or persistent cancer, overall survival status, primary site, molecular subtype, and MYC genomic alterations. PA biogenesis gene sets were associated with tumor-intrinsic features while myeloid cells in HPV+ HNSCs were enriched in PA catabolism, regulatory, transport, putrescine, and spermidine gene set expression. PA gene set expression also correlated with IFNγ or cytotoxic T-cell single-sample gene set enrichment analysis (ssGSEA) scores across TCGA tumor types. PA transport ssGSEA scores were associated with poor survival whereas putrescine ssGSEA scores portended better survival for several tumor types. Thi melanomas enriched in PA synthesis or combined gene set expression exhibited worse anti-PD-1 responses. These data address hurdles to antitumor immunity warranting further investigation of divergent PA metabolism in the TME. Significance: Despite the presence of tumor-infiltrating lymphocytes and antigen, antitumor immunity is often insufficient in tumor control. We leverage HPV-related head and neck cancers to identify metabolic challenges to antitumor immune responses. PA metabolism is associated with tumor-intrinsic features while the myeloid compartment exhibits enriched PA regulatory gene expression.


Introduction
Features promoting antitumor immunity include the presence of cytotoxic tumor-infiltrating lymphocytes (TIL) and antigen (e.g., viral, tumor neoantigens; refs. 1-4). However, not all immune cell-infiltrated, immunogenic tumors exhibit favorable antitumor immunity. The tumor microenvironment (TME) diminishes antitumor T-cell function through recruitment of tolerogenic cell types, nutrient depletion, and the creation of an acidic, hypoxic microenviron-

T-Cell Receptor Diversity Analysis
T-cell receptor (TCR) diversity was assayed and reported in the Supplementary Data of Thorsson and colleagues (10) We used the Shannon entropy and Richness variables from TCGA HNSC data to test the relationship between polyamine (PA) pathway gene set expression and TCR diversity.

Single-Sample Gene Set Enrichment Analysis
Single-sample gene set enrichment analysis (ssGSEA; v10.0.3) was implemented in GenePattern (RRID:SCR_003201) to estimate PA pathway and immune gene set scores (Supplementary Table S1; ref. 11). Default parameters were used with rank normalization for all ssGSEAs.

Cellular Abundance Estimates from Bulk RNA-sequencing Data
We utilized CIBERSORT in "Impute Cell Expression" mode on TCGA transcripts per million (TPM) RNA sequencing (RNA-seq) data from patients with HNSC to infer relative cellular proportions. For the CIBERSORT analysis, we used a HNSC reference single-cell (sc) RNA-seq dataset (12) in TPM normalization space to define proportions of tumor cells, macrophages, fibroblasts, CD8 + , and CD4 + T cells in bulk RNA-seq data. TCGA HNSC TPM RNA-seq data were used for this analysis to keep the bulk RNA-seq and reference matrix in the same normalization space per CIBERSORT recommendations.

T-Cell Infiltration Stratification
T-cell infiltration scores from ssGSEA were generated using CD8 + T-cell and cytotoxic T cell (CTL) gene sets described by Bindea and colleagues (ref. 13; Supplementary Table S1). We scaled ssGSEA scores from each gene set.

Samples were dichotomized [T cell-enriched (Thi) vs. T cell-depleted (Tlo)]
using an upper quartile cutoff of ssGSEA scores for both CTL and CD8 + T-cell gene sets (Supplementary Table S2). Tumors in the highest ssGSEA score quartile for either the CD8 + T cell or CTL signature were categorized as Thi and the remainder Tlo (Supplementary Table S2). ssGSEA stratification of T-cell infiltration status was consistent with inferred CD8 + T-cell abundance from computational microdissection with CIBERSORT (Wilcoxon rank-sum of CD8 + T-cell abundance between Thi and Tlo, P < 0.001; Supplementary  Fig. S1; ref. 14). ssGSEA stratification was used for all downstream analyses for (i) comparability with other studies using the Bindea immune cell gene sets to infer immune cell responses; and (ii) to use a consistent, gene set-driven approach, agnostic to the tissue of origin.

Metabolic Gene Curation
We defined a set of 2,520 genes implicated in metabolism using gene sets from Broad Institute's Molecular Signature Database (MSigDB; ref. 15) and Shaul and colleagues (16).

Survival Analysis
We utilized univariate Cox regression analysis to test associations between PA pathway scores and risk of mortality.

Molecular Subtype Classification
To define molecular subtypes using TCGA HNSC expression data, we utilized an R script kindly provided by the Fertig and Seiwert labs based on prior work which implements a correlation-based nearest centroid technique (17). Subtypes were assigned for the entire TCGA HPV + and HPV − HNSC dataset including basal, classical, or mesenchymal subtypes.

HPV Integration
Using data from Parfenov and colleagues, we assigned TCGA HPV + HNSC viral integration status (18).

scRNA-Seq Analysis
HPV + HNSC TIL data from Cillo and colleagues (21) were downloaded, preprocessed, normalized, and scaled using Seurat (v4.0. 1; refs. 22, 23). Data were mapped onto a single-cell reference dataset to identify immune cell subsets (22) which were used to evaluate expression of PA pathway enzymes across cell types. The "AddModuleScore" function was used to determine PA gene set (i.e., "module") scores using the gene sets defined below. A similar process was used for HPV − HNSC scRNA-seq data which were TPM normalized and were preprocessed, log 2 transformed, and scaled (12). distributed data. χ 2 tests were used to evaluate independence between groups with expected cell counts ≥5. Fisher exact tests were used to test for independence with any expected cell count <5. To account for multiple hypothesis testing, the FDR was controlled using the method of Benjamini and Hochberg. Pearson correlation analyses were performed in R using the WGCNA software package (RRID:SCR_003302; ref. 25). An alpha of 0.05 was used as a threshold for statistical significance, except in the case of multiple hypothesis testing where a q value of 0.25 was used. *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001; ****, P ≤ 0.0001; ns, P > 0.05.

Ethics Approval and Consent to Participate
Data were obtained from publicly available databases.

Metabolic Genes Enriched Among High-risk, HPV + T Cell-infiltrated HNSCs
To identify metabolic features impairing the antitumor immune response in an otherwise favorable viral antigen positive, T cell-enriched TME, we performed a metabolism focused transcriptional network analysis (8) comparing TCGA HPV + Thi HNSCs stratified by a high-risk HPV + HNSC gene signature (i.e., C1; ref. 27). We hypothesized that T cell-infiltrated HPV + C1 HNSCs exhibit a poor prognosis secondary to aberrant metabolic features. The C1 signature is driven primarily by tumor cell-intrinsic expression. Notably, immune-related signatures are not differentially expressed between the C1 (worse prognosis) and C2 (better prognosis) HPV + HNSC subsets. The HPV + C1 signature demonstrates an intermediate gene expression profile between HPV − and HPV + tumors with better survival (i.e., HPV + C2) and has lower expression of the HPV E1ˆE4 splicing isoform, a feature associated with resistance to radiation.
To identify metabolic gene network differences between TCGA HPV + Thi HN-SCs stratified by the C1 signature, we generated a score across tumors taking the cumulative expression across C1 genes (CDA, DFNA, KRT, MDFI, UPP, DPF, RGS, TUBB, PPPRB, RHOD, CPA, PNLIPRP) for each sample and stratifying them into high, intermediate, and low strata. The distribution of C1 strata by T-cell infiltration status is shown in Table 1. Nine percent of Thi tumors and 55% of Tlo tumors were C1 high (χ 2 test, P < 0.001).
Focusing our analysis on 2,520 metabolism-related genes derived from MSigDB  Table S5) among the HPV + Thi C1 high tumors. DE results were used as input for network analysis focused on identification of the most differing metabolic subnetworks between C1hi versus C1lo tumors (Fig. 1C). Consistent with the DAVID pathway-level analysis, we observed a subnetwork of PA metabolism gene enrichment among the C1hi tumors including SMOX, SRM, and MTAP. Notably, this PA subnetwork illustrated a close association with lactate (LDHA), kynurenine (KYNU, IDO), tryptophan (WARS), beta-alanine (UPB), and α-ketoglutarate (IDH) metabolism ( Fig. 1C; Supplementary Fig. S2). Quantification of PA metabolism gene expression was performed using ssGSEA as a function of C1 expression strata. A set of 35 curated genes involved in PA metabolism and transport were quantified across TCGA HPV + Thi HNSCs using ssGSEA demonstrating lower PA ssGSEA scores among the C1lo tumors relative to the C1 intermediate or C1hi tumors (Fig. 1D).
We also compared DE among HPV + Thi tumors based on their molecular subtype (classical vs. immune/mesenchymal; Supplementary Table S6). Among HPV + HNSCs, the immune/mesenchymal subtype has a better prognosis than classical subtype tumors (17). The immune/mesenchymal subtype is characterized by immune markers (e.g., CDA, ICOS) and mesenchymal markers (e.g., VIM, SA). The most distinctive transcriptional feature of the classical subtype in both HPV + and HPV − HNSCs is enrichment for PA metabolism gene expression (17). Molecular subtype across TCGA HNSCs was defined using a correlation-based, nearest centroid classification approach (17). Table 1 shows the distribution of molecular subtypes given T-cell status across HPV + HNSCs. Eighty-three percent of Thi tumors and 12% of Tlo tumors were classified as mesenchymal (χ 2 test, P < 0.001; Table 1). No HPV + Thi tumors were classified as basal. The basal subtype is characterized by a predominance of HPV − HNSCs, hypoxia, and EGFR/HER signaling (17).

PA Synthesis and Transport Genes are Associated with Worse Clinical and Molecular Features Among Patients with HPV + HNSC
To characterize the extent to which PA metabolism genes are related to clinical and molecular features among HPV + HNSCs, we performed hierarchical clustering of gene expression using the 35 PA metabolism-related genes introduced above. Hierarchical clustering revealed three sample clusters ( Fig. 2A, column clusters). Oropharyngeal tumors made up most of the cohort (54/97) followed by oral cavity cancers (32/97; Table 2). Clusters 1 and 2 consisted of 87% and 78% oropharyngeal tumors versus cluster 3 which was enriched in oral cavity tumors (68%; Fisher exact test, P < 0.001). Clusters 1 and 2 were predominantly Thi (69% (27/39) and 61% (11/18), respectively; Fig. 2A; Table 2). Cluster 3 tumors were mainly Tlo (80% (32/40)). The distribution of Thi and Tlo tumors by primary site is shown in Table 1. Oropharyngeal tumors comprised 78% of Thi tumors and 35% of Tlo tumors. Tumor stage did not differ between clusters nor did smoking status ( Fig. 2A; Table 2).
Recurrent/Metastatic (R/M), tumor status, and overall survival status were worse in cluster 3. Thirty-two percent of cluster 3 subjects experienced R/M compared with 18% and 11% of clusters 1 and 2, respectively (Fisher exact test, Fig. 2A;  Fig. 2A; Table 2]. While the mesenchymal subtype may be associated with the degree of cancer-associated fibroblasts (12), there was a strong association between mesenchymal subtype and Thi status (Table 1).

PA Gene Expression and Tumor-intrinsic Features
First, we tested whether PA gene expression is a function of tumor-intrinsic features (i.e., molecular subtype, HPV status, HPV integration, or TMB).
We defined PA biogenesis-specific gene sets based on prior knowledge (Table 3; ref. 31). We included broader PA biogenesis-defined gene sets (i.e., synthesis, catabolism, regulatory, and transport) as well as metabolite specific gene sets (i.e., putrescine, spermidine, and hypusine) given the unique cellular functions of these respective metabolites. PAOX was included in both the putrescine and spermidine synthesis gene sets as it oxidizes N-acetylated PAs to generate putrescine and spermidine. The SAT gene product catalyzes PA acetylation. AMD is included in the spermidine synthesis gene set as it encodes S-adenosylmethionine decarboxylase which produces the aminopropyl donor, S-adenosylmethionine, necessary for spermidine synthesis. Hypothesizing that PA metabolites can be immunosuppressive, the "combined" gene set was informatically and empirically defined using PA genes that were negatively correlated with both Bindea cytotoxic lymphocyte gene set (CTL) and REAC-TOME IFNγ gene set ssGSEA scores across multiple TCGA T cell-infiltrated tumor types (Supplementary Figs. S4A and S4B; ref. 32). The transporters included have been previously reported as putative PA transporters (31,(33)(34)(35)(36)(37)(38)(39)(40)(41)(42)(43)(44).
Hypothesizing that PA metabolism gene expression depends on molecular subtype, we did not identify differences in enrichment of the PA gene sets among the HPV + HNSCs (Fig. 3A). Among the HPV − Thi HNSCs, ssGSEA scores from several PA gene sets were significantly higher among the basal AACRJournals.org Cancer Res Commun; 2(7) July 2022   compared with the classical or mesenchymal subtypes including the synthesis, catabolism, regulatory, spermidine, and hypusine gene sets. In contrast, PA-combined gene set ssGSEA scores were lower among the basal tumors compared with the classical or mesenchymal HPV − HNSCs. HPV + Tlo HNSCs also lacked differences in PA ssGSEA scores except for transport which was significantly higher among basal tumors whereas putrescine ssGSEA scores were higher among the classical and mesenchymal tumors than the basal tumors ( Supplementary Fig. S5).
PA pathway expression was characterized on the basis of HPV status. Prior studies identified that viral replication depends on PAs (45) and that some viruses may stimulate PA synthesis (46,47). Moreover, nicotine drives ODC expression (48). PA synthesis, catabolism, transport, and spermidine ssGSEA scores were higher among HPV − Thi versus HPV + Thi HNSCs (Fig. 3B). PA synthesis and combined gene set ssGSEA scores were higher among HPV − smokers compared with nonsmokers whereas regulatory ssGSEA scores were higher among the nonsmokers (Supplementary Fig. S6). HPV + smokers demonstrated higher PA synthesis, transport, and combined ssGSEA scores than nonsmokers (Supplementary Fig. S6).
We examined whether HPV integration impacts PA pathway expression evaluating the association between PA pathway ssGSEA scores and HPV integration (18). Defining tumors as HPV integrated or HPV nonintegrated, we observed greater synthesis, putrescine, and hypusine ssGSEA scores among the HPV + Thi HNSCs with nonintegrated HPV whereas there were no differences in the integration status among the HPV + Tlo HNSCs ( Fig. 3C; Supplementary  Fig. S5C).
Finally, we tested the relationship between TMB and PA gene set expression given that PAs are involved in epigenetic regulation and DNA stabilization.
TMBlo HPV + Tlo HNSCs had higher regulatory ssGSEA scores than TMBhi HPV + Tlo HNSCs (Supplementary Fig. S5D). Taken together, these data reveal evidence that PA gene set expression varies with molecular subtype, HPV status, integration status, and TMB among HPV + Thi HNSCs. While these data reveal associations between PA gene set expression and tumor features, we were also interested in the extent to which tumor-infiltrating immune cells express PA genes. This could reveal insight into strategic targeting of PA metabolism to optimize antitumor immunity while diminishing tumor cell proliferation and viability.

PA Gene Set Expression Among HNSC Intratumoral Immune Populations
Given the importance of PA expression to T-cell and macrophage function (49,50), we utilized scRNA-seq data to examine PA expression across HPV + HNSC intratumoral immune populations ( Fig. 4; ref. 21). We generated singlecell PA pathway enrichment scores using the gene sets defined above which we projected onto single cells (Fig. 4A-C). PA catabolism, regulatory, putrescine, and spermidine gene sets were significantly enriched in CD16 + and CD14 + monocytes relative to the lymphoid lineages ( Fig. 4C; Supplementary Tables S12 and S13  Fig. S8; Supplementary Tables S15 and S16). In contrast, HPV − HNSC tumor cells had greater transport and combined pathway gene set expression than T cells, macrophages, or fibroblasts and higher synthesis gene set expression than T cells or fibroblasts (Supplementary Fig. S8; Supplementary Tables S15 and S16).
CIBERSORT was used to infer the cellular composition of the TME bulk RNAseq data from TCGA HPV + Thi HNSCs. These data were used to test the correlation between cell-type abundance and ssGSEA scores for PA pathway expression (Fig. 4D). After adjusting for multiple hypothesis testing, we observed a nonstatistically significant positive correlation between macrophage abundance and regulatory gene set expression (r = 0.39, FDR q-value = 0.29). Fibroblasts exhibited a strong negative correlation with hypusine gene set expression (r = −0.55, q-value = 0.003). Collectively, these data demonstrate relatively greater enrichment of PA catabolism and regulatory gene set expression in the myeloid compartment relative to the lymphoid compartment.

Pan-Cancer PA Gene Set Expression and Immune Response
We tested the relationship between PA pathway gene set expression and T-cell receptor (TCR) diversity (10). We compared richness and Shannon entropy scores between PA gene set strata among TCGA HPV + HNSCs. TCR richness and Shannon entropy were negatively associated with PA synthesis (Kruskal-Wallis test, P = 0.004 and P = 0.006, respectively; Supplementary Fig. S9). Other PA pathway gene set scores were not associated with TCR clonality among these tumors except putrescine and hypusine scores which were associated with higher richness and Shannon scores among the We further hypothesized that tumor cell PA synthesis and/or transport gene set expression is associated with worse antitumor T-cell function as inferred by IFNγ or cytotoxic lymphocyte ssGSEA scores. Correlation analyses were performed between the REACTOME IFNγ signaling gene set or the Bindea CTL ssGSEA scores and PA pathway ssGSEA scores (Fig. 5A).  (Fig. 5A). Hypothesizing that PAs contribute to immunosuppression, we sought to characterize the association between PA pathway ssGSEA scores and survival across cancers.

Pan-Cancer PA Gene Set Expression and Outcomes in T Cell-enriched Tumors
We performed Cox regression testing the association between PA pathway ss-GSEA scores and risk of mortality in T cell-enriched tumors. High transport scores were associated with a greater risk of mortality among Bladder urothelial carcinoma (BLCA), HNSC, and HPV + Cervical and endocervical cancers (CESC) TCGA cohorts. Combined PA pathway gene set scores were associated with higher mortality among the BLCA, Lung adenocarcinoma (LUAD), and HPV + CESC cohorts, with a trend toward worse survival among the Prostate adenocarcinoma (PRAD), Skin Cutaneous Melanoma (SKCM), HPV + HNSC and CESC tumors (Fig. 5B). Interestingly, high putrescine ssGSEA scores were associated with better survival among HNSC, KIRC, LUAD, LUSC, and HPV + CESC Thi tumors. In contrast, PA combined pathway gene set scores were not associated with increased risk of mortality across Tlo tumors (Supplementary Fig. S10). However, higher PA transport scores were associated with worse survival among Tlo HNSC, SKCM, and BLCAs.
Given the negative association between the PA combined pathway ssGSEA scores with effector lymphocyte gene set scores and prognosis, we hypothesized AACRJournals.org Cancer Res Commun; 2(7) July 2022  that PA expression affects response to ICI. Specifically, we were interested in the association with ICI response based on PA biosynthetic and transport pathway expression (given the availability of commercially available inhibitors of these functions) among a subset of T cell-enriched, nivolumab-treated melanomas (26). High combined PA pathway ssGSEA scores were associated with worse survival (Fig. 5C). There was no difference in survival based on the synthesis or transport ssGSEA scores. Response to ICI was greatest among patients with low combined or synthetic gene set ssGSEA scores (Fig. 5D).

Discussion
To identify metabolic features impairing the antitumor immune response in an otherwise favorable TME, we leveraged T cell-infiltrated, immunogenic, HPV + HNSC. Using this strategy, PA gene set expression was associated with aggressive molecular phenotypes, diminished antitumor immunity, poor prognosis across cancer types, and a poor response to immunotherapy among melanomas. These data demonstrated both tumor-dependent and stromal differences in PA metabolism gene expression. As PAs are necessary for tumor proliferation and function, T-cell activation and differentiation, and macrophage differentiation, the PA metabolism axis represents a therapeutic target for leveraging divergent PA metabolic needs between tumor and immune cells.
PAs, a family of low molecular weight polycations, regulate cell processes from proliferation and adaptive immunity (51,52) to epigenetic modifications (53,54), metabolite availability (55,56), transcriptional regulation (57), and chromatin stabilization (58). Cells primarily synthesize PAs and acquire them from the TME. Given the essential role of PAs in tumor and T-cell function (49), this pathway may be rationally targeted to diminish tumor growth and/or enhance antitumor immunity. In recent work, investigators identified the dependence of CD4 + T-cell differentiation on PA metabolism, for example demonstrating that T H 17 cells and T regs may rely more on transport than PA synthesis (59,60). In our scRNA-seq analyses among tumor TILs from HPV + HNSCs, we did not observe a difference in PA transport gene set expression between T regs and CD4 + CTL, CD4 + T CM , or CD4 + T EM cells (Supplementary Table S13).
In contrast, we identified relatively greater PA spermidine and putrescine gene set expression among the T regs compared with CD4 + T CM and CD4 + T EM cells. We did not test PA gene set expression in other CD4 + T-cell lineages as this was beyond the scope of this study, but it would be intriguing to compare the effect of PA metabolism on CD4 + T-cell differentiation in the context of the TME.
The metabolic milieu of the TME diminishes antitumor immune control by producing inhibitory metabolites, depleting essential nutrients, and creating a hypoxic and acidic ecosystem. Here, PA gene set expression was upregulated in high-risk, T cell-infiltrated HPV + HNSCs. Gene expression clusters with T cell-infiltrated tumors were enriched in PA catabolism and regulation genes (i.e., SAT, OAZ-3). PA gene expression varied by molecular subtype, HPV status, HPV integration status, and TMB status. High TMB HPV + Thi HN-SCs were associated with higher PA synthesis, transport, and spermidine gene set scores. At the single-cell level in HPV + HNSCs, PA catabolism, regulatory, transport, putrescine, and spermidine gene set ssGSEA scores were enriched in the myeloid compartment relative to lymphoid lineages. In the pan-cancer analysis, we observed negative correlations between PA synthesis, transport, combined, or hypusine gene set ssGSEA scores with effector lymphocyte function ssGSEA scores. The question persists of whether immunosuppressive immune populations leverage PA metabolism to permit tumor growth and whether there is a feedback loop of PA-dependent tumor-intrinsic paracrine function promoting immunosuppressive immune function.
PA blockade therapy (AMXT1501/DFMO) using difluoromethylornithine (DFMO), an ornithine decarboxylase inhibitor, and AMXT1501 (31, 61), a PA transport inhibitor, demonstrates excellent responses in diffuse intrinsic pontine glioma (62) or MYCN transgenic mice models (31). AMXT1501/DFMO shows activity in immunocompetent, but not immunodeficient mouse tumor models (63). In contrast, monotherapy is less effective than combined therapy, likely secondary to compensatory mechanisms. This may account for the limited responses noted to date with PA monotherapy (64,65). In is intriguing to speculate that PA expression in the TME may serve as a biomarker for patient selection in employing agents of the PA pathway such as in the case of patients with HPV + HNSC with increased T-cell infiltrates and high PA levels.
Disruption of PA homeostasis in the TME may affect the function of infiltrating immune cells. Several metabolites contribute to PA synthesis including methionine, glutamine, arginine, and proline. About 30% of PAs are derived from glutamine in activated T cells, though PA synthesis blockade prevents T-cell expansion (49). In macrophages, inhibition of either PA synthesis or hypusination diminishes oxidative phosphorylation and prevents alternative macrophage differentiation (50). PAs are immunosuppressive, and tumors may leverage PAs for intrinsically mediated immune evasion by inducing autophagy (66), favorable epigenetic alterations (53), or stabilizing their DNA from the effects of cytotoxic lymphocytes (58).
Future studies will need to address some of the limitations of this study including the small sample size of HPV + HNSCs, limiting our power for survival analyses. Cohorts powered to detect differences in survival as a function of metabolite enrichment may further elucidate mechanisms diminishing antitumor immunity. Defining the extent to which PA metabolism diminishes antitumor immunity either because of exposure to supraphysiologic intracellular PA levels and/or through cell-intrinsic mechanisms that promote tumor cell survival will help overcome hurdles to immunotherapy. Finally, the role of PA transport in T-cell function is unknown. The degree to which putative PA transporters regulate nutrient transport and antitumor immunity requires further investigation. These data aim to uncover fundamental insights into the effect of metabolites in the TME and antitumor immune responses and spur a line of inquiry investigating the extent to which PA metabolism impairs antitumor immunity.