Introduction

Tumor antigen encountered by T cells and subsequent tumor-specific cytotoxic T lymphocyte (CTL) activation are essential, but alone are insufficient for an effective antitumor immune response [1]. Active uptake of nutrients and their metabolism are also critically important for not only meeting the enhanced energy requirements of the expanding T cells, but also for rewiring T cell differentiation in the tumor microenvironment [2]. Amino acids are the building blocks of proteins that are the essential structural and functional elements of activated T cells, and amino acid metabolites also play a key role in T cell activation and proliferation [3, 4]. Emerging experimental data from both human cancer patients and tumor-bearing mice indicate that amino acids also generate various metabolites that regulate various signaling pathways to control T cell activation, expansion, exhaustion, and effector function [5, 6]. Furthermore, amino acid metabolism, particularly the slc7a5-mediated amino acid transportation and subsequent metabolism in T cells are essential for chimeric antigen receptor (CAR) T cell proliferation and function. The amino acid-dependent metabolic reprogramming is therefore critical for effective T cell anti-tumor immune response in the tumor microenvironment [7].

Tryptophan (Trp) is an essential neutral amino acid and is a component of most proteins. However, only a small fraction of intracellular Trp is used for protein biosynthesis. The majority of intracellular Trp is metabolized through three metabolic pathways: the kynurenine (Kyn), 5-hydrotryptamine, and the indole pathways, to generate bioactive metabolites [8]. Among these three metabolic pathways, the Kyn pathway metabolizes more than 95% of intracellular Trp [9], in which Trp is catalyzed by Trp-2,3-dioxygenase, indoleamine-2,3-dioxygenase 1, and indoleamine-2,3-dioxygenase 2 to generate Kyn and other metabolites [5, 8]. Kyn is a bioactive ligand that directly activates cytoplasmic transcription factor aryl hydrocarbon receptor (AhR) [10, 11] that is then translocated to the nucleus to activate transcription of target genes [12, 13]. Trp metabolites improve CTL differentiation and function [6] and enhance the efficacy of chemotherapy and immune checkpoint inhibitor [14]. On the other hand, Trp metabolites also promote T cell exhaustion and suppress adaptive anti-tumor immunity to facilitate tumor immune escape [13, 15]. In addition, Trp metabolites activate tumor-associated macrophages to indirectly suppress T cell activation to promote tumor immune evasion [16]. The Trp metabolic pathway may therefore execute both tumor promotion and suppression function via a cellular context-dependent mechanism. Extracellular Trp is transported into cells by the neutral amino acid transporter system, which consists of two heterodimers: Slc3a2-Slc7a5 and Slc3a2-Slc7a8 [5, 17]. Both Slc3a2 and Slc7a5 are required for amino acid transport and function in plasmacytoid dendritic cells (pDCs) activation [18] and is a rate-limiting transporter of leucine in T cells [19, 20]. However, the specific amino acid transporters and their function in the Trp metabolic pathway in tumor-reactive T cells remains elusive.

We aimed at determining tumor cell regulation of the metabolic reprogramming of tumor antigen-activated T cells in the tumor microenvironment. We determined that the tumor cells induce Trp transporter Slc7a5 upregulation in tumor-specific T cells to increase T cell intracellular Trp level, which is essential for tumor cell-specific T cell activation. Mechanistically, we demonstrated that the Slc7a5-Trp metabolic axis activates AhR to up-regulate FasL expression to maintain T cell anti-tumor lytic activity to suppress tumor development. Our findings argue that the Slc7a5-Trp metabolic pathway boosts T cell activation and effector function in human cancer therapy.

Materials and methods

Mice

C57BL/6 and BALB/c mice were obtained from the Huafukang (Beijing, China). The Slc7a5fl/fl mice (B6.129P2-Slc7a5tm1.1Daca/J) contains the LoxP sites flanking exon 1 of the Slc7a5 (solute carrier family 7 (cationic amino acid transporter, y+ system), member 5) gene [20] and were obtained from the Jackson Laboratory (stock #027252). CD4-Cre mice [Tg(Cd4-cre)1Cwi/BfluJ)] were obtained from the Jackson Laboratory (stock # 017336). Mice with Slc7a5 deletion only in T cells were created by crossing the Slc7a5fl/fl mice to CD4-Cre mice. Both male and female mice were used in this study. AT3 breast tumor cells were injected to female mice. B16-F10 melanoma cell line was derived from a male mouse and was therefore injected to male mice.

Tumor cells

The murine mammary carcinoma cell line 4T1 and melanoma cell line B16-F10 were obtained from American Type Culture Collection (Manassas, VA). The murine malignant mesothelioma cell line AB1 was obtained from Sigma-Aldrich (St Luis, MO). Murine mammary tumor cell line AT3 was established from the primary mammary gland carcinoma of a female MTAG [mouse mammary tumor virus (MMTV) long terminal repeat polyoma virus middle T antigen transgenic mouse] mouse [21]. AT3 cell line was provided by Dr. Scott Abrams (Roswell Park Comprehensive Cancer Center, Buffalo, NY). All cells were tested bi-monthly for mycoplasma contamination and all cells were mycoplasma-negative at the time of the experiments.

Tumor antigen-specific CTL

The tumor antigen/AH1 peptide-specific T cells (2/20 CTLs) were produced and maintained as described in the literature and provided by Dr. Scott Abrans [22, 23]. The CTLs (1.33 × 105 cells/mL) were cultured in RPMI 1640 medium with 10% FBS and AH1 peptide (SPSYVYHQF, 1 μg/mL), recombinant IL-2 protein (10 Units/mL), and lethally irradiated spleen cells of BALB/c mouse (3.3 × 105 cells/ml) weekly. The CTLs were tested bimonthly for mycoplasma contamination and were mycoplasma-free at time of experiments.

CTL proliferation assay

The 2/20 CTLs were stimulated with the AH1 peptide and feeder cells as previous described [24]. The stimulated CTLs were collected seven days after stimulation and purified using lymphocyte separation medium. The purified CTLs were then labelled with CSFE using the CellTrace CFSE Proliferation Kit (Cat# 34554, Life Technologies Corp, Eugene, OR). The lablled CTLs were co-cultured with AB1 and 4T1 tumor cells, respectively at a 1:1 ratio. The co-culture mixtures were collected at day 1, 2, and 3, stained with anti-CD8-specific antibody and analyzed by flow cytometry. The CD8+ CTLs were gated and analyzed for CFSE intensity dilution.

CTL lytic activity assay

Tumor cells (1 × 105 cells/well) were mixed with the 2/20 CTLs at various ratios and cultured in the U bottom 96-well plate (Corning, Corning, NY) in 100 μl RPMI 1640 medium plus 10% FBS. JPH203 (Selleckchem, Houston, TX. Cat# S8667) were added to the culture. Floating cells were collected. Adherent cells were detached with Trypsin and mixed with the floating cells. Cells were stained with CD8-FITC and Annexin V-APC (Biolegend, San Diego, CA) in Annexin V-binding buffer for 30 min, stained with PI for 5 min, and analyzed by flow cytometry. The CD8- tumor cells were gated for Annexin V+ and/or PI+ cells.

Cytokine and chemokine assays

After thawing on ice, samples were centrifuged for 10 min @14,000 x g and clear sample was used for the assay. Biolegend recommends mixing standards with the Matrix C they provide, which is designed to mimic serum matrix. This allows for a more accurate estimation of the final concentration of cytokines in the samples as plasma components may inhibit the signal for some cytokines. The Standards in Assay Buffer were diluted 2x with the Matrix C provided in the kit.Serum samples were diluted 2x with an Assay Buffer provided in the kit.

Metabolomics

Metabolite extraction and metabolomic analysis were carried out in the National Institutes of Health West Coast Metabolomics Center at the University of California Davis (Davis, CA). Data were acquired using ALEX-CIS GCTOF mass spectrometry with a Restek corporation Rtx-5Sil MS column and processed using ChromaTOF software version 2.32 (Leco, St. Joseph, MI) and the BinBase algorithm, as described previously [25]. Metabolite set enrichment analysis (MSEA) of metabolomics data was conducted using the web-based platform MetaboAnalyst 6.0 according to the online instructions [26].

ATAC-sequencing analysis

2/20 CTLs were isolated from the culture mixture using mouse CD45 nanobeads (Biolegend, San Diego, CA). Nuclei were prepared from the purified cells and used for the ATAC-seq library construction using the Omni-ATAC-seq protocol. Nuclei were incubated with 2.5 mL Nextera Tn5 transposase (Illumina) in 50 mL 1X transposition reaction at 37oC for 1 hr. The transposition reaction mixtures were purified with the DNA Clean and Concentrator Kit (Zymo Research, Irvine CA), and amplified for 11-13 cycles using NEBNext HighFidelity 2X Master Mix (New England Biolabs, Ipswich, MA) and Nextera Index primers (Illumina, San Diego, CA). The libraries were size-selected and sequenced on a HiSeq instrument in a pair-end 150 cycle run. Raw reads generated in this study in fastq format and were first examined by FastQC v0.11.8 for quality control purpose. Adaptor and quality trimming of the raw sequencing reads was performed using Trim Glore! v0.6.3. Trimmed reads were aligned to mouse genome mm10 using Bowtie2 v2.3.4.1. Reads aligned to the mitochondria genome were filtered using samtools v1.8. PCR duplicates were removed using MarkDuplicates from Picard v2.18.2. The aligned reads overlapping with ENCODE blacklist regions (version 2) were removed using samtools. ATAC-seq peak calling was performed using MACS2 v2.1.1.20160309. BigWig files were generated by bedGraphtoBigwig and visualized in Integrated genome viewer (IGV) v2.8.3. The bigwig track plot for individual locus was generated using karyoploteR v1.16 in R v3.6.3.

Bulk RNA sequencing analysis

Total RNA was isolated from T cells using the Direct-zol RNA microprep kit according to manufacturer’s instructions (Zymo Research, Irvine CA). Bulk RNA sequencing was performed in Novogen (Sacramento, CA). Reads were aligned using STARv2.5 and reference genome (mm10) and gene model annotation files from Ensembl. Reads were quantified using HTSeqv0.6.1. Differential gene expression analysis was done using DESeq2 v1.36. Ranked-log2 fold changes from DESeq2 analysis were used as inputs for analysis. GSEA was performed using GSEA4.0 (Broad) with gene-set permutation. GO pathway enrichment was performed with clusterProfiler V4.0. Heatmap was generated using ComplexHeatmap v2.12.0.

scRNA-sequencing analysis

AT3 tumor cells (5×105 cells/mouse) were injected to the #3 mammary gland of female C57BL/6 WT mice (n = 3) and Slc7a5.TKO mice (n = 3). Twenty days later, tumors were dissected from the tumor-bearing mice, digested with collagenase solution, passed through a 100 μM cell strainer. The tumor digests were washed with PBS + 0.5% BSA, resuspended in PBS + 0.5% BSA, and incubated with mouse CD45 nanobeads (Biolegend). CD45+ cells were isolated by using a magnetic stand, resuspended in PBS (without Ca or Mg) with 0.04% BSA and freshly used for the scRNA-seq using the 10X Genomics (10X Genomics. Pleasanton, CA). Cell viability was assessed with Cellometer Auto 2000 (Nexcelom) using an AOPI staining. Samples with viability >80% were loaded in the Chromoium X instrument (10X Genomics) to capture approximately 10 × 103 targeted cells per sample using Chromium Next GEM Single Cell 3ʹ Reagent Kit v3.1 Dual Index form 10X Genomics. GEMs (Gel Beads-in-emulsion) were generated by partitioning oil, combining barcoded Single Cell 3’ v3.1 Gel Beads, a Master Mix containing cells and Partitioning Oil onto Chromium Next GEM Chip G. scRNA-seq libraries were generated according to the manufacturer’s instructions. Enzymatic fragmentation and size selection are used to optimize the cDNA amplicon size. P5, P7, i7 and i5 sample indexes, and TruSeq Read 2 (read 2 primer sequence) are added via End Repair, A-tailing, Adaptor Ligation, and PCR. The final libraries contain the P5 and P7 primers used in Illumina amplification. QC (quality control) analysis was performed prior sequencing with 2100 Bioanalyzer and 2200 TapeStation (Agilent, Santa Clara, CA, USA). Library concentration was assessed prior sequencing using Qubit Fluorometer (Thermo Fisher). Sequence was performed with Novaseq6000 System Illumina (Illumina Inc. San Diego CA) platform following 10X Genomics guideline; read lengths U28 | I10 | I10 | Y90 and sequencing depth 40’000 paired-end reads per cell.

scRNA-sequencing bioinformatics analysis

The feature count matrices were generated using Cell Ranger v7.1.0 and the Cell Ranger’s version of mm10 reference genome. The output filtered feature counts matrices were further processed and analyzed using R v 4.3.1 and Seurat package v4.9.9. Low-quality cells were removed from each sample if they contain fewer than 500 unique molecular identifiers (UMIs), fewer than 100 features, counts exceeding the sum of the median and 3 times the median absolute deviation for either UMIs or features, or more than 15% mitochondrial UMIs. The cell multiplets were removed using the DoubletFinder package v2.0.3. The Seurat SCTransform method was employed for normalization and variance stabilization of molecular count data. The Seurat objects of all samples were integrated for clustering using Seurat’s canonical correlation analysis (CCA). Clustering was completed by using the RunPCA(), FindNeighbors() and FindClusters() functions with the first 70 principal components and a clustering resolution of 0.3. Nonlinear dimensionality reduction was performed using t-distributed Stochastic Neighbor Embedding (tSNE) technique with the RunTSNE (?) function implemented in Seurat package. Cluster-specific markers were identified by applying the FindAllMarkers function in Seurat, utilizing the default Wilcoxon Rank Sum test. Cell types were assigned to clusters by comparing annotation results from scMRMA v1.0 with the implemented in the package PanglaoDB reference and CIPR v0.1.0 with the Immunological Genome Project reference implemented in the package. Additionally, we conducted a detailed evaluation of the cluster-specific markers to confirm cell-type assignments. Gene set enrichment scores were calculated using the UCell v2.4.0 package. For differential expression (DE) analysis, scRNA-seq cont matrix was converted to pseudo-bulk matrix using the Aggregate Expression (?) function from the Seurat package and subsequently analyzed with the DESeq2 package v1.40.1.

Tumor mouse models

AT3 tumor cells were injected to the mammary gland of female C57BL/6 mice to establish orthotopic breast tumor. B16-F10 cells were injected subcutaneously to the right flanks of male C57BL/6 mice to establish orthotopic melanoma. Tumor size was measured with a digital caliber and by weighing the dissected tumor. Tumor lung metastases were quantified by injecting India ink (10% India ink in 0.1% ammonium hydroxide in water) into the lungs via the trachea. The inflated lungs were rinsed in water and fixed a glass scintillation vial containing Fekete’s solution (50% Ethanol, 6% Formaldehyde, and 3% glacial acetic acid). Tumor nodules were counted under a dissecting microscope. For tumor-infiltrating immune cell analysis, tumors were dissected from the tumor-bearing mice, digested with collagenase solution [Collagenase (Sigma-Aldrich), 1 mg/mL, Hyaluronidase (Sigma-Aldrich), 0.1 mg/mL, and DNase I: 30 U/mL]. The tumor digests were then passed through a 100 μM cell strainer (Corning, Corning, NY), incubated with red cell lysis buffer to remove red cells, washed in PBS, stained fluorescent dye-conjugated CD4-, CD8-, and CD11b-specific mAbs (Biolegend), and analyzed on a BD FACSCalibur Flow cytometer with CellQuest Pro (BD Biosciences, San Diego, CA) or a BD LSRFortessa cell analyzer with BD Diva 8.01 (BD Biosciences). All flow cytometry data was analyzed with FlowJo v10.6.0 (BD Biosciences).

FasL blockade immunotherapy

The AT3 and B16-F10 tumor-bearing mice were treated with IgG or anti-FasL monoclonal antibody clone MFL3 (Bio X cells, Lebanon, NH) by intraperitoneal injection. The tumor-bearing mice were treated once every 3 days at 200 μg IgG or MFL3/mouse for 4-5 times. Tumor size was measured using a digital caliber and tumor weight was determined in a scale. Tumor lung metastases were quantified using the India ink inflation method as described.

Chromatin Immunoprecipitation (ChIP) Assay

ChIP assays were carried out using the Chromatin Immunoprecipitation Assay Kit (Cat# 17-295, Merck, NJ, USA). Briefly, cells (1 × 106 cells) were harvested, cross linked by 1% formaldehyde, washed in PBS, resuspended in SDS lysis buffer and sonicated followed by centrifuge. The sonicated cell supernatant was diluted 10 fold in ChIP dilution buffer (reserve 5% diluent as input), then added AhR antibody and incubated overnight. The next day, the beads were added into cell supernatant and incubated in 4°C for 1 h. The beads were then washed by low salt immune complex wash buffer, high salt immune complex wash buffer, LiCl immune complex wash buffer, TE buffer in sequence, and then eluted by 500 µL elution buffer. The eluants were reversed and recovered by phenol/chloroform/isoamyl alcohol. The cd95 promoter DNA was detected by Real-Time Quantitative PCR using promoter DNA-specific primers.

Tumor cell proliferation and apoptosis analysis

Tumor cell proliferation in the tumor tissues was analyzed using IHC with Ki67-specific antibody as previously described [27]. Tumor cell apoptosis in the tumor tissues was analyzed using in situ TUNEL assay as previously described [27].

Datasets

Human breast cancer scRNA-Seq datasets [28] were downloaded and analyzed using R package. The FASLG, FAS, and SLC7A5 transcript levels in human cells is based on human scRNA-seq datasets retrieved from Human Protein Atlas (HPA) that contains 81 cell types from 31 datasets from the Single Cell Expression Atlas, the Human Cell Atlas, the Gene Expression Omnibus, the Allen Brain Map, European Genome-phenome Archive, and the Tabula Sapiens. The transcript level is presented as the consensus normalized transcripts per million (nTPM).

Statistical analysis

Statistical analysis was performed using ANOVA and paired Student’s t-test using GraphPad. A p < 0.05 is considered as statistically significant.

Results

Tumor cells activate tumor-specific CTLs with increased intracellular Trp

To assess the direct role of tumor cells in regulating the metabolic reprogramming in tumor-reactive T cells, we made use of a defined co-culture model of tumor cell-specific CTLs and cognate antigen-expressing tumor cells. In this model, the tumor cell-specific 2/20 CTLs recognize antigen AH1 peptide-expressing tumor cells [22]. AH1 is H-2Ld-restricted peptide derived from the endogenous mouse leukemia provirus gp70 protein and characterized as the immunodominant antigen in mouse tumor cell lines recognized by CTLs in mice immunized with autologous tumor cells [22, 29]. AH1 peptide is therefore a natural tumor antigen derived from self-protein gp70 in mouse tumor cell lines. AB1 and 4T1 are well-characterized mouse mesothelioma and breast tumor cell lines that express gp70 protein. AB1 and 4T1 tumor cells are therefore specific targets of 2/20 CTLs. The AB1/4T1 tumor cell and the 2/20 CTL co-culture model thus allows the analysis of the direct effect of tumor cells on tumor-activated CTLs. 2/20 CTLs are CD8+PD-1+gp70 tetramer+ CTLs (Fig. S1A). As expected, 2/20 CTLs lyse AB1 and 4T1 tumor cells in a dose-dependent manner (Fig. S1B). AB1 and 4T1 tumor cells respond to 2/20 CTLs to up-regulate PD-L1 (Fig. 1A). 2/20 CTLs responded to tumor cells to up-regulate PD-1 (Fig. 1B) and to proliferate rapidly and continuously in vitro in the 3-day time period analyzed (Fig. 1C). Analysis of the tumor cell and 2/20 CTL co-culture supernatant revealed that tumor cell activated 2/20 CTLs secrete high level of IFNγ, TNFα, type I IFNs (Fig. 1D), T cell cytokines (Fig. 1E), and chemokines (Fig. 1F & G). Our findings determine that this tumor cells and 2/20 CTL co-culture model resembles T cell-tumor interactions in human cancer.

Fig. 1: Tumor cell and tumor-specific T cell interaction leads to T cell activation.
figure 1

A The tumor-specific 2/20 CTL phenotype. B CTL responds to tumor cells by activating PD-1 expression. C Tumor cell and tumor-specific CTL co-culture results in rapid CTL proliferation in vitro. Day1= one day after tumor and CTL co-culture. Day2=two days after tumor and CTL co-culture. Day3=three days after tumor cell and CTL co-culture. D–G Cytokine and chemokine profiles of tumor-activated CTLs in vitro. 2/20 CTLs were cultured with 4T1 and AB1 tumor cells at 1:1 ratio, respectively. Culture supernatants from tumor cell alone, CTLs alone, and tumor cell-CTL cocultures were collected at days 1 and 3 and measured for the indicated cytokines.

Full size image

To profile the direct effect of tumor cells on CTL metabolism during T cell activation, we performed metabolomics profiling (Fig. 2A–E) and determined that both tumor antigen and tumor cells regulate unique metabolic pathways in tumor-specific CTLs (Fig. 2B, D). In antigen AH1-activated CTLs, the most significant change is increased glucose uptake, decreased intracellular cysteine (Cys), Trp, 3-phosphaglycerate, aconitic acid, and adenosine monophosphate (AMP) (Fig. 2B). In the tumor cell-activated CTLs, there are broader changes in metabolites (Fig. 2D). In contrast to the AH1-activated CTLs, glucose level is diminished and levels of Trp and 3-phosphaglycerate increased in tumor-activated CTLs. The intracellular levels of asparagine, aspartic acid, Cys, glutamic acid, glutamine, fructose-1,6-bisphosphate, lactic acid, orotic acid, cytidine, xanthosine are also decreased (Fig. 2C and E). Compared to the tumor-activated CTLs, AH1-activated CTLs show increased anaerobic glycolysis (lactic acid), TCA cycle (citric acid) and glucose, whereas the tumor cell-activated CTLs show increased Trp and 3- phosphoglycerate (Fig. 2C and E). These observations indicate that tumor cells induce a broad change in metabolites with depleted intracellular glucose and diminished glycolysis-TCA cycle in the antigen-activated CTLs.

Fig. 2: Tumor-specific CTLs respond to tumor cells by increasing tryptophan accumulation.
figure 2

A Experimental design of metabolomics, RNA-Seq, and ATAC-Seq analysis of tumor-specific CTLs. B Metabolite set enrichment analysis (MSEA) of metabolites increased (left) or decreased (right) in antigen (AH1 peptide)-activated 2/20 CTLs compared to resting 2/20 CTLs. C Relative levels of amino acids, glycolysis and TCA cycle intermediates, and PPP metabolites and nucleotides in antigen peptide-activated 2/20 CTLs compared to resting 2/20 CTLs. D Metabolite set enrichment analysis (MSEA) of metabolites increased (left) or decreased (right) in tumor cell-activated CTLs as compared to the resting CTLs. E Relative levels of amino acids, glycolysis-TCA cycle intermediates, and PPP metabolites in tumor cell-activated CTLs as compared to the resting CTLs.

Full size image

Trp transporter Slc7a5 is transcriptionally activated in activated CTLs

We next performed genome-wide ATAC-Seq analysis of AH1 peptide-activated and tumor cell-activated 2/20 CTLs using resting 2/20 CTLs as control. The ATAC-seq analysis identified 16,097 and 24,079 differential peaks representing increased or decreased chromatin accessibility in AH1 peptide- and tumor-activated 2/20 CTLs compared to resting CTLs, respectively (Fig. S2). The 7982 more peaks as identified by ATAC-seq in the tumor cell-activated CTLs indicate that tumor cells regulate CTL chromatin remodeling in addition to the tumor antigen/AH1-induced chromatin accessibility change. In AH1-activated CTLs, clustering analysis grouped the peaks into 4 clusters. Cluster 1 constitutes the largest cluster with smaller peaks in the AH1-activated CTLs but much larger peaks in the tumor-activated CTLs, whereas clusters 2-4 peaks have stronger signals in AH1-activated CTLs than tumor cell-activated CTLs and resting CTLs (Fig. S2). In the tumor cell-activated CTLs, clustering analysis also grouped the peaks into 4 clusters. Clusters 1 and 2 represent peaks with increased ATAC-seq intensities and sizes in tumor cell-activated CTLs and further strengthen in AH1-activated CTLs compared to resting CTLs. However, cluster 3, consisting of the largest number of peaks, appears to be only induced in tumor cell-activated but not in AH1-activated CTLs (Fig. S2). Taken together, our findings indicate that tumor cells enhance the antigen-induced chromatin accessibility in both the peak number (increased chromatin accessibility sites) and peak height (greater chromatin accessibility) in the tumor-antigen-activated CTLs (Fig. S2).

Functional clustering revealed that a large set of genes with increased chromatin accessibility belongs to the solute carrier (SLC) superfamily (Fig. 3A–C). Parallel genome-wide gene expression analysis by RNA-seq revealed that both transcriptional activation and repression are associated with the increase in the respective gene chromatin accessibility (Fig. 3A, B). In the AH1-activated CTLs, approximately half of the increased chromatin accessibility is linked to transcriptional repression and another half is linked to transcriptional activation (Fig. 3A, D). In the tumor cell-activated CTLs, the majority of increased chromatin accessibility is linked to transcriptional activation (Fig. 3B, D). Further analysis of SLC gene cluster revealed that Slc7a5 gene has increased chromatin accessibility and expression level in both AH1-activated and tumor cell-activated CTLs as compared to resting CTLs (Fig. 3D). However, AH1-activated CTLs have a greater elevation in Slc7a5 expression than tumor-activated CTLs (Fig. 3D). Analysis of the Slc7a5 gene locus indicates that, compared to tumor antigen/AH1 only, tumor cells induced chromatin accessibility increase in three unique sites and one of these three sites had AP-1 motif (Fig. 3E). In summary, our findings determine that tumor-specific CTLs up-regulate Slc7a5 expression and increase intracellular Trp level in response to tumor cell-induced activation.

Fig. 3: Slc7a5 is transcriptionally activated in tumor cell-activated CTLs.
figure 3

A Heatmaps illustrating differential chromatin accessibility at 2 kb windows centered at the summit of the ATAC-seq peaks identified between AH1-activated 2/20 CTLs and resting 2/20 CTLs (first three columns) and gene expression log2 fold changes at 24 hours between these two samples (4th column on the right). Line graphs on top of the heatmaps demonstrate the average accessibility profiles of clusters 1-4, respectively. The box plot on top of the expression heatmap represents the average log2 fold changes of clusters 1-4. The ATAC-seq peak signals represent the average signals of two biological replicates. For genes associated with multiple ATAC-seq peaks, only peaks with the highest log2 fold changes of differential peak signals were shown. B Heatmaps illustrating differential chromatin accessibility at 2 kb windows centered at the summit of the ATAC-seq peaks identified between tumor cell-activated 2/20 CTLs and resting 2/20 CTLs samples (first three columns) and gene expression log2 fold changes at 24 hours between these two samples (4th column on the right). C GSEA analysis results demonstrate the enrichment of up-regulated transporter genes in AH1-activated and tumor cell-activated 2/20 CTLs as compared with resting 2/20 CTLs. 142 differentially expressed transporter genes were used as a gene set and compared with rank-sorted gene expression from matched samples. The results suggest that more transporter genes are up regulated. D Heatmap illustrating the differential expression patterns of the 142 transporter genes at various time points of 2/20 CTL activation by antigen AH1 and tumor cells. E Wiggle plot of ATAC-seq peaks in Slc7a5 gene locus. The colors of the tracks represent the sample groups. Green, resting 2/20 CTLs; red, AH1-activated 2/20 CTLs; and blue, tumor cell-activated 2/20 CTLs. The unique motif peaks in tumor activated CTLs are indicated with the blue arrows. Red arrows point to peaks in activated CTLs.

Full size image

Slc7a5 is essential for CTL intracellular Trp level and tumor lytic activity in vitro

Our above findings determined that tumor-specific CTLs respond to tumor cell encounter by increasing intracellular Trp level and improving Slc7a5 transcriptional activation. To determine whether the increased intracellular Trp level depends on Slc7a5 in tumor-activated CTLs, we co-cultured 4T1 and AB1 tumor cells with 2/20 CTLs, respectively, in the presence of JPH203, a Slc7a5-specific inhibitor. The rationale is that if Slc7a5 is essential for Trp uptake in CTL response to tumor cells to enter the activation phase, inhibition of Slc7a5 should decrease Trp level in tumor-activated CTLs. Indeed, JPH203 treatment decreased Trp level in both 4T1 and AB1 tumor cell-activated 2/20 CTLs (Fig. 4A–D). In fact, among the amino acids detected, JPH203 only inhibited Trp level in both 4T1- and AB1-activated CTLs (Fig. 4A). JPH203 increased the intracellular levels of glucose, glucose-6-phosphate, aconitic acid, and ribulose-5-phosphate in both tumor cell-activated CTLs (Fig. 4A–D).

Fig. 4: Slc7a5 is essential for CTL tryptophan uptake in tumor cell-activated CTLs in vitro.
figure 4

A–D 2/20 CTLs were co-cultured with 4T1 and AB1 tumor cells, respectively in the presence of JPH203 (10 μM) for 24 h, and analyzed for the indicated metabolites. E–H 2/20 CTLs were co-cultured with 4T1 and AB1 tumor cells, respectively, in the presence of JPH203 at the indicated doses for 6 h. The CTL and tumor co-culture mixtures were stained with CD8-specific mAb, Annexin V, and PI. CD8+ and CD8- cells were gated for cell death of the CTLs (E, F) and tumor cells (G, H).

Full size image

To determine the function of Slc7a5 in CTL anti-tumor cytotoxicity, we used the 4T1-2/20 CTL and AB1-2/20 CTL co-culture models in the presence of JPH203 and analyzed CTL lytic activity. A short time course was used to unmask the saturation of CTL lytic activity. JPH203 showed no cytotoxicity to 2/20 CTLs in the efficacious concentration (Fig. 4E, F), and inhibited 2/20 CTL lytic activity in killing the tumor cells at a dose-dependent manner (Fig. 4G & H). In addition, JPH203 showed no cytotoxicity to the tumor cells in the absence of CTLs (Fig. 4H). Taken together, these findings determined that Slc7a5 transports Trp in tumor-activated CTLs and the Slc7a5-Trp axis enhances CTL anti-tumor cytotoxicity.

Slc7a5 is essential for T cell activation and host cancer immunosurveillance in mice

To determine whether the above findings can be translated to the tumor microenvironment in vivo, we created mice with Slc7a5 deletion only in T cells (Slc7a5.TKO mice). Slc7a5.TKO mice have normal T cell, B cell, and myeloid cell differentiation as compared to WT littermate control mice (Fig. S3A–C). AT3 tumor cells were injected into the mammary gland of Slc7a5.TKO and littermate control WT mice (WT mice) to establish orthotopic breast tumor. The transplanted AT3 primary tumors grew in similar rate in WT and Slc7a5.TKO mice (Fig. 5A, B). However, the AT3 primary tumor metastasized to the lung in the tumor-bearing Slc7a5.TKO mice, but no lung metastases developed in the WT mice (Fig. 5C). To determine that this phenomenon is not tumor type-specific, we injected melanoma B16-F10 cells subcutaneously to establish orthotopic melanoma. B16-F10 tumor grew only in 4 of the 7 WT mice, but all 7 Slc7a5.TKO mice developed B16-F10 primary tumor (Fig. 5D, E). Furthermore, the B16-F10 primary tumor grew significantly faster in Slc7a5.TKO mice as compared to the WT mice (Fig. 5D & E). Analysis of tumor cell proliferation revealed that the AT3 tumor cells have similar proliferation rate in WT and SLC7a5.TKO mice (Fig. S4A, B). However, AT3 tumor in the WT mice exhibited significantly higher apoptosis rate than the tumor in the Slc7a5.TKO mice (Fig. S4C, D).

Fig. 5: Slc7a5 is essential for TIL expansion and tumor growth control in mice.
figure 5

A AT3 cells were injected orthotopically in breast in WT (n = 9) and Slc7a5.TKO (n = 8) C57BL/6 mice. Shown is tumor image. B Tumor size and weight as in A were quantified and presented. C The tumor-bearing mice as in A were sacrificed after 14 days and the lungs were inflated by India ink. The tumor nodules were quantified and presented. D B16F10 cells were injected subcutaneously in WT (n = 4) and Slc7a5.TKO (n = 7) C57BL/6 mice. Shown is tumor image. E Tumor size and weight as in D were quantified and presented. F Spleens were collected from AT3 tumor-bearing WT and Slc7a5.TKO C57BL/6 mice as in A and analyzed by flow cytometry for CD8+, CD4+ and CD11b+ cells. Shown are representative dot plots. G Quantification of indicated immune cells as in (F). H Tumor tissues were collected from AT3 tumor-bearing WT and Slc7a5.TKO C57BL/6 mice as in A and analyzed by flow cytometry for CD8+, CD4+ and CD11b+ cells. Shown are representative dot plots. I Quantification of indicated immune cells as in (H). J Spleens were collected from B16F10 tumor-bearing WT and Slc7a5.TKO C57BL/6 mice as in D and analyzed by flow cytometry for CD8+ and CD4+ cells. Shown are representative dot plots. K Quantification of indicated immune cells as in (J). L Tumor tissues were collected from B16F10 tumor-bearing WT and Slc7a5.TKO C57BL/6 mice as in D and analyzed by flow cytometry for CD8+ and CD4+cells. Shown are representative dot plots. M Quantification of indicated immune cells as in (L).

Full size image

In AT3 tumor-bearing mice, the CD4+ T cell level is not different in spleens of WT and Slc7a5.TKO mice and CD8+ T cell level is lower in Slc7a5.TKO mice than in WT mice. Myeloid cell level is not different between WT and Slc7a5.TKO mice (Fig. 5F, G). In B16-F10 tumor-bearing mice, CD8+ T cell level is not different in spleens of WT and Slc7a5.TKO mice and CD4+ T cell level is lower in Slc7a5.TKO mice than in WT mice (Fig. 5J, K). Analysis of the tumors revealed that both tumor-infiltrating CD4+ and CD8+ T cells are diminished in Slc7a5.TKO mice as compared to WT mice in both AT3 tumor-bearing (Fig. 5H, I) and B16-F10-tumor-bearing (Fig. 5L, M) mice. These findings indicate that Slc7a5 is essential for T cell activation and function in response to tumor.

Slc7a5 regulates activation of the AhR metabolic pathway in tumor-infiltrating T cells in mice

To elucidate the cellular and molecular mechanisms underlying Slc7a5 function in regulating the tumor cell-induced T cell activation and function, we dissected tumor tissues from AT3 tumor-bearing WT and Slc7a5.TKO mice, isolated CD45+ tumor-infiltrating immune cells, and analyzed the isolated cells by single-cell RNA sequencing (scRNA-seq). A total of 127,621 cells passed quality control and were analyzed by scRNA-seq. Uniform manifold approximation and projection (UMAP) visualization identified 10 unique cell clusters/subpopulations in the tumor-infiltrating immune cells (Fig. 6A) with unique gene signatures (Fig. S5). Consistent with the findings by flow cytometry analysis (Fig. 5H, I), T cells constitute a small population of immune cells in the tumor microenvironment, and its level is lower in Slc7a5.TKO mice than in WT mice (Fig. 6A). Although total tumor-infiltrating myeloid cell levels are not different between WT and Slc7a5.TKO mice (Fig. 5I), host T cell Slc7a5 deficiency decreased DC and neutrophils, and increased monocytes in the tumor microenvironment (Fig. 6A).

Fig. 6: Slc7a5 regulates TCR signaling and the AhR metabolic pathway in TILs in mice.
figure 6

A UMAP plot of major cell subpopulations isolated from orthotopic breast tumor-bearing mice, colored by identified cell clusters (right panel). B Dot plot of signaling pathway activation as identified by expression of gene signatures in the indicated cell subpopulations in WT and Slc7a5.TKO (KO) mice. C Violin plot of signaling pathway activation as identified by expression of gene signatures in the indicated cell subpopulations in WT and Slc7a5.TKO mice. Dot (D) and violin (E) plots of expression of the indicated genes in the indicated cell subpopulations in WT and Slc7a5.TKO mice.

Full size image

Analysis of cellular signaling pathways revealed that the TCR signaling pathway and AhR metabolic pathway were impaired in tumor-infiltrating T cells in Slc7a5.TKO mice as compared to WT mice. However, the activation of several signaling and metabolic pathways with known functions in T cells cell activation, including the mTORC1, pyruvate metabolism and TCA cycle, IFN-I, IFNγ were not affected by Slc7a5 deficiency in T cells (Fig. 6B, C). Analysis of individual gene expression levels revealed that the expression levels of PD-1 and FasL were lower in tumor-infiltrating T cells in Slc7a5.TKO mice as compared to WT mice (Fig. 6D, E). IFNγ expression level was also lower in tumor-infiltrating T cells in Slc7a5.TKO mice (Fig. 6D, E). Taken together, our findings indicate that Slc7a5 deficiency in T cells impairs TCR and AhR signaling pathway in the cellular level, and decreased expression of IFNγ, PD-1 and FasL in the molecular level in T cells in the tumor microenvironment.

AhR binds to the Faslg promoter to activate FasL expression in T cells to control tumor growth in mice

AhR is a transcription factor that is in an inactive form in the cytoplasm, but translocates to nucleus to bind to gene promoters to activate transcription once activated [12]. AhR is known to bind to the Pdcd1 promoter to activate PD-1 transcription [13]. As expected, both AhR signaling pathway and PD-1 expression is diminished in Slc7a5-deficient T cells (Fig. 6D, E). FasL expression is also diminished in Slc7a5-deficient tumor-infiltrating T cells (Fig. 6D, E), which leads us to hypothesize that AhR binds to the Faslg promoter to activate FasL expression. To test this hypothesis, we analyzed the Faslg locus in the tumor cell-activated 2/20 CTLs and identified three tumor cell-induced chromatin accessibility peaks (Fig. 7A). DNA sequence analysis of the Faslg promoter identified an AhR-binding consensus sequence (Fig. 7B). CD8+ T cells were then isolated from AT3 tumor-bearing WT and Slc7a5.TKO mice and analyzed by ChIP using an AhR-specific antibody. Indeed, AhR binds to the Faslg promoter in tumor-infiltrating CD8+ T cells in WT mice but not in SLC7a5.TKO mice (Fig. 7B, C).

Fig. 7: AhR activates FasL expression to suppress tumor growth and metastasis in vivo.
figure 7

A Wiggle plot of ATAC-seq peaks in the Faslg gene locus. Peaks observed in the tumor-activated CTLs are highlighted with arrows. B The Falg gene promoter structure with the predicted AhR-binding consensus sequence element (left panel). C Tumor-infiltrating T cells were isolated from AT3 tumor-bearing WT and SLC7A5.TKO mice. Shown is purity of the isolated CD8+ T cells from the tumor tissues. The purified CD8+ T cells were then subjected to ChIP using anti-AhR antibody. The input DNA was used as positive control and the IgG was used as negative control. D, E Breast tumor AT3 cells were injected orthotopically into mice. Sixteen mice with similar tumor size were randomly divided into two groups and treated with IgG or anti-FasL antibody every 3 days for four times. Shown are tumor images (D) and tumor size and weight (E) at the end of experiment. F The tumor-bearing mice as shown in (C) were inflated with ink to inflate lungs and tumor nodules in the lungs were quantified. G, H Melanoma B16F10 cells were injected subcutaneously to mice. Twelve mice with similar size of tumors were randomly divided into two groups and treated with IgG or anti-FasL mAb, respectively, every 3 days for 4 times. Shown are tumor image (G) and tumor size and weight (H). I TUNEL analysis of melanoma B16F10 tumors treated with IgG or anti-FasL antibody as in (G). Shown are representative images. Red arrows point to apoptotic cells. The apoptotic cells were quantified and presented at the right.

Full size image

FasL is a ligand of activated T cells for executing CTL cytotoxicity [30, 31], including anti-tumor cytotoxicity [23, 32,33,34,35]. Our above findings that tumor growth and progression were increased in Slc7a5.TKO mice and FasL expression is decreased in tumor-infiltrating T cells in Slc7a7.TKO mice suggest that diminished FasL expression might be responsible at least in part for the increased tumor growth and progression. To test this hypothesis, we treated AT3 tumor-bearing mice with FasL neutralization mAb. FasL blockade significantly increased AT3 tumor growth and AT3 tumor lung metastasis (Fig. 7D–F). We then extended this study to the B16-F10 mouse model. FasL blockade also significantly increased B16-F10 tumor growth in vivo (Fig. 7G, H). Analysis of tumor cell apoptosis revealed that FasL blockade significantly decreased tumor cell apoptosis in vivo (Fig. 7I). Our findings therefore determine that Slc7a5-Trp metabolic pathway activates AhR to up-regulate FasL expression in tumor-infiltrating CTLs to control tumor growth in mice.

FasL expression in T cells correlates with cancer patient response to nivolumab immunotherapy and survival

To determine the human cancer relevance, we extended our study to human cancer datasets. We retrieved FASLG scRNA-seq datasets from the Human Protein Atlas (HPA) and Genotype-Tissue Expression (GTEx) datasets [36]. FasL is selectively expressed in NK cells and T cells (Fig. S6). In human breast cancer patients treated with pembrolizumab [28], PD-1 and FasL expression levels are higher in tumor-infiltrating T cells in responders than in non-responders (Fig. 8A–E). Furthermore, Slc7a5 expression level is positively correlated with FasL expression level in tumor-infiltrating T cells in breast cancer patients (Fig. 8F). Analysis of TCGA datasets revealed that breast and melanoma cancer patients who had high FasL expression levels had significantly longer overall and disease-free survival (Fig. 8G, H). More importantly, analysis of datasets of pan-cancer (thirty-three types of human cancer) revealed that patients with high FasL expression level exhibit significantly longer overall and disease-free survival (Fig. 8I). Taken together, these findings determine that the SLC7A5-FasL axis mediates human cancer patient response to immune checkpoint inhibitor immunotherapy and improve patient survival.

Fig. 8: FasL expression is correlated with human cancer patient response to PD-1 blockade immunotherapy and patient survival.
figure 8

A Published human breast cancer patient scRNA datasets were extracted downloaded from the database [28] and analyzed using R package. UMAP plot major cell subpopulations in responders and non-responders of breast cancer patients treated with pembrolizumab. B–E UMAP plot of the major cell populations as shown in B, showing PD-1 (C, D) and FasL (E, F) expression level in TILs. Red dots represent PD-1+ (C) and FasL+ (E) TILs. PD-1 (D) and FasL (F) expression level is shown as violin plots. F Correlation between FasL expression and Slc7a5 expression in tumor-infiltrating T cells in human breast cancer patients. G TCGA dataset analysis showing correlation between FasL expression level and patient survival in breast cancer. H TCGA dataset analysis showing correlation between FasL expression level and patient survival in melanoma. I TCGA dataset analysis showing FasL expression level and patient survival in pan-cancer (33 types of human cancers).

Full size image

Discussion

Leucine is an amino acid that is transported to T cells through Slc7a5 during T cell activation by TCR engagement in vitro [19]. The Slc7a5-Leucine metabolic pathway is essential for T cell differentiation during viral infection in mice [20]. In this study, we observed that increased level of intracellular Trp is a key phenotype of tumor cell-activated CTLs in vitro and this increase in intracellular Trp is through Slc7a5. Mice with Slc7a5 deficiency only in T cells exhibited diminished T cell activation and expansion when challenged with tumor. These findings identified the Slc7a5-Trp axis as a metabolic pathway that controls tumor-specific T cell activation in response to tumor. Slc7a5 is the catalytic subunit of the system L1-type amino acid transporter and forms a heterodimer with subunit Slc3a2 [5, 17] to transport large neutral amino acids, including leucine, isoleucine, histidine, methionine, phenylamine, tyrosine, tryptophan, and valine in mammalian cells [5, 18]. The mechanism underlying Trp transport through Slc7a5 in tumor cell-activated CTLs is not clear and requires further studies.

In T cells, Slc7a5 is the rate-limiting transporter subunit that transport leucine to induce T cell metabolic reprogramming to enhance antigen-activated TCR signaling to activate Myc and mTORC1 signaling pathway to regulate T cell activation and differentiation during a viral infection [20]. The Myc and mTORC1 signaling pathways are therefore TCR downstream signaling pathways [37, 38], which are essential for TCR initiated and the Slc7a5-Leucine metabolic pathway-regulated T cell activation and differentiation during viral infection [20]. Slc7a5 also transports methionine to reprogram a H3K27me3-dependent transcriptional program to regulate T cell stemness [39]. In this study, we observed that tumor cells activate the TCR signaling pathway in tumor-infiltrating T cells in tumor-bearing mice. However, activation of the Myc and mTORC1 signaling pathways does not depend on Slc7a5. Our findings thus indicate that Slc7a5 transports different amino acids into T cells to rewire different T cell metabolic programs to control T cell activation and function in different cellular context.

Unlike the TCR-mediated canonical signaling pathway that activates Myc and mTORC1 signaling pathways in T cells, we observed instead that the AhR signaling pathway, a pathway that is associated with type I regulatory T cell differentiation, T cell exhaustion and tumor immune evasion [13, 16, 40,41,42], is regulated by Slc7a5 in tumor cell-activated T cells. Trp metabolites are potent ligands for AhR [15, 43]. AhR is known to act as a transcription factor once activated by Trp metabolites in T cells [13, 44]. Although compelling experimental data indicate that AhR activates the immune suppressive pathways in the tumor microenvironment [13, 15, 16, 43, 45, 46], emerging experimental data indicate that AhR may also functions as a cell intrinsic factor for T cell immunity [6, 46, 47] and Trp metabolites enhance efficacy of checkpoint inhibitor immunotherapy and chemotherapy [14, 48]. In addition, loss of AhR in mice promoted tumor development through decreased T cell infiltration in the tumor microenvironment [49]. In this study, we determined that the Slc7a5-Trp metabolic pathway activates AhR to up-regulate FasL expression in tumor-infiltrating T cells. T cells kill tumor cells via the Fas-FasL [30, 31, 35, 50, 51] and perforin-granzyme B [33, 52] pathways. Compelling literatures indicate that FasL is essential for T cell-mediated host cancer immunosurveillance to control tumor growth in vivo [23, 34, 53]. Emerging literature demonstrates that the Fas-FasL cytotoxic pathway is also essential for CAR-T cell efficacy [54,55,56,57]. In addition, FasL plays a key role in suppression of tumor lung metastasis [23]. Our findings thus suggest that in addition to regulating T cell activation and expansion, the Slc7a5-Trp metabolic pathway also activates AhR to up-regulate FasL expression in tumor-infiltrating T cells to control tumor growth and progression in mice.

Although Slc7a5 is necessary for host cancer immunosurveillance in both the AT3 mammary tumor-bearing mice and B16-F10 melanoma tumor-bearing mice, the activity of T cells significantly differs in these two types of tumor models. The AT3 primary tumor size does not change in the absence of Slc7a5, but its suppressive activity of tumor metastasis is impaired, whereas B16-F10 melanoma primary tumor multiplicity and growth are significantly reduced in mice with Slc7a5 deficiency in T cells. Furthermore, the density of tumor infiltrating, but not spleen resident, T cells are decreased in Slc7a5.TKO mice. These findings indicate that rather than a defect in T cell clonal expansion during the priming phase in secondary lymphoid tissues, there might be a defect in T cell recruitment and/or survival into tumor site in Slc7a5.TKO mice.

Slc7a5 is up-regulated in tumor cells and promotes tumor growth [58], which provides a rationale to inhibit Slc7a5 in human cancer therapy [59]. Slc7a5 inhibitor is currently being tested in the clinic. However, current clinical trials indicate that Slc7a5 inhibitor has no efficacy in human cancer patients. Our findings indicate that Slc7a5 is also essential for activation and effector function of tumor-infiltrating CTLs. Emerging experimental data indicate that engineering T cells to express SLC7A5 alongside CARs leads to enhanced CAR-T proliferation in the tumor microenvironment [60,61,62] and therefore could be a promising new direction in adoptive and CAR-T cell cancer immunotherapy [62, 63]. Our findings thus argue that Slc7a5 should be activated to boost T cell activation in the tumor microenvironment in human cancer therapy. However, AhR also promotes differentiation of the immune suppressive type I regulatory T cells [40,41,42]. The potential translation of our findings to the treatment of cancer in humans may be complex and depends on the trade-off between increased Trp-AhR activation in CTLs, the potential suppressive effect of regulatory CD4+ T cells, and the promotion of tumor cell proliferation by AhR.

Our findings indicate that tumor cell factor(s) induce increased Trp uptake and accumulation in tumor-activated CTLs. The tumor-specific CTLs respond to tumor cells by rapidly producing various cytokines, chemokines, and growth factors. It is therefore likely that factor(s) produced by CTLs promote Slc7a5 uptake of Trp. It is also possible that tumor cells-produced factor(s) facilitate Slc7a5-dependent Trp uptake in CTLs. Further studies are required to identify these factors.