Inactivation of tumor-infiltrating lymphocytes (TIL) is one of the mechanisms mitigating antitumor immunity during tumor onset and progression. Epigenetic abnormalities are regarded as a major culprit contributing to the dysfunction of TILs within tumor microenvironments. In this study, we used a murine model of melanoma to discover that Tet2 inactivation significantly enhances the antitumor activity of TILs with an efficacy comparable to immune checkpoint inhibition imposed by anti-PD-L1 treatment. Single-cell RNA-sequencing analysis suggested that Tet2-deficient TILs exhibit effector-like features. Transcriptomic and ATAC-sequencing analysis showed that Tet2 ablation reshapes chromatin accessibility and favors binding of transcription factors geared toward CD8+ T-cell activation. Furthermore, the ETS family of transcription factors contributed to augmented CD8+ T-cell function following Tet2 depletion. Overall, our study establishes that Tet2 constitutes one of the epigenetic barriers that account for dysfunction of TILs and that Tet2 inactivation could promote antitumor immunity to suppress tumor growth.
This study suggests that ablation of TET2+ from TILs could promote their antitumor function by reshaping chromatin accessibility for key transcription factors and enhancing the transcription of genes essential for antitumor activity.
Dynamic epigenetic alterations have been observed in CD8+ tumor-infiltrating lymphocytes (TIL) during tumor progression (1, 2). Epigenetic plasticity is one of the critical factors governing the expression of key genes that define the status of TILs during tumor development. Recent studies have shown that DNA methylation plays an essential role in regulating CD8+ T-cell immune response during chronic infection and tumorigenesis (3–6). Both effector and inhibitory genes are tightly controlled by DNA methylation during CD8+ expansion, activation, and exhaustion (7). For example, the demethylation of Gzmb, Ifng, and Pdcd1 loci is essential for active expression of these genes in CD8+ effector T cells during acute infection (3). In addition, Ghoneim and colleagues reported that blockage of de novo DNA methylation via Dnmt3a deletion resulted in prolonged CD8+ T-cell immunity during persistent immune response (4). Clearly, DNA methylation-associated epigenetic reprogramming is critical for the adaptive immune response of CD8+ T cells.
The reversal of DNA methylation, or DNA demethylation, is catalyzed by the ten–eleven translocation (TET) protein family (TET1, 2, 3) of dioxygenases (8–10), which successively oxidize 5-methylcytosine (5mC) to yield 5-hydroxymethylcytosine (5hmC), 5-formylcytosine (5fC), and 5-carboxylcytosine (5caC; refs. 8–10). Genetic depletion of Tet could impair chromatin accessibility and perturb the binding of transcription factors (TF) to their genomic targets in cells of the immune systems, including B cells, T cells, and macrophages (11–14). Among all the three Tet proteins, Tet2 is most frequently mutated in the hematopoietic and immune systems (15–17). Tet2-deficient hematopoietic stem cells, lymphoid, or myeloid cells exhibited aberrant immune activities when responding to external stimulations (18–20). For example, Tet2 ablation in CD8+ T cells promotes memory differentiation and enhances pathogen control (6). In addition, disruption of TET2 in both chimeric antigen receptor (CAR) T cells and tumor-infiltrating myeloid cells (13, 14) improves immunotherapy efficiency, suggesting the beneficial effects of TET2 inactivation during antitumor immunotherapy. At the molecular level, we and other groups have shown that Tet-mediated DNA methylation oxidation regulates gene expression by controlling chromatin accessibility and TF binding during embryonic development (11, 21). However, the role of Tet2 in TILs during antitumor immunity has not been systematically defined.
In this study, we utilized a murine B16-OVA and Tet2-deficient OT-I mouse model to investigate the function of Tet2 in TILs. We unveiled that Tet2-deficient CD8+ T cells exhibited enhanced cytotoxicity to tumor cells at an early stage of tumorigenesis, which resulted in significantly enhanced antitumor efficiency. Tet2-deficient TILs showed similar antitumor activity to immune checkpoint inhibitors, for example, anti-PDL-1. Transcriptomic and epigenomic studies revealed that Tet2 inactivation reshaped the chromatin accessibility at the genomic binding sites of several key transcription factors (e.g., the bZIP and ETS TF families), thereby enhancing the transcription of genes encoding effector molecules (e.g., Tnf and Prf1) to promote the antitumor efficiency of TILs and suppress melanoma growth in vivo.
Materials and Methods
C57BL/6J mice (The Jackson Laboratory) and Tet2 knockout (Tet2KO) mice (22) were crossed with OT-1 mice [C57BL/6-Tg(TcraTcrb)1100Mjb/J; The Jackson Laboratory] bearing transgenic T-cell receptors that recognize ovalbumin (OVA257–264) in the context of CD8 co-receptor interaction with MHC class I. Crossed mice in age between 6 and 20 weeks were used for CD8+ T-cell isolation. Six to 12-week-old CD45.1+ mice (B6.SJL-Ptprca Pepcb/BoyJ; The Jackson Laboratory) were challenged with B16-OVA mouse melanoma tumor cells and then used as recipients for adoptive T-cell transfer experiments. All mice were maintained in the animal facility at the Institute of Biosciences and Technology, Texas A&M University. All animal studies were approved by the Institutional Animal Care Use Committee of the Texas A&M University Institute of Biosciences and Technology.
Mouse T-cell isolation and ex vivo activation
CD8+ T cells were isolated using a mouse CD8a+ T-Cell Isolation Kit (Miltenyi Biotec). Briefly, harvested spleens and lymph nodes from WT-OT-I and Tet2KO-OT-I mice were grounded and filtered with 70-μm cell strainers (Falcon) to remove debris. The suspended cells were treated with an ammonium–chloride–potassium (ACK) lysis buffer to lyse red blood cells and residual cells were purified for total CD8+ T-cell isolation using the T-Cell Isolation Kit according to the manufacturer's protocol. Purified CD8+ T cells were activated by anti-CD3 (clone 17A2; InVivoMAb), anti-CD28 (clone 37.51; InVivoMAb) and a high dose of mouse IL2 (100 unit/mL; eBioscience), and cultured in multiwell plates precoated with a goat anti-hamster IgG (H+L; Invitrogen). After 2 days of culture, T cells were removed from the coated plate and seeded with a low dose of mouse IL2 (20 units/mL; eBioscience) for 24 hours before injection.
The B16-OVA mouse model of melanoma and adoptive T-cell transfer
B16-OVA mouse melanoma cells were obtained from Dr. Anjana Rao's laboratory in La Jolla Institute (La Jolla, CA; ref. 23) and were cultured in complete DMEM, and passaged at least two times before injection. The cells were tested negative for potential Mycoplasma contamination on a monthly basis. Cultured B16-OVA cells were trypsinized and washed with 1× PBS and diluted at 6 million cells/mL. Prepared B16-OVA cells (300,000 cell/mice in 50 μL PBS) were injected intradermally into the dorsal flanks of CD45.1+ recipient mice. Tumor sizes were measured and recorded every day. Ten to twelve days after B16-OVA inoculation, in vitro cultured and activated WT-OT-I and Tet2KO-OT-I CD8+ T cells were retro-orbitally injected into tumor bearing CD45.1+ mice (2 million cells/mice). Mice were sacrificed 3 and 8 days after adoptive T-cell transfer. Anti–PD-L1 Ab (clone no. 10F.9G2; BioXCell, 200 μg/mice) or control anti-KLH rat IgG2b (BioXCell) were injected (intraperitoneally) in the corresponding groups at day 3 and day 6 after adoptive transfer.
CD8+ T-cell isolation
Spleens, peripheral blood, and tumor mass were dissected at the indicated time points after adoptive T-cell transfer. Spleens and peripheral blood undergo red blood cell lysis by using an ACK lysis buffer and used for downstream flow cytometry analysis. Dissected tumors were dissociated using ultrafine scissors and tweezers, and then incubated in RPMI media with Liberase TL (100 μg/mL; Roche) for 15 minutes at 37°C. The same amount of Liberase was added into the sample and incubated for additional 10 minutes at 37°C. The dissociated tissues were then passed through 70 μm cell strainers (Falcon) and washed with PBS for three times. Cells were then labeled with a mouse anti-CD63-biotin antibody (Miltenyi Biotec) followed by the negative selection using anti-biotin microbeads (Miltenyi Biotec). The negatively selected cells were then stained with corresponding cell surface markers for flow cytometry analysis or cell sorting.
Cell surface marker staining and in vitro restimulation for cytokine staining
In vitro activated CD8+ T cells (before injection, day 0 control) and purified TILs were stained with corresponding cell surface markers (dilution of 1:100 to 1:200) in FACS buffer (1× PBS with 0.1% FBS and 2 mmol/L EDTA) for 20 minutes. Stained cells were washed twice using FACS buffer. For cytokine staining, the same sets of cells were seeded in 48-well plates, followed by incubation with T-cell media containing 10 nmol/L of phorbol 12-myristate 13-acetate (PMA; Sigma-Aldrich), 500 nmol/L of ionomycin (Thermo Fisher Scientific), and 1 μg/mL of brefeldin (BioLegend) for 4 hours at 37°C. Then the cells were stained with corresponding cell surface antibodies (CD45.1, CD45.2, and CD8) followed by fixation and permeabilization with BD Cytofix/Cytoperm Fixation/Permeabilization Solution Kit (BD Biosciences). Fixed and permeabilized cells were then stained with cytokine antibodies and proceed for flow cytometry analysis. All flow cytometry analysis was performed using LSRII (BD Biosciences) and the data were analyzed using FlowJo v.10 (TreeStar Inc.).
In vitro cytotoxicity assay
B16-OVA cells (250,000 cell/well) were incubated with a Cell Proliferation Dye conjugated with APC (eBioscience) and seeded on 24-well plates for 1 hour at 37°C. Then, the culture medium was removed and 4× amount of in vitro activated CD8+ T cells with complete T-cell media were added into labeled B16-OVA cells for 4 and 8 hours at 37°C. Cocultured cells were then washed twice with 1× PBS and dead cells were stained with the CellEvent Caspase-3/7 Green Detection Reagent (Invitrogen) according to the manufacturer's instructions. Double-stained populations were examined by flow cytometry analysis.
For cell viability, proliferation, and cell-cycle analysis, CD8+ T cells were cocultured with prestained B16-OVA cells for 4 hours before analysis. After coculture, the cell mixture was stained with an anti-CD8a antibody (BioLegend). For cell viability analysis, propidium iodide (PI; BD Pharmingen, 5 μL) was incubated with the cell mixture for 5 minutes and the dead cell population was measured by the PI-positive population within CD8+ gated cells. For cell proliferation analysis, the cell mixture was fixed and permeabilized using Transcription Factor Buffer Set (BD Pharmingen), followed by incubating with anti-Ki-67 antibody (BioLegend). Actively proliferating cells were detected by the Ki-67 positive population within CD8+ gated cells. For cell-cycle analysis, 20 μmol/L of BrdU was incubated with the cultured cell mixture for 2 hours followed by an additional 2 hours incubation. Then the cell mixture was stained with an anti-CD8a antibody and fixed/permeabilized using Transcription Factor Buffer Set (BD Pharmingen). Fixed cell mixture was then incubated with 20 μg DNase I (Sigma-Aldrich) at 37°C for 1 hour. Then BrdU antibody was used to capture BrdU-pulsed cells and 7-AAD (Tonbo Bioscience) was used to stain DNA.
Pharmacologic inhibition treatment
YK-4–279 (Tocris Bioscience, 0.1 μmol/L) and TK216 (MedChemExpress, 0.5 nmol/L) were used in both individual B16-OVA or CD8+ T-cell culture, and coculture experiments.
RNA extraction, reverse transcription, and qRT-PCR
Isolated TILs were lysed with RLT buffer (Qiagen) and homogenized using 16G of syringe five times. Then total RNA was purified using AllPrep DNA/RNA Micro Kit (Qiagen) following manufacturer's instruction. The concentration of RNA was measured by Qubit 4 Fluorometer with Qubit RNA High Sensitivity Assay Kit. Purified RNA (500 ng) was reverse transcribed using ABscript II cDNA First Strand Synthesis Kit (ABclonal). Gene expression was quantified on ViiA 7 Real-Time PCR System (Applied Biosystems) using 2× Universal SYBR Green Fast qPCR Mix (ABclonal). Two-step cycling program was used with initial denaturation for 10 minutes at 95°C, followed by 40 cycles with denaturation for 10 seconds at 95°C and annealing/elongation at 60°C for 30 seconds. The total mRNA amount was normalized to endogenous GAPDH mRNA. Primers used for qPCR are listed below:
ATAC-seq library preparation and data analysis
ATAC-seq library preparation was performed as described previously (24). In brief, nuclei were isolated from 50,000 cells in a lysis buffer (10 mmol/L Tris-HCl, pH 7.4, 10 mmol/L NaCl, 3 mmol/L MgCl2, 0.1% IGEPAL CA-630), then transposition reaction was performed using an Illumina Nextera DNA Library Preparation Kit at 37°C for 30 minutes. Tagmented DNA fragments were purified using a MinElute PCR Purification Kit (Qiagen) and purified DNA were amplified with a KAPA Real-Time Library Amplification Kit (Roche). Amplified libraries were then purified with AmpureXP beads. The size of DNA libraries was measured by Bioanalyzer using an Agilent High-Sensitivity DNA Kit (Agilent). The concentration of DNA Libraries was measured by the Qubit 4 Fluorometer with a Qubit dsDNA High Sensitivity Assay Kit (Thermo Fisher Scientific). Equimolarly combined libraries were sequenced on NextSeq 500 (Illumina) with the NextSeq 500/550 High Output Kit v2.5 (80PE).
Adaptor trimming of raw reads were performed by TrimGalore v0.5.0 with default parameters, and high-quality (Q ≥ 20) reads were uniquely aligned to mm10 reference genome using Bowtie2 with “–very-sensitive” options. Only uniquely mapped reads were finally extracted for downstream analysis. The resulting alignment of each sample (with two biological replicates) was analyzed by Genrich v.0.5 (https://github.com/jsh58/Genrich) with ATAC-seq mode (option: -j, -q 0.05, -d 150). The options to remove PCR duplicates (-r) and to discard alignments to chrM (-e chrM) were used to call chromatin accessible peak regions. Bedtools merge was used to count the reads that fell into non-overlapped peak regions, and the significantly differential chromatin accessible regions were detected using DESeq2 with normalized peak signals (fold change ≥ 2; FDR < 0.05). Venn diagram was plotted using R package ggplot2 to find group-specific chromatin changed regions. Motif annotation of differential chromatin accessible regions was performed using HOMER software. GREAT (25) analysis was used to perform the functional annotation of differential chromatin accessible regions. The number of mapped reads and mapped ratios are listed in Supplementary Table S1.
To detect the chromatin accessibility across the transcription factors, published BATF and ETS1 chromatin immunoprecipitation (ChIP)-seq data were downloaded from GSE54191 and GSE56393, respectively. High-quality reads were mapped to mm10 reference using Bowtie2 with “-very-sensitive” options, and the binding regions were detected using Macs2 with default parameters. The profile and heatmaps of chromatin accessibility across upstream and downstream of BATF binding loci were generated using the computeMatrix and plotHeatmap scripts from the deepTools package. Bam2wig.py was used to transform the bam file to normalized bigwig files (option: -t 2000000000).
RNA-seq library preparation and data analysis
PolyA tailed mRNA was enriched using NEBNext Poly(A) mRNA Magnetic Isolation Module (NEB). A NEBNext Ultra II Directional RNA Library Prep Kit for Illumina (NEB) was used to generate RNA-sequencing (RNS-seq) libraries. Then adaptor for Illumine sequencing system was ligated into cDNA, followed by final library amplification with illumina TruSeq single index (NEB). Constructed libraries were purified with AmpureXP beads. Size and concentration of generated libraries were verified as described above.
To process the sequencing data, low-quality bases and adaptors were trimmed using TrimGalore v.0.5.0 with default parameters (https://github.com/FelixKrueger/TrimGalore). Clean reads of RNA-seq data were aligned to mm10 reference genome using Hisat2 (version, 2.1.0) with default parameters and only uniquely mapped reads were used for downstream analysis. Count matrix for each gene was generated using htseq-count (HTSeq package). DESeq2 was used to identify significantly differentially expressed genes (DEG) in WT and Tet2KO samples between different time points (fold change ≥ 2; FDR < 0.05). In-house R scripts were used to plot the volcano and scatter plots for DEGs. The Gene Set Enrichment Analysis (GSEA) was performed for the functional enrichment analysis of DEGs. Hierarchical cluster analysis of the union DEGs was used to determine group-specific signature genes. The number of mapped reads and mapped ratios were listed in Supplementary Table S1.
Single-cell RNA-seq (scRNA-seq) libraries were generated using the Chromium Single-Cell 3′ Reagent V2 Kit (10× Genomics) according to the manufacturer's instructions. Briefly, single-cell GEM was generated and barcoded in a Chromium Controller (10× Genomics). Then RNA transcripts from single cells were reverse transcribed, amplified, and fragmented. The library size and concentration were measured using Agilent Bioanalyzer 2100 (Agilent) and Qubit 4 Fluorometer with Qubit dsDNA High-Sensitivity Assay Kit as described above. Libraries were subjected to illumina NextSeq 500 system using the NextSeq 500 High Output v2 Kit (Illumina) with a paired end, dual indexing (26/8/0/58-bp) format. The mkfastq in Cellranger (version 3.0.2) was first used to demultiplex the raw sequencing data into fastq files. Cellranger was used to align raw fastq files to mm10 reference genome and perform barcode and UMI counting. The count matrix for each gene of each cell was taken as an input file for R package Seruat (version 2.3). Reads with the same UMI were combined and then annotated to ensembl genes (mm10). To ensure the data quality, the genes detected in less than 10 cells and the cells with less than 20 genes were filtered out first from each dataset. We used the first eight principle components to perform cell cluster and t-distributed stochastic neighbor embedding (t-SNE) with resolution = 0.4. To identify the marker genes, differential expression analysis was performed by FindAllMarkers function with Wilcoxon rank sum test. Violin plots were generated using Seurat Vlnplot function. The raw reads of different CD8+ TILs cells (bulk RNA-seq dataset) were downloaded from GSE122969, and the data analysis was similar to RNA-seq data of this study. The heatmap of scaled gene expression level across top 20 significant marker genes for each cluster was plotted by R package ComplexHeatmap.
The sequencing datasets have been deposited into NCBI BioProject under the accession number GSE152496.
Tet2 deletion enhances the antitumor activity of CD8+ TILs
To study the function of Tet2 in CD8+ TILs, we crossed Tet2KO mice (22) with OT-I mice (26) to generate a Tet2KO-OT-I mouse model (Supplementary Fig. S1A), which could recognize the chicken ovalbumin (OVA) antigen (27) expressed on the surface of B16 murine melanoma cells (B16-OVA). In this model, engineered B16-OVA melanoma cells could be recognized by adoptively transferred OT-I OVA specific CD8+ T cells during tumor onset and progression (28, 29). As shown in Fig. 1A, the B16-OVA melanoma cells were intradermally injected in the B6.SJL-Ptprca Pepcb/BoyJ recipient mice (30) expressing CD45.1 isoform in their lymphocytes. In parallel, the CD8+ T cells were purified from wild-type (WT)-OT-I or Tet2KO-OT-I mice with CD45.2 expression in their lymphocytes, and cultured ex vivo for expansion for up to 72 h. Next, the purified CD45.2+CD8+OT-I T cells (WT or Tet2KO; 2 million cells/mice) were adoptively transferred to the tumor-bearing CD45.1+ recipient mice at 10–12 days after tumor inoculation via retro-orbital injection. To compare the tumor-suppressive effects between immune checkpoint inhibitors and Tet2 ablation, we further treated the mice with an anti-PD-L1 inhibitor or an IgG2b antibody (as control). In mice transferred with Tet2KO TILs, we observed a pronounced reduction of tumor size when compared with the WT group (Fig. 1B; Supplementary S1B, red line). The similar suppressive effect was also noted in mice transferred with WT TILs and treated with anti-PD-L1 antibody (Fig. 1B; Supplementary S1B, green line). Surprisingly, the anti-PD-L1 treatment had minor effects on mice transferred with Tet2KO TILs compared with the control group transferred with Tet2KO TILs (Fig. 1B; Supplementary S1B, blue vs. red lines). These data suggested that Tet2 depletion in CD8+ T cells exhibits similar tumor-suppressive effects as anti-PD-L1 treatment does to melanoma. However, the combination of Tet2 loss in CD8+ T cells with anti-PD-L1 treatment does not seem to further benefit the antitumor efficacy.
We then moved on to analyze the distribution of donor CD45.2+CD8+ T cells at 8 days after adoptive transfer in the melanoma-bearing CD45.1+ recipient mice (Fig. 1C; Supplementary Figs. S1C and S1D). Within the tumor foci, we observed a significant increase of the fraction of CD45.2+CD8+OT-I TILs in the Tet2KO group when compared with that in the WT group. Interestingly, the expansion of Tet2KO TILs was not evident in the similar recipient mice treated with anti-PD-L1 (Fig. 1C), suggesting that anti-PD-L1 treatment might exert inhibitory effects on Tet2KO TILs expansion. To further track the adoptively transferred CD8+ T cells, we measured the CD45.2+CD8+OT-I T cells in the tumor, spleen, and peripheral blood (PB; Fig. 1D). We observed a marked increase of CD45.2+CD8+OT-I T cells (∼1.5- to 2.5-fold) in all these tissues for the Tet2KO group, which is consistent with the previous observation that Tet2-deficient lymphocytes exhibit an expansion advantage (6, 19). In addition, within all the injected CD45.2+CD8+OT-I T cells, Tet2KO CD8+ T cells tend to be enriched more at tumor foci when compared with the WT group (Fig. 1E). Interestingly, in addition to CD45.2+CD8+OT-I T cells, we observed significantly higher endogenous CD45.1+CD8+OT-I T cells within the tumor foci, but not in lymphocytes or PB of the recipient mice transferred with Tet2KO CD8+ T cell when compared with the mice transferred with WT CD8+ T cells (Supplementary Figs. S1D and S1E). These findings suggest that the tumor suppressive effect of Tet2KO TILs might involve both autonomous and nonautonomous cellular effects.
Tet2KO TILs exhibit enhanced immune response and cytotoxicity at the early stage
To further characterize TILs in the B16-OVA mouse model of melanoma, we isolated the tumors at 3 and 8 days after CD8+ T-cell injection, purified CD8+ TILs, and performed flow cytometry analysis on the signature marks within both the CD45.2+ (transferred) and CD45.1+ (endogenous) populations. After adoptive transfer in the Tet2KO group, we observed a significant increase in the production of IFNγ and TNFα, two signature cytokines reflecting the effector function of CD8+ T cells (Fig. 2A), at day 3 but not at day 8, compared with the WT group. A similar trend was noted for the PD-1, Tim-3 positive population (Fig. 2B). These data suggested that deletion of Tet2 enhances the antitumor efficiency of TILs at an early stage. In parallel, we measured IFNγ, TNFα PD-1, and Tim-3 levels in the endogenous CD45.1+CD8+ TILs (Supplementary Figs. S2A and S2B). No significant difference of IFNγ or TNFα production in endogenous CD45.1+CD8+ TILs was detected between the WT and Tet2KO groups (Supplementary Fig. S2A). We observed elevated PD-1 or Tim-3 positive population in mice transferred with Tet2KO CD45.2+CD8+ TILs at 3 days after adoptive transfer (Supplementary Fig. S2B). These results indicate that the production of pro-inflammatory cytokines, for example, IFNγ and TNFα, by Tet2KO CD8+ TILs might be one of the factors that promote the activation of endogenous CD45.1+CD8+ TILs in the recipient mice.
To further confirm the cytotoxicity effects of Tet2KO CD8+ T cells, we performed the in vitro coculture experiment. As shown in Fig. 2C, we purified CD8+ T cells from WT-OT-I or Tet2KO-OT-I and cultured them for 2 days. In parallel, we labeled in vitro cultured B16-OVA cells with cell proliferation dyes, which can be used to separate them from cocultured CD8+ T cells in the following experiment (Supplementary Fig. S2C). Then we cocultured in vitro activated CD8+ T cells and in vitro labeled B16-OVA cells for 4 to 8 hours. The cytotoxicity effects were then measured by the caspase-3/7 activity in labeled B16-OVA cells (Supplementary Fig. S2C). With this assay, we observed that B16-OVA cells cocultured with Tet2KO CD8+ T cells displayed significant higher caspase-3/7 activity than WT CD8+ T cells at both 4 and 8 hours after coculture (Fig. 2D). In parallel, we observed increased cell proliferation (Supplementary Figs. S2D and S2E) and decreased cell death (Fig. 2E) in Tet2KO CD8+ T cells when cocultured with B16-OVA cells compared with the WT group. These data confirmed that Tet2KO CD8+ T cells displayed higher cytotoxic effects than WT CD8+ T cells toward B16-OVA melanoma cells in vitro and this cytotoxic effect is an early event.
scRNA-seq reveals enhanced activation of Tet2KO TILs
To further identify the cell populations that might contribute to antitumor immunity, we performed scRNA-seq analysis in CD45.2+CD8+ TILs in four experimental groups (WT, Tet2KO, WT + anti-PD-L1, and Tet2KO + anti-PD-L1; Fig. 3A; Supplementary Fig. S3A; Supplementary Table S1) 8 days after adoptive transfer. We observed three major groups of cells: TILs (CD8+), monocytes (F4/80+), and B16-OVA tumor cells (CD63+; Fig. 3A; Supplementary Fig. S3B). We performed cross-comparison analysis between WT and Tet2KO without anti-PD-L1 treatment, WT with and without anti-PD-L1 treatment, and Tet2KO with and without anti-PD-L1 treatment (Fig. 3A). We found that the Tet2KO group exhibited significant reduction in the B16-OVA tumor cell population compared with the WT group (Fig. 3A, left). Similarly, in the WT group, anti-PD-L1 treatment also led to a notable decrease of B16-OVA tumor cells compared with the untreated group (Fig. 3A, middle). For the Tet2KO group, anti-PD-L1 treatment did not seem to cause major changes in cell populations (Fig. 3A, right). These results echoed the findings shown in Fig. 1, implying that Tet2KO TILs indeed suppress B16-OVA cancer cell growth. To further study how Tet2 deletion influences CD8+ TILs, we selected CD8a-expressing cells for further cluster analysis (Fig. 3B). We identified five major clusters in selected CD8+ TILs (Fig. 3B; Supplementary Table S2). The distribution of cell numbers within four experimental groups was different in five clusters (Fig. 3C). Cluster 0 showed similar cellular distribution among the four groups, whereas anti-PD-L1 treatment resulted in significant decrease of the cell number in both WT and Tet2KO groups in clusters 1 and 2. Tet2KO resulted in increased cell numbers in clusters 3 and 4; whereas anti-PD-L1 treatment augmented the cell number only in the WT group for clusters 3 and 4, but not in the Tet2KO group (Fig. 3C). To further elucidate the cell identities within these five clusters, we mined publicly available RNA-seq data from the purified subsets of CD8+ T cells, including naïve, effector, memory CD8+ T cell, PD1+Tim3+ double positive (DP) CD8+ TILs with and without anti-PD1 treatment (31). Then we selected the top 20 expressed genes within each cluster and examined the expression level of these genes in purified cell types using the published RNA-seq data (31). On the basis of gene expression patterns, the cells within clusters 0, 1, and 2 had high levels of Pdcd1, Ctla4, and Havcr2 with exhaustion-like features. In contrast, cells in clusters 3 and 4 exhibited effector-like features with high expression of Tnf and NF-κB signaling-related genes (Fig. 3D and E; Supplementary Table S2). This analysis demonstrated that, although a subset of Tet2-deficient TILs showed similar exhaustion-like patterns as WT, some Tet2KO TILs displayed enhanced effector-like features and might contribute to the enhanced antitumor immunity. Interestingly, we observed that anti-PD-L1 treatment exerted similar effects on WT and Tet2KO TILs bearing exhaustion-like features (cluster 0–2), but had less effects on Tet2KO TILs that showed effector-like features (cluster 3–4), which is consistent with the overall phenotype shown in Fig. 1C.
Tet2 deletion enhances the transcription of tumor-suppressive genes in TILs
From scRNA-seq analysis, we noted that the CD45.2+CD8+ selection method still retains B16-OVA tumor cells and monocytes with highly expressed CD63 (Fig. 3A; Supplementary Fig. S3B). In addition, we observed significant changes of TNFα and IFNγ production at 3 days after T-cell transfer between WT and Tet2KO TILs (Fig. 2B). Therefore, we further purified CD45.2+ CD63−CD8+ T cells from tumor tissues at day 0 and 3 after adoptive transfer for further RNA-seq (transcriptome) and ATAC-seq (chromatin accessibility) analysis (Fig. 4A; Supplementary Fig. S4A; Supplementary Tables S1, S3, and S4). We observed massive transcriptomic and epigenetic remodeling events from day 0 to day 3 after adoptive transfer in both WT and Tet2KO TILs (Supplementary Fig. S4B). Consistent with previous reports (11, 21, 32), we observed a strong positive association between transcriptional changes and chromatin accessibility in the analyzed TILs (Supplementary Fig. S4B), suggesting that epigenetic remodeling is one of the critical regulators contributing to transcriptional outputs during antitumor immunity.
To further examine the function of Tet2 during this process, we first compared DEGs between day 0 and 3 TILs and then compared the DEGs between WT and Tet2KO groups, anticipating to identify WT- or Tet2KO-specific DEGs during antitumor immunity at different days (Fig. 4B; Supplementary Fig. S4C; Supplementary Table S3). The GSEA analysis (33) showed that the genes specifically upregulated in Tet2KO TILs were enriched with cytokine production and immune signaling pathways (e.g., IFNγ responding genes; Fig. 4C–E; Supplementary Fig. S4D; Supplementary Table S3). Genes specifically downregulated in Tet2KO TILs were associated with cell cycle and proliferation (e.g., centromere genes and Cdks; Fig. 4C; Supplementary Fig. S4D; Supplementary Table S3). In WT TILs, genes that were specifically up- or downregulated are associated with certain house-keeping functions, such as metabolism (e.g., NADH-Ubiquinone Oxidoreductase), and RNA processing (e.g., WD repeat domains; Fig. 4C; Supplementary Fig. S4D; Supplementary Table S3). Some of the genes identified from RNA-seq analysis were further validated using real-time qPCR analysis (Fig. 4D). For example, we observed over a five-fold increase of the expression of Prf1 in Tet2KO CD8+ T cells compare with the WT group (Fig. 4D and E), which is consistent with the observed enhancement of cytotoxicity (Fig. 2D). Interestingly, we also detected several key transcription factors, such as Fos and Klf4, that were differentially expressed in Tet2KO CD8+ T cells (Fig. 4F), suggesting that Tet2 might regulate key transcription factors in CD8+ T cells and indirectly induce the transcriptional alterations during antitumor immunity. Collectively, our findings suggest that Tet2 ablation in TILs might lead to increased expression of cytokines and other immunomodulatory genes to enhance antitumor immunity.
Tet2 knockout reshapes the chromatin accessibility in TILs
Next, we performed similar comparative analysis on chromatin accessibility by using ATAC-seq (Fig. 5A; Supplementary Fig. S5A and S5B; Supplementary Table S4). We first identified differentially enriched regions between day 0 and 3 in WT or Tet2KO TILs. We observed a marked increase in chromatin accessibility from day 0 to day 3 in WT TILs, but to a lesser extent in Tet2KO TILs (Fig. 5A; Supplementary Fig. S5A). To further investigate how Tet2 regulates the function of TILs during this process, we identified differential chromatin accessible regions that were specifically altered in WT- or Tet2KO groups (Fig. 5A; Supplementary Fig. S5B). Genomic Regions Enrichment of Annotations Tool (GREAT) analysis (25) showed that these regions are close to the genes essential for T-cell activation and immune response (Supplementary Fig. S5C). Because chromatin accessible regions are known to be critical for TF binding (34), we analyzed the enrichment of TF motifs within the identified WT- or Tet2-specific differential accessible regions (Fig. 5B). We found that bZIP and ETS motifs were most prominently enriched within WT- and Tet2-specific differential accessible regions, respectively (Fig. 5B and C). WT-specific open genomic regions were mainly enriched with the motifs for the bZIP TF family, including BATF and AP-1; whereas Tet2-specific accessible genomic regions were enriched with motifs for the ETS TF family, including ETS1 and ELFs (Fig. 5B). To further validate this observation, we compared our own ATAC-seq data with the published BATF (bZIP family) ChIP-seq data collected in lymphocytes (35). We focused on the chromatin accessibility states of the BATF binding regions in both WT and Tet2KO TILs (Fig. 5C). At day 0, the chromatin accessibility of the BATF binding regions in Tet2KO CD8+ T cells was significantly higher than of the WT group. At day 3, the chromatin accessibility of the same regions was reduced in the Tet2KO group, but not in the WT group. These findings are consistent with the motif enrichment analysis results, both highlighting that Tet2 deletion reshapes the chromatin accessibility of selected TFs in TILs. Furthermore, genes that showed positive correlations between their expression and chromatin accessibility in CD8+ TILs were found to exhibit differential enrichment of TF motifs (Fig. 5D; Supplementary Fig. S5D). For example, BATF was enriched at the Slc14a1 promoters, and we observed relatively higher gene expression and chromatin accessibility of Slac14a1 in WT TILs when compared with Tet2KO TILs (Fig. 5E, top). On the other hand, Irf2 gene exhibited both higher expression and chromatin accessibility in Tet2KO TILs 3 days after adoptive transfer than in WT TILs, and we observed ETS1 enrichment, but not BATF at the promoter of Irf2 (Fig. 5E, bottom). To further test whether Tet2KO TILs exhibited enhanced transcriptional activity of the ETS family members, we applied two ETS Inhibitors, YK-4–279 and TK-216 (36), to both WT and Tet2KO CD8+ OT-I T cells that were cocultured with B16-OVA tumor cells as shown in Fig. 2C, and assessed the degrees of tumor cell apoptosis by using caspase-3/7 staining. We noted that both YK-4–279 and TK-216 treatment led to a significant reduction in B16-OVA tumor cell apoptosis induced by Tet2KO CD8+ OT-I T cells, but had minor effects on B16-OVA tumor cells cocultured with WT CD8+ OT-I T cells (Fig. 5F). As negative control, YK-4–279 and TK-216 treatment per se did not induce major cytotoxicity in B16-OVA cells (Supplementary Fig. S5E). The YK-4–279 and TK-216 mediated suppressive effect on tumor killing might be due to the inhibition of the cytokine production in Tet2KO CD8+ T cells (Supplementary Fig. S5F). Taken together, these data suggest that Tet2 deletion remodels the chromatin accessibility and transcriptional activity of selected TFs to impact gene expression in TILs to affect antitumor immunity.
TET2 is one of the frequently mutated genes in the hematopoietic and immune systems and is a well-known tumor suppressor (15, 17). In addition, TET2 mutations are often observed in elderly individuals with clonal hematopoiesis (CH; ref. 16), suggesting that TET2 mutations are also present in the progeny of hematopoietic stem and progenitor cells (HSCP), such as monocytes and lymphocytes, and subsequently alters their functions during innate and adaptive immune response. Indeed, recent studies have shown that CH-associated Tet2 loss-of-function in HSPCs accelerates atherosclerosis due to the increased inflammasome activity in Tet2-deficient macrophages (37). On the other hand, the deletion of TET2 in CAR T cells and tumor-associated macrophages (TAM) suppresses the tumor growth and significantly benefits the antitumor immunotherapy (13, 14). In this study, we investigated the effects of Tet2 LOF in CD8+ TILs during tumor onset and progression. Our study provides additional evidence to support the beneficial role of Tet2 deficiency in the immune system to combat cancer. Mechanistically, we have unveiled that deletion of Tet2 reshapes the chromatin accessibility at genomic binding regions of several key transcription factors, for example, BATF and ETS1, in CD8+ TILs, thereby enhancing its antitumor immunity and suppressing melanoma growth in vivo.
Previous studies using an acute viral infection model have demonstrated that Tet2 loss promotes the early onset of memory CD8+ T cells and enhances their antiviral activity (6). In our study, we have shown that Tet2-deficient CD8+ TILs display more effective antitumor activity than the WT CD8+ TILs. The deletion of Tet2 augments the expansion of CD8+ T cells not only in tumors, but also in spleens and peripheral blood, which is consistent with previous observations made in Tet2 deficient CD4+ T cells and HSPCs (18, 19). In addition, we have found that more CD8+ T cells tend to be enriched within the tumor foci in the Tet2KO group than in the WT group. This effect was also observed in CD45.1+CD8+ T cells in the recipient mice transferred with Tet2KO CD8+ T cells, suggesting that Tet2 inactivation could promote the enrichment of CD8+ T cells within tumor foci to enhance the antitumor efficiency. In our study, we also observed the up-regulation of Ifn-γ and Tnf-α in Tet2KO TILs, which might mediate the enrichment of CD45.1+ TILs in recipient mice transferred with Tet2KO TILs. Further studies are needed to clarify the underlying mechanism. Nonetheless, our exciting findings establish that Tet2 inactivation not only renders CD8+ TILs more potent in tumor killing, but also leads to a more immunogenic tumor microenvironment to enhance the overall antitumor immunity.
Chromatin reprogramming in the tumor microenvironment is one of the critical barriers blocking the antitumor function of CD8+ TILs (2). The chromatin reprogramming process is reversible during the early stage of tumor development and gradually becomes irreversible in the late stage (2). In our study, we observed that Tet2 deficiency leads to massive chromatin reprogramming in TILs and enhances their antitumor efficiency at an early stage during tumor development. Compared with WT TILs, Tet2-deficient TILs bear higher chromatin accessibility at the ETS (ETS1) family binding regions but show reduced accessibility in the bZIP family binding regions, such as Batf. Pharmacologic inhibition of the ETS TF family members reduced the antitumor efficiency of Tet2KO CD8+ T cells, but not WT CD8+ T cells, further confirming that Tet2 deficiency reprograms the transcriptional activity of selected TFs. These changes in chromatin accessibility and transcriptional activity might lead to upregulation of genes that are essential for the effector function of TILs, including genes involved in TNFA and IFNG pathways. In addition to changes in chromatin accessibility arising from Tet2 depletion in CD8+ TILs, Tet2 loss might abrogate its interaction with negative regulators of gene transcription, such as Sin3a (38) and Hdac2 (20), thereby contributing to augmented transcriptional activation in Tet2 deficient cells. Furthermore, the DNA binding of many TFs might be directly influenced by DNA methylation and its oxidative derivatives (5hmC, 5fC, and 5caC) generated by the Tet protein family (39–41). Therefore, Tet2 deficiency induced alterations in DNA methylation and its oxidation could add another layer of control over TFs binding to theirs genomic targets in CD8+ TILs. Further studies are needed to explore these possibilities in CD8+ TILs.
In summary, our study highlights the inhibitory role of Tet2 in regulating CD8+ TILs during antitumor immunity. Tet2 inactivation significantly enhances the effector-like function of TILs to promote antitumor immunity by reshaping the chromatin accessibility in TILs at the early stage during tumor development.
No disclosures were reported.
M. Lee: Resources, data curation, formal analysis, investigation, methodology, writing–review and editing. J. Li: Data curation, software, formal analysis, validation, visualization, methodology, writing–review and editing. J. Li: Conceptualization, resources, data curation, software, formal analysis, supervision, writing–review and editing. S. Fang: Formal analysis, investigation, methodology. J. Zhang: Validation, investigation, methodology, writing–review and editing. A.T.T. Vo: Investigation, methodology. W. Han: Validation, investigation, methodology. H. Zeng: Validation, investigation, methodology. S. Isgandarova: Validation, investigation, methodology. M. Martinez-Moczygemba: Resources, supervision, validation, investigation, methodology. Y. Zhou: Conceptualization, resources, data curation, supervision, funding acquisition, writing–original draft, project administration, writing–review and editing. D. Sun: Data curation, software, formal analysis, supervision. Y. Huang: Conceptualization, resources, data curation, formal analysis, supervision, funding acquisition, writing–original draft, project administration, writing–review and editing.
We are grateful for Dr. Jianjun Shen and the MD Anderson Cancer Center next-generation sequencing core at Smithville (CPRIT RP120348 and RP170002), and the Epigenetic core at the Institute of Biosciences and Technology at the Texas A&M University. This work was supported by grants from Cancer Prevention and Research Institute of Texas (RP180131 to D. Sun; RP190581 to M. Moczygemba), NIH grants (R01HL134780, R21GM138824, and R01HL146852 to Y. Huang; R01GM112003, R21GM132778, and R01CA232017 to Y. Zhou), the Welch Foundation (BE-1913-20190330 to Y. Zhou), the American Cancer Society (RSG-18-043-01-LIB to Y. Huang; RSG-16-215-01-TBE to Y. Zhou), and by an allocation from the Texas A&M University start-up funds (Y. Huang).
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.