Distinct T cell functions enable efficient immunoediting and prevent tumor emergence of developing sarcomas

Abstract

T cells edit tumors by eliminating neoantigen-expressing tumor cells. Yet, how and when this is achieved remains uncertain. Using a murine sarcoma model with fluorescent neoantigens, we found that tumors developed later and in fewer T cell-sufficient mice (∼53% penetrance) than T cell-deficient mice (∼100%). With T cells, all emergent tumor cells had silenced neoantigens, but neoantigen-negative tumor cells were also present in every T cell-deficient mouse. This suggested silencing was necessary but not sufficient for outgrowth. Genetic removal of neoantigens restored tumor penetrance if implemented on day 5 post-tumor initiation, but not day 10, because CD8⁺ and CD4⁺ T cells infiltrated the tissue and eliminated most neoantigen-positive and -negative tumor cells within 8 days. Single-cell analyses on day-7 tumors showed oncogenic changes including increased proliferation and T cell-dependent upregulation of the IFNγ-response gene Cd274 (PD-L1). T cell-depletion rescued both neoantigen-positive and -negative cells, while IFNγ blockade rescued only negative cells. This shows that T cells efficiently edit sarcomas of neoantigens and prevent early tumors via IFNγ-independent and IFNγ-dependent (bystander) mechanisms.

Keywords: GEMMs; Immunoediting; T cell elimination; escape; immunogenicity; immunosurveillance; neoantigen silencing; pre-emergent tumor; sarcoma; tumor clonality.

Figure 1.. Efficient T cell-mediated immunoediting is a feature of sarcomas in the KP^FRT model.

(A) Lentiviral construct design containing FLPo recombinase and GP33/66 antigen embedded within a loop of mClover protein. (B) Schematic representation of initiating a transformed/tumor cell post infection (PI) in KP^FRT mice. (C) Experimental schedule using lentivirus made from (A). (D) Tumor emergence in KP^FRT mice treated with anti-CD4/CD8 depletion antibodies (T cell depleted, top) vs non-depleted (Control) and Rag1^−/− (Rag KO, bottom) vs Rag1^+/+ (Control) PI. Data are pooled from 2–3 independent experiments from each condition. Remaining Rag Het/WT mice without tumors at 150 days were treated with anti-CD4/CD8 depletion (as indicated by arrows) antibodies once a week for 3 weeks and monitored for an additional 50 days. (E) Sarcoma scoring system of 0 – 4 assessing the tumor emergence and growth in KP^FRT mice with or without T cell depletion. (F) FACS plots showing cell lines with the highest mCl-GP33/66 expression per condition. Note: plots chosen to illustrate possible outcomes and do not reflect mean % expression of groups. Quantification of mCl-GP33/66 expression in cell lines from indicated conditions. P values determined by Kruskal-Wallis test. (G) Updated lentiviral construct design containing FLPo recombinase and GP33/66 antigen embedded within a loop of mClover protein that includes one promoter and two gene products separated by 2A self-cleaving peptide. (H) Schematic representation of initiating a transformed/tumor cell post infection (PI) in KP^FRT Ai65/CreER mice. (I) Tumor emergence in KP^FRT mice using the updated lentivirus. (J) FACS plots and quantification showing representative mCl-GP33/66 expression of tumor cells isolated from Control (Rag Het/WT) mice compared to that from Rag KO mice. Samples are pregated on live singlets that are CD45⁻ CD31⁻ TER119⁻ tdTomato⁺ (see Figure S2 for gating strategy) from dissociated tibia muscle. Data shown as mean ± SD of the %mCl-GP33/66⁺ population (of its parent Lin⁻ Tom⁺ cells). P value determined by Mann-Whitney test. (K) Mean fluorescence intensity of mCl-GP33/66 in tumors from Control or Rag KO mice from (J). Bar plots show data as mean ± SD of MFI values from each biological replicate. P value determined by Mann-Whitney test. See also Figures S1, and Tables S1–S2.

Figure 2.. T cell immunoediting drives clonal loss amongst emergent tumors.

(A) Lenti-x integration analysis workflow to assess 6 T cell depleted lines and 6 control lines. 20–26 bacterial colonies per cell line from TOPO TA cloning were isolated and sequenced. (B) Integration sites were validated by designing primers for each integrant found and their respective chromosomal site and performing PCR on original gDNA. Primers are listed in Table S2. (C) Pie graphs representing the number of putative integration sites per cell line. Center numbers represent the day of tumor emergence prior to cell line derivation. (D) Quantification and statistical analysis of (C) as ratio of # unique integrations found from total colonies assessed per line. Data shown as mean ± SD. P value determined using Mann-Whitney test. (E) Sequence analysis of Kras-G12D amplification or loss of WT Kras in sarcoma cell lines determined as the ratio of Mutant to WT allele frequency, averaged from 3–4 sequencing reads per cell line. See also Figure S2 and Table S1.

Figure 3.. T cell-mediated immunoediting occurs en force within 10 days of cancer initiation.

(A) Floxed lentiviral construct containing FLPo recombinase and mClover-GP33/66. (B) Schematic representation of initiating a transformed/tumor cell with the floxed lentiviral construct post infection (PI) in KP^FRT Ai65/CreER mice. Upon integration, the loxP site in the 3’LTR is duplicated, flanking the FLPo-mCl-GP33/66 insert. FLPo-mCl-GP33/66 is removed following tamoxifen (Tam)-induced Cre-mediated recombination. (C) Experimental schedule using lentivirus made from (A). (D) Tumor emergence after Tam treatment (neoantigen removal) at days 5, 10, or 20 PI. Graph represents combined data from 2–4 independent experiments per condition. P values determined by log rank test. (E) FACS plots and quantification showing % mCl-GP33/66 expression in cell lines from T cell depleted mice (n=17), Control mice (without depletion; n=6) and mice receiving Tam beginning at D5 PI (n=10). Data shown as mean ± SD. (F) Quantification of (E). P values determined by Kruskal-Wallis test. (F) PCR for Flpo from genomic DNA isolated from tumor cells initiated with the Floxed lentivirus and receiving Tam beginning at D5 PI showing successful Cre-mediated recombination. See also Figure S2 and Tables S3 and S4.

Figure 4.. The antigen-specific T cell immunoediting response peaks around 8 days post cancer initiation.

(A-C) FACS plots of a time course between day 5 to day 9 post cancer initiation using the CMV-Flpo/PGK-mCl-GP33/66 lentivirus where tibia muscle were dissociated into single cells and analyzed using spectral flow cytometry for (A) tdTomato⁺ tumor cells (pregated on live singlet CD45⁻ CD31⁻ TER119⁻; n=6–10 mice per time point), (B) CD8⁺ T cells (pregated on live singlet CD45⁺ CD4⁻; n=−6 mice per time point), and (C) CD4⁺ T cells (pregated on live singlet CD45⁺ CD8⁻; n=3–6 mice per time point; data are representative of 3 independent experimental replicates) using KP^FRT Ai65/CreER mice pretreated with tamoxifen. P values obtained using Kruskal-Wallis test. (D-F) FACS plots of a time course between day 5 to day 9 post cancer initiation using the CMV-Flpo/PGK-eGFP-GP33/66 lentivirus where tibia muscle were dissociated into single cells and analyzed using spectral flow cytometry for (D) tdTomato⁺ cancer cells (pregated on live singlet CD45⁻; n=7–10 mice per time point; combined from the independent CD8⁺ T cell and CD4⁺ T cell experiments), (E) CD44^hi H2-D^b-GP33 tetramer⁺ cells (pregated on live singlet CD45⁺ Thy1.2⁺ CD4⁻ CD8⁺; n=4–5 mice per time point), and (F) CD44^hi I-A^b-GP66 tetramer⁺ cells (pregated on live singlet CD45⁺ Thy1.2⁺ CD4⁺ CD8⁻; n=3–5 mice per time point; data are representative of 2 independent experimental replicates) using KP^FRT Ai65/CreER mice pretreated with tamoxifen. (G) Quantification of total Tomato⁺ tumor cells and Neoantigen⁺ (NeoAg⁺) Tomato⁺ cells at day 20 dissociated from the tibia muscle of Control mice (n=8) or mice with CD4⁺ T cell depletion (α-CD4; n=5), CD8⁺ T cell depletion (α-CD8; n=10), or from RAG KO mice (n=8). Data are representative of 4 independent experimental replicates. P values determined by Kruskal-Wallis test. Horizontal lines represent mean, each dot represents a biological replicate. See also Figure S3.

Figure 5:. Nascent immunogenic sarcoma cells acutely undergoing T cell immunoediting have enrichment of interferon γ (IFNγ) response/PD-L1 (Cd274) and elevated proliferation.

(A) Uniform manifold approximation and projection (UMAP) plots of unsupervised clustering of scRNA-seq data from dissociated and sorted Tomato⁺ cells (pregated on live singlet CD45⁻ CD31⁻ TER119⁻) in plots stratified by Cluster IDs, 0–7. (B-D) Unbiased pathway enrichment using the MSigDB2020 signature on the Enrichr tool to define transcriptional signatures overrepresented (B) globally (pseudo-bulk) that are driven by T cells, (C) specifically in muscle satellite cells (Cav1⁺) from KP mice that are driven by T cells, and (D) specifically in muscle satellite cells from KP mice relative to P only mice. All hits displayed show an adjusted P-value of >0.05. (E) Gene expression heatmaps (color displaying normalized expression z-score) of markers to define major cluster labels seen in (A) including Ptprc (CD45; immune cells), Tnni2 (troponin I2; mature myofibers), and Cav1 (caveolin 1; satellite cells). (F) Gene expression heatmaps of proliferation markers, Mki67, Pclaf, and Cenpa. (G) Gene expression heatmaps of IFNγ-responsive genes that modulate T cell interactions, Cd274 (PD-L1; ligand for the T cell checkpoint receptor, PD-1), H2-D1 (MHC Class I complex gene), and B2m (MHC Class I complex gene). (H) Violin plots of Cd274/PD-L1, Mki67, and H2-D1 stratified by Cluster ID. (I) Proportionate representation of cells from each sample group within cluster 2 (KI67⁺), cluster 3 (Cd274/PD-L1⁺), and cluster 4 (Tnni2⁺). (J) Pseudobulk expression analysis of MHC Class I complex genes, H2-D1 and B2m, of cells from KP mice with and without T cells. P value determined by Wilcoxon test. (K) quantitation of total Tom⁺ cells, PD-L1⁺ cells, and Ki67⁺ cells (pregated on live singlet CD31⁻CD45⁻TER119⁻tdTomato⁺ cells) at day 7 post initiation ± the GFP-GP33/66 neoantigens (Lenti-Flpo-mCl-GP33/66 vs. Lenti-Flpo) and ± T cells (Rag⁺ vs. Rag KO). P value determined by Mann-Whitney test. Horizontal line represents mean values and individual dots represent biological replicates. See also Figures S8–S10.

Figure 6.. T cells efficiently eliminate neoantigen-expressing tumor cells in pre-emergent stage through IFNγ-dependent and independent mechanisms.

(A) Representative FACS plots of Tomato and mCl-GP33/66 (neoantigen⁺ or negative) expressing tumor cells in the muscle 5- and 20-days PI. Plots are pregated on live singlet Tomato⁺ Lin⁻ (live singlet CD45⁻ CD31⁻ TER119⁻ CD11b⁻ cells) population. (B) Median total of tumor cells per condition displayed as neoantigen⁺ (green, on top) and negative fractions (red, on bottom). (C) Quantification of frequency of neoantigen⁺ Tomato⁺ tumor cells in the muscle. (D) Quantification of total number of tumor cells (Tom⁺). (E) Quantification of total number of neoantigen⁺ tumor cells (also Tom⁺). Data pooled from 2–4 independent experiments per condition (Day 5 Control, N=16; day 5 T cell depleted, N=13; day 20 Control, N=32; day 20 T cell depleted, N=23; day 20 IFNγ blockade, N=16). P value determined by one-way ANOVA. Horizontal lines represent median, each dot represents biological replicates. See also Figure S11.