Abstract
Despite breakthroughs in cancer immunotherapy, most tumor-reactive T cells cannot persist in solid tumors due to an immunosuppressive environment. We developed Tres (tumor-resilient T cell, https://resilience.ccr.cancer.gov/), a computational model utilizing single-cell transcriptomic data to identify signatures of T cells that are resilient to immunosuppressive signals, such as transforming growth factor-β1, tumor necrosis factor-related apoptosis-inducing ligand and prostaglandin E2. Tres reliably predicts clinical responses to immunotherapy in melanoma, lung cancer, triple-negative breast cancer and B cell malignancies using bulk T cell transcriptomic data from pre-treatment tumors from patients who received immune-checkpoint inhibitors (n = 38), infusion products for chimeric antigen receptor T cell therapies (n = 34) and pre-manufacture samples for chimeric antigen receptor T cell or tumor-infiltrating lymphocyte therapies (n = 84). Further, Tres identified FIBP, whose functions are largely unknown, as the top negative marker of tumor-resilient T cells across many solid tumor types. FIBP knockouts in murine and human donor CD8+ T cells significantly enhanced T cell-mediated cancer killing in in vitro co-cultures. Further, Fibp knockout in murine T cells potentiated the in vivo efficacy of adoptive cell transfer in the B16 tumor model. Fibp knockout T cells exhibit reduced cholesterol metabolism, which inhibits effector T cell function. These results demonstrate the utility of Tres in identifying biomarkers of T cell effectiveness and potential therapeutic targets for immunotherapies in solid tumors.
This is a preview of subscription content, access via your institution
Access options
Access Nature and 54 other Nature Portfolio journals
Get Nature+, our best-value online-access subscription
$29.99 / 30 days
cancel any time
Subscribe to this journal
Receive 12 print issues and online access
$209.00 per year
only $17.42 per issue
Buy this article
- Purchase on Springer Link
- Instant access to full article PDF
Prices may be subject to local taxes which are calculated during checkout
Similar content being viewed by others
Data availability
The RNA-seq data generated by this study have been deposited to the NCBI Gene Expression Omnibus with accession ID GSE186428. All flow analysis raw data are available in FCS format at the Flowrepository76 with accession IDs as listed in Supplementary Table 10 and source data tables for each figure. All gene scores computed by the Tres model are available at https://resilience.ccr.cancer.gov.
For public datasets, single-cell RNA-seq data collected from public studies for training Tres signatures are listed in Supplementary Table 1. T cell transcriptomic data for validating the Tres model are available for download as processed data at https://github.com/data2intelligence/Tres_main. Bulk-tumor transcriptomic datasets of ICI pretreatment tumors are listed in Supplementary Table 4. Public ChIP-seq profiles of SREBF2 were originally available from GSE31477 and binding peak calls reprocessed by the Cistrome pipeline are available at http://cistrome.org. The CCLE cancer cell line expression profiles are available from https://depmap.org. Genome-wide CRISPR screen data on human donor T cells are downloaded from supplementary materials of the original studies12,13 and GSE174255. The MILE leukemia dataset is available from GSE13159.
For public databases, TIDE is available at http://tide.dfci.harvard.edu. Cistrome is available at http://cistrome.org. MSigDB, which provides KEGG gene sets (v.7.4) of cell cycle and DNA replication, is available at https://www.gsea-msigdb.org.
Source data for all main and extended data figures are provided. Raw western blots are shown in Supplementary Figs. 1, 4 and 5. Raw Incucyte images are shown in Supplementary Fig. 2. Source data are provided with this paper.
Code availability
The interactive Tres server is available at https://resilience.ccr.cancer.gov. The source code and test data for the Tres model and T cell efficacy prediction are available at https://github.com/data2intelligence/Tres. The programs for reproducing major results, including computing Tres signatures and predicting immunotherapy response, are available at https://github.com/data2intelligence/Tres_main. Source data are provided with this paper.
Change history
11 August 2022
A Correction to this paper has been published: https://doi.org/10.1038/s41591-022-01997-8
References
Ribas, A. & Wolchok, J. D. Cancer immunotherapy using checkpoint blockade. Science https://doi.org/10.1126/science.aar4060 (2018).
Rosenberg, S. A. et al. Durable complete responses in heavily pretreated patients with metastatic melanoma using T-cell transfer immunotherapy. Clin. Cancer Res. https://doi.org/10.1158/1078-0432.CCR-11-0116 (2011).
Robbins, P. F. et al. Tumor regression in patients with metastatic synovial cell sarcoma and melanoma using genetically engineered lymphocytes reactive with NY-ESO-1. J. Clin. Oncol. https://doi.org/10.1200/JCO.2010.32.2537 (2011).
Guedan, S., Ruella, M. & June, C. H. Emerging cellular therapies for cancer. Annu. Rev. Immunol. https://doi.org/10.1146/annurev-immunol-042718-041407 (2019).
Martinez, M. & Moon, E. K. CAR T cells for solid tumors: new strategies for finding, infiltrating, and surviving in the tumor microenvironment. Front. Immunol. https://doi.org/10.3389/fimmu.2019.00128 (2019).
Sharma, P., Hu-Lieskovan, S., Wargo, J. A. & Ribas, A. Primary, adaptive, and acquired resistance to cancer immunotherapy. Cell 168, 707–723 (2017).
Tschumi, B. O. et al. CART cells are prone to Fas- and DR5-mediated cell death. J. Immunother. Cancer 6, 71 (2018).
Green, D. R., Droin, N. & Pinkoski, M. Activation-induced cell death in T cells. Immunol. Rev. https://doi.org/10.1034/j.1600-065x.2003.00051.x (2003).
Sade-Feldman, M. et al. Defining T cell states associated with response to checkpoint immunotherapy in melanoma. Cell 175, 998–1013 (2018).
Caushi, J. X. et al. Transcriptional programs of neoantigen-specific TIL in anti-PD-1-treated lung cancers. Nature 596, 126–132 (2021).
Suvà, M. L. & Tirosh, I. Single-cell RNA sequencing in cancer: lessons learned and emerging challenges. Mol. Cell 75, 7–12 (2019).
Shifrut, E. et al. Genome-wide CRISPR screens in primary human T cells reveal key regulators of immune function. Cell. https://doi.org/10.1016/j.cell.2018.10.024 (2018).
Schmidt, R. et al. CRISPR activation and interference screens decode stimulation responses in primary human T cells. Science. https://doi.org/10.1126/science.abj4008 (2022).
Legut, M. et al. A genome-scale screen for synthetic drivers of T cell proliferation. Nature. https://doi.org/10.1038/s41586-022-04494-7 (2022).
Zhou, P. et al. In vivo discovery of immunotherapy targets in the tumour microenvironment. Nature 506, 52–57 (2014).
Dong, M. B. et al. Systematic immunotherapy target discovery using genome-scale in vivo CRISPR screens in CD8 T cells. Cell 178, 1189–1204 (2019).
Gicobi, J. K., Barham, W. & Dong, H. Immune resilience in response to cancer therapy. Cancer Immunol. Immunother. 69, 2165–2167 (2020).
Jerby-Arnon, L. et al. A cancer cell program promotes T cell exclusion and resistance to checkpoint blockade. Cell 175, 984–997 (2018).
Jiang, P. et al. Systematic investigation of cytokine signaling activity at the tissue and single-cell levels. Nat. Methods 18, 1181–1191 (2021).
Kanehisa, M., Sato, Y., Kawashima, M., Furumichi, M. & Tanabe, M. KEGG as a reference resource for gene and protein annotation. Nucleic Acids Res. 44, D457–D462 (2016).
Wehbi, V. L. & Taskén, K. Molecular mechanisms for cAMP-mediated immunoregulation in T cells – role of anchored protein kinase A signaling units. Front. Immunol. https://doi.org/10.3389/fimmu.2016.00222 (2016).
Naing, A. et al. PEGylated IL-10 (pegilodecakin) induces systemic immune activation, CD8 T cell invigoration and polyclonal T cell expansion in cancer patients. Cancer Cell 34, 775–791 (2018).
Santra, S. & Ghosh, S. K. Interleukin-4 is effective in restoring cytotoxic T cell activity that declines during in vivo progression of a murine B lymphoma. Cancer Immunol. Immunother. https://doi.org/10.1007/s002620050385 (1997).
Krishna, S. et al. Stem-like CD8 T cells mediate response of adoptive cell immunotherapy against human cancer. Science 370, 1328–1334 (2020).
Subramanian, A. et al. Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles. Proc. Natl Acad. Sci. USA 102, 15545–15550 (2005).
Lee, N. et al. HLA-E is a major ligand for the natural killer inhibitory receptor CD94/NKG2A. Proc. Natl Acad. Sci. USA 95, 5199–5204 (1998).
Lu, L. et al. Regulation of activated CD4+ T cells by NK cells via the Qa-1-NKG2A inhibitory pathway. Immunity 26, 593–604 (2007).
Soderquest, K. et al. Cutting edge: CD8+ T cell priming in the absence of NK cells leads to enhanced memory responses. J. Immunol. 186, 3304–3308 (2011).
Azizi, E. et al. Single-cell map of diverse immune phenotypes in the breast tumor microenvironment. Cell. https://doi.org/10.1016/j.cell.2018.05.060 (2018).
Wu, S. Z. et al. A single-cell and spatially resolved atlas of human breast cancers. Nat. Genet. https://doi.org/10.1038/s41588-021-00911-1 (2021).
Steen, C. B. et al. The landscape of tumor cell states and ecosystems in diffuse large B cell lymphoma. Cancer Cell. https://doi.org/10.1016/j.ccell.2021.08.011 (2021).
Mathewson, N. D. et al. Inhibitory CD161 receptor identified in glioma-infiltrating T. cells by single-cell analysis. Cell. https://doi.org/10.1016/j.cell.2021.01.022 (2021).
Puram, S. V. et al. Single-cell transcriptomic analysis of primary and metastatic tumor ecosystems in head and neck cancer. Cell. https://doi.org/10.1016/j.cell.2017.10.044 (2017).
Ma, L. et al. Tumor cell biodiversity drives microenvironmental reprogramming in liver cancer. Cancer Cell. https://doi.org/10.1016/j.ccell.2019.08.007 (2019).
Liu, Y. et al. Tumour heterogeneity and intercellular networks of nasopharyngeal carcinoma at single cell resolution. Nat. Commun. https://doi.org/10.1038/s41467-021-21043-4 (2021).
Lambrechts, D. et al. Phenotype molding of stromal cells in the lung tumor microenvironment. Nat. Med. 24, 1277–1289 (2018).
Young et al. Single-cell transcriptomes from human kidneys reveal the cellular identity of renal tumors. Science. https://doi.org/10.1126/science.aat1699 (2018).
Durante, M. A. et al. Single-cell analysis reveals new evolutionary complexity in uveal melanoma. Nat. Commun. https://doi.org/10.1038/s41467-019-14256-1 (2020).
Zheng, L. et al. Pan-cancer single-cell landscape of tumor-infiltrating T cells. Science. https://doi.org/10.1126/science.abe6474 (2021).
Zhang, Y. et al. Single-cell analyses reveal key immune cell subsets associated with response to PD-L1 blockade in triple-negative breast cancer. Cancer Cell https://doi.org/10.1016/j.ccell.2021.09.010 (2021).
Yost, K. E. et al. Clonal replacement of tumor-specific T cells following PD-1 blockade. Nat. Med. 25, 1251–1259 (2019).
Gattinoni, L. et al. A human memory T cell subset with stem cell-like properties. Nat. Med. https://doi.org/10.1038/nm.2446 (2011).
Doering, T. A. et al. Network analysis reveals centrally connected genes and pathways involved in CD8+ T cell exhaustion versus memory. Immunity. https://doi.org/10.1016/j.immuni.2012.08.021 (2012).
Good, C. R. et al. An NK-like CAR T. cell transition in CAR T cell dysfunction. Cell https://doi.org/10.1016/j.cell.2021.11.016 (2021).
Giordano, M. et al. Molecular profiling of CD8 T cells in autochthonous melanoma identifies Maf as driver of exhaustion. EMBO J. 34, 2042–2058 (2015).
Lauss, M. et al. Mutational and putative neoantigen load predict clinical benefit of adoptive T cell therapy in melanoma. Nat. Commun. 8, 1738 (2017).
Fraietta, J. A. et al. Determinants of response and resistance to CD19 chimeric antigen receptor (CAR) T cell therapy of chronic lymphocytic leukemia. Nat. Med. 24, 563–571 (2018).
Chen, G. M. et al. Integrative bulk and single-cell profiling of premanufacture T-cell populations reveals factors mediating long-term persistence of CAR T-cell therapy. Cancer Discov. 11, 2186–2199 (2021).
Lu, S. et al. Comparison of biomarker modalities for predicting response to PD-1/PD-L1 checkpoint blockade: a systematic review and meta-analysis. JAMA Oncol. 5, 1195–1204 (2019).
Litchfield, K. et al. Meta-analysis of tumor- and T cell-intrinsic mechanisms of sensitization to checkpoint inhibition. Cell. https://doi.org/10.1016/j.cell.2021.01.002 (2021).
Kolpakova, E. et al. Cloning of an intracellular protein that binds selectively to mitogenic acidic fibroblast growth factor. Biochem. J. 336, 213–222 (1998).
Huang, Y.-F. et al. FIBP knockdown attenuates growth and enhances chemotherapy in colorectal cancer via regulating GSK3β-related pathways. Oncogenesis 7, 77 (2018).
Hanada, K.-I., Yu, Z., Chappell, G. R., Park, A. S. & Restifo, N. P. An effective mouse model for adoptive cancer immunotherapy targeting neoantigens. JCI Insight. https://doi.org/10.1172/jci.insight.124405 (2019).
Patel, S. J. et al. Identification of essential genes for cancer immunotherapy. Nature 548, 537–542 (2017).
Chu, V. T. et al. Increasing the efficiency of homology-directed repair for CRISPR-Cas9-induced precise gene editing in mammalian cells. Nat. Biotechnol. 33, 543–548 (2015).
Luo, J., Yang, H. & Song, B. L. Mechanisms and regulation of cholesterol homeostasis. Nat. Rev. Mol. Cell Biol. https://doi.org/10.1038/s41580-019-0190-7 (2020).
Kohlmann, A. et al. An international standardization programme towards the application of gene expression profiling in routine leukaemia diagnostics: the Microarray Innovations in LEukemia study prephase. Br. J. Haematol. 142, 802–807 (2008).
Jiang, P., Freedman, M. L., Liu, J. S. & Liu, X. S. Inference of transcriptional regulation in cancers. Proc. Natl Acad. Sci. USA. https://doi.org/10.1073/pnas.1424272112 (2015).
Brown, M. S. & Goldstein, J. L. The SREBP pathway: regulation of cholesterol metabolism by proteolysis of a membrane-bound transcription factor. Cell. https://doi.org/10.1016/S0092-8674(00)80213-5 (1997).
Stromnes, I. M. et al. Abrogating Cbl-b in effector CD8(+) T cells improves the efficacy of adoptive therapy of leukemia in mice. J. Clin. Invest. 120, 3722–3734 (2010).
Chen, Y. et al. Cholesterol inhibits TCR signaling by directly restricting TCR-CD3 core tunnel motility. Molecular Cell https://doi.org/10.1016/j.molcel.2022.02.017 (2022).
Ma, X. et al. Cholesterol induces CD8 T cell exhaustion in the tumor microenvironment. Cell Metab. 30, 143–156 (2019).
Rodrigues, N. V. et al. Low-density lipoprotein uptake inhibits the activation and antitumor functions of human Vγ9Vδ2 T cells. Cancer Immunol. Res. 6, 448–457 (2018).
Blank, N. et al. Atorvastatin inhibits T cell activation through 3-hydroxy-3-methylglutaryl coenzyme A reductase without decreasing cholesterol synthesis. J. Immunol. 179, 3613–3621 (2007).
James, G., Witten, D., Hastie, T. & Tibshirani, R. An Introduction to Statistical Learning: with Applications in R (Springer, 2013).
Kidani, Y. et al. Sterol regulatory element–binding proteins are essential for the metabolic programming of effector T cells and adaptive immunity. Nat. Immunol. 14, 489–499 (2013).
Marchler-Bauer, A. et al. CDD/SPARCLE: functional classification of proteins via subfamily domain architectures. Nucleic Acids Res. 45, D200–D203 (2017).
Jiang, P. et al. Signatures of T cell dysfunction and exclusion predict cancer immunotherapy response. Nat. Med. 24, 1550–1558 (2018).
Curtis, C. et al. The genomic and transcriptomic architecture of 2,000 breast tumours reveals novel subgroups. Nature 486, 346–352 (2012).
Xi, H. et al. A human skeletal muscle atlas identifies the trajectories of stem and progenitor cells across development and from human pluripotent stem cells. Cell Stem Cell. https://doi.org/10.1016/j.stem.2020.04.017 (2020).
Silva-Filho, J. L., Caruso-Neves, C. & Pinheiro, A. A. S. IL-4: an important cytokine in determining the fate of T cells. Biophys. Rev. https://doi.org/10.1007/s12551-013-0133-z (2014).
Emmerich, J. et al. IL-10 directly activates and expands tumor-resident CD8(+) T cells without de novo infiltration from secondary lymphoid organs. Cancer Res. 72, 3570–3581 (2012).
Hanna, B. S. et al. Interleukin-10 receptor signaling promotes the maintenance of a PD-1int TCF-1+CD8+ T cell population that sustains anti-tumor immunity. Immunity https://doi.org/10.1016/j.immuni.2021.11.004 (2021).
Schaaf, M. B., Garg, A. D. & Agostinis, P. Defining the role of the tumor vasculature in antitumor immunity and immunotherapy. Cell Death Dis. 9, 115 (2018).
Pan, D. et al. A major chromatin regulator determines resistance of tumor cells to T cell-mediated killing. Science. https://doi.org/10.1126/science.aao1710 (2018).
Spidlen, J., Breuer, K. & Brinkman, R. Preparing a minimum information about a flow cytometry experiment (MIFlowCyt) compliant manuscript using the International Society for Advancement of Cytometry (ISAC) FCS file repository (FlowRepository.org). Curr. Protoc. Cytom. https://doi.org/10.1002/0471142956.cy1018s61 (2012).
Ghandi, M. et al. Next-generation characterization of the Cancer Cell Line Encyclopedia. Nature 569, 503–508 (2019).
Duffy, D. J. et al. Integrative omics reveals MYCN as a global suppressor of cellular signalling and enables network-based therapeutic target discovery in neuroblastoma. Oncotarget 6, 43182–43201 (2015).
Zheng, R. et al. Cistrome Data Browser: expanded datasets and new tools for gene regulatory analysis. Nucleic Acids Res. 47, D729–D735 (2019).
Acknowledgements
We thank S. Rosenberg, F. Lowery, K. Sri and M. Carrington for their helpful discussions. This work utilized the computational resources of the NIH HPC Biowulf cluster. This work is supported by the intramural budget allocation from the NCI, NIH, awarded to P.J. Y.Z. is partially supported by a Bau Tsu Zung Bau Kwan Yeu Hing Research and Clinical Fellowship from the University of Hong Kong. T.V. is supported by the NCI iCURE and T2I Postdoc fellowship.
Author information
Authors and Affiliations
Contributions
P.J. and Y.Z. designed the study and wrote the manuscript. P.J. performed all computational analysis and created the Tres package and web server. Y.Z. and T.V. performed wet-laboratory experiments. R.K. provided human donor T cells. C.S. and C.D. provided Pmel-1 TCR transgenic mice. R.P. deployed the Tres server under the government domain. K.A. and X.G. provided the laboratory space. L.G., J.H., F.L., D.P., T.N., Z.C., C.W. and G.M. participated in discussions.
Corresponding author
Ethics declarations
Competing interests
NCI, NIH has submitted a provisional patent application no. 63/246,743 for using FIBP KO to enhance cellular immunotherapy efficacies, with P.J. and Y.Z. as inventors. All other authors declare no competing interests.
Peer review
Peer review information
Nature Medicine thanks Alexander Huang and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. Saheli Sadanand was the primary editor on this article and managed its editorial process and peer review in collaboration with the rest of the editorial team.
Additional information
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Extended data
Extended Data Fig. 1 Quantifying immunosuppression of T cells in single-cell data.
a, Correlations between immunosuppressive signaling and T cell proliferation or cytotoxicity, shown with the violin plot smoothed by a kernel density estimator (n = 168 per group). Each dot represents a tumor. The p-value was from the two-sided Wilcoxon rank-sum test. b, Association between TGF-β1 signaling and SMAD transcription factor (TF) activities, an example from a melanoma tumor18. Each dot presents a CD8 + T cell. The Y-axis presents the TGF-β1 activities. The X-axis presents the SMAD TF activities. The p-value was from the two-sided Wilcoxon rank-sum test, comparing TGF-β1 activities between T cells with positive and negative SMAD activities. c, Receiver operating characteristic (ROC) curve for the quality of signaling activities. At different thresholds, the ROC curve presents false-positive rates against true-positive rates of predicting SMAD or REL activities through TGF-β1 or TRAIL signaling activities, respectively. The diagonal line represents random expectation. d, Quality of signaling activities across single-cell cohorts. Each dot presents a single-cell cohort. The area under the ROC curve (AUC) measures the quality of TGF-β1 or TRAIL signaling, shown with violin plots smoothed by a density kernel estimator. P-values were from the two-sided Wilcoxon signed-rank test, comparing group values with 0.5 (random expectation). e, Correlation between PGE2 signaling and cAMP pathway activities, an example of a melanoma tumor18. Each dot represents a CD8 + T cell. The X-axis shows the PGE2 signaling activity. The Y-axis shows the cAMP geneset value. f, Correlations between PGE2 signaling and cAMP activities. Each dot represents a tumor. The correlations were shown through box plots. The thick line represents the median value. The bottom and top of the boxes are the 25th and 75th percentiles (interquartile range). Whiskers encompass 1.5 times the interquartile range. The p-value was from the two-sided Wilcoxon signed-rank test, comparing group medians and zero. g, Gene expression of immunosuppressive signals. Each dot represents a patient. Y-axis presents the average gene expression across all cells for each lineage (X-axis). All lineages were ranked by the median expression and only the top ten profiles were shown in box plots as panel f.
Extended Data Fig. 2 Modeling transcriptomic signatures of tumor-resilient T cells.
a, Variable interaction test. The correlation between immunosuppression (through TRAIL or PGE2 signaling) and CD8 + T cell proliferation is plotted as Fig. 1c. Individual cells were split into high or low groups according to the expression of IL7R or FIBP (Cutoff = 2). P-values and t-values (Tres gene scores) were from the two-sided interaction test using continuous values without any cutoffs. b, Distribution of interaction test p-values, an example from a melanoma tumor #9818. Two-sided student t-test p-values were computed for every gene and shown by a histogram. c, Tres gene scores of T cell function markers. Each dot represents a tumor. Tres scores computed for each signal are shown through violin plots smoothed by a kernel density estimator (n = 139, 137, 132 for TGF-β1, TRAIL, PGE2 groups). P-values were from the two-sided Wilcoxon signed-rank test comparing group values and zero. d, Tres gene scores of immunosuppressive signal receptors, shown as panel c. Scores for FIBP and IL7R are attached as references. e, Similarity of Tres scores from different single-cell cohorts. The hierarchical clustering is based on Pearson correlations among median Tres scores among all patients and immunosuppressive signals in each cohort. f, Receiver operating characteristic (ROC) curve for the quality of Tres score. The ROC curve presents false-positive rates against true-positive rates of predicting T-cell persistence markers through Tres gene scores at different thresholds, using the example data in Fig. 1e. The diagonal line represents random expectation. The area under the ROC curve (AUC) is a quantitative measure of Tres score quality. g, Quality of Tres scores. Each dot represents a tumor with sample counts labeled beneath each group. The AUC in each cohort was shown through violin plots, smoothed by a kernel density estimator. P-values (labeled above each group) were computed through the two-sided Wilcoxon signed-rank test, comparing the group values and the random expectation 0.5. h, ROC curves for Tres scores computed using in vitro cell line data from the CCLE project77. Results from the two most frequent lineages (with more than 100 cell lines) are shown as panel f.
Extended Data Fig. 3 Geneset enrichment analysis for Tres gene scores.
a, Gene set enrichment analysis (GSEA)25, an example on αβ T cell proliferation. X-axis presents the gene ranked by median Tres scores (bottom Y-axis) across all tumors and immunosuppressive signals. Y-axis on the top presents enrichment scores at each gene rank. The P-value is computed through the two-sided permutation test (n = 1000 randomizations). HLA-E is the first gene annotated in the pathway. b, Top-three enriched biological processes. For each cancer type, enrichment of gene ontology pathways was analyzed for median Tres scores among all patients. Top-three enriched and depleted pathways for Tres scores related to TGF-β1, TRAIL and PGE2 signaling were shown together. c, Tres scores of non-classical MHC class I molecules. Each dot represents a value from a tumor. Tres scores computed for each signal are shown through violin plots smoothed by a kernel density estimator (n = 139, 137, 132 for TGF-β1, TRAIL, PGE2 groups). P-values were from the two-sided Wilcoxon signed-rank test comparing group values and zero.
Extended Data Fig. 4 Control analyses of the median Tres signature in predicting immunotherapy responses.
a, Prediction through correlations between the median Tres signature and T-cell bulk transcriptomic profiles. b, Tres score correlations with profiles from post-treatment tumors. Each point represents an ICI-treated tumor. Y-axis presents the correlation between the Tres signature and T-cell expression profiles. P-values were from the two-sided Wilcoxon rank-sum test comparing group values. The thick line represents the median value. The bottom and top of the boxes are the 25th and 75th percentiles (interquartile range). Whiskers encompass 1.5 times the interquartile range. c, Tres scores indicate T-cell clinical efficacy. The ROC curve presents false-positive rates against true-positive rates of predicting whether T cells are from responders or non-responders. The performance of diverse signatures was compared. d, Comparison among T-cell signatures in predicting clinical response. The negative predictive value was shown for T-cell efficacy signatures in Supplementary Table 3. All box-plots have the same format as panel b (n = 6 independent datasets per box). e, Lack of associations between Tres score correlations and adoptive cell therapy efficacy in tumors with T-cell infiltration lower than average. The survival of patients upon adoptive T cell transfer was shown for tumors with positive or negative Tres score correlations. P-values were from the two-sided Wald test using continuous values. f, The performance of Tres on predicting ICI outcomes using bulk data. Each dot represents a pretreatment tumor transcriptomics cohort listed in Supplementary Table 4. The first group presents the results using all samples. The other two groups present results using tumors with positive or negative CTL levels. The Y-axis presents Cox-PH risk z-scores as the association between overall survival and Tres signature correlations. Box plots have the same format as panel b. P-values were computed through the two-sided Wilcoxon signed-rank test. g, Tres prediction performance on different combinations of treatments and sample sites in a triple-negative breast cancer study40. The area under the ROC curve (AUC) and confidence intervals were shown with 0.5 as the random expectation. h, Tres prediction performance in tumors when immunosuppressive signals are lower than average. The data and box-plots are shown as panel f.
Extended Data Fig. 5 High FIBP levels in T cells are associated with inferior antitumor efficacy.
a, FIBP is the first among all genes based on the average rank across all signatures in Fig. 3a. b, High FIBP levels in bulk tumors indicate T-cell dysfunction, generated from the TIDE server68. For each neuroblastoma tumor78, the cytotoxic T-lymphocyte (CTL) infiltration level was estimated as the average expression level of CD8A, CD8B, GZMA, GZMB, and PRF1. Each Kaplan–Meier plot presents tumors in two groups: ‘CTL Top’ has above-median CTL values among all samples, while ‘CTL Bottom’ has values below the median. Samples were split according to the FIBP expression level through the best separation criterion maximizing the difference of associations between CTL and survival outcome between two groups. P-values are computed through the two-sided interaction t-test between CTL and FIBP using continuous values without any cutoffs. c, Only Fibp knockout (KO) promoted T-cell mediated cancer killing. Each gene was knocked out through a mixture of three independent gRNAs in Pmel-1 mouse T cells. The abundance of B16-mhgp100 co-cultured with gene KO T cells was measured as the intensity of tdTomato. The Y-axis presents the difference between each target gene KO and Rosa26 control KO. Each point presents the median and standard error of the mean as error bars (n = 3 cell-culture replicates). d, CRISPR screen phenotypes of FIBP compared to CBLB. Each dot represents a gRNA. The Y-axis shows the log2 fold-change (logFC) of gRNA abundances between target and control conditions. Each condition presents mean and standard deviations as error bars across all independent gRNAs (n = 4 for the TCR stimulation screen12, n = 6 for the IL-2 production screen13).
Extended Data Fig. 6 In vitro phenotypes of T cells after gene knockouts.
a, Efficacy of human donor T cells on killing A375 or Mel624 cells. The T-cell efficacy was measured as the relative red fluorescent protein (RFP) intensity (Y-axis) of cancer cells across time points (X-axis) as Fig. 4d. The culture conditions include three control gRNAs (AAVS1 for human and Rosa26 for mouse), parental T cells, and cancer cells cultured without T cells. For each condition, the mean and standard deviation (error bars) values were shown (n = 3 cell-culture replicates). b, Result from Pmel-1 TCR T cells on killing B16-mhgp100 cells, shown as panel a. c, Representative CFSE signals of T cell proliferation assay in Pmel-1 CD8 + T cells. The proliferative ability of Pmel-1 T cells were compared in Fibp knockout versus control cells (Rosa26 KO) after restimulation with anti-CD3/28 antibodies for 4 days. Representative CFSE signals from TCR stimulated and unstimulated T cells were shown. d, T cell proliferation assay in Pmel-1 CD8 + T cells. The mean and standard deviation values (n = 3 independent gRNAs) of the proliferation index for each gene were calculated using Flowjo software, with gating strategy in panel e. The comparison was through the two-sided Wilcoxon rank-sum test. e, Gating strategy for CFSE staining analysis. The forward-scatter area (FSC-A) and side-scatter area (SSC-A) are used for dead cell and debris removal. The forward-scatter height (FSC-H), width (FSC-W), side scatter height (SSC-H), width (SSC-W) are used to gate the single cells. We include the unstimulated T cells to define the background populations. Cells from all biological replicates in panel d are pooled together for demonstration. f, Early activation measured 12 h after TCR stimulation. CD69 level was determined by flow cytometry in FIBP and control KO cells in mouse and human. The left panel shows a representative histogram of signal intensity. The right panel shows the mean and standard deviations as error bars (n = 3 independent gRNAs per gene) for marker positive fractions. The comparisons between groups were through the two-sided Wilcoxon rank-sum test.
Extended Data Fig. 7 Analysis of T cells in adoptive cell transfer.
a, Tumor size in different groups randomized at day zero. All mice were randomized into different groups to achieve an even initial tumor size (n = 15 for Rosa26, Cblb, and Fibp knockout (KO) groups, n = 5 for parental and No treatment groups). The tumor area distribution was shown through violin plots smoothed with a kernel density estimator. b, Tumor size at different days post-adoptive cell transfer (ACT) for groups in panel a. c, Average tumor size in negative control groups. The tumor area mean was shown when no mice reached endpoints, with standard errors as error bars. d, Kaplan Meier curves of mice survival durations. The fraction of mice that did not reach the endpoint was shown on different days after ACT. The comparison of survival durations between no-treatment and parental groups was through the two-sided log-rank test. e, In-vivo flow analysis of T-cell phenotype markers. Marker positive fractions of T cells with gene KOs were shown with mean values and standard deviations as error bars (tumor counts labeled under each group, growth curves available in Supplementary Fig. 3). Different groups were compared through the one-sided Wilcoxon rank-sum test, and none of them achieved statistical significance. f, Gating strategy of in-vivo flow analysis. The forward scatter height (FSC-H) and side scatter height (SSC-H) are used for dead cell and debris removal. The forward scatter height (FSC-H), width (FSC-W), side scatter height (SSC-H), and width (SSC-W) are used to gate the single cells. DAPI was used to eliminate dead cells, and CD8 + cells were gated. Then marker positive cells (CD69 in this example) were gated on the basis of the non-staining control.
Extended Data Fig. 8 Differentially-expressed genes upon FIBP knockout are enriched for cholesterol metabolism functions.
a, Correlation between FIBP and ABCA1 among acute lymphoblastic leukemia samples (n = 58) from the MILE project57. b, Mean and standard error (error bars) of correlations (n = 22 leukemia subtypes per bar) between FIBP and gene members in the cholesterol metabolism pathway. Gene members that positively or negatively regulate cholesterol metabolism were shown in red or blue colors, respectively. The comparison between the positive and negative groups is through the two-sided Wilcoxon rank-sum test. c, Differential expression values for SREBF2 ChIP-Seq targets and non-targets. The Cistrome database contains two public human ChIP-Seq profiles from B lymphocytes and HepG2 cells79. The regulatory score on each target gene is computed by the RABIT framework58. Target genes are those with regulatory scores larger than 0.5. None target genes are those with zero regulatory scores. The gene count in each group is labeled above each boxplot. The thick line represents the median value. The bottom and top of the boxes are the 25th and 75th percentiles (interquartile range). Whiskers encompass 1.5 times the interquartile range. The comparison between groups is through the two-sided Wilcoxon rank-sum test. d, RT–qPCR validation of essential enzymes in cholesterol synthesis in mouse T cells. Normalized expression levels were shown for Rosa26 control and Fibp knockout conditions. The mean values were shown (n = 2 independent gRNAs). e, RT–qPCR levels of cholesterol metabolic regulators in mouse T cells with Fibp and vector control overexpression. The mean and standard deviation (error bars) were shown (n = 3 cell-culture replicates).
Extended Data Fig. 9 FIBP knockout reduces cholesterol levels in T cells.
a, Gating strategy of in-vivo Filipin III flow analysis. The forward scatter Area (FSC-A) and side scatter area (SSC-A) are used for dead cell and debris removal. The forward scatter height (FSC-H), width (FSC-W), side scatter height (SSC-H), and width (SSC-W) are used to gate the single cells. Live CD8 + T subset is gated on non-staining negative control and single staining positive control. The Filipin-III level was measured as the median fluorescence density. Cells from one example tumor are shown for demonstration. b, Cholesterol levels in Fibp versus Rosa26 knockout (KO) T cells via Filipin III staining, with gating strategy in panel c. The left panel shows representative histograms of signal intensity. The right panel presents mean and standard deviation as error bars (n = 3 independent gRNAs). The p-value was from the two-sided Wilcoxon rank-sum test. c, Gating strategy of in vitro Filipin III analysis. The procedure is the same as panel a, except that the T-cell gating step is not involved as samples are in-vitro cultures of CD8 + T cells. d, Cholesterol levels via oxidation-based quantification. Each KO group has replicates from three independent gRNAs. Each overexpression group has replicates from three cell cultures. The data were presented and compared in the same way as panel b. e, LDL uptake after lipoprotein deprivation. CD8 + T cells with Fibp and Rosa26 control KOs were treated with Dil-LDL after lipoprotein deprivation for 16 h. The LDL uptake was determined using flow cytometry. The left panel shows representative histograms. The right panel presents mean and standard deviation as error bars (n = 3 independent gRNAs).
Extended Data Fig. 10 Simvastatin does not inhibit cholesterol levels in activated T cells.
(a, b), Cholesterol levels upon simvastatin treatments. T cells were treated with simvastatin concurrently with anti-CD3/28 activation (a, During) or 72 hours after anti-CD3/28 activation (b, Post). The median Filipin III intensity was measured with flow cytometry at different simvastatin concentrations, with a representative plot on the left and the median intensity on the right. The gating strategy follows the same scheme in Extended Data Fig. 9c. (c, d), Simvastatin inhibits T-cell proliferation. For each condition, the T-cell expansion index was computed through CFSE staining and flow cytometry analysis. The data is shown in the same way as panels a, b.
Supplementary information
Supplementary Information
Supplementary Figures and Tables
Source data
Source Data Fig. 1
Statistical Source Data.
Source Data Fig. 2
Statistical Source Data.
Source Data Fig. 3
Statistical Source Data.
Source Data Fig. 4
Statistical Source Data and Supplementary Figure indices of unprocessed western blots and live-cell imaging.
Source Data Fig. 5
Statistical Source Data and flow repository IDs.
Source Data Fig. 6
Statistical Source Data, Supplementary Figure indices of unprocessed western blots, and flow repository IDs.
Source Data Extended Data Fig. 1
Statistical Source Data.
Source Data Extended Data Fig. 2
Statistical Source Data.
Source Data Extended Data Fig. 3
Statistical Source Data.
Source Data Extended Data Fig. 4
Statistical Source Data.
Source Data Extended Data Fig. 5
Statistical Source Data.
Source Data Extended Data Fig. 6
Statistical Source Data and flow repository ID.
Source Data Extended Data Fig. 7
Statistical Source Data and flow repository ID.
Source Data Extended Data Fig. 8
Statistical Source Data.
Source Data Extended Data Fig. 9
Statistical Source Data and flow repository ID.
Source Data Extended Data Fig. 10
Statistical Source Data and flow repository ID.
Rights and permissions
Springer Nature or its licensor holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.
About this article
Cite this article
Zhang, Y., Vu, T., Palmer, D.C. et al. A T cell resilience model associated with response to immunotherapy in multiple tumor types. Nat Med 28, 1421–1431 (2022). https://doi.org/10.1038/s41591-022-01799-y
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1038/s41591-022-01799-y
This article is cited by
-
Integrating single-cell multi-omics and prior biological knowledge for a functional characterization of the immune system
Nature Immunology (2024)
-
Stroma-infiltrating T cell spatiotypes define immunotherapy outcomes in adolescent and young adult patients with melanoma
Nature Communications (2024)
-
Identifying microRNAs associated with tumor immunotherapy response using an interpretable machine learning model
Scientific Reports (2024)
-
Allergen-induced CD11c + dendritic cell pyroptosis aggravates allergic rhinitis
Cell Communication and Signaling (2023)
-
Personalized tumor combination therapy optimization using the single-cell transcriptome
Genome Medicine (2023)