科技科学

In vivo genome-wide CRISPR screens of human T cells in solid tumours

Nature作者：Qi Liu2026年8月12日正文已收录本站

Data availability

All CRISPR screening data generated in this study are provided in Supplementary Tables 1, 2 and 5. The crRNA sequences used are listed in Supplementary Table 3. All RNA sequencing data are provided in Supplementary Table 6. The Gene Expression Omnibus accession number for RNA sequencing raw data is GSE330227. The human reference genome GRCh38.p14 was used for RNA sequencing alignment and analysis and is available (https://www.gencodegenes.org/human/release_45.html). Source data are available with this paper.

References

  1. Carnevale, J. et al. RASA2 ablation in T cells boosts antigen sensitivity and long-term function. Nature 609, 174–182 (2022).

    Article  ADS  CAS  PubMed  PubMed Central  Google Scholar 

  2. Freitas, K. A. et al. Enhanced T cell effector activity by targeting the Mediator kinase module. Science 378, eabn5647 (2022).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  3. Brudno, J. N., Maus, M. V. & Hinrichs, C. S. CAR T cells and T-cell therapies for cancer: a translational science review. JAMA 332, 1924–1935 (2024).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  4. Du, B. et al. CAR-T therapy in solid tumors. Cancer Cell 43, 665–679 (2025).

    Article  CAS  PubMed  Google Scholar 

  5. Anderson, K. G., Stromnes, I. M. & Greenberg, P. D. Obstacles posed by the tumor microenvironment to T cell activity: a case for synergistic therapies. Cancer Cell 31, 311–325 (2017).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  6. Arner, E. N. & Rathmell, J. C. Metabolic programming and immune suppression in the tumor microenvironment. Cancer Cell 41, 421–433 (2023).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  7. Shifrut, E. et al. Genome-wide CRISPR screens in primary human T cells reveal key regulators of immune function. Cell 175, 1958–1971.e15 (2018).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  8. Belk, J. A. et al. Genome-wide CRISPR screens of T cell exhaustion identify chromatin remodeling factors that limit T cell persistence. Cancer Cell 40, 768–786.e7 (2022).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  9. Knudsen, N. H. et al. In vivo CRISPR screens identify modifiers of CAR T cell function in myeloma. Nature 646, 953–962 (2025).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  10. Datlinger, P. et al. Systematic discovery of CRISPR-boosted CAR T cell immunotherapies. Nature 646, 963–972 (2025).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  11. Wei, J. et al. Targeting Regnase-1 programs long-lived effector T cells for cancer therapy. Nature 576, 471–476 (2019).

    Article  ADS  CAS  PubMed  PubMed Central  Google Scholar 

  12. Siska, P. J. et al. Mitochondrial dysregulation and glycolytic insufficiency functionally impair CD8 T cells infiltrating human renal cell carcinoma. JCI Insight 2, e93411 (2017).

    Article  PubMed  PubMed Central  Google Scholar 

  13. Klysz, D. D. et al. Inosine induces stemness features in CAR-T cells and enhances potency. Cancer Cell 42, 266–282.e8 (2024).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  14. Sutra Del Galy, A. et al. In vivo genome-wide CRISPR screens identify SOCS1 as intrinsic checkpoint of CD4+ TH1 cell response. Sci. Immunol. 6, eabe8219 (2021).

    Article  PubMed  Google Scholar 

  15. Si, J. et al. Hematopoietic progenitor kinase 1 (HPK1) mediates T cell dysfunction and is a druggable target for T cell-based immunotherapies. Cancer Cell 38, 551–566.e11 (2020).

    Article  CAS  PubMed  Google Scholar 

  16. Morotti, M. et al. PGE2 inhibits TIL expansion by disrupting IL-2 signalling and mitochondrial function. Nature 629, 426–434 (2024).

    Article  ADS  CAS  PubMed  PubMed Central  Google Scholar 

  17. Luo, C. et al. Integrated computational analysis identifies therapeutic targets with dual action in cancer cells and T cells. Immunity 58, 745–765.e9 (2025).

    Article  CAS  PubMed  Google Scholar 

  18. Lacher, S. B. et al. PGE2 limits effector expansion of tumour-infiltrating stem-like CD8+ T cells. Nature 629, 417–425 (2024).

    Article  ADS  CAS  PubMed  PubMed Central  Google Scholar 

  19. Wang, B. & Bian, Q. SATB1 prevents immune cell infiltration by regulating chromatin organization and gene expression of a chemokine gene cluster in T cells. Commun. Biol. 7, 1304 (2024).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  20. Verdeil, G. MAF drives CD8+ T-cell exhaustion. Oncoimmunology 5, e1082707 (2016).

    Article  PubMed  Google Scholar 

  21. Kim, Y. J. et al. The transcription factor Mef2d regulates B:T synapse-dependent GC-TF differentiation and IL-21-mediated humoral immunity. Sci. Immunol. 8, eadf2248 (2023).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  22. Giordano, M. et al. The tumor necrosis factor α-induced protein 3 (TNFAIP3, A20) imposes a brake on antitumor activity of CD8 T cells. Proc. Natl Acad. Sci. USA 111, 11115–11120 (2014).

    Article  ADS  CAS  PubMed  PubMed Central  Google Scholar 

  23. Szeto, A. C. H. et al. Mef2d potentiates type-2 immune responses and allergic lung inflammation. Science 384, eadl0370 (2024).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  24. Dahabieh, M. S. et al. The prostacyclin receptor PTGIR is a NRF2-dependent regulator of CD8+ T cell exhaustion. Nat. Immunol. 26, 1139–1151 (2025).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  25. Pham, T. H., Okada, T., Matloubian, M., Lo, C. G. & Cyster, J. G. S1P1 receptor signaling overrides retention mediated by Gαi-coupled receptors to promote T cell egress. Immunity 28, 122–133 (2008).

    Article  CAS  PubMed  Google Scholar 

  26. Arnon, T. I. et al. GRK2-dependent S1PR1 desensitization is required for lymphocytes to overcome their attraction to blood. Science 333, 1898–1903 (2011).

    Article  ADS  CAS  PubMed  PubMed Central  Google Scholar 

  27. Muppidi, J. R., Lu, E. & Cyster, J. G. The G protein-coupled receptor P2RY8 and follicular dendritic cells promote germinal center confinement of B cells, whereas S1PR3 can contribute to their dissemination. J. Exp. Med. 212, 2213–2222 (2015).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  28. Muppidi, J. R. et al. Loss of signalling via Gα13 in germinal centre B-cell-derived lymphoma. Nature 516, 254–258 (2014).

    Article  ADS  CAS  PubMed  PubMed Central  Google Scholar 

  29. Feig, C. et al. Targeting CXCL12 from FAP-expressing carcinoma-associated fibroblasts synergizes with anti-PD-L1 immunotherapy in pancreatic cancer. Proc. Natl Acad. Sci. USA 110, 20212–20217 (2013).

    Article  ADS  CAS  PubMed  PubMed Central  Google Scholar 

  30. McCandless, E. E., Zhang, B., Diamond, M. S. & Klein, R. S. CXCR4 antagonism increases T cell trafficking in the central nervous system and improves survival from West Nile virus encephalitis. Proc. Natl Acad. Sci. USA 105, 11270–11275 (2008).

    Article  ADS  CAS  PubMed  PubMed Central  Google Scholar 

  31. Kabashima, K. et al. Prostaglandin E2-EP4 signaling initiates skin immune responses by promoting migration and maturation of Langerhans cells. Nat. Med. 9, 744–749 (2003).

    Article  CAS  PubMed  Google Scholar 

  32. Lu, E., Wolfreys, F. D., Muppidi, J. R., Xu, Y. & Cyster, J. G. S-geranylgeranyl-L-glutathione is a ligand for human B cell-confinement receptor P2RY8. Nature 567, 244–248 (2019).

    Article  ADS  CAS  PubMed  PubMed Central  Google Scholar 

  33. Gallman, A. E. et al. Abcc1 and Ggt5 support lymphocyte guidance through export and catabolism of S-geranylgeranyl-l-glutathione. Sci. Immunol. 6, eabg1101 (2021).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  34. Jung, I. Y. et al. CRISPR/Cas9-mediated knockout of DGK improves antitumor activities of human T cells. Cancer Res. 78, 4692–4703 (2018).

    Article  CAS  PubMed  Google Scholar 

  35. Wang, H. et al. Cbl-b inhibition improves manufacturing efficiency and antitumoral efficacy of anti-CD19 CAR-T cells. Int. Immunopharmacol. 147, 113971 (2025).

    Article  CAS  PubMed  Google Scholar 

  36. Schmidt, R. et al. CRISPR activation and interference screens decode stimulation responses in primary human T cells. Science 375, eabj4008 (2022).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  37. Postler, T. S. A most versatile kinase: the catalytic subunit of PKA in T-cell biology. Int. Rev. Cell Mol. Biol. 361, 301–318 (2021).

    Article  CAS  PubMed  Google Scholar 

  38. Zhang, M. et al. G protein-coupled receptors (GPCRs): advances in structures, mechanisms, and drug discovery. Signal Transduct. Target. Ther. 9, 88 (2024).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  39. Wu, V. H. et al. The GPCR-Gαs-PKA signaling axis promotes T cell dysfunction and cancer immunotherapy failure. Nat. Immunol. 24, 1318–1330 (2023).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  40. Ben Nasr, M. et al. Glucagon-like peptide 1 receptor is a T cell-negative costimulatory molecule. Cell Metab. 36, 1302–1319.e12 (2024).

    Article  PubMed  Google Scholar 

  41. Zeng, S. et al. Local TSH/TSHR signaling promotes CD8+ T cell exhaustion and immune evasion in colorectal carcinoma. Cancer Commun. 44, 1287–1310 (2024).

    Article  Google Scholar 

  42. Giuffrida, L. et al. CRISPR/Cas9 mediated deletion of the adenosine A2A receptor enhances CAR T cell efficacy. Nat. Commun. 12, 3236 (2021).

    Article  ADS  CAS  PubMed  PubMed Central  Google Scholar 

  43. Li, S. et al. In silico and pharmacological evaluation of GPR65 as a cancer immunotherapy target regulating T-cell functions. Front. Immunol. 15, 1483258 (2024).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  44. Qiao, G. et al. β-Adrenergic signaling blocks murine CD8+ T-cell metabolic reprogramming during activation: a mechanism for immunosuppression by adrenergic stress. Cancer Immunol. Immunother. 68, 11–22 (2019).

    Article  CAS  PubMed  Google Scholar 

  45. Nehama, D. et al. B7-H3-redirected chimeric antigen receptor T cells target glioblastoma and neurospheres. EBioMedicine 47, 33–43 (2019).

    Article  PubMed  PubMed Central  Google Scholar 

  46. Vitanza, N. A. et al. Intracerebroventricular B7-H3-targeting CAR T cells for diffuse intrinsic pontine glioma: a phase 1 trial. Nat. Med. 31, 861–868 (2025).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  47. Bengsch, B. et al. Epigenomic-guided mass cytometry profiling reveals disease-specific features of exhausted CD8 T cells. Immunity 48, 1029–1045.e5 (2018).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  48. Qi, C. et al. Claudin18.2-specific CAR T cells in gastrointestinal cancers: phase 1 trial interim results. Nat. Med. 28, 1189–1198 (2022).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  49. Goudy, L. et al. Integrated epigenetic and genetic programming of primary human T cells. Nat. Biotechnol. https://doi.org/10.1038/s41587-025-02856-w (2025).

  50. Barnkob, M. B. et al. Semaphorin 3A causes immune suppression by inducing cytoskeletal paralysis in tumour-specific CD8+ T cells. Nat. Commun. 15, 3173 (2024).

    Article  ADS  CAS  PubMed  PubMed Central  Google Scholar 

  51. Matas-Rico, E. et al. Autotaxin impedes anti-tumor immunity by suppressing chemotaxis and tumor infiltration of CD8+ T cells. Cell Rep. 37, 110013 (2021).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  52. Carlson, C. R. et al. Delineation of type I protein kinase A-selective signaling events using an RI anchoring disruptor. J. Biol. Chem. 281, 21535–21545 (2006).

    Article  CAS  PubMed  Google Scholar 

  53. Newick, K. et al. Augmentation of CAR T-cell trafficking and antitumor efficacy by blocking protein kinase A localization. Cancer Immunol. Res. 4, 541–551 (2016).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  54. Wunderlich, M. et al. Improved multilineage human hematopoietic reconstitution and function in NSGS mice. PLoS ONE 13, e0209034 (2018).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  55. Shindyapin, V. V. et al. Myeloid cell mobilization and recruitment by human mesothelioma in NSG-SGM3 mice. Cells 13, 2135 (2024).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  56. Billerbeck, E. et al. Development of human CD4+FoxP3+ regulatory T cells in human stem cell factor-, granulocyte-macrophage colony-stimulating factor-, and interleukin-3-expressing NOD-SCID IL2Rγnull humanized mice. Blood 117, 3076–3086 (2011).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  57. Zorn, J. A. et al. Humanization of a strategic CD3 epitope enables evaluation of clinical T-cell engagers in a fully immunocompetent in vivo model. Sci. Rep. 12, 3530 (2022).

    Article  ADS  CAS  PubMed  PubMed Central  Google Scholar 

  58. Siegler, E. L. et al. Efficient gene editing of CART cells with CRISPR-Cas12a for enhanced antitumor efficacy. Blood 136, 6–7 (2020).

    Article  Google Scholar 

  59. Tieu, V. et al. A versatile CRISPR-Cas13d platform for multiplexed transcriptomic regulation and metabolic engineering in primary human T cells. Cell 187, 1278–1295.e20 (2024).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  60. Takacsi-Nagy, O. et al. Evolutionarily guided transcription factor design programs novel T cell states. Preprint at bioRxiv https://doi.org/10.1101/2024.11.06.622344 (2024).

Download references

Acknowledgements

Q.L. acknowledges support from the Outstanding Doctoral Graduates Development Scholarship of Shanghai Jiao Tong University. We thank members of the Carnevale, Marson and Eyquem laboratories and T. Tolpa for their valuable support; F. Zhu for helpful discussions and assistance with bulk RNA sequencing data analysis; and, the Center for Advanced Technology at UCSF and the Mouse Metabolism Core (supported by NORC grant P30DK098722) for technical assistance. Figures were created, in part, with BioRender (https://biorender.com).

Funding

J.C. received funding from the NIH (K08CA252605 and R01CA309039), the Parker Institute for Cancer Immunotherapy, the Burroughs Wellcome Fund, the Mark Foundation for Cancer Research, CRISPR Cures for Cancer, the V Foundation, the Lydia Preisler Shorenstein Donor Advised Fund, the Pascarella Scholars Fund, the RunforAmma, and the UCSF Living Therapeutics Initiative. F.P. is the Connie Bob Lurie Fellow of Damon Runyon Cancer Research Foundation (DRG-2510-23). A.M. received funding from the Simons Foundation, the Lloyd J. Old STAR Award (Cancer Research Institute), the Parker Institute for Cancer Immunotherapy, the Innovative Genomics Institute, the Larry L. Hillblom Foundation (grant 2020-D-002-NET), the Arc Institute, the Byers family, K. Jordan and the CRISPR Cures for Cancer Initiative. J.E. received funding from the Parker Institute for Cancer Immunotherapy and CRISPR Cures for Cancer. G.M.A. is supported by NIH grant K08CA259610 and the Lydia Preisler Shorenstein Donor Advised Fund. J.C., A.M., G.M.A., B.R.S., S.P.B., J.E. and Z.G. are grateful for support from the Weill Cancer Hub West.

Author information

Authors and Affiliations

  1. Department of Medicine, University of California San Francisco, San Francisco, CA, USA

    Qi Liu, Peixin Amy Chen, Esha Urs, Maya M. Arce, Charlotte H. Wang, Jun Yan, Vinh Q. Nguyen, Zhongmei Li, Jin Seo, Nupura Kale, Yikai Luo, Laine Goudy, Haixia Zhong, Jae Hyung Jung, Greg M. Allen, Justin Eyquem, Alexander Marson & Julia Carnevale

  2. Gladstone-UCSF Institute of Genomic Immunology, San Francisco, CA, USA

    Qi Liu, Peixin Amy Chen, Esha Urs, Shimin Zhang, Maya M. Arce, Charlotte H. Wang, Jun Yan, Vinh Q. Nguyen, Zhongmei Li, Jin Seo, Nupura Kale, Yikai Luo, Laine Goudy, Chandrima Modak, Emma Dann, Jae Hyung Jung, Brian R. Shy, Justin Eyquem, Stacie E. Dodgson, Alexander Marson & Julia Carnevale

  3. UCSF CoLabs, University of California San Francisco, San Francisco, CA, USA

    Vinh Q. Nguyen

  4. Department of Surgery, University of California San Francisco, San Francisco, CA, USA

    Vinh Q. Nguyen

  5. Diabetes Center, University of California San Francisco, San Francisco, CA, USA

    Vinh Q. Nguyen

  6. Department of Microbiology and Immunology and Howard Hughes Medical Institute, University of California San Francisco, San Francisco, CA, USA

    Fanglue Peng, Taylor N. LaFlam & Jason G. Cyster

  7. Division of Pediatric Rheumatology, Department of Pediatrics, University of California San Francisco, San Francisco, CA, USA

    Taylor N. LaFlam

  8. Department of Genetics, Stanford University School of Medicine, Stanford, CA, USA

    Emma Dann

  9. Department of Surgery, Stanford University School of Medicine, Stanford, CA, USA

    Amanda Kirane

  10. Division of Oncology, Department of Medicine, Stanford University School of Medicine, Stanford, CA, USA

    Allison Betof Warner

  11. Division of Immunology and Rheumatology, Department of Medicine, Stanford University, Stanford, CA, USA

    Boi Bryant Quach & Zinaida Good

  12. Division of Computational Medicine, Department of Medicine, Stanford University, Stanford, CA, USA

    Boi Bryant Quach & Zinaida Good

  13. Parker Institute for Cancer Immunotherapy, San Francisco, CA, USA

    Zinaida Good, Greg M. Allen, Justin Eyquem, Alexander Marson & Julia Carnevale

  14. Weill Cancer Hub West, Stanford University and University of California, San Francisco, CA, USA

    Zinaida Good, Brian R. Shy, Sagar P. Bapat, Greg M. Allen, Justin Eyquem, Alexander Marson & Julia Carnevale

  15. Department of Laboratory Medicine, University of California San Francisco, San Francisco, CA, USA

    Brian R. Shy

  16. UCSF Helen Diller Family Comprehensive Cancer Center, University of California San Francisco, San Francisco, CA, USA

    Brian R. Shy, Greg M. Allen, Alexander Marson & Julia Carnevale

  17. Gray Faculty of Medical and Health Sciences, Tel Aviv University, Tel Aviv, Israel

    Eric Shifrut

  18. George S. Wise Faculty of Life Sciences, Tel Aviv University, Tel Aviv, Israel

    Eric Shifrut

  19. Dotan Center for Advanced Therapies, Tel Aviv Sourasky Medical Center, Tel Aviv, Israel

    Eric Shifrut

  20. Department of Pathology, University of California San Francisco, San Francisco, CA, USA

    Sagar P. Bapat

  21. Department of OB/Gyn, Center for Reproductive Sciences, University of California San Francisco, San Francisco, CA, USA

    Katherine Fuh

  22. Institute for Human Genetics (IHG), University of California San Francisco, San Francisco, CA, USA

    Alexander Marson

  23. Department of Microbiology and Immunology, University of California San Francisco, San Francisco, CA, USA

    Alexander Marson

  24. Innovative Genomics Institute, University of California Berkeley, Berkeley, CA, USA

    Alexander Marson

Authors

  1. Qi Liu
  2. Peixin Amy Chen
  3. Esha Urs
  4. Shimin Zhang
  5. Maya M. Arce
  6. Charlotte H. Wang
  7. Jun Yan
  8. Vinh Q. Nguyen
  9. Zhongmei Li
  10. Jin Seo
  11. Nupura Kale
  12. Fanglue Peng
  13. Yikai Luo
  14. Laine Goudy
  15. Taylor N. LaFlam
  16. Haixia Zhong
  17. Chandrima Modak
  18. Emma Dann
  19. Jae Hyung Jung
  20. Amanda Kirane
  21. Allison Betof Warner
  22. Boi Bryant Quach
  23. Zinaida Good
  24. Brian R. Shy
  25. Eric Shifrut
  26. Sagar P. Bapat
  27. Greg M. Allen
  28. Justin Eyquem
  29. Katherine Fuh
  30. Stacie E. Dodgson
  31. Jason G. Cyster
  32. Alexander Marson
  33. Julia Carnevale

Contributions

Q.L., J.C. and A.M. conceived and designed the study. Q.L. developed the in vivo screening platform, performed the CRISPR screens and primary T cell in vitro and in vivo experiments, and analysed the data. P.A.C., E.U., S.Z., Z.L. and J.S. assisted with the in vitro and in vivo CAR T validations. J.Y., C.M. and E.D. assisted with the sub-library screening. F.P., Y.L. and T.N.L. assisted with the migration assays and immunofluorescent staining. V.Q.N. assisted with cell sorting for the screens. N.K. generated the Brunello library virus and, together with L.G., assisted with the CRISPRoff assay. C.H.W. conducted the P2RY8 evolutionary analysis. H.Z., G.M.A. and J.H.J. provided the B7H3-BBz CAR and CLDN18.2-28z CAR constructs, respectively. K.F. provided ascites samples from patients with ovarian cancer. A.K., A.B.W., B.B.Q. and Z.G. provided TILs from one patient with melanoma, the data from which are included in the peer review file. Q.L., M.M.A., C.H.W. and E.S. performed the computational analyses. B.R.S., S.P.B., G.M.A., J.E. and J.G.C. helped design the assays and discussed the results. Q.L. prepared the figures with input from J.C., S.E.D. and A.M. J.C., S.E.D. and Q.L. wrote the manuscript, with input and editing from all authors.

Corresponding authors

Correspondence to Qi Liu or Julia Carnevale.

Ethics declarations

Competing interests

J.C. and Q.L. are authors of a patent application related to this paper. A.M. is a cofounder of Site Tx, Arsenal Biosciences, Spotlight Therapeutics and Survey Genomics; serves on the boards of directors at Site Tx and Survey Genomics; is a member of the scientific advisory boards of Site Tx, Arsenal Biosciences, Cellanome, Survey Genomics, NewLimit, Amgen, Tenaya and Network Bio; owns stock in Arsenal Biosciences, Site Tx, Cellanome, Spotlight Therapeutics, NewLimit, Survey Genomics, Tenaya, Lightcast and Network Bio; has received fees from Site Tx, Arsenal Biosciences, Cellanome, Spotlight Therapeutics, NewLimit, AbbVie, Gilead, Pfizer, 23andMe, Amgen, Network Bio, PACT Pharma, Juno Therapeutics, Tenaya, Lightcast, Trizell, Vertex, Merck, Genentech, GLG, ClearView Healthcare, AlphaSights and Rupert Case Management; is an investor in and informal advisor to Offline Ventures; and is a client of EPIQ. The Marson laboratory has received research support from the Parker Institute for Cancer Immunotherapy, the Emerson Collective, the Arc Institute, Juno Therapeutics, Epinomics, Sanofi, GlaxoSmithKline, Gilead and Anthem, and reagents from Genscript, 10X, Ultima and Illumina. S.E.D. is a shareholder of Site Tx.

Peer review

Peer review information

Nature thanks Joseph Fraietta and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. Peer reviewer reports are available.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Extended data figures and tables

Extended Data Fig. 1 Clonal selection and evaluation to establish a model system with high T cell recovery from tumours for screening.

a. Schematic of the anti-CD3 scFv-expressing tumour model for intratumoral enrichment of human T cells. b. Workflow depicting the generation of anti-CD3 scFv overexpressing A375 single-cell clones. c. Flow cytometric analysis of anti-CD3 scFv surface expression on bulk-transduced vs untransduced A375 cells. d. Flow cytometric analysis of anti-CD3 scFv expression across 70 selected single-cell clones. e. Scatter plot comparing Incucyte-based cancer cell killing rates of A375 cell lines in the presence of primary human T cells plotted vs. rate of tumour growth of each line in the absence of T cells. Tumour lines include wild-type A375 cells and 27 of the anti-CD3 scFv-overexpressing A375 single-cell clones selected from (d). For clones with comparable anti-CD3 scFv expression levels, a single representative clone was selected for further analysis. Labeled clones were assessed in follow-up experiments. f-j. A375low, A375medium, and A375high single-cell clones from (e) were subcutaneously engrafted into NSG mice, followed by intravenous infusion of 3 M primary human T cells. Shown are the experimental timeline (f), a visual of tumour size at day 20 post-engraftment (g), TIL number at day 20 (h), and expression of CD39 and CD69 in CD4+ and CD8+ TIL (i-j,). n = 4 mice per group. Note that the A375high tumours were fully eradicated, and thus we were unable to phenotype any T cells from those tumours in h-j. The data shown here were generated from the same in vivo validation experiment as Fig. 1b and therefore share the same A375low group. P values determined using two-tailed unpaired Student’s t-test (h-j). **P < 0.01, ****P < 0.0001. Box plots show the median (centre line), 25th–75th percentiles (box bounds), minimum and maximum values (whiskers), and all individual data points (h-j).

Source data

Extended Data Fig. 2 Phenotypic analysis of T cells from CD3 scFv-expressing tumours compared to spleens.

a-f. The A375low single-cell clone was subcutaneously engrafted into NSG mice followed by intravenous infusion of primary human T cells. T cell phenotypes in spleens and tumours were analyzed 9 days after T cell infusion. Shown are MFI of CD154 and CD137 in CD8+ T cells (a); MFI of PD-1, CD39, CD154, CD69, CD137, and Ki-67 in CD4+ T cells (b); frequencies of TNF-α+ and IFN-γ+ CD4+ T cells (c); frequencies of Tscm and Tem CD4+ T cells (d); statistical analysis of ECAR and OCR (e), and MFI of MitoTracker™ Red CMXRos staining (f) in T cells isolated from spleen and tumour (n = 4 mice for a-d and f; n = 5 technical replicates for e, representative of three mice); corresponding CD8 analyses are shown in Fig. 1. g-i. In vivo competition assay of AAVS1 control and ZC3H12A-KO primary human T cells in A375low tumour-bearing mice. Shown are the experimental timeline (g); flow cytometry gating strategy (h) and relative proportions of T cell populations recovered from spleens and tumours from a second donor (i). Left: BFP_AAVS1 control sgRNA mixed with GFP_ZC3H12A sgRNA (n = 5 mice); Right: GFP_AAVS1 control sgRNA mixed with BFP-ZC3H12A sgRNA (n = 3 mice). Data from the other donor are shown in Fig. 1j. “Single cells” and “CD45” plots in (h) are displayed using a biexponential (hyperlog) scale to allow visualization of events with low or negative fluorescence values. j. Flow cytometry gating strategy for sorting the highest (CD39high) and lowest (CD39low) 20% of TIL from A375low tumour-bearing NSG mice for the experiment shown in Fig. 1k. P values determined using two-tailed unpaired Student’s t-test (a-d, f, i). **P < 0.01, ***P < 0.001, ****P < 0.0001. Data are mean ± s.e.m. (i). Box plots show the median (centre line), 25th–75th percentiles (box bounds), minimum and maximum values (whiskers), and all individual data points (a-f).

Source data

Extended Data Fig. 3 Quality control and downstream validation analyses identifying P2RY8 as a regulator of T cell tumour infiltration.

a. In vivo competition assay of AAVS1 control and ZC3H12A-KO primary human T cells analyzed 7 days (left) and 9 days (right) after infusion. n = 3 mice per group. b. Distribution and GINI index of sgRNA read counts from input, spleen, and tumour samples across two donors in the in vivo abundance screens. c. Pearson correlation (r) of normalized sgRNA read counts comparing tumour versus input and spleen versus input across two donors. d. KEGG pathway enrichment analysis of dropout genes in the tumour and spleen abundance screens, P values were calculated using a one-sided hypergeometric test and adjusted for multiple comparisons using the Benjamini–Hochberg method (adjusted P value, padj). The top 20 enriched pathways (ranked by adjusted P value) were selected independently for each screen. Pathways associated with diseases (e.g., cancer, infection, neurodegeneration) were removed to focus on canonical signaling and metabolic pathways. Dot size indicates the number of genes involved; color reflects statistical significance (−log10 adjusted P value). e. GO pathway analysis showing enriched pathways among the top enriched and depleted gene hits from the tumour abundance screen. Dot size indicates gene count. P values were calculated using a one-sided hypergeometric test and adjusted for multiple comparisons using the Benjamini–Hochberg method. f. GO biological process analysis showing enriched pathways from tumour and spleen abundance screens. GO enrichment by one-sided hypergeometric test (clusterProfiler v4.10.1); significance based on the q value < 0.0005. Dot size represents -log10(q value). g. Enriched gene ranks comparing tumour versus spleen across two donors, P values determined using one-sided MAGeCK RRA test. Genes depleted in the tumour versus input comparison were excluded. The top 30 ranked genes are highlighted as black dots. Genes with known roles in GPCR signaling or previously characterized functions in cell trafficking and migration are labeled in red. h. Representative flow cytometry plots from the P2RY8 ligand bioassay using CXCL12 and a mixture of WT and P2RY8-GFP+ WEHI-231 cells, corresponding to Fig. 3d. i. CRISPR editing efficiency of P2RY8 in primary human T cells measured by ICE analysis (left, n = 2 donors) and flow cytometric analysis of P2RY8 protein expression (right, representative of 3 donors). j. Relative GGG levels in the supernatants of the indicated tumour cell lines compared to media control, as assessed by mass spectrometry analysis (n = 3 technical replicates). k. Flow cytometric histograms of CD19-28z CAR expression in AAVS1 and P2RY8-KO CAR T cells before transfer into tumour-bearing NSG mice. l-m. A375 tumour-bearing NSG mice treated with NY-ESO-1 TCR T cells 9 days after tumour implantation. Tumour growth (l), and CD39 expression in tumour-infiltrating TCR+CD8+ T cells (m) are shown. n = 5 mice per group. P values determined using two-tailed unpaired Student’s t-test (a, m). **P < 0.001; NS, not significant. Data are mean ± s.e.m. (a). Box plots show the median (centre line), 25th–75th percentiles (box bounds), minimum and maximum values (whiskers), and all individual data points (m). Dashed lines, individual mice; solid, group mean (l).

Source data

Extended Data Fig. 4 IFNγ-based genome-wide in vivo CRISPR screen nominates genes regulating intratumoral T cell effector function.

a. Flow cytometry gating strategy for sorting the top 20% IFNγ high (IFNγhigh) and bottom 20% IFNγ low (IFNγlow) TIL from A375low tumour-bearing NSG mice across 2 human T cell donors for the screen. TIL from the same donor were pooled for sorting. b. Distribution and GINI Index of sgRNA read counts from deep sequencing of TIL from IFNγhigh and IFNγlow populations across two donors in the in vivo IFNγ screens. c. GSEA of hallmark interferon gamma response and hallmark allograft rejection pathways in IFNγhigh versus IFNγlow dropout genes. Statistical significance was determined using Benjamini–Hochberg multiple test correction. Data were analyzed across 2 human T cell donors from the IFNγ screen. d. Comparison of in vivo primary human T cell screens, plotting log2FC values from IFNγ screen (IFNγhigh versus IFNγlow) on the Y axis vs. log2FC values from in vivo tumour abundance screen (tumour versus input) on the X axis. Data were analyzed across 2 human T cell donors. P values determined using one-sided MAGeCK RRA test. Genes shown in bold are those that also appear in the IFNγ-in vivo only section of the comparison plot in Fig. 4c. P < 0.05 and abs (log2FC) ≥ 0.2. e. Flow cytometric histograms of surface CD19-28z-CAR expression level in AAVS1, GNAS-KO and STUB1-KO CD19-28z CAR T cells prior to transfer into tumour-bearing NSG mice. f. CRISPR editing efficiency of the GNAS and STUB1 genes in primary human T cells, as determined by knockout score from ICE analysis (n = 2 donors).

Source data

Extended Data Fig. 5 In vivo sub-library IFNγ screens validate candidate regulators across tumour models.

a. Composition of the sub-library. b-c. In vivo IFNγ screens using NY-ESO-1 TCR T cells in wild-type A375 tumour-bearing NSG mice. Tumour-infiltrating T cells were isolated and sorted into the top 20% IFNγhigh and bottom 20% IFNγlow populations. Shown are correlation plots of gene-level log2 fold changes between IFNγhigh and IFNγlow populations across independent replicates from a single T-cell donor (b, n = 2 technical replicates per donor, 2 donors) and between two different T cell donors (c). Each replicate represents pooled T cells from two mice. r, Pearson’s correlation coefficient. d-f. In vivo sub-library IFNγ screens the in A375low tumour model. A375low tumour cells were subcutaneously engrafted into NSG mice followed by transfer of sub-library-edited polyclonal T cells. Tumour-infiltrating T cells were isolated and sorted into the top 20% IFNγhigh and bottom 20% IFNγlow populations. Shown are correlation plots comparing gene-level log2 fold changes across different T cell donors (d, n = 3 donors) and independent mice from a single donor (e, n = 3 mice). f, Volcano plots show enriched and depleted gene candidates associated with IFNγ expression in TIL, P values determined one-sided MAGeCK RRA test. Three T-cell donors and 8 mice were used for the screens; each mouse was sequenced individually. g. Comparison of gene-level enrichment between the A375low model and the NY-ESO-1 TCR T cell model. Each point represents one gene. h. Gene-level log2 fold change z-scores for GNAS across individual mice in the A375low model and independent replicates in the wild-type A375 model.

Extended Data Fig. 6 Analysis of GPCR genes upstream of Gαs across in vivo CRISPR screens and testing in preclinical models.

a-b. AAVS1 and GNAS-KO T cells were treated with a combination of the four agonists shown in Fig. 4g for 2 days, data shown are log2FC of MFI normalized to untreated AAVS1 control of indicated activation markers (a, n = 2 donors) and statistical analysis of OCR and ECAR (b, n = 4 technical replicates, representative of three donors). c. Incucyte-based killing assay comparing GNAS-KO and AAVS1 control HER2-28z CAR TIL in the presence (right) or absence (left) of the same four-agonist combination shown in Fig. 4g. TIL were isolated from the ascites of a second ovarian cancer patient (patient 2) and engineered to express HER2-28z CAR prior to co-culture with SKOV3 cancer cells. d. sgRNA log2 fold changes for GPCR genes upstream of Gαs and gene rank values from the in vivo IFNγ screen and tumour abundance screen. e. Tumour growth (left) and tumour size at day 90 post CAR T infusion (right) in CD19-A549 tumour-bearing NSG mice treated with AAVS1 control, GNAS-KO, PTGER4-KO, or GPR65-KO CD19-28z CAR T cells, 9 days after tumour injection (n = 6 mice per group). The data shown here were generated from the same in vivo validation experiment as Figs. 3h and 4e; shared groups are shown in different figures for narrative clarity. P values determined using two-tailed unpaired Student’s t-test in (e). **P < 0.01. Data are mean ± s.e.m. Dashed lines, individual mice; solid, group mean (e).

Source data

Extended Data Fig. 7 GNAS ablation in human T cells enhances tumour control in multiple solid tumour preclinical models.

a. CAR expression in AAVS1 and GNAS-KO CD19-28z CAR T cells from 2 human donors before transfer into NSG mice. b. Tumour growth (left) and survival (right) of NSG mice bearing subcutaneous CD19-A375 melanoma tumours treated with CD19-28z CAR T cells. n = 6 mice per group. Data are from one donor; the second donor is shown in Fig. 5a. c. Absolute number of CAR+ T cells per gram of WT-A375 or CD19-A375 tumours from the rechallenge experiment shown in Fig. 5b. n = 4 mice. d-f. Absolute tumour weight (d), CD4/CD8 proportion in CAR+ cells (e) and CAR+ CD4 and CD8 cell number per gram of tumour (f) from the experiment shown in Fig. 5c. n = 5 mice per group. g. MES-SA tumour growth upon tumour rechallenges. Naïve NSG mice were newly challenged and surviving mice from Fig. 5g that experienced complete responses were rechallenged with MES-SA tumour cells on day 70 post initial B7H3-BBz CAR T cell infusion. n = 5 naïve mice and 7 surviving GNAS KO-treated mice. h-i. Tumour growth (h) and survival (i) of NSG mice bearing MES-SA uterine sarcoma tumours treated with the indicated doses of AAVS1 or GNAS-KO B7H3-BBz CAR T cells. n = 6 mice for TRAC KO and 7 mice for all other groups. Highest-dose data are also shown in Fig. 5g. j-l. Transcriptomic profiling of B7H3-BBz CAR T cells recovered from MES-SA tumours. Heatmap showing differentially expressed genes between AAVS1 control and GNAS-KO CAR T cells (j, n = 6 replicates from 3 donors per group; each replicate represents pooled TIL from 2 mice). GSEA with Benjamini–Hochberg multiple test correction showing reduced enrichment of Tex signature genes in GNAS KO CAR T cells compared to AAVS1 controls (k). FPKM + 1 values of selected inhibitory molecules, lines connecting paired samples from the same donor (l), n = 3 donors per group. m-p. 1.5 × 106 OE19 esophageal adenocarcinoma cells were subcutaneously engrafted into NSG mice at day 0, followed by intravenous infusion of 0.5 × 106 (low dose) CLDN18.2 CAR T cells at day 8. Tumour growth curves (m), absolute number of CLDN18.2 CAR T cells per gram of tumour (n), CD4/CD8 proportions among CAR T cells (o) and CD39 expression in tumour-infiltrating CAR T cells (p) were analyzed on day 30 post CAR T cell infusion. n = 5 mice per group. q-x. Lung metastasis model using luciferase-A549 NSCLC cells and B7H3-BBz CAR T cells from an additional donor. Shown are: tumour growth measured by IVIS (q), lung masses and lung images along with representative lung H&E staining at day 90 after CAR T cell infusion (r); CAR+ frequencies (s), CD4/CD8 proportions (t), activation marker expression (u), ki-67 levels (v), cytokine production (w), and CD39 expression (x) in T cells isolated from lung. n = 6 mice per group. P values determined using two-way ANOVA in (b, g, h, m, q) for tumour growth analysis, Log-rank (Mantel-Cox) test in (b, i) for mice survival analysis, and two-tailed unpaired Student’s t-test in (c–f, n–p, r–x) and two-tailed paired Student’s t-test in (l). *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. Data are mean ± s.e.m. (c, h). Box plots show the median (centre line), 25th–75th percentiles (box bounds), minimum and maximum values (whiskers), and all individual data points (d-f, n-p, r, s-x). Dashed lines, individual mice; solid, group mean (b, g, m, q).

Source data

Extended Data Fig. 8 Consistent performance of GNAS-KO CAR T cells relative to other candidate edits across preclinical models.

a-b. CD19-A375 melanoma cells were subcutaneously engrafted into NSG mice, followed by intravenous infusion of a low dose of CD19 CAR-28z T cells. The experimental timeline (a, top), averaged tumour growth curves (a, bottom), and individual tumour growth curves (b) are shown. n = 5-6 mice per group as indicated. c-f. NSG mice were intravenously engrafted with luciferase-expressing A549 cells, followed by intravenous infusion of a low dose of B7H3-BBz CAR T cells (c, top left), averaged tumour growth curves (c, bottom left), survival curves (c, right), lung weights (d), lung images (e) and representative bioluminescent imaging of tumour burden in mouse lungs collected on day 99 post infusion (f). n = 6. g-h. MES-SA uterine sarcoma cells were subcutaneously engrafted into NSG mice, followed by intravenous infusion of a low dose of B7H3-BBz CAR T cells (g, top), averaged tumour growth curves (g, bottom), and individual tumour growth curves (h) are shown (n = 4-6 mice per group as indicated). i. Incucyte-based cytotoxicity assay of B7H3-BBz CAR T cells treated with different indicated gene edits. T cell killing was assessed in the absence (left) or presence (right) of the same four-agonist combination shown in Fig. 4g. n = 4 technical replicates per group, representative of three donors. P values determined using two-way ANOVA in (a, c, g) for tumour growth analysis, Log-rank (Mantel-Cox) test in (c) for mice survival analysis, and two-tailed unpaired Student’s t-test in (d). *P < 0.05, **P < 0.01, ****P < 0.0001. NS, not significant. Data are mean ± s.e.m. in (a, d, g, i) and geometric mean ± geometric sd in (c).

Source data

Extended Data Fig. 9 GNAS-KO human T cells show no evidence of dysregulated proliferation, antigen-independent killing, or in vivo toxicity.

a. Cell viability (left) and normalized fold change in absolute T cell numbers (right) of CD19-28z CAR T cells cultured with or without cytokines over time. Fold change was normalized to T cell number on day 0. n = 4 donors. b. Normalized growth of wild-type A375 melanoma cells cultured alone or co-cultured with AAVS1 or GNAS-KO CD19-28z CAR T cells at a 1:1 ratio. Tumour growth was monitored by Incucyte imaging of mKate+ A375 cells. c. Comparison of differentially expressed genes identified by RNA-seq in GNAS KO versus AAVS1 control CD19-28z CAR T cells in vitro in the presence (y-axis) or absence (x-axis) of the four-agonist combination shown in Fig. 4g. RNA-seq was performed using CAR T cells derived from three human donors and two independent GNAS sgRNAs. d. Incucyte-based cytotoxicity assay of CD19-28z CAR T cells treated with six independent CRISPRoff sgRNAs targeting GNAS or AAVS1 control. T cell killing was assessed in the absence (left) or presence (right) of the four-agonist combination shown in Fig. 4g, n = 5 technical replicates per group, representative of two donors. e-f. Flow cytometric histograms of CD19-28z CAR and endogenous TCR expression (e), and body weight of mice (f) from the experiment shown in Fig. 5f. g-h. Flow cytometric histograms of B7H3-BBz knock-in CAR and endogenous TCR expression (g), and body weights of mice (h) from the experiment shown in Fig. 5g. i-m. Flow cytometric histograms of B7H3-BBz knock-in CAR and endogenous TCR expression (i) and body weights (j) from the experiment in Fig. 5i. H&E staining of tissues (k, representative of 4 mice per group) and spleen images (l) from mice collected on day 90 after B7H3-BBz CAR T cell infusion, from the experiment in Extended Data Fig. 7q. Spleen and lung images (m) from mice collected on day 156 after B7H3-BBz CAR T cell infusion, from the experiment in Fig. 5i. Data are mean ± s.e.m. Dashed lines, individual mice; solid, group mean (f-j).

Source data

Extended Data Fig. 10 GNAS and P2RY8-GNAS double deletion enhances tumour infiltration, and the double KO shows no evidence of increased in vivo toxicity.

a. Immunofluorescence images of CAR T cell infiltration. Subcutaneously CD19-A549 tumour-bearing NSG mice with 5 × 105 CD19-28z CAR T cells were divided into three CAR T cell treatment groups: AAVS1 control, GNAS KO, and P2RY8-GNAS double KO. Tumours were harvested at day 35 after CD19-28z CAR T cell injection, and immunofluorescence staining was performed to detect infiltrating CAR T cells (cyan). T cells were labeled with anti-human CD8α-biotin and anti-human CD4-biotin, followed by streptavidin-BV421, which marks the transferred human CAR T cells in this model. Tumour cells were identified by mKate transgene expression (red). Scale bar, 200 μm. n = 4 AAVS1, 5 for other groups. b. Body weight changes from the experiment shown in Fig. 5m. n = 6 per group. c. Tumour growth (left) and body weight (right) of CD19-AsPC-1 tumour-bearing NSG mice treated with AAVS1 control or P2RY8-GNAS double KO CD19-28z CAR-T cells, beginning 25 days after tumour injection. n = 5 mice per group AAVS1/P2RY8-GNAS double KO. This experiment was performed concurrently with the experiments shown in Figs. 3i and 5f, and therefore share the same AAVS1 control group. d-e. In vivo safety assessment of HER2-28z CAR TIL in an ovarian cancer model. NSG mice were engrafted intraperitoneally with 0.75 × 106 luciferase-expressing SKOV3 cells (day 0) and treated with 1 × 106 HER2-28z CAR TIL (day 16). This was a tumour-clearing dose, which enabled long term safety assessment. Shown are tumour growth by bioluminescence imaging (d, left), body weight changes (d, right), as well as H&E staining of tissues collected on day 153 (e, representative of 3 mice per group). TIL: tumour-infiltrating lymphocytes from an ovarian cancer patient, n = 6 mice per group. f. Schematic summarizing the in vivo CRISPR screening platform and mechanisms regulating human T cell function in solid tumours. Dashed lines, individual mice; solid, group mean (c-d).

Source data

Supplementary information

Supplementary Figure 1 (download PDF )

Representative gating strategies for flow cytometric analysis of T cells isolated from tumour-bearing NSG mice.

Reporting Summary (download PDF )

Supplementary Tables (download ZIP )

Supplementary Table 1. MAGeCK analysis results from in vivo genome-wide CRISPR knockout abundance screens. Supplementary Table 2. MAGeCK analysis results from in vivo genome-wide CRISPR knockout IFNγ screens. Supplementary Table 3. crRNA sequences used in this study. Supplementary Table 4. Information on primary human T cell donors used in in vivo screening and validation studies. Supplementary Table 5. sgRNA list and screening results from the sub-library screening. Supplementary Table 6. Differential gene expression analysis from RNA-seq.

Peer Review File (download PDF )

Source data

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Liu, Q., Chen, P.A., Urs, E. et al. In vivo genome-wide CRISPR screens of human T cells in solid tumours. Nature (2026). https://doi.org/10.1038/s41586-026-10906-9

Download citation

  • Received:

  • Accepted:

  • Published:

  • Version of record:

  • DOI: https://doi.org/10.1038/s41586-026-10906-9