Multimodal stimulation screens reveal unique and shared genes limiting T cell fitness

Summary Genes limiting T cell antitumor activity may serve as therapeutic targets. It has not been systematically studied whether there are regulators that uniquely or broadly contribute to T cell fitness. We perform genome-scale CRISPR-Cas9 knockout screens in primary CD8 T cells to uncover genes negatively impacting fitness upon three modes of stimulation: (1) intense, triggering activation-induced cell death (AICD); (2) acute, triggering expansion; (3) chronic, causing dysfunction. Besides established regulators, we uncover genes controlling T cell fitness either specifically or commonly upon differential stimulation. Dap5 ablation, ranking highly in all three screens, increases translation while enhancing tumor killing. Loss of Icam1-mediated homotypic T cell clustering amplifies cell expansion and effector functions after both acute and intense stimulation. Lastly, Ctbp1 inactivation induces functional T cell persistence exclusively upon chronic stimulation. Our results functionally annotate fitness regulators based on their unique or shared contribution to traits limiting T cell antitumor activity.


In brief
Lin et al. perform multimodal genomewide CRISPR knockout screens in primary CD8 T cells for genes controlling fitness upon differential stimulation.They identify Dap5, Icam1, and Ctbp1, which are functionally annotated and characterized based on their unique or shared contribution to traits limiting T cell antitumor activity.

Primary CD8 + T cells Tumor cells
In vitro functional readout

A D E F G C B
8][29] This can be driven by distinct stimulation contexts, resulting in cell differentiation into several states.][32] Tumor-specific T cells face repetitive antigenic stimulation in the TME, which can be intense and trigger activation-induced cell death (AICD).3][44] Therefore, blocking AICD may improve T cell survival in tumors. 45][49] Hence, harnessing cell proliferation represents an opportunity to obtain sufficient tumor-reactive T cells.
Chronic antigen stimulation accounts for a third signal hampering T cell fitness, limiting ''effector-persistence'' or dysfunction.][81] Thus, interfering with genes limiting effector persistence under chronic stimulation may allow for more durable immunotherapy responses.
The multifactorial causes of T cells losing their fitness upon differential stimulation are the subject of intense study, as they impede tumor control.While several players were identified previously by CRISPR screening,   it has not yet been systematically addressed whether critical factors control only specific aspects or simultaneously regulate multiple T cell fitness features. This i what we set out to study here, in an unbiased, genome-wide fashion, investigating three different stimulation modalities, namely: intense, acute, and chronic stimulation.

RESULTS
Multimodal function-based genome-wide CRISPR knockout screens for genes contributing to T cell fitness upon differential stimulation We set out to recapitulate the aforementioned three key processes determining T cell antitumor activity.First, in immunocompetent mice bearing ovalbumin (OVA)-expressing melanomas, we identified endogenous OVA-specific CD8 tumor-infiltrating lymphocytes (TILs) showing reduced viability compared to non-specific T cells (Figure S1A), in agreement with AICD by intense antigen stimulation.Second, in an ACT mouse model, we observed increased proliferation of transferred OVA-specific T cells in tumors compared to spleens 3 days after ACT (Figure S1B), indicating rapid proliferation upon antigen stimulation.Third, transferred T cells expressed higher levels of inhibitory receptors in tumors compared to spleens at tumor endpoint, showing exhaustion induced by chronic antigen stimulation (Figure S1C).
These results prompted us to use unbiased CRISPR-Cas9 knockout screens to uncover genes in primary murine CD8 T cells that contribute to fitness loss upon differential stimulation.The TME comprises several factors influencing T cell responses.To identify players regulating specific effector traits under defined stimulation contexts, to avoid confounders, and to ensure robust library coverages, we designed three independent genome-scale CRISPR-Cas9 knockout screens: intense (two successive 24 h-TCR stimulations, enriching for sgRNAs promoting survival), acute (single 24 h-TCR stimulation followed by 4 days proliferation, enriching for sgRNAs promoting proliferation), and chronic (repetitive tumor-antigen stimulation for 11 days, enriching for sgRNAs promoting persistence) (Figure 1A).
We crossed Cas9-GFP 108 mice with OT-I mice 109 and isolated naive OT-I/Cas9 cells.After 48 h-priming with anti-CD3 and anti-CD28, a genome-wide sgRNA library 110 was retrovirally transduced.Cells were pharmacologically selected for 6 days.A t 0 library reference sample was harvested 24 h post-selection and a pre-reactivation sample (primed effector cells) was taken right before the start of all screens to confirm dropout of essentialgene-targeting sgRNAs. 111(Figures S1D and S1E; Table S1).
For the intense stimulation/survival screen, library-containing cells (referring to CD8 effector cells, unless otherwise specified) were challenged twice with 24 h-anti-CD3 stimulation, causing cell viability to drop progressively, indicating strong survival pressure (Figure S1F).For the acute stimulation/proliferation screen, one-time 24 h-anti-CD3 stimulation was applied.Stimulated cells were cultured for additional 3 days, allowing cell  S1).(D) Enrichment of individual sgRNAs targeting genes identified from published T cell screens.Numbers above plots indicate signed -Log 10 (MAGeCK score).(E) GSEA of GO biological process from screen hits (Table S1).FDR: false discovery rate.(F) GSEA of CD8 lineage gene sets 107 from screen hits (Tables S1 and S5).NES: normalized effect size.(G) Numbers of overlapping genes from top 50 hits of each screen.Genes are listed by average effect size (Table S1).expansion (Figure S1G).For the chronic stimulation/persistence screen, T cells were continuously stimulated by D4M.OVA mouse melanoma cells 112 for 11 days at a fixed T cell:tumor cell ratio.A resting group was included where medium was refreshed without tumor cells, avoiding screening for proliferation regulators independent of chronic stimulation.As expected, 11 days chronic stimulation triggered the upregulation of multiple exhaustion markers (Figure S1H), [56][57][58]113,114 while inducing apoptosis 115 (Figure S1I) and terminal differentiation 116 (Figure S1J). Furtherore, these cells exhibited reduced cytotoxicity compared to resting cells upon restimulation (Figures S1K and  S1L), adopting a dysfunction phenotype. Sysematic flow cytometry analysis of cells under all screen conditions confirmed and extended the previously described phenotypes (Figures 1B  and S1M).Although we cannot exclude confounding signals contributing to the final population in each setting, the characterization of our screen settings supports their key phenotypes (survival, proliferation, and persistence) serving as discriminating factors.
Cells were collected at each screen endpoint, genomic DNA was isolated, and sgRNAs were PCR-amplified and sequenced.sgRNA enrichment from output samples was compared to either pre-reactivation (intense and acute screens) or resting (chronic screen) samples by MAGeCK analysis 117 (Figures 1C; Table S1).We identified several regulators discovered previously, including Arid1a, Rasa2, Ccnc, and Zc3h12a (alias Regnase-1 118 ). 90,92,102,104Moreover, sgRNAs targeting Fas, a key positive regulator of cell death and apoptosis, were enriched particularly in the intense stimulation screen 34,36 (Figures 1C and  1D), all illustrating the screen robustness.Gene ontology (GO) term 119,120 gene set enrichment analysis (GSEA) 121,122 with MAGeCK-ranked hits showed enrichment of expected biological processes related to apoptosis and proliferation for the intense and acute stimulation screens, respectively, and to activation for both (Figures 1E; Table S1).As the GO term database lacks exhaustion signatures, we derived gene sets from published sin-gle-cell RNA-seq (scRNA-seq) data 107 (Table S5).Highlighting the relevance of the chronic stimulation screen, the exhaustion signature (Chronic_>D15) was enriched exclusively in the chronic setting (Figures 1F; Tables S1 and S5).Integrating the top 50 enriched genes from all screens, we identified four shared hits: Trp53, Eif4g2 (alias Dap5 123 ), Serf2, and Sh2d3c, whose depletion positively influenced T cell fitness upon all three stimulation modes (Figures 1G; Table S1).

Dap5 inactivation alleviates global inhibition of effector T cell fitness and enhances tumor-killing capacity
To assess clinical relevance of overlapping hits-Trp53, Dap5, Serf2, and Sh2d3c-(Figure 2A) from all three screens, we queried a cohort of patients with melanoma receiving TIL therapy, 124,125 where RNA-seq was performed for TIL products prior to infusion (Besser M.J., RNA-seq data unpublished).We stratified patients receiving TIL with the highest and lowest expression of the indicated genes.Patients receiving TILs expressing low DAP5 or SERF2 showed significantly longer overall survival (OS) (Figures 2B; Table S2).No significant effect was seen in TILs with low SH2D3C or TP53 expression (Figure S2A; Table S2).Querying the role of DAP5 and SERF2 in cell exhaustion, we extended our analysis to 49 publicly available scRNAseq data from pan-cancer cohorts from the TISCH database. 126xpression of both genes was significantly higher in the exhausted CD8 subset than in conventional CD8 cells (Figure S2B; Table S2).Thus, we identified two fitness genes, DAP5 and SERF2, whose ablation enhances T cell fitness under all three stimulation types, and whose expression levels correlate with exhaustion and unfavorable TIL response.
To validate Dap5, OT-I/Cas9 cells were retrovirally transduced with sgRNA targeting Dap5.Dap5-KO T cells showed significantly higher cell count under all stimulation settings (Figure 2C), including chronic CD3 stimulation (Figure S2C), confirming that Dap5 inactivation improves general T cell fitness.It was important to determine whether Dap5-KO impacts effector functions.S2).(I) Absolute Log 2 (fold-change) of CD3 stimulation-induced upregulated and downregulated genes, related to (H) (Table S2).(J) Flow cytometry analysis on cells, with or without 24 h-CD3 stimulation, analyzed with one-way ANOVA, followed by a Tukey post-hoc test (n = 9 biological replicates).We co-cultured Dap5-KO or Ctrl T cells with different murine melanoma cell lines expressing TCR-matched antigens (OT-I:OVA 109 or Pmel:gp100 127,128 ).In line with our screen result, Dap5-KO T cell number increased when cultured with tumor cells (Figures 2D and S2D), accompanied by a superior tumor cell elimination (Figures 2E and S2E).This result demonstrates an enhanced effector function upon Dap5 ablation, which is not restricted to TCR-antigen specificity (Figure S2E).To extend these observations, we inactivated DAP5 in human CD8 cells expressing MART-1 TCR 129 for functional assessment (Figure 2F).We observed that its inactivation in human T cells also resulted in increased cell numbers upon CD3 stimulation (Figure S2F), and in co-culture with MART-1 tumor cells (Figure 2G).These data indicate that Dap5 inactivation enhances T cell fitness and efficacy under various stimulation conditions.
1][132][133][134][135] To understand its role in regulating translation in T cells, we first assessed the effect of Dap5 ablation on global translation.We incubated Ctrl and Dap5-KO T cells with methionine analog L-homopropargylglycine (HPG) that is incorporated into newly synthesized proteins, to determine overall translation rate.Dap5-KO increased global translation (Figure S2G).Consistently, a strongly reduced level of the translation repressor 4E-BP1 136 was observed (Figure S2H), together suggesting that DAP5 may act as a translational inhibitor in T cells.
To dissect whether this increase in translation is driven by a subset of highly translated mRNAs or rather a global effect, we performed polysome and total mRNA sequencing.A comparison between the polysome and translational profiling revealed similar overall translation efficiency between Ctrl and Dap5-KO cells (Figure S2I; Table S2), suggesting a global increase in translation.We compared the transcriptomic profiles of Dap5-KO and Ctrl T cells with or without CD3 stimulation, showing increased expression of cell cycle genes alongside a moderate immune activation program (Figures 2H; Table S2).Furthermore, Dap5-KO cells had dampened global transcriptional changes upon stimulation (Figures 2I; Table S2), accompanied by a significantly lower FAS expression (Figure 2J).By combining DAPI and annexin V staining, we observed a significant reduction of apoptosis in Dap5-KO cells (Figures 2K and S2J), accompanied by a slight increase in viability after TCR stimulation (Figure S2K).Antibody-mediated blockade of FasL benefited sgCtrl-expressing cells upon stimulation while having a minor effect on Dap5deficient cells, supporting the notion that Fas signaling is diminished in Dap5-KO cells, thereby aiding their survival (Figure S2L).Dap5 ablation also resulted in 2-fold downregulation of PD-1 (Figure 2L), but did not affect IFNg production upon stimulation (Figure S2M).
These results prompted us to investigate whether Dap5 inactivation benefits T cell function in vivo, where multiple challenges must be overcome for effective tumor control.We first examined whether Dap5 ablation increases T cell numbers within the tumor in an in vivo competition experiment (Figure 2M).In line with our in vitro data, Dap5-KO cells were more abundant in tumors compared to Ctrl cells (Figures 2N and S2N), with increased proliferative activity (Figure S2O).To assess the clinical potential of Dap5 inactivation, we carried out an ACT therapy in a B16.OVA melanoma mouse model.Mice underwent total body irradiation (TBI) as a lymphodepleting regimen before receiving either Ctrl or Dap5-KO T cells, 137 and tumor growth and survival were monitored (Figure 2O).Mice that received Dap5-KO cells showed improved tumor control (Figures 2P and S2P) and survival (Figure S2Q).Our observations both in vitro and in vivo demonstrate that Dap5-KO in effector cells enhances global translation while suppressing FAS expression, together contributing to improving fitness upon stimulation, boosting their antitumor efficacy.
Loss of Icam1-mediated homotypic T cell interactions amplifies CD8 T cell expansion and improves effector functions shortly after TCR stimulation Next, we characterized overlapping hits from the acute and intense stimulation screens (Figure 3A).Their ablation rendered T cells more resistant to AICD while boosting proliferation upon Figure 3. Loss of Icam1-mediated homotypic T cell interactions amplifies CD8 T cell expansion and improves effector functions shortly after TCR stimulation (A) Overlapping genes from top 50 hits from each screen, genes are ranked by average effect size.(B) STRING protein-protein interaction analysis of shared targets from the two boxes in (A) (32 genes).Interactions include direct (physical) and indirect (functional) associations.(C) Kaplan-Meier OS curves of patients receiving TIL therapy 124,125 S3).(H) Proteomic STRING enrichment analysis of differentially expressed proteins comparing Icam1-KO with Ctrl T cells after 24 h-CD3 stimulation, showing top enriched GO biological process (ranked by enrichment strength (Log 10 (observed/expected)) with FDR < 0.1 (Table S3).stimulation, producing a sufficient cell pool facilitating tumor control.For prioritization, we performed STRING analysis 138 with all genes in this category (n = 32).This identified multiple genes involved in cell-cell interaction or extravasation pathways (Figures 3B and S3A; Table S3), suggesting a crucial role of cellcell interaction in regulating T cell fitness upon TCR stimulation.Among candidates involved in cell-cell interaction regulation, integrin subunit alpha l (Itgal), encoding the integrin alpha L chain that represents one-half of the lymphocyte function-associated antigen 1 (LFA1) heterodimer, 139 and its ligand ICAM1 140 showed up as top enriched hits.4][145][146][147][148] Patients receiving TILs with lower ICAM1 expression showed significantly longer OS (Figures 3C; Table S2), implying its clinical relevance in cancer immunotherapy.
To validate the role of homotypic T cell interaction regulated by Icam1 upon TCR stimulation, Icam1 was ablated from T cells and stimulated with CD3 antibody for 24 h.Immediately after, as reported, Ctrl cells formed dense, homotypic cell-cell aggregates 147 but not Icam1-KO cells (Figure 3D).After both intense and acute stimulation, Icam1-KO T cells showed higher viable cell counts (Figure 3E).More importantly, their tumor cell-killing ability was significantly enhanced (Figure 3F), indicating a positive impact of Icam1 inactivation on T cell effector function.
To dissect the underlying mechanism, we performed transcriptomic analysis on Icam1-KO T cells upon TCR stimulation.Prior to stimulation, no difference was observed with Ctrl cells.However, upon stimulation, gene transcripts involved in effector function were markedly increased in Icam1-KO cells, indicating a stronger effector phenotype (Figures 3G; Table S3).Simultaneous proteomic analyses showed a strong enrichment of processes involved in T cell activation in Icam1-KO cells (Figure 3H; Table S3), in line with the transcriptomic profiles.Additional flow cytometry analysis revealed enhanced cytokine production following TCR stimulation by Icam1-KO (Figure 3I).Notably, Icam1-KO cells did not show higher PD-1/LAG3 co-expression compared to control cells 7 days post-stimulation (Figure S3B).Furthermore, higher viable cell counts were accompanied by an increased KLRG1-/CD62L+ population of Icam1-KO cells 7 days after stimulation (Figure S3C), suggesting a higher potential for memory precursor development. 149,150o understand whether the induced effector function triggered by Icam1-KO is achieved by targeting cell-cell interaction, we studied the different roles of the extracellular and intracellular domains of ICAM1 in T cells.We re-expressed either wild-type Icam1 (wtIcam1) or a Icam1 mutant lacking its intracellular domain (dcytIcam1) in Icam1-KO cells.After stimulation, dcytI-cam1 cells exhibited a similar clustering phenotype as Ctrl cells (Figure S3D), indicating that the intracellular domain of ICAM1 does not contribute to cell-cell interaction.To determine whether the extracellular domain alone is sufficient for reversing the positive effects of Icam1 loss, we assessed viable cell counts 4 days after CD3 stimulation.Of note, wtIcam1 cells showed higher ICAM1 expression levels than parental cells (Figure S3E), which led to T cell hyperclustering, even prior to stimulation (Figure S3D).This also led to impaired cell viability (Figure 3J) and a reduced memory precursor population (Figure S3F).These results reveal that the clustering behavior of T cells correlates with their cell viability, effector function, and phenotype.
To translate these findings to a more clinically relevant setting, it is noteworthy that humans express five ICAM family members (ICAM1-5).2][153][154] In contrast, murine CD8 cells express almost exclusively Icam1, and the Icam3 gene was inactivated in mice during evolution (Figure S3G). 155Therefore, we generated single, double, and triple KOs in human CD8 T cells. 1 week after CD3 stimulation, the double (ICAM1 and ICAM3) and triple (ICAM1, ICAM2, and ICAM3) knockouts resulted in significant increases in viable cell counts (Figures 3K and S3H).
Based on our data with human T cells, we reasoned that targeting LFA1 by a blocking antibody (CD11a) may prevent clustering, which is likely more efficient than blocking ICAM1 (CD54) alone.Indeed, treatment of human primary CD8 cells with CD11a antibody resulted in better expansion, outperforming CD54 antibody treatment (Figure 3L).Thus, preventing homotypic T cell interactions by targeting the ICAM-LFA1 axis recapitulates the improved effector phenotype achieved by genetic manipulation, which may merit exploration for cell therapy.
Ctbp1 ablation induces T cell persistence exclusively under chronic stimulation, associated with reduced ZEB2/T-bet-dependent terminal differentiation Lastly, we wished to characterize genes that, instead, uniquely contribute to only a single T cell fitness setting: chronic stimulation.From the top hits exclusively identified in the chronic stimulation screen (Figure 4A), multiple genes have been reported previously, either with potential for cancer immunotherapy (Regnase-1, 102 Cblb, 159 and Ccnc 92 ) or associated with exhaustion (Cd69) 160 or terminal differentiation (Zeb2). 158This encouraged us to focus on this group to identify targets that upon inactivation could prolong tumor-specific T cell persistence and sustain effector function.
We took the top 25 genes from MAGeCK analysis, from which we prioritized the validation of the 10 genes showing the largest log 2 fold-change (LFC) and scoring with 4/4 sgRNAs in the library (sgRNAs ranking below the alpha cutoff in the MAGeCK analysis) 117 (Figures 4B and S4A).Regnase-1-KO cells were strongly enriched, corroborating the screen system. 102][163] sgRNAs perturbing the top 10 genes were transduced into OT-I/Cas9 cells.Cells were then either chronically stimulated by adding fresh D4M.OVA cells at a fixed T cell:tumor cell ratio or refreshed without adding tumor cells (''resting'') for 11 days, identical to the screen setting (Figure 1A).8/10 of the hits were validated by increased viable cell counts (Figure 4C).From the parallel analysis in the resting condition, depletion of four hits (including Tp53) resulted in either increased or decreased viable cell counts in the absence of chronic tumor-antigen stimulation (Figure S4B), which were excluded from further analysis.
For prioritization, T cells ablated for top validated hits (stimulation-dependent) were exposed to longer and stronger chronic tumor-antigen stimulation for 3 weeks, causing clear exhaustion S4C-S4E).Ctbp1-KO cells had acquired the most pronounced increase in cell count (Figure 4D).This effect was confirmed with additional sgRNAs (Figure 4E), chronic CD3 stimulation (Figure S4F), across tumor types (Figure 4F), and a different matching antigen-TCR pair (Figure 4G).Corroborating these results in human CD8 cells, we found that CTBP1 ablation again caused a stronger persistence phenotype after chronic stimulation with antigen-matched tumor cells (Figures 4H  and S4G).
To understand the role of CTBP1 in T cell persistence, we performed transcriptomic profiling after 3 weeks of chronic stimulation.Many DNA replication and cell cycle regulating genes were upregulated in Ctbp1-KO T cells, whereas genes involved in cell differentiation and responses were downregulated (Figures 4I; Table S4).In line with our chronic stimulation screen results, we found that an exhaustion signature (Figure 1F) 107 was negatively enriched in Ctbp1-KO cells (Figure 4J), implying that Ctbp1 inactivation slows down cell exhaustion upon chronic stimulation.
Since CTBP1 functions as a transcriptional corepressor, 164 we set out to identify potential interactors in T cells upon stimulation.We performed immunoprecipitation mass spectrometry (IP-MS) with buffers differing in stringency, identifying several potential CTBP1 interactors.As expected, ZEB1 and ZEB2, previously established CTBP1 interactors, 165,166 showed up as top hits (Figures 4K and S4H; Table S4).They are reciprocally expressed during CD8 T cell development; 167 Zeb2 is crucial in promoting terminal effector differentiation, whereas Zeb1 is required for maintaining the homeostasis of memory cells. 157,158,168,169gether with our IP-MS data, these results suggest an important role for CTBP1 in regulating ZEB1/ZEB2-mediated T cell differentiation.To investigate whether ZEB1 and ZEB2 control the phenotype of Ctbp1-KO T cells, we performed GSEA analysis.An enrichment of both Zeb1-knockout (AIGNER_ZEB1_ TARGETS) 156 and Zeb2-knockout signatures 157 was found (Figures 4L; Table S5).Moreover, Ctbp1-KO T cells adopted a less terminal differentiated effector phenotype, as measured by gene sets from two independent studies on effector differentiation in either a chronic 73 (Figure S4I; Table S4) or acute 170 (Figure S4J; Table S4) LCMV infection mouse model.These data suggest that CTBP1 cooperates with ZEB2 to regulate effector terminal differentiation.
Next, we investigated how CTBP1 affects terminal differentiation and effector status of CD8 T cells.We observed that T-box transcription factor Tbx21 (T-bet), a key regulator of antigeninduced effector function, [171][172][173] was one of the top upregulated genes in the transcriptomic analysis (Figure S4K).This was confirmed in Ctbp1-KO cells after chronic stimulation (Figure 4M).T-bet, together with ZEB2, drives terminal differentiation by promoting terminally differentiated effector (TE) genes while repressing memory precursor (MP) genes. 158Using previously reported signatures, 158 we found that 3 weeks after chronic tumor-antigen stimulation, multiple T-bet-repressed/ZEB2dependent MP genes (Figures 4N and 4O; Table S4), as well as T-bet-induced/ZEB2-independent TE genes (Figures S4L and S4M; Table S4), were significantly higher expressed in Ctbp1-KO cells.These results indicate a collaborative role of CTBP1 together with T-bet and ZEB2 in regulating T cell (A) Overlapping genes from top 50 hits of each screen, listing top exclusive genes from the chronic stimulation screen (ranked by effect size).Genes selected for validation are in bold.(B) -Log 10 (MAGeCK score) for all genes in the chronic stimulation screen.(C) In vitro validation of top-ranking hits exclusively from the chronic stimulation screen, showing relative viable T cell count after 11 days chronic D4M.OVA stimulation as in the screen.Top 25 genes with 4/4 enriched sgRNAs were re-ranked by effect size (LFC, log 2 (fold change)), and top 10 genes were selected for validation.Cell count fold-change was normalized to resting condition (Figure S4B).Analyzed with one-way ANOVA, followed by a Dunnett post-hoc test from three biological replicates with two different sgRNAs per replicate (n = 3x2).(D) Relative viable cell counts of T cells expressing indicated sgRNAs after prolonged (3 weeks) chronic D4M.OVA stimulation.Genes with significantly increased cell count after chronic stimulation (Figure 4C), but without proliferation (dis-) advantage (+/À 25% change) under resting condition (Figure S4B), were selected for prolonged chronic in vitro stimulation.Analyzed with one-way ANOVA, followed by a Dunnett post-hoc test from three biological replicates with two different sgRNAs per replicate (n = 3x2).S4).(J) GSEA of exhaustion signature, Chronic_>D15 (UP in Tex, as in Figure 1F), comparing Ctbp1-KO to Ctrl T cells after chronic stimulation.(K) Immunoprecipitation mass spectrometry (IP-MS) analysis of CTBP1 from wt OT-I/Cas9 cells after CD3 stimulation (n = 2 independent experiments with different IP buffers, see also Figure S4H; Table S4).Proteins identified from both independent IP-MS are in pink.1% Triton X-100 IP buffer was used.(L) GSEA of ZEB1-KO_UP (AIGNER_ZEB1_TARGETS) 156 and ZEB2-KO_UP 157 signatures (UP in ZEB1 or ZEB2 KO cells), comparing Ctrl and Ctbp1-KO T cells after chronic stimulation (Table S5).(M) Flow cytometry analysis of indicated T cells after 3 weeks chronic stimulation, analyzed with two-tailed paired t test (n = 4 biological replicates).(N) Expression of memory precursor (MP) signature genes known to be repressed by T-bet but either dependent or independent of ZEB2 regulation, 158 related to I) (Table S4).(legend continued on next page) differentiation: Ctbp1-KO induces T-bet expression, thereby enhancing effector function.In contrast, Ctbp1-KO restrains ZEB2's inhibitory function on MP genes to promote effector terminal differentiation.Thus, we hypothesize that Ctbp1-KO in matured effector cells endows them with a hybrid phenotype with enhanced effector function but delayed terminal differentiation, prolonging functional-effector persistence.

Blocking CTBP1-mediated terminal T cell differentiation preserves T cell effector function and enables long-term tumor control
To examine the functionality of Ctbp1-KO T cells, and to test our hypothesis, we determined their tumor-killing capacity and effector phenotype after 3 weeks chronic tumor-antigen stimulation in vitro.Ctbp1-KO T cells showed improved tumor-eliminating capacity post-chronic stimulation, but not in a resting condition (Figures 5A and S5A), consistent with the chronic stimulation screen and extending the prioritization results (Figure S4B).This was paralleled by enhanced cytokine production (Figures 5B and S5B), stronger cell proliferation capacity upon restimulation (Figure 5C) and upregulation of IL-2 receptor (Figure S5C).Moreover, although more activated, Ctbp1-KO cells did not show a pronounced exhaustion phenotype (Figures 5D  and S5D), and they were more resistant to cell apoptosis (Figure 5E).On the other hand, several terminal differentiation markers were lower expressed (Figures 5F and S5E), 64,116 whereas central memory markers [174][175][176][177][178][179][180] were upregulated (Figure 5G, S5F and S5G) in Ctbp1-KO cells after chronic stimulation.These data support our hypothesis that depleting Ctbp1 in effector T cells causes them to retain a hyperactivated yet less terminal differentiated status, indicating the potential of creating long-lasting effectors.This prompted us to set up an in vivo model to study prolonged tumor-antigen stimulation in immune-competent mice.As the classic ACT tumor model can be influenced by a short-term proliferation advantage, we developed a prolonged tumor-antigenstimulation ACT model.Transferred T cells were challenged in vivo by multiple rounds of irradiated-tumor cell injection prior to viable tumor cell transplantation, extending the chronic stimulation duration (Figure 5H).This allowed us to focus on the long-term persistence of transferred T cells, minimizing the proliferation confounder during early expansion.To avoid possible rejection of OT-I/Cas9 cells, Cas9-expressing recipient mice (C57BL/6J background) were used.7 days post viable tumor cell injection, transferred T cells were harvested from tumors and lymph nodes.Ctbp1-KO cells isolated from tumors, but not from lymph nodes, showed an increased population of less terminally differentiated cells, displaying central memory marker expression (Figures 5I and S5H), in line with our observation in vitro.Furthermore, mice receiving Ctbp1-KO T cells showed better tumor control 20 days after viable tumor cell injection (44 days post ACT) (Figures 5J and 5K).We observed 9/18 complete responses for ACT with Ctbp1-KO cells compared to 1/18 for control cells (Figure S5I), resulting in significantly longer overall tumor-free survival (Figure 5L).
Lastly, to investigate a potential role of CTBP1 in regulating CD8 cells in cancer immunotherapy, we analyzed scRNA-seq data from TILs of patients with melanoma treated with ICB. 181e found a significant correlation between low CTBP1 expression in CD8 TILs and favorable ICB response (Figure S5J).Similarly, we observed a trend of better survival when patients received TIL expressing low levels of CTBP1 (Figure S5K; Table S2, Besser TILs cohort). 124,125In line with our findings for DAP5 and SERF2, CTBP1 expression was significantly higher in the exhausted CD8 subset than in conventional CD8 cells in pan-cancer cohorts from the TISCH database 126 (Figures 5M; Table S2).Both published patient data and our own in vivo data support our finding that Ctbp1 inactivation in effector T cells reinforces their effector function, delaying cells from terminal differentiation and exhaustion.The enhanced effector persistence allows for improved tumor control and prolonged survival in a chronic tumor-antigen-stimulation mouse model, meriting therapeutic exploration of Ctbp1 inactivation for T cell therapy.

DISCUSSION
In this study, we uncovered and compared genes either exclusively, or commonly, contributing to T cell fitness under different modes of TCR stimulation.As the complex and dynamic nature of the TME has proven challenging to single out key factors, while maintaining high library coverage in vivo, we opted for a multimodal functional screen approach at genome-scale in defined settings.We performed three genome-wide CRISPR-Cas9 knockout functional screens in CD8 T cells upon different stimulations: intense, acute, and chronic, covering key aspects of effector biology, namely: survival, proliferation, and persistence.We identified several regulators previously reported by (M) CTBP1 expression in CD8Tex and CD8T cells from 49 scRNA-seq datasets (pan-cancer).Expression level was directly derived from TISCH2 website analysis 126 (Table S2).TPM: transcripts per million.Analyzed with Wilcoxon test (n = 49 independent datasets).
Error bars indicate SD unless otherwise specified.*p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001.others, which not only confirmed their critical roles in controlling T cell antitumor efficacy, but extend those data by demonstrating their differential involvement in common, or specific, aspects determining T cell fitness.Furthermore, we uncovered, validated, and characterized several regulators not previously reported, which harness T cell fitness under either common (Dap5 and Icam1), or exclusive (Ctbp1) T cell-stimulating conditions (Figure 6).
We identified Dap5 as a critical negative regulator of T cell fitness under all three stimulation conditions.Its inactivation protects cells from cell apoptosis immediately following TCR stimulation, thereby increasing proliferation, cumulatively allowing for improved tumor control in vivo.3][184][185] Dap5 ablation resulted in a global increase of translation, accompanied by a decline in 4E-BP1 protein.Polysome profiling indicated that the vast majority of mRNA was translated more efficiently.Cell cycle-regulating genes, such as Ccnb1, Mki67, Ccne2, and Cenpe, [186][187][188][189] were induced in Dap5-KO T cells both preand post-stimulation, whereas activation-induced immunosuppressive genes, such as Nr4a1, Pdcd1, Fas, and Tnfsf4, 34,56,190,191 were suppressed; suppression of both PD-1 and FAS was confirmed at the protein level.These observations may explain the phenotype induced by Dap5 inactivation: an activated cell cycle program at baseline allows cells to achieve their effector status, resulting in attenuated activation and protection from AICD and dysfunction.Simultaneously, the increased capacity for global translation by Dap5 ablation fuels the already activated and rapidly dividing cells.These traits are consistent with the observed fitness benefit upon all three stimulations.In combination with our TIL data, our results merit exploring the therapeutic benefit of lowering Dap5 expression for T cell therapy.
Next, we focused on the group of genes involved in both intense and acute signaling; their perturbation protects cells from AICD while enhancing cell proliferation shortly after TCR stimulation, another feature of potential relevance for T cell therapies.The hits include multiple regulators of cell-cell interactions. 192,193Specifically, sgRNAs targeting Icam1 and Itgal (encoding an LFA1 subunit), and Fermt3 (integrin activator) 194 were all highly enriched in the screens.Thus, interrupting cellcell interactions is beneficial for the expansion of effector cells right after TCR stimulation, consistent with previous data. 147echanistically, perturbing ICAM1 surface expression leads to enhanced effector function.Icam1-KO effector cells exhibit stronger cytotoxicity, as judged by their transcriptional profile and functional readouts.This was not limited to mouse cells, as ICAM1 and ICAM3 co-depletion from human T cells produced a similar phenotype.These results suggest that disrupting ICAM1-mediated homotypic clustering enables T cells to proliferate more, while being less susceptible to undergo death.Our data predicts that pharmacologic interference with ICAM1/3 in human T cells may have translational value.This is supported by our clinical evidence showing that patients receiving TILs with low ICAM1 expression have a better prognosis.As no ICAM3 antibody is currently available, we blocked the ICAM-LFA1 interaction using an LFA1 antibody, resulting in enhanced cell expansion and effector function.However, given the essential role of LFA1 in T cell extravasation from the endothelial compartment, its therapeutic targeting may be challenging. 195,196Notably, ICAM-LFA1-mediated clustering enables mutual costimulation, resulting in density-dependent self-regulation of proliferation and apoptosis. 148The inhibitory signal is likely dominated during CD3 stimulation, where rapid expansion and high-density culture happen.Therefore, blocking ICAM-LFA1 interaction may be beneficial for ex vivo T cell expansion where the inhibitory signal may exert a major influence.
Lastly, we identified and characterized genes contributing exclusively to chronic stimulation.,198 Several of our top hits were identified in previous screens (Ccnc, 92 and Regnase-1 102 ), are known to regulate T cell development (Zeb2, 157 Cd69, 160 and Ets1 199 ), or were translated into a clinical target (Cblb 159 ).Moreover, disrupting genes involved in mTORC1 regulation led to T cell expansion (Lamtor4, Rraga, and Flcn), 200 in line with the finding that mTOR inhibition regulates stem-like CD8 cell development and exhaustion during chronic infection. 201We found that Ctbp1 ablation in T cells enhanced persistence and effector function upon chronic tumor stimulation, both in vitro and in vivo.3][204][205] However, its role in T cell differentiation is unknown.Our transcriptomic data suggest that upon chronic stimulation, Ctbp1 inactivation pushes cells into a relatively active state accompanied by stalled terminal differentiation.Together with the IP-MS analysis, the results indicate a collaborative role of CTBP1, together with ZEB2, in regulating T-bet/ZEB2-induced effector terminal differentiation. 158This was supported by GSEA analysis, as well as in vitro and in vivo functional and phenotypic validation.Due to the oncogenic activity of Ctbp1 in tumor development, its pharmacologic targeting may come with double benefit, increasing functional effector persistence for immunotherapy, while impacting on tumor cell growth.
In summary, we report an unbiased discovery of genes contributing either to individual or common fitness traits upon T cell stimulation.While confirming previously established regulators, we report several unknown genes and characterize their differential involvement in T cell fitness traits, specifically Dap5, Icam1, and Ctbp1.These screen hits merit preclinical exploration: whereas for some pharmacologic strategies may be developed (like antibodies for ICAM1-3, or small molecule inhibitors for DAP5 and CTBP1), we envisage also a shorter route to clinical translation, namely by genetic perturbation in T cell products for adoptive transfer, as is currently being explored for CAR T cells. 206,207Our comprehensive screens for different aspects of T cell fitness also provide the community with considerable functionally annotated gene lists for increasing our understanding of T cell stimulation.The computational interface we include with this manuscript may facilitate this exploration: https://rhpc.nki.nl/sites/hithub/app/.

METHOD DETAILS Murine CD8 T cell isolation and in vitro cultures
Spleens from male or female OT-I/Cas9 mice were harvested and mechanically dissociated using a 100 mm and 70 mm cell strainer (Corning).The cell suspension was washed by centrifugation at 1000 xg using an isolation buffer (0.1% BSA in PBS).Red blood cells were lysed using a red blood cell lysis buffer (155 mM NH 4 Cl, 10 mM NaHCO 3 , 0.1 mM EDTA in distilled water; all Sigma) for five minutes.Cells were then washed once in PBS and once in isolation buffer and resuspended in isolation buffer.CD8 T cell isolation was performed using the Dynabeads Untouched Mouse CD8 Cells kit (11417D, Invitrogen) according to manufacturer's instructions.Isolated naı ¨ve CD8 T cells were then resuspended in mouse CD8 T cell medium (RPMI, 10% FBS, 100 U/ml of Penicillin-Streptomycin, 10 ng/ml IL-2 (12340026, ImmunoTools), 0.5 ng/ml IL-7 (12340075, ImmunoTools), 1 ng/ml IL-15 (12340155, ImmunoTools) and 50 mM 2-mercaptoethanol (Merck)) at a concentration of 1x10 6 cells/ml and primed using plate-bound CD3 antibody (0.25 mg/ 2x10 6 cells, clone 145-2C11, Thermo Fisher Scientific) and CD28 antibody (2.5 mg/ 2x10 6 cells, clone 37.51, Thermo Fisher Scientific) for 48 h.T cells were then either retrovirally transduced (see below) or maintained daily at a density of 1x10 6 cells/ml for approximately 10 days before performing experiments (T cell stimulation by CD3 antibody or tumor-antigen).

Construction of retroviral vectors
To generate the retroviral sgRNA vector, the sgRNA cassette of the lentiCRISPR v2 (#52961, Addgene) was modified to replace the BsmbI sites with BbsI sites and subsequently cloned into the pMSCV puro backbone (634401, Clonetech) by restriction cloning.sgRNAs targeting genes of interest were either taken from the Brie library or generated using CHOPCHOP 226 and cloned into the pMSCVpuro-sgRNA backbone by Golden-Gate cloning. 227The mAmetrine expressing retroviral sgRNA vector, pMSCVpuro-sgRNA-mAmetrine, was generated by inserting the mAmetrine fluorescent protein fragment after a mouse PGK promoter, upstream of the puromycin resistance sequence.The re-expression of wildtype or mutated ICAM1 in Icam1-KO cells was generated by inserting either a wildtype Icam1 fragment or an Icam1 fragment lacking the intracellular domain into a pMSCVpuro-sgRNA backbone after a mouse PGK promoter, upstream of the puromycin resistance sequence.For retroviral library construction, the sgRNA cassette of the Brie library (#73633, Addgene) was amplified by PCR and cloned into the pMSCVpuro backbone by restriction cloning.See Table S6 for oligonucleotide sequences used for sequencing and generating CRISPR-Cas9-mediated knockouts.

Retrovirus production and transduction of murine CD8 T cells
For retrovirus production, three million Platinum-E cells were seeded in a 10cm dish.After 24 h, these cells were transfected by polyethyleneimine (45 mg / 10 mg DNA, Polysciences) with 5 mg of pCL-ECO plasmid (#12371, Addgene) and 5 mg of the pMSCVpuro-sgRNA retroviral vectors.After another 24 h, the medium was replaced by Opti-MEM (Thermo Fisher Scientific) containing 2% FBS, 100 U/ml of Penicillin-Streptomycin. 24h later, the supernatant containing retrovirus was harvested, filtered through a 0.45 mm filter and stored at 4 C. Fresh medium was added to Platinum-E cells.The next day, supernatant was again harvested and filtered, combined with the supernatant of the first harvest and concentrated 10-20 times by spin-filter centrifugation (100 kDa pore size, Merck).The concentrated supernatant was snap frozen and stored at -80 C until used.For murine CD8 T cell transduction, 48 h CD3/CD28 antibody-primed T cells were harvested, one million primed OT-I/Cas9 T cells were mixed with 1 mL concentrated retroviral supernatant in a non-tissue culture treated 24-well plate pre-coated with Retronectin (25 mg/well, TB T100B, Takara).Cells were then spinfected at 3000 xg, 25 C for 1.5 h with minimum acceleration and no brake.After centrifugation, the plate was placed in the incubator.T cells were refreshed with mouse CD8 T cell medium 24 h after at the concentration of 1x10 cells/ml medium.48 h after spinfection, puromycin (4 mg/mL, Sigma) was added to the medium and cells were selected for at least 6d before starting experiments.

Genome-wide CRISPR screens (3 different settings) and MAGeCK analysis
Naive OT-I/Cas9 T cells were isolated, primed for 48 h using plate-bound CD3 antibody (0.25 mg/ 2x10 6 cells, Thermo Fisher Scientific) and CD28 antibody (2.5 mg/ 2x10 6 cells, Thermo Fisher Scientific) and transduced with the genome-wide Brie sgRNA library with at least a 1000x library representation.1d after puromycin selection we harvested a library reference sample (t 0 ).T cells were then cultured for an additional days.8 days post-transduction, we harvested a Pre-reactivation reference sample.The remainder the transduced cell pool was resuspended in T cell stimulation medium (RPMI, 10% FBS, 100 U/ml of Penicillin-Streptomycin, 10 ng/ml IL-2 (12340026, ImmunoTools) and 50 mM 2-mercaptoethanol) at cell density of 1x10 6 cells/ml.Cells were then stimulated under three different conditions: (1) Intense: selected CD8 T cells were stimulated with plate-bound CD3 antibody (1.25 ug/ 2x10 6 cells) in non-tissue culture 24-well plates (CD3 stimulation plate).After 24 h, the same procedure was repeated by transferring cells to new CD3 stimulation plates.24 h after the second-round of stimulation, dead cells were removed with the Dead Cell Removal kit according to manufacturer's instructions (130-090-101, Miltenyi) and cells were harvested for analysis.(2) Acute: selected CD8 T cells were stimulated with CD3 antibody (1.25 ug/ 2x10 6 cells) on 24-well stimulation plates for 24 h.Cells were then removed from the plates and refreshed daily at 1x10 6 cells/ml with T cell stimulation medium for another 3 days before harvesting.(3) Chronic: for chronically stimulated samples, selected CD8 T cells were co-cultured with fresh D4M.OVA tumor cells in T cell stimulation medium for 11 days (11 times).Fresh tumor cells were added to T cells daily at a fixed T cell: tumor cell ratio.For the resting condition, cells were refreshed daily with T cell stimulation medium without adding tumor cells.Both stimulated and resting cells were harvested for analysis after 11 days (11 times) tumor stimulation.Final samples were collected by washing the T cells twice with PBS, after which the cell pellet was snap frozen and stored at -80 C until use.Genomic DNA was isolated using the Blood and Cell culture MAXI Kit (13362, Qiagen), according to manufacturer's instructions.sgRNAs were amplified using a one-step barcoding PCR with NEBNext High Fidelity 2X PCR Master Mix (M0541L, NEB) and the forward and reverse primers (Table S6).The hexa-N nucleotide stretch contains a unique barcode identifying each sample following deep sequencing.MAGeCK (v0.5.7) 117 was used to perform the analysis of the screen.To assess the depletion of core essential genes we compared the Pre-reactivation reference sample to the t 0 library reference sample.We used shared core essential genes for all cell lines tested in the DepMap projects from the Broad and Sanger institutes as references. 111,228,229We filtered out from this list genes that were not expressed in our T cells (Table S1).

In vitro T cell stimulation and viability assay
For short-term CD3 stimulation, activated murine CD8 T cells were either rested or stimulated with plate-bound CD3 antibody (1.25 ug/ 2x10 6 cells) in T cell stimulation medium for 24 h for one (acute stimulation) or two rounds (intense stimulation).For short-term tumor-antigen-stimulation, activated T cells were challenged once with tumor cells expressing matching antigens in the T cell stimulation medium.For chronic CD3-stimulation, T cells were stimulated with plate-bound CD3 antibody (1.25 ug/ 2x10 6 cells) in T cell stimulation medium.Cells were passed onto fresh coated plates every other day for 8 days.For extended chronic tumor-antigenstimulation, matched fresh tumor cells were added to the T cell cultures every other day for a total of 3 weeks.For short-term CD3-stimulation of human CD8 T cells, activated T cells were stimulated with plate-bound CD3 antibody (5 mg/ 2x10 6 cells, 16-0037-85, Bioscience) for 24 h.For human CD8 T cells chronic tumor stimulation, activated MART-1 specific T cells were co-cultured with D10 melanoma cells every other day for at least 3 weeks.
T cell viability was analyzed after stimulation at the moment indicated in the figure legend.Viable cells were analyzed by CASY counter or flow cytometry according to the staining of DAPI or LIVE/DEADÔ Fixable Near-IR Dead Cell Stain Kit (L34976, Thermo Fisher Scientific).To obtain precise cell counts for flow cytometry analysis, Sphero AccuCount blank particles 5.26 mm (ACBP-50-10, Spherotec) were added to the samples.Data was processed using FlowJo software (BD Biosciences).

T cell-tumor co-culture cytotoxicity assay
Resting or stimulated mouse CD8 T cells were co-cultured with matching tumor cells at fixed T cell: tumor ratio.Days of co-culture depends on different T cells or tumor cell lines used in each experiment, as indicated in the figure legend.After co-culture, T cells were removed and remaining tumor cells were analyzed.T cell cytotoxicity was assessed by tumor colony formation in which the remaining tumor cells were fixed and stained for 1 h using crystal violet solution containing 0.1% crystal violet (CV, Sigma) and 50% methanol (Honeywell).For quantification, the remaining crystal violet was solubilized in 10% acetic acid (Sigma).Absorbance of this solution was measured on an Infinite 200 Pro spectrophotometer (Tecan) at 595 nm.

HPG translation assay
Ctrl and Dap5-KO T cells were used to assess the translational activity using the Click-iTÔ HPG Alexa FluorÔ 594 Protein Synthesis Assay Kit (C10429, Thermo Fisher Scientific) according to manufacturer's recommendations.In brief, 2x10 6 CD8 T cells were stimulated with CD3 antibody for 24 h as described above.Cells were then collected and washed with 1 ml pre-warmed methionine-free DMEM and resuspended in 1 ml pre-warmed methionine-free DMEM containing 50 mM Click-iTÒ HPG.The samples were then incubated for 30 min at 37 C. Following the incubation time, cells were washed with PBS and transferred into a 96-well V-bottom plate.Samples were then permeabilized using the Fixation & Permeabilization Buffer Set (88-8824-00, Thermo Fisher Scientific).Cells were washed twice with 200 ml 3% BSA/PBS. 100 ml of Click-iT reaction cocktail was added per sample and incubated for 30 min at room temperature.The reaction cocktail was then removed and samples were washed with 100 ml Click-iT reaction rinse buffer and finally taken up in 200 ml FACS buffer for sample acquisition using flow cytometry.
Western blot CD8 T cells were centrifuged at 1000 xg for 5 minutes and supernatant was removed.Cells were then washed with PBS twice before resuspending them in an appropriate volume of RIPA lysis buffer (50mM TRIS pH 8.0, 150mM NaCl, 1% Nonidet P40, 0.5% sodium deoxycholate, 0.1% SDS) supplemented with HALT Protease and Phosphatase inhibitor cocktail (78444, Fisher Scientific).Lysis was carried out on ice for 30 minutes.Samples were then centrifuged at 17000 xg and whole cell lysates were collected.Bio-Rad protein assay (500-0006, Bio-Rad) was used to quantify the protein content of each lysate.Protein concentrations were equalized and immunoblot samples were prepared by addition of 4xLDS sample buffer (15484379, Fisher Scientific) containing 10% b-Mercaptoethanol (final concentration 2.5%) and subsequent incubation of the samples at 95 C for five minutes.Proteins in lysates were size-separated on 4-12% Bis-Tris polyacrylamide-SDS gels (Invitrogen) and transferred to iBlotÔ Transfer Stack (Invitrogen).Blots were blocked using 4% BSA in 0.2% Tween-20 in PBS.Blocked membranes were incubated with primary antibodies overnight.Immunoblots were developed using the Super Signal West Dura Extended Duration Substrate (34075, Thermo Fisher Scientific).The luminescence signal was captured by the Bio-Rad ChemiDoc imaging system.See key resources table for antibody list.

Flow cytometry
For surface protein staining, samples were collected and cells were spun down in V-bottom 96-well plates by centrifugation at 1000 xg for 5 min and washed twice with FACS buffer (0.1% BSA/PBS).Antibodies against surface markers of interest were diluted in the FACS buffer according to the manufacturer's instructions.Washed cells were then resuspended in 50 ml staining solution containing antibodies for 30 min on ice in dark.Following the staining step, cells were washed twice with FACS buffer by centrifugation at 1000 xg for 5 min.After washing, cells were resuspended in the FACS buffer for data acquisition.Dead cells were identified by positive DAPI (BD), or the LIVE/DEAD Fixable Near-IR (L34976, Thermo Fisher Scientific).For intracellular cytokine staining, samples were stimulated with 20 ng/ml PMA (Sigma) and 1 ug/ml Ionomycin (Sigma) for 4-5 h. 1 h after PMA/Ionomycin stimulation, GolgiPlug (555029, BD Bioscience) was added according to manufacturer's instructions to block the secretion of intracellular protein.Intracellular staining was performed with Foxp3/transcription factor staining buffer set after surface staining according to the manufacturer's instructions (00-5523-00, eBioscience).Annexin V staining was conducted in combination with Annexin binding buffer (A13202, Thermo Fisher Scientific) according to manufacturer's instructions.For surface and intracellular protein expression analysis, LSRFortessa Flow Cytometer or an LSR II Flow Cytometer (both BD) were used.Flow cytometry antibodies used in this study are listed in key resources table .Secreted cytokine measurements in the cell culture supernatant of reactivated T cells were performed using the mouse IL2, TNF and IFNg Cytometric Bead Array Flex set (558297, 558299, 558296, BD Biosciences) following manufacturer's instructions.Flow cytometric analysis for CBA assay was performed using an iQue Screener PLUS (Intellicyte, Sartorius).All flow cytometric data was processed using FlowJo software (BD Biosciences).

Immunoprecipitation mass spectrometry and sample preparation
Activated murine CD8 T cells were stimulated with CD3 antibody (1.25 ug/ 2x10 6 cells) stimulation plates for 24 h.Cells were then harvested, washed twice with PBS and lysed in immunoprecipitation (IP) lysis buffer for 30 minutes.Two different IP lysis buffers were used for two independent experiments: Triton-X-100 IP buffer (30 mM Tris-HCl pH 7.4, 120 mM NaCl, 2 mM EDTA, 2 mM KCl, 1% Triton X-100) or NP40 IP buffer (50 mM Tris-HCl pH 7.4, 150 mM NaCl, 2 mM MgCl2, 0.1% NP-40), which both were supplemented with HALT Protease and Phosphatase inhibitor cocktail.The lysate was then centrifuged at 17000 xg for 10 minutes.Protein-containing supernatant was harvested and quantified.8 mg of protein per sample was incubated with CTBP1 antibody (8684, CST) or isotype control (10500C, Invitrogen) and kept on a rotator for 2 h at 4 C.After incubation, pre-washed protein A beads (1614013, Bio-Rad) were added and incubated for another 2 h (Triton-X-100 IP buffer experiment) or overnight (NP40 IP buffer experiment).Beads were washed twice in the IP lysis buffer and once in PBS after immunoprecipitation.Washed beads were resuspended in 1x S-Trap lysis buffer and heated at 95 C for 7 min.in the presence of 20 mM DTT. Supernantants were transferred to new 1.5 mL tubes, after which proteins were alkylated with 40 mM iodoacetamide (30 min.at RT in the dark).Finally, proteins were digested o/n with 2 mg trypsin (Sigma-Aldrich) on S-Trap Micro spin columns according to the manufacturer's instructions (ProtiFi, NY, USA).
Peptides were eluted, vacuum dried and stored at -80 C until LC-MS/MS analysis.LC-MS/MS was performed using the same instrumentation and setup as described above, with the exception that a 90-min.gradient containing a 70-minute linear increase from 7% to 29% solvent B was applied for peptide separation.Immunoprecipitation mass spectrometry data were analyzed by MaxQuant (version 1.6.17.0) using standard settings with 'match between runs' selected.MS/MS data were searched against the Mus Musculus Swissprot database (17,042 entries, release 2020_07) complemented with a list of common contaminants.The maximum allowed mass tolerance was 4.5 ppm in the main search and 20 ppm for fragment ion masses.False discovery rates for peptide and protein identification were set to 1%.Trypsin/P was chosen as cleavage specificity allowing two missed cleavages.Carbamidomethylation on cysteines and methionine oxidation were set as fixed and variable modifications, respectively.LFQ intensities were log2-transformed in Perseus (version 1.6.14.0) (REF) after which protein abundance values were filtered for at least two valid values (out of 3) in at least one condition.Missing values were replaced by imputation based a normal distribution, using a width of 0.3 and a downshift of 1.8.Differentially expressed proteins were determined using a t-test (thresholds: p<0.05 and 2 Log LFQ abundance ratio < -1.0 ^> 1.0).
In vivo competition assay sgCtrl and sgDap5-expressing OT-I/Cas9 T cells were generated as described above.sgCtrl and sgDap5-expressing T cells were stained with either CellTrace Violet (C34557, Thermo Fisher Scientific) or CellTrace Far Red (C34564, Thermo Fisher Scientific).Staining was performed according to manufacturer's recommendations.A parallel experimental arm with swapped staining colors was included.sgCtrl and sgDap5 T cells were then mixed at a 1:1 ratio.5 x10 6 mixed T cells were transferred into B16F10-dOVA tumor-bearing male or female C57BL/6 mice 9 days after tumor transplantation.After 3 days, tumors and spleens were harvested.Samples were processed into single cell suspensions, stained for LIVE/DEAD and CD8 and analyzed by flow cytometry.
In vivo prolonged chronic tumor stimulation experiment sgCtrl-mAmetrine and sgCtbp1-mAmetrine-expressing OT-I/Cas9 T cells were generated as described above, using the pMSCVpuro-sgRNA-mAmetrine vector. 1 x 10 6 100 Gy irradiated (irr-) B16F10-OVA cells were first intraperitoneally (i.p.) injected to male or female Cas9-EGFP mice (C57BL/6 background), followed by i.v.injection of 5 x 10 6 sgCtrl-mAmetrine or sgCtbp1-mAmetrine-expressing OT-I/Cas9 T cells.100.000U hIL-2 (Proleukin, Novartis) i.p. injection was given twice per day in the first 3 consecutive days.Mice receiving T cells were challenged again with 1 x 10 6 100 Gy irr-B16F10-OVA 16d after the first irr-tumor challenge.On d22, 0.5 x 10 6 healthy B16F10-OVA cells were s.c.injected to the mice.7 days after healthy tumor injection, sentinel mice were sacrificed, transferred T cells isolated from tumors and spleens were stained and analyzed by flow cytometry.Tumor growth was followed by measuring tumor volume three times weekly, and survival was measured according to tumor volume endpoint.
T cell isolation from murine spleens, tumors and lymph nodes Tumors were harvested and cut into small pieces, incubated in 5 mL tumor digestion medium (RPMI, 2% FBS, 10 U/mL DNAse I (Sigma), 200 U/mL collagenase type IV (Life Technologies)) at 37 C for 30 min while shaking.After digestion, tumor digests were then passed through 70 mm cell strainers (Corning), washed once with RPMI containing 10% FBS and once with PBS.Spleens were mechanically dissociated with syringes in RPMI containing 10% FBS and passed through 70 mm cell strainers (Corning).Samples were washed once with PBS and incubated in 2 ml red blood cell lysis buffer (155 mM NH 4 Cl, 10 mM NaHCO 3 , 0.1 mM EDTA in distilled water; all Sigma) for 5 min, followed by twice PBS wash.Lymph nodes were mashed in RPMI containing 10% FBS, and passed through 70 mm cell strainers (Corning), washed once with PBS.All samples were resuspended in the buffer used in downstream experiments.
Gene set enrichment analysis GSEA analysis of screen results: Gene ontology term enrichment of biological process (GOBP) gene sets from each screen was performed by using GSEA software (v4.1.0) 121,122on the whole MAGecK output gene list, ranked by signed -Log 10 (MAGeCK Score).All gene sets that were either negatively or positively enriched (FDR <=0.25) in at least one screen were included.Go terms were then clustered using REVIGO. 211GO terms containing keywords or are related to ''APOPTOSIS, CELL DEATH, T CELL, LYMPHOCYTE, ACTIVATION, PROLIFERATION and ADHESION'' are included, while GO terms with irrelevant cell types (if mentioned in the name) were excluded.Heat map shows the -log 10 (FDR) value of each GO term enrichment.
For the analysis of CD8 T cell lineage gene sets enrichment from acute resolving and chronic viral infection models, all gene sets were derived from the scRNA-Seq data published in Nat Immunol.2022. 107Top 50 differentially expressed genes in each cluster family (based on time frame and LCMV models) were used to generate gene sets used in this study: Acute_D8 (including CTL/ EFF/MP clusters), Acute_D15 (including Trans I/Trans II/Trans CTL clusters), Acute_D30 (including Mem cluster), Chronic_>D15 (including Eff-like/prolif I/prolif II/pre-Exh clusters) and Chronic_>D15 (including Exh-Int/Exh-Prog/Exh-Term/ Exh-HSP/ Exh-Term Gzma/ Exh-KLR clusters).The enrichment of the five CD8 T cell lineage gene sets from each screen output was performed by using GSEA software (v4.1.0)as described above.And the 3 more relevant gene sets: Acute_D8, Acute_D15, Chronic_>D15 are shown.
For GSEA analysis on transcriptomic data of sgCtrl and sgCtbp1-expressing OT-I/Cas9 cells after chronic stimulation, expression differences of the whole gene list ranked by stat values from the DESeq2 output was used as input.The Chronic_>D15 gene set was derived as described above.The ZEB1-KO_UP gene set (AIGNER_ZEB1_TARGETS) was taken directly from MSigDB database. 156he ZEB2-KO_UP gene set was derived from published database, 157 genes with differential expression between ZEB2-deficient and -sufficient P14 CD8 T cells after LCMV infection are taken (29 upregulated genes).Gene sets ''DAY8_EFFECTOR_VS_DAY30_ EXHAUSTED_CD8_TCELL_LCMV_CLONE13_UP'' (GSE41867), 73 ''KAECH_DAY8_EFF_VS_DAY15_EFF_CD8_TCELL_UP'' and ''KAECH_DAY8_EFF_VS_MEMORY_CD8_TCELL_UP'' (GSE100001) 170 were taken directly from MSigDB database as mentioned in the figure legend.

Proteomic analysis and sample preparation
For differential protein expression analysis of sgIcam1 or sgCtrl-expressing OT-I/Cas9 cells, cells were stimulated with CD3 antibody for 24 h.Right after stimulation, cells were collected, washed twice with PBS and snap frozen.For protein digestion, frozen cell pellets were lysed in boiling Guanidine (GuHCl) lysis buffer. 230Protein concentration was quantified and diluted to 2 M GuHCl, and samples were digested twice (4 h and overnight) with trypsin (Sigma-Aldrich) at 37 C at an enzyme/substrate ratio 1:75.Digestion was quenched by the addition of TFA (final concentration 1%), after which the peptides were desalted on a Sep-Pak C18 cartridge (Waters).The eluates were vacuum dried and prior to mass spectrometry analysis, peptides were reconstituted again in 2% formic acid.Peptide mixtures were analyzed by nanoLC-MS/MS on an Q Exactive HF-X Hybrid Quadrupole-Orbitrap Mass Spectrometer equipped with an EASY-NLC 1200 system (Thermo Fisher Scientific).Samples were directly loaded onto the analytical column (ReproSil-Pur 120 C18-AQ, 1.9 mm, 75 mm 3 500 mm, packed in-house) and eluted at a constant flow of 250 nl/min.Solvent A was 0.1% formic acid/water and solvent B was 0.1% formic acid/80% acetonitrile.For single-run proteome analysis, a 3 h gradient was employed containing a linear increase from 5% to 27% solvent B, followed by a 15-minute wash.
Proteome data were analyzed by MaxQuant (version 1.6.10.43) 215using standard settings.MS/MS data were searched against the Mus Musculus Swissprot database (17,027 entries, release 2020_02) complemented with a list of common contaminants.The maximum allowed mass tolerance was 4.5 ppm in the main search and 20 ppm for fragment ion masses.False discovery rates for peptide and protein identification were set to 1%.Trypsin/P was chosen as cleavage specificity allowing two missed cleavages.Carbamidomethylation was set as a fixed modification, while oxidation was used as variable modification.For Proteome data, LFQ intensities were log 2 -transformed in Perseus (version 1.6.10.43). 216Differentially expressed proteins were determined using t test (minimal threshold: FDR: 5% and S0: 0.1).

Protein-protein association and functional enrichment by STRING analysis
For the analysis of screen overlapping hits, 4 genes from the all-overlapping group (intense/acute/chronic) and 28 genes from the intense/acute overlapping group were taken as input for the STRING 138 analysis for protein-protein interaction and functional enrichment analysis (32 genes in total).Genes involved in cell-cell interaction and extravasation are highlighted in the protein-protein association network.Enrichment of the top 15 GOBP gene sets (ranked by Strength, FDR< 0.01) was plotted.
For enrichment analysis of differentially expressed proteins between Ctrl and Icam1-KO OT-I/Cas9 T cells after 24 h CD3 antibody stimulation, proteins with significant (-log(p-value)>=1.3)difference in fold expression between reactivated Icam1-KO and control cells were used as input.Analysis was performed by STRING analysis with input either Up in Icam1 or Down in Icam1 (Up in WT) protein list.Gene sets with FDR<=0.1 were taken.And very small pathways (<15 genes) were excluded because of redundancy with larger pathways and too large pathways (>200 genes) were excluded since they are overly general.For comprehensive interpretation of enriched pathways, biological processes were only shown if the term ''positive regulation of'' or ''negative regulation of'' is mentioned.

Transcriptomic (RNAseq) analyses and sample preparation
For differential gene expression analysis between ctrl and gene-of-interest-KO T cells, cells were either stimulated or rested as described in the figure legend.To harvest the samples, CD8 T cells were sorted (for tumor stimulated samples) and pelleted, washed twice with PBS and resuspended in RLT buffer (Qiagen).The total RNA was isolated using the RNeasy Mini Kit (Qiagen), including an on-column DNase digestion (Qiagen), according to the manufacturer's instructions.cDNA libraries were generated using the TruSeq Stranded mRNA sample preparation kit (Illumina Inc.) according to the manufacturer's instructions and were sequenced on a HiSeq2500 or NextSeq 550 system (Illumina Inc.).Sequenced samples were mapped to the mouse genome (Mus.musculus.GRCm38) using STAR (v2.6.0c) in two-pass mode with default settings.Read counts were computed using HTSeq-count with default settings (0.11.4), 213 normalization and statistical analysis of differential gene expression was performed using DESeq2 (v1.30.0).A sequencing batch effect is taken into account in the DESeq model by using the batch as a covariate. 214lysome profiling analysis and sample preparation Polysomal RNA isolation was performed as described previously. 231Briefly, Sucrose gradients for separation of polysomes were usually prepared by gentle sequential addition of 2.2 ml of the different sucrose solutions (e.i., 47, 37, 27, 17 and 7% in Tris-HCl pH 7.5 (20 mM), MgCl 2 (10 mM) and KCl (100 mM), supplemented with 2 mM DTT (10197777001, Sigma), Ribosafe RNase inhibitor (1 ml/ml, BIO-65027, Bioline) and CHX (100 mg/ml, 239763, Sigma) into a 12 mL tube (Beckman, 9/16 3 3 1/2 in.) and left overnight at 4 C to achieve continuous gradient prior to the centrifugation.Cells were treated with 100 mg/mL CHX and harvest after washing with PBS with CHX and lysed.The lysates were centrifuged 1300 xg for 10 min at 4 C and the supernatants were transferred into new tubes.From the cleared lysates, 500 mL was loaded on top of each gradient, mounted on SW41TI rotor and centrifuged at 36000 rpm for 2 hr at 4 C. Following the centrifugation, each gradient was split into 15 equal fractions of 760 ml.Fractions 9-13 were collected for RNA isolation using TRIzol reagent (Thermo Fisher Scientific) according to the manufacturer's instructions, polyA selected and followed by RNA library preparation as described above for mRNA-Seq.

Assessment of global translation efficiencies
We analyzed the generated RNA-Seq and polysome-Seq datasets in the following way.For both, initially, quality control was performed using the FastQC tool.Then, transcript quantifications were performed by Salmon, 217 using the protein-coding transcript sequences from gencode vM21 annotation.All dataset-specific differential analyses (gene expression or polysome occupancy) were performed in an R environment, using the DESeq2 package. 214Differential translation efficiency analyses were performed using the RiboDiff tool, 218 for which the input consisted of salmon-based transcript quantifications of primary transcripts that are determined based on Ensembl 96 APPRIS annotation.Genes with low sequencing depth were excluded from the translation efficiency analysis.

Survival analysis of patients receiving TIL therapy
The primary data are from a TIL trial conducted at the Sheba Medical Center (Trial number: NCT00287131 and NCT03166397, Tel Hashomer, Israel) 124,125 and all patients (including both male and female) gave written informed consent.RNA was extracted from infused TIL products using Tri Reagent (#T9424, Sigma-Aldrich) according to the manufacturer's protocol.RNA-Seq libraries were prepared with Illumina's Ribo Zero Gold and TruSeq stranded library prep kits and sequenced on the Illumina HiSeq2500 platform using paired-end sequencing with read length of 23125-150 bps.Reads were aligned to the human genome reference build hg38 using STAR aligner 212 and were quantified with FeatureCounts. 219After filtration of lowly expressed genes (counts below 10 in more than 90% of samples), raw counts were normalized in the R environment according to the LIMMA pipeline. 220For survival analysis, we compared between the upper (top 33.3%) and lower thirds (bottom 33.3%) of patients, according to the expression of indicated genes.Kaplan-Meier plots were generated using the survival and survminer R packages.P-values for survival analysis were computed using the log-rank test.

Single cell analysis
For the Pan-cancer CD8 TILs expression of DAP5, SERF2 and CTBP1, data was downloaded from TISCH2 database (Table S2). 126tudies containing both cell types of CD8 T cells (CD8T) and Exhausted CD8 T cells (CD8Tex) are included in the analysis (total 49 datasets).The expression level (log(TPM/10+1) of DAP5, SERF2 and CTBP1 in CD8T and CD8Tex cell types from each study were taken directly from the TISCH2 website, and the average expression was calculated.
For the analysis of CTBP1 expression in CD8 T cells from responders and non-responders treated with ICB, the gene expression data and metadata information were downloaded from the TISCH2 database (GSE120575) 181 and analyzed using Seurat (v4.3.0). 221cRNA-Seq for CTBP1 expression in CD8 T cells in the context of responders and non-responders.Initially, metadata information (patient IDs and responses) was added to the gene expression data based on cell IDs.Cells with low read count (<200) were removed from the samples, followed by standard single cell analysis pipeline: normalization, scaling, dimension reduction with Uniform Manifold Approximation and Projection (UMAP) and clustering.The annotation of CD8 T cells from the sequencing data was performed by Cluster Identity Predictor (CIPR) 222 based on the average expression of genes in the clusters.Identified CD8 T cells were analyzed for their CTBP1 expression.Averages of CTBP1 expression within responding and non-responding patient cohorts were compared.

QUANTIFICATION AND STATISTICAL ANALYSIS
Details of the statistical analyses performed on each experiment are indicated in the respective figure legends.For biological experiments (non-omics), analyses were performed by Prism (Graphpad Software Inc., v8.4.3).Unless indicated, when comparing two groups, a Two-tailed Student's t test was used for normally distributed data, and a two-tailed Mann-Whitney test was used for not normally distributed data.When comparing more than one group to the control group, one-way ANOVA with Holm-Sidak's multiple comparisons test was performed when data is normally distributed, or Kruskal-Wallis test with Dunn's post hoc test was used when data was not normally distributed.Tukey's post-hoc analysis was used for multiple comparisons between all groups.Data distribution normality was analyzed by Shapiro-Wilk test.P value lower than 0.05 was considered as statistically significant.For in vivo experiments, sample size estimation for experimental study design was calculated by G*Power. 232

ADDITIONAL RESOURCES
Screen hits from this study can be visualized via the reader interface: https://rhpc.nki.nl/sites/hithub/app/S1). 1 E) Density plots of Z-scores of the Broad/Sanger common essential genes compared to non-essential genes when comparing the t0/library reference sample to the prereactivation sample as in (D).

Figure 1 .
Figure 1.Multimodal function-based genome-wide CRISPR knockout screens for genes contributing to T cell fitness upon differential stimulation (A) T cell stimulation screens setup.(B) Marker expression heatmap from flow cytometry analysis of T cells stimulated with indicated conditions as in (A).Z score indicates the fold change to resting cells.(C) MAGeCK analysis of screen results (TableS1).(D) Enrichment of individual sgRNAs targeting genes identified from published T cell screens.Numbers above plots indicate signed -Log 10 (MAGeCK score).(E) GSEA of GO biological process from screen hits (TableS1).FDR: false discovery rate.(F) GSEA of CD8 lineage gene sets107 from screen hits (TablesS1 and S5).NES: normalized effect size.(G) Numbers of overlapping genes from top 50 hits of each screen.Genes are listed by average effect size (TableS1).
(K) Representative flow cytometry plots (n = 2 biological replicates) showing apoptotic T cells.(L) Flow cytometry analysis of T cells 4 days after CD3 stimulation, analyzed with Mann-Whitney test (n = 4 biological replicates).(M) Outline of in vivo competition assay.(N) Left: Flow cytometry plot showing T cell mixes, input or isolated from tumors 3 days after ACT.Right: Quantification of in vivo competition assay, analyzed with two-tailed paired t test (n = 5 mice/group).(O) Outline of ACT tumor model.(P) B16.OVA tumor growth in mice treated with either Ctrl or Dap5-KO T cells, as in (O), analyzed with a two-tailed unpaired t test per time point.Error bars represent SEM (n = 9 mice/group).Error bars indicate SD, unless otherwise specified.*p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001.

Figure 5 .
Figure 5. Blocking CTBP1-mediated terminal T cell differentiation preserves T cell effector function and enables long-term tumor control (A) Crystal violet (CV) staining quantification of viable B16.OVA tumor cells after 4 days co-cultured with equal amounts of indicated OT-I/Cas9 cells that were rested or chronically stimulated with tumor cells for 3 weeks.Analyzed with two-tailed paired t test (n = 3 biological replicates).

(
B) Flow cytometry analysis of IFNg and TNF double-positive population of indicated OT-I/Cas9 cells after 3 weeks D4M.OVA stimulation.Cells were re-stimulated with PMA/Ionomycin prior to analysis.Left: representative plot.Right: Quantification.Analyzed with Mann-Whitney test (n = 7 biological replicates).(C) As in (B), showing Ki67 expression, analyzed with two-tailed paired t test (n = 4 biological replicates).(D-G) Flow cytometry analyses of indicated marker expression on Ctrl and Ctbp1-KO T cells after 3 weeks chronic D4M.OVA stimulation, analyzed with two-tailed paired t test (D, E, G) or Mann-Whitney test (F).Data points indicate biological replicates.(H) Outline of in vivo prolonged tumor antigen stimulation ACT experiment, related to Figures 5I-5L, S5H, and S5I.(I) Flow cytometry analyses of marker expression on transferred T cells isolated from tumors 7 days after viable tumor cell transplantation, as in (H).KLRG1-/ CD127+ cells are considered less terminally differentiated.Analyzed with two-tailed unpaired t test (n = 6 mice/group).(J) Tumor size 20 days after viable tumor injection, as in (H) (when first mouse dropped out at tumor endpoint), showing data pooled from 2 independent experiments.Analyzed with two-tailed unpaired t test.(n = 18 mice/group).(K) Measurement of tumor outgrowth as in (H), analyzed with two-tailed unpaired t test (n = 18 mice/group).Error bars indicate SEM.(L) Kaplan-Meier plot depicting the survival of B16.OVA tumor-bearing mice as in (H), analyzed with regular log rank test (n = 18 mice/group).

Figure 6 .
Figure 6.Unique and shared genes limiting T cell fitness identified in multimodal stimulation screens (Left) When effector T cells receive TCR stimulation, they undergo rapid proliferation accompanied by AICD, limiting expansion.When antigen-stimulation persists, cells eventually become terminally differentiated, apoptotic, or dysfunctional.(Right) Intense, acute, and chronic stimulation screens reveal factors regulating either common or specific T cell fitness traits.Dap5 depletion in activated T cells stimulates global mRNA translation, upregulates cell cycle gene activity, and suppresses FAS expression, allowing cell pool expansion under all three stimulation conditions.Icam1 ablation or Icam-LFA1 interaction blockade prevents T cell hyperclustering upon stimulation, allowing increased exposure to stimulation signals.This contributes to their stronger cytotoxicity and expansion, especially after intense and acute stimulation.On the contrary, Ctbp1 depletion does not influence T cell expansion in the short run, but benefits their long-term persistence and functionality exclusively under chronic stimulation.It exerts this effect by hindering CTBP1/ZEB2/T-bet co-regulated effector terminal differentiation.

Figure S1 :Figure 1 A
Figure S1: Multimodal function-based genome-wide CRISPR knockout screens for genes contributing to T cell fitness upon differential stimulation, related to Figure 1 A) Percentage of live cells among SIINFEKL-tetramer-positive and -negative CD8 T cells from OVA + or OVA -MeVa2.1 tumors.Statistical analysis was performed by twotailed paired t-test (n=5 mice/ group).B) Flow cytometry analysis of CellTrace Violet (CTV) staining on day 3 after transfer of OT-I/Cas9 T cells, isolated from B16.OVA tumor-bearing C57BL/6 mice.Statistical analysis was performed by two-tailed paired t-test (n=9 mice/ group).C) Flow cytometry analysis of PD-1 and LAG3 surface expression on day 14 after transfer of OT-I/Cas9 T cells, isolated from B16.OVA tumor-bearing C57BL/6 mice.Statistical analysis was performed by two-tailed paired t-test (n=4 biological replicates).D) Signed -Log10(MAGeCK score) of gene inactivations comparing the t0/library reference sample with the pre-reactivation sample.Highlighted are the Broad/Sanger common essential genes (TableS1).1

F
) Viable T cell count after intensive CD3 stimulation.T cells were stimulated with CD3 antibody every 24h.Viable cells were analyzed by flow cytometry.Statistics were performed with one-way ANOVA, followed by a Dunnett post-hoc test (n=5 biological replicates).G) Histograms depicting the CTV dilution of activated OT-I/Cas9 T cells at either 24, 48, 72 and 96h post CD3 stimulation.H-J) Flow cytometry analysis of surface marker expression (H, J) and Annexin V staining (I) on OT-I/Cas9 T cells daily stimulated with D4M cell line with or without OVA expression for 12d.Resting cells were refreshed with medium only.Statistical analysis was performed with one-way ANOVA, followed by a Tukey post-hoc test (n=4 biological replicates).K) Schematic outline of in vitro T cell-tumor co-culture killing assay.L) Quantification of viable tumor cells from T cell-tumor co-culture killing assay.D4M.OVA tumor cells were co-cultured for 2d with OT-I/Cas9 T cells that had been either rested or chronically stimulated with D4M.OVA for 2wk.Viable tumor cells were stained by CV and quantified by acetic acid solubilization.Resting, refresh medium only.Statistical analysis was performed by two-tailed paired t-test (n=4 biological replicates).M) Flow cytometry analysis of marker expression from T cells stimulated with conditions performed in three different screens as in Fig.1A.Statistical analysis was performed by one-way ANOVA with Holm-Sidak's multiple comparisons test (n=5 biological replicates).Error bars indicate SD. * P<0.05; ** P<0.01; *** P<0.001; **** P<0.0001.
B CRISPR-mediated knockout in human T cells B In vitro T cell stimulation and viability assay B T cell-tumor co-culture cytotoxicity assay B HPG translation assay B Western blot B Flow cytometry B Antibody blocking experiments B Immunoprecipitation mass spectrometry and sample preparation B In vivo tumor growth experiment B In vivo competition assay B In vivo prolonged chronic tumor stimulation experiment B review