Letter https://doi.org/10.1038/s41586-019-0985-x

NR4A transcription factors limit CAR T cell function in solid tumours ­­­­­­­­­­­Joyce Chen1,2,3,4*, Isaac F. López-Moyado1,4,5, Hyungseok Seo1, Chan-Wang J. Lio1, Laura J. Hempleman1, Takashi Sekiya6, Akihiko Yoshimura7, James P. Scott-Browne1,4,9,10* & Anjana Rao1,3,4,8,10*

T cells expressing chimeric antigen receptors (CAR T cells) targeting CD8+CD45.1+ OT-I cells (specific for chicken ovalbumin (OVA) human CD19 (hCD19) have shown clinical efficacy against B cell SIINFEKL peptide presented by H-2Kb; Extended Data Fig. 2a, b) malignancies1,2. CAR T cells have been less effective against solid showed similar tumour growth rates (Extended Data Fig. 2c); low tumours3–5, in part because they enter a hyporesponsive (‘exhausted’ numbers of CAR T cells were transferred to minimize tumour rejec- or ‘dysfunctional’) state6–9 triggered by chronic antigen stimulation tion (Extended Data Fig. 2d). Eight days after adoptive transfer, CAR and characterized by upregulation of inhibitory receptors and loss of and OT-I tumour-infiltrating lymphocytes (TILs) (Fig. 1b, Extended effector function. To investigate the function of CAR T cells in solid Data Fig. 2b, e, f) comprised around 18% and 9%, respectively, of CD8+ tumours, we transferred hCD19-reactive CAR T cells into hCD19+ TILs (Fig. 1c) and exhibited similar proportions of PD-1hiTIM3hi cells tumour-bearing mice. CD8+CAR+ tumour-infiltrating lymphocytes compared to endogenous TILs (Fig. 1b, Extended Data Fig. 2b). All and CD8+ endogenous tumour-infiltrating lymphocytes expressing TILs produced low levels of TNF and IFNγ upon re-stimulation with the inhibitory receptors PD-1 and TIM3 exhibited similar profiles PMA and ionomycin (Fig. 1d, e), confirming their decreased function. of gene expression and chromatin accessibility, associated with The transcriptional profiles of ‘highly exhausted’ PD-1hiTIM3hi secondary activation of nuclear receptor transcription factors CAR TILs (population A, Fig. 1b) were similar to those of endoge- NR4A1 (also known as NUR77), NR4A2 (NURR1) and NR4A3 nous PD-1hiTIM3hi TILs (population C, Fig. 1b), but distinct from (NOR1) by the initiating transcription factor NFAT (nuclear factor those of CAR and endogenous ‘antigen-specific memory precursor’ of activated T cells)10–12. CD8+ T cells from humans with cancer PD-1hiTIM3lo TILs18 (populations B and D) and naive-like endog- or chronic viral infections13–15 expressed high levels of NR4A enous PD-1loTIM3lo TILs (population E) (Fig. 2a, Extended Data transcription factors and displayed enrichment of NR4A-binding Fig. 2g, Supplementary Table 1). The chromatin accessibility profiles motifs in accessible chromatin regions. CAR T cells lacking all of endogenous, OT-I and CAR PD-1hiTIM3hi and PD-1hiTIM3lo TIL three NR4A transcription factors (Nr4a triple knockout) promoted subsets (A–D, F) were similar to one another, but distinct from those tumour regression and prolonged the survival of tumour-bearing of PD-1loTIM3lo endogenous TILs (E), which resembled naive CD8+ mice. Nr4a triple knockout CAR tumour-infiltrating lymphocytes T cells (Fig. 2b, Extended Data Fig. 3). PD-1hiTIM3lo TILs (B, D) resem- displayed phenotypes and gene expression profiles characteristic of bled memory-precursor CD8+ T cells7,11,12, with accessible regions CD8+ effector T cells, and chromatin regions uniquely accessible showing substantial enrichment for consensus TCF1 motifs (Fig. 2b, in Nr4a triple knockout CAR tumour-infiltrating lymphocytes cluster 6). Regions that were selectively accessible in PD-1hi populations compared to wild type were enriched for binding motifs for NF-κB (A–D, F) were enriched for consensus NR4A (nuclear receptor, NR) and AP-1, transcription factors involved in activation of T cells. We as well as NFAT, NFκB, bZIP and IRF:bZIP motifs (Fig. 2b, clusters identify NR4A transcription factors as having an important role in 8, 9). NR4A protein expression was higher in PD-1hiTIM3hi than in the cell-intrinsic program of T cell hyporesponsiveness and point to PD-1hiTIM3lo TILs (Fig. 2c, Extended Data Fig. 4a–d). NR4A inhibition as a promising strategy for cancer immunotherapy. Single-cell RNA-sequencing (RNA-seq) data from human CD8+ Mouse B16-OVA melanoma, EL4 thymoma and MC38 colon adeno- TILs provided further justification for studies of NR4A. In CD8+ TILs carcinoma cell lines were engineered to express hCD19 (Extended Data infiltrating a human melanoma14, NR4A1 and NR4A2 expression Fig. 1a). B16-OVA-hCD19 cells stably maintained hCD19 expression showed a strong positive correlation with PDCD1 (PD-1) and HAVCR2 after growth in syngeneic C57BL/6J mice for 18 days and subsequent (TIM3) expression, and NR4A3 showed a moderate positive correlation culture for 7 days ex vivo (Extended Data Fig. 1a, right). B16-OVA and (Fig. 2d). PDCD1 and HAVCR2 expression correlated positively with B16-OVA-hCD19 cells grew at the same rate in vivo, indicating that TIGIT, CD38, CTLA4, JUN, TOX, TOX2 and IRF4 and negatively with the hCD19 antigen did not cause tumour rejection (Extended Data TCF7 (Extended Data Fig. 4e–g, Supplementary Table 2). Additionally, Fig. 1b, left). On the basis of tumour growth rate, we inoculated mice NR4A, NFAT, bZIP and IRF:bZIP motifs were enriched in regions with 500,000 B16-OVA-hCD19 tumour cells (Extended Data Fig. 1b, uniquely accessible in CD8+PD-1hi TILs from human melanoma and right). Mouse CD8+ T cells retrovirally transduced with a second- non-small cell lung cancer13 and in HIV-antigen-specific CD8+ T cells generation CAR against hCD1916,17 exhibited a transduction efficiency of from infected humans15 (Fig. 2e, cluster 9). The upregulation of NR4A 95.5 ± 4.0% (mean ± s.d.) (Extended Data Fig. 1c, d), produced TNF and family members and enrichment of NR4A binding motifs in differen- IFNγ upon re-stimulation with EL4-hCD19 cells, and exhibited dose-de- tially accessible regions of chronically stimulated human and mouse pendent lysis of B16-OVA-hCD19 cells (Extended Data Fig. 1e–g). CAR CD8+PD-1hi T cells11,12,15,19 led us to focus on NR4A family members T cells did not express higher surface levels of PD-1, TIM3 or LAG3 than as potential transcriptional effectors of CD8+ T cell exhaustion. mock-transduced cells under resting conditions (Extended Data Fig. 1h). The three NR4A proteins are essential for regulatory T cell develop- C57BL/6J mice bearing B16-OVA-hCD19 tumours and adoptively ment20, indicating redundant function. We compared NR4A-sufficient transferred with CD8+CD45.1+Thy1.1+ CAR T cells (Fig. 1a, b) or (wild type) CAR TILs with NR4A triple knockout (Nr4a TKO) CAR

1Division of Signaling and Gene Expression, La Jolla Institute for Immunology, La Jolla, CA, USA. 2Biomedical Sciences Graduate Program, School of Medicine, University of California, San Diego, La Jolla, CA, USA. 3Department of Pharmacology, University of California, San Diego, La Jolla, CA, USA. 4Sanford Consortium for Regenerative Medicine, La Jolla, CA, USA. 5Bioinformatics and Systems Biology Graduate Program, University of California, San Diego, La Jolla, CA, USA. 6Department of Immune Regulation, National Center for Global Health and Medicine, Ichikawa, Chiba, Japan. 7Department of Microbiology and Immunology, Keio University School of Medicine, Tokyo, Japan. 8Moores Cancer Center, University of California, San Diego, La Jolla, CA, USA. 9Present address: Department of 10 Biomedical Research, National Jewish Health, Denver, CO, USA. These authors contributed equally: James P. Scott-Browne, Anjana Rao. *e-mail: [email protected]; [email protected]; [email protected]

NAtUre| www.nature.com/nature RESEARCH Letter

IFN TNF TNF and IFN a d γ P = 0.1723 γ CAR ** ** Inoculate adoptive *** ** tumour transfer Day 21: 50 * *** 50 ** isolate TILs 60 ** C57BL/6J 40 40 30 40 30 mice Day 13 20 20 20 b Day 8 after adoptive transfer: CAR CD8+ T cells 10 10 0 0 0 + + Percentage of cells CAR CAR CAR All CD8 T cells CAR CD8 producing cytokines CAR Endo.OT-I CAR Endo.OT-I CAREndo.OT-I 5 in vitro in vitro in vitro 2.14 25.4 10 105 No stimulation PMA + ionomycin 104 A + 103 B 68.4 Tumour-in ltrating CD8 T cells 104 e 0 CAR Endogenous OT-I In vitro CAR CD8+ TILs CD8+ TILs CD8+ TILs CD8+ T cells + 5 No stimulation CAR CD8 Endogenous CD8+ 10 0.31 0.26 0.35 0.17 0.24 0.25 0.068 0.051 103 105 104 D 3 0 104 10 Endogenous PD-1: BV421 C + CD8 3 10 70.2 0 90.8 8.63 93.8 5.70 92.7 6.80 99.3 0.61 72.3 0.077 PMA + ionomcyin 0 105 1.60 4.89 1.20 1.56 0.99 1.75 15.7 39.4 CD45.1:PeCy7 0103 104 105 E 15.2 104 Thy1.1: PerCP–Cy5.5 0 103 104 105 103 TIM3: PE CAR Endogenous c 0 OT-I Endogenous 60.4 33.1 76.5 20.8 73.3 24.0 37.2 7.76 3 4 5 3 4 5 3 4 5 3 4 5

TNF: BV421 0 10 10 10 010 10 10 0 10 10 10 0 10 10 10 020406080 100 IFN : APC Percentage of all CD8+ T cells γ Fig. 1 | CAR, OT-I and endogenous CD8+ TILs isolated from B16-OVA- show mean values with data points for 6, 5 and 11 independent experiments hCD19 tumours exhibit similar phenotypes. a, Experimental design to for CAR, OT-I and endogenous TILs, respectively. d, Quantification assess CAR and endogenous TILs; 1.5 × 106 CAR T cells were adoptively of cytokine production after re-stimulation of CAR, OT-I and transferred into C57BL/6J mice 13 days after tumour inoculation. b, Left, endogenous (endo.) CD8+ TILs, compared to cultured CD8+ CAR T cells representative flow cytometry plot identifying CD8+CD45.2+ endogenous re-stimulated with PMA and ionomycin or left unstimulated. Bars show TILs and CD8+CD45.1+Thy1.1+ CAR TILs (Thy1.1 encoded in the CAR mean values with data points for 3 independent experiments. All P values retroviral vector). Right, flow cytometry plots showing PD-1 and TIM3 were calculated using two-tailed unpaired t-tests with Welch’s correction, surface expression on CD8+ CAR and endogenous TILs. c, Bar graph *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001. e, Representative showing the percentage of CAR and OT-I TILs in total CD8+ TILs. Bars flow cytometry plots of cytokine production after re-stimulation. TILs (Fig. 3a), using Rag1-deficient recipient mice to avoid poten- Nr4a3−/− mice with both CAR and Cre retroviruses, and naive CD8+ tial rejection. Nr4a TKO and control wild-type CAR T cells were T cells from Nr4a1fl/flNr4a2fl/flNr4a3+/+ mice with CAR and empty obtained by transducing naive CD8+ T cells from Nr4a1fl/flNr4a2fl/fl retroviruses, respectively (Extended Data Fig. 5a–c). Compared to

a 20 c (C) Endo. NR4A1 NR4A2 NR4A3 PD-1hiTIM3hi (E) Endo. CAR PD-1hi TIM3hi TILs (A) lo lo ** TILs PD-1 TIM3 400 * 400 ** 500 10 CAR PD-1hiTIM3lo TILs (B) TILs 300 300 400 300 hi hi (A) CAR 200 200 Endogenous PD-1 TIM3 TILs (C) MFI 200 PD-1hiTIM3hi 100 100 Endogenous PD-1hiTIM3lo TILs (D) 0 100 TILs 0 0 0 Endogenous PD-1loTIM3lo TILs (E) (D) Endo. P = 0.0616 * * hi lo 300 400 600 6 –10 PD-1 TIM3 TILs * * * d PC2: 13% variance (B) CAR 200 P = 0.4441300 * 400 P = 0.1389 NR4A1 hi lo PD-1 TIM3 TILs MFI 200 5 100 200 –20 0420 0 100 4 PC1: 71% variance 0 0 0 b Mouse CD8+ Tcell ATAC-seq e Human CD8+ Tcell ATAC-seq 3 1.5 1.0 32,394 differentially 15,986 differentially 2 0.5 accessible regions accessible regions 0 3,349 2,486 3,658 2,865 4,057 2,833 5,390 4,146 3,610 3,579 3,644 624 1,652 1,340 1,581 1,217 1,369 980 Naive 6 lo lo NR4A2 Endo PD-1 TIM3 +2 Naive +2 Endo PD-1hiTIM3lo 5 hi lo hi CAR PD-1 TIM3 (TIL) PD-1 (TIL) Endo PD-1hiTIM3hi Effector (virus) 4 OT-I PD-1hiTIM3hi Effector memory VCR2 (signal in (signal in 3.5

2 hi hi 2 CAR PD-1 TIM3 (bulk) HA 3 3.0 2.5 log log –2 –2 Effector Central memory 2.0 ATAC-seq signa l ATAC-seq signa l sample vs mean ) (LCMVarm) sample vs mean ) Memory (bulk) 2 1.5 hi hi Exhausted PD-1 TIM3 (LCMVCl13) 6 123456789 123456789 (HIV,chronic) NR4A3 bZIP (AP-1) bZIP (AP-1) ETS (Ets1-distal) 5 +2 ETS (Ets1-distal) +2 HMG (TCF7) HMG (TCF7) IRF:bZIP (IRF:BATF) IRF:bZIP (IRF:BATF) 4 IRF (IRF4) IRF (IRF4) NR (NUR77) 3 1.00 RHD,bZIP (NFAT:AP1) NR (NUR77) 0.75 RHD (NFAT) RHD,bZIP (NFAT:AP1) 0.50

vs all peaks) vs all peaks) 0.25 RHD (NFkB-p65) –2 RHD (NFAT) 2 (frequency in cluster –2 (frequency in cluster 2 T-box (TBET) 2 Runt (RUNX) 2 3 4 5 log log Known motif enrichment Zf (Klf4) Known motif enrichment T-box (TBET) 123456789 1 23456789 PDCD1 Fig. 2 | CAR and endogenous CD8+ TILs exhibit similar gene expression for CAR comparisons (top) were calculated using two-tailed paired t-tests; and chromatin accessibility profiles. a, Principal component analysis of P values for endogenous comparisons (bottom) were calculated using row- RNA sequencing (RNA-seq) data from CAR PD-1hiTIM3hi (A) and matching one-way ANOVA with Greenhouse–Geisser correction and Tukey’s PD-1hiTIM3lo (B) TILs and endogenous PD-1hiTIM3hi (C), PD-1hiTIM3lo (D) multiple comparisons tests; for both calculations, *P ≤ 0.05, **P ≤ 0.01, or PD-1loTIM3lo (E) TILs. Data represent 3 independent experiments, each ***P ≤ 0.001, ****P ≤ 0.0001. Data show mean ± s.d. and individual values using TILs pooled from 9–14 mice. b, Top, heatmap of mouse CD8+ T cell from three independent experiments, each using TILs pooled from 9–14 assay for transposase-accessible chromatin using sequencing (ATAC-seq) mice. d, Scatterplots of RNA-seq data showing expression of PDCD1 (x axis) + 14 data showing difference in signal (measured as log2(normalized read counts)) and HAVCR2 (y axis) in single cells of human CD8 TILs , with expression from mean for 9 k-means clusters. Bottom, heatmap of motif enrichment of the indicated NR4A genes shown in the colour scale. Each dot represents a analysis. Data shown for one representative member of transcription factor single cell. e, Top, human CD8+ T cell ATAC-seq data from PD-1hi TILs (two families enriched in at least one cluster compared to all accessible regions. samples from melanoma, one sample from non-small cell lung tumour13) Representative motifs with a fold enrichment greater than 1.5, a P value and antigen-specific CD8+ T cells from individuals with HIV15 showing –10 less than 1 × 10 , and found in at least 5% of regions are shown. difference in signal (measured as log2(normalized read counts)) from mean c, Quantification of NR4A expression (mean fluorescence intensity); P values for 9 k-means clusters. Bottom, heatmap of motif enrichment analysis.

NAtUre| www.nature.com/nature Letter RESEARCH control tumour-bearing mice adoptively transferred with wild-type a d + CD8 CAR T cells, tumour-bearing mice adoptively transferred with Inoculate Adoptive Inoculate Adoptive + tumour transfer tumour transfer Day 21: Nr4a TKO CD8 CAR T cells showed pronounced tumour regression Day 90 isolate –/– –/– and enhanced survival (Fig. 3a–c). Differences in tumour size were Rag1 mice Rag1 mice TILs Day 7 Day 13 apparent as early as day 21 after tumour inoculation (Fig. 3b, bottom). 3.0 ) 5 b 2 PBS 10 WT 2.5 e TKO Nr4a TKO CAR T cells promoted tumour rejection and prolonged m WT 4 survival even in immunocompetent recipient mice (Extended Data 2.0 10 1.5 Nr4a TKO 3 area (c 10 Fig. 5d–g). Thus, NR4A transcription factors suppress tumour rejec- 55.3% 22.5% 1.0 our tion in the CAR T cell model. m 0.5 0 To assess NR4A redundancy, we evaluated the anti-tumour effects of Tu 0.0 0103 104 105 0103 104 105 + 0102030405060708090 PD-1: BV421 3.0 TIM3: PE

CD8 CAR T cells lacking individual NR4A proteins (Extended Data ) 2

Fig. 6a). Nr4a TKO CAR T cells showed greater anti-tumour activity m 2.5 WT PD-1 2.0 Nr4a 25,000 than CAR T cells lacking NR4A1, NR4A2 or NR4A3 (Extended Data TKO * 20,000 area (c 1.5 Fig. 6b–d). Moreover, retroviral expression of any NR4A transcription 15,000

1.0 MFI + our 10,000 factor in CD8 T cells (Extended Data Fig. 7a) resulted in increased m cell counts

0.5 Normalize d

Tu 5,000 expression of inhibitory surface receptors and decreased cytokine pro- 0.0 3 4 5 0 0102030405060708090 010 10 10 duction upon re-stimulation (Extended Data Fig. 7b–d). In principal 3.0 PD-1: BV421 WT TKO ) 2 2.5 component analyses of RNA-seq data, the majority of the variance m WT TIM3 2.0 Nr4a TKO (78%) was between cells expressing any NR4A transcription factor 80 **** 1.5

area (c 60 compared with cells expressing the empty vector control (Extended cell s 1.0 + 40 our m

Data Fig. 7e, Supplementary Table 3). In both RNA-seq and ATAC-seq, 0.5 IM3 cell counts Normalize d T 20 Tu pairwise comparisons showed few if any differences between NR4A 0.0 Percentage of 0102030405060708090 0103 104 105 0 TIM3: PE WT TKO family members (Extended Data Fig. 7f, g). Thus, the three NR4A Days after tumour inoculation proteins induce similar changes in transcriptional and chromatin acces- No stimulation PMA + ionomycin + Tumour sizes on f 105 2.22 1.08 11.14.84 sibility profiles in CD8 T cells. day 21 after inoculation 104 To assess phenotypic and genome-wide changes associated with **** 103 3.0 * anti-tumour function, we modified experimental conditions to delay ) 0 2 **** 94.9 1.77 80.1 4.02 WT

m 2.5 tumour regression (Fig. 3d). Tumour sizes and TIL recoveries were 105 3.34 12.9 7.09 36.9 Nr4a 2.0 similar between Nr4a TKO and wild type (Extended Data Fig. 8a–c). 104 TKO 1.5 Eight days after adoptive transfer, Nr4a TKO TILs showed a mild but area (c 103

1.0 : APC

mour 0 statistically significant decrease in PD-1 expression compared to wild- 0.5 51.4 32.3 29.0 27.0 IF N γ type TILs, and the total Nr4a TKO PD-1hi population was markedly Tu 0.0 0103 104105 0103 104 105 PBS WT TKO TNF: BV421 skewed towards low TIM3 expression (Fig. 3e). Moreover, the percent- c 100 IFNγ age of cells expressing TNF or both IFNγ and TNF after re-stimulation IFNγ TNF and TNF P = 0.0832 was significantly higher in Nr4a TKO compared to wild-type TILs 75 100 100 ** 100 ** (Fig. 3f). TIM3lo Nr4a TKO CAR TILs were noticeably skewed towards 80 80 80 50 low TCF1 expression (Extended Data Fig. 8d, top); the TIM3loTCF1lo ****P < 0.0001 60 60 60 40 40 40 lo hi 25 population, which is different from the TIM3 TCF1 memory pre- 20 20 20 18,21–23 Percentage survival

cursor population that expands after PD-1 blockade , may be Percentage of cells 0 0 0

0 producing cytokines responsible for increased effector function. There was no significant 0102030405060708090 Days after tumour inoculation difference in the mean fluorescence intensities of TCF1, T-bet or Eomes

PBS WT Nr4a TKO ionomyci n ionomyci n ionomyci n ionomyci n ionomyci n ionomyci n

(Extended Data Fig. 8d, bottom). A + A + A + A + A + A + No stimulation No stimulation No stimulation No stimulation No stimulation No stimulation

Nr4a TKO TILs showed increased expression of genes related to PM PM PM PM PM PM effector function; that is, mRNAs encoding effector proteins (IL-2Rα, Fig. 3 | NR4A-deficient CAR TILs promote tumour regression and TNF and granzymes) were upregulated. Genes expressed in naive or prolong survival. a, Experimental design; 3 × 106 wild-type or Nr4a memory T cells compared to effector populations (for example, Sell and TKO CAR T cells were adoptively transferred into Rag1−/− mice 7 days Ccr7) were downregulated and inhibitory surface receptors typically after tumour inoculation. PBS was injected as a control. b, Top, tumour upregulated in hyporesponsive T cells (Pdcd1, Havcr2, Cd244, Tigit and growth in individual mice. Bottom, tumour sizes of individual mice at Cd38) were downregulated in Nr4a TKO compared to wild-type TILs day 21 (mean ± s.d.); P values calculated using an ordinary one-way (Fig. 4a, Extended Data Fig. 8e–g, Supplementary Table 4). Gene set ANOVA with Tukey’s multiple comparisons test. c, Survival curves; < enrichment analysis24 comparing RNA-seq data from Nr4a TKO versus ****P 0.0001 calculated using log-rank (Mantel–Cox) test. Surviving mouse numbers at days 7, 21 and 90 were n = 21, 14 and 0 wild-type TILs against gene sets from effector, memory and exhausted 11 (PBS); n = 35, 25 and 1 (wild type); n = 39, 36 and 27 (Nr4a TKO). populations from LCMV-infected mice supported these conclusions. d, Experimental design; 1.5 × 106 wild-type or Nr4a TKO CAR T cells To identify transcriptional targets of individual NR4A proteins, we were adoptively transferred into Rag1−/− mice 13 days after tumour clustered genes differentially expressed in Nr4a TKO compared to wild- inoculation and then analysed 8 days later. e, Surface expression of PD-1 type TILs (Fig. 4a, Supplementary Table 4), by changes in gene expression and TIM3 on CAR+NGFR+ cells with a set level of CAR expression when NR4A was ectopically expressed (Fig. 4b, Supplementary Table 3). (103–104). Representative flow cytometry plots (top), histograms Clusters 1 and 2 contain genes downregulated in the absence of NR4A (middle and bottom, left) and mean ± s.d. and individual values (right) and upregulated in NR4A-expressing cells (Pdcd1, Havcr2 and Cd244 of 6 independent experiments, each using TILs pooled from 3–8 mice. in cluster 1; Tox, Tigit and Cd38 in cluster 2). Cluster 4 contains genes P values were calculated using two-tailed paired t-tests with Welch’s upregulated in Nr4a TKO TILs and downregulated in NR4A-expressing correction. f, Top, representative flow cytometry plots for TNF and IFNγ production. Bottom, quantification showing the mean ± s.d. and cells (Tnf and Il21). Previous publications have shown that Runx3 is a + 25 individual values of 5 independent experiments, each using TILs pooled downstream target of NR4A1 in CD8 T cell development and that from 3–8 mice. IL-2 was not detectable above background (not shown). 26 its overexpression contributes to tumour regression , but Runx3 was P values were calculated using two-tailed paired t-tests between not differentially expressed in Nr4a TKO compared to wild-type TILs. stimulated wild-type and Nr4a TKO CAR TILs. *P ≤ 0.05, **P ≤ 0.01, As expected, a substantial fraction of regions (36%) with lower ***P ≤ 0.001, ****P ≤ 0.0001. accessibility in Nr4a TKO TILs contained NR4A-binding motifs.

NAtUre| www.nature.com/nature RESEARCH Letter

In vitro ATAC-seq density a 6 Nr4a TKO > WT b c 536 log2(counts per kb) Il1r2 3 Il21 Pdcd1, 10

4 Cdkn1a Cluster Nr4a TKO NR4A 1 NR4A 2 NR4A 3 Batf3 Cd70 Havcr2, 1,787 Gzmg Gzmc 1 2 Gzmf Cd244 10 scale) 2 Il2 3 2 2 Gzme Il2ra Tnf Nfkbia scale)

2 2 1 Tbx21 Ifng Tox, 10 0 1 Tcf7 Pdcd1 3 Tigit, Cd101 0 Cd38 Tox 0 Cd38 10 Tigit –2 Cd7 Ccr7 Ccl5 –1 754 Havcr2 4 Tnf, Nr4a TKO CAR TILs 10–1 Cd244 –4 Sell –2 5 Il21 10–1 100 101 102 103

Fold change (log Il10ra –3 Wild-type CAR TILs 540 Fold change (log –6 6 WT > Nr4a TKO Nr4a TKO >WT 6912 15 –302 7 AP-1 (bZIP), 1 × 10 , 71% Mean of normalized counts (log2 scale)

–16 d 20 kb Empty NFkB-p65 (RHD), 1 × 10 , 25% Peaks CAR T wild type NR4A1 WT > Nr4a TKO CAR T Nr4a TKO NR4A2 –118 Endogenous PD-1hiTIM3hi NR4A (NR), 1 × 10 , 36% OT-I PD-1hiTIM3hi NR4A3 –12 CAR TPD-1hiTIM3hi Normalized cell count s NFAT (RHD), 1 × 10 , 11% 0103 104 105 hi lo Endogenous PD-1 TIM3 hNGFR: FITC CAR TPD-1hiTIM3lo Fold vs background Endogenous PD-1loTIM3lo Effector e function Naive 0.10 NR4A1: 4.4x NR4A2: 2.8x LCMVArm effector 0.08 NR4A3: 5.2x bZIP, LCMVArm memory 0.06 LCMVCl13PD-1hiTIM3hi NFκB 0.04 In vitrorv empty NFAT 0.02 In vitrorv NR4A1 Percentage inpu t 0.00 In vitrorv NR4A2 NR4A, other Inhibitory transcription factors; receptors, In vitrorv NR4A3 gene expression and (PD-1, Empt y Empt y In vitrorv mock NR4A 1 NR4A 2 NR4A 3 NR4A 1 NR4A 2 NR4A 3 chromatin changes TIM3) Foxp3 Pdcd1 In vitro rv NFAT-CA-RIT CNS2 –23kb Pdcd1 (neg. ctrl) enhancer Fig. 4 | Gene expression and chromatin accessibility profiles indicate Nr4a deletion and downregulated after NR4A overexpression (for increased effector function of Nr4a TKO compared to wild-type CAR example, Tnf, Il21) are indicated. c, Scatterplot of pairwise comparison of TILs. a, Mean average plots of genes differentially expressed in Nr4a ATAC-seq density (Tn5 insertions per kb) between Nr4a TKO and wild- TKO versus wild-type CAR TILs; P values calculated using Wald test type CAR TILs, showing differentially accessible regions and associated (as implemented in DESeq2), and adjusted using the Benjamini–Hochberg de novo identified motifs. d, Left, genome browser view of the Pdcd1 locus method. Differentially expressed genes (adjusted P < 0.1, fold change in all previously mentioned ATAC-seq samples and CA-RIT-NFAT1- (log2 scale) ≥ 1 or ≤ −1) are highlighted; selected genes are labelled. transduced cells. Grey bar, exhaustion-specific enhancer around 23 kb b, Heatmap of genes with opposing expression changes between Nr4a 5′ of the Pdcd1 transcription start site. Top right, histogram of NR4A deletion and NR4A overexpression. Fold change values (log2 scale) of expression in cells expressing HA-tagged NR4A1, NR4A2 or NR4A3 genes differentially expressed in Nr4a TKO relative to wild-type CAR TILs (representative of two biological replicates). Bottom right, ChIP–qPCR were compared to corresponding values in cells ectopically expressing showing enrichment of HA-tagged NR4A over background at the Pdcd1 NR4A1, NR4A2 or NR4A3, and 7 k-means clusters were identified. enhancer (technical replicates from one of two independent experiments). Genes downregulated after Nr4a deletion and upregulated after NR4A e, Proposed role of NR4A in T cells chronically stimulated with antigen. overexpression (for example, Pdcd1, Havcr2, Tox), or upregulated after

A smaller subset contained NFAT-binding sites without an adjacent AP-1 bZIP and NFκB motifs (Extended Data Fig. 9a, middle panel, blue lines; site, suggesting that NR4A maintains the accessibility of ‘exhaustion- and additional examples Ccr6 and Ifng). TNF and IL-21, cytokines related’ regions10 that bind NFAT without AP-1 (Fig. 4c). Regions more involved in effector functions30, are more highly expressed in Nr4a accessible in Nr4a TKO compared to wild-type TILs were enriched TKO compared to wild-type TILs (Fig. 4a). Two bZIP-motif-containing for consensus bZIP (71%) and Rel/NFκB (25%) binding motifs, con- regions of the Il21 promoter gain accessibility and the Tnf locus shows firming the established role of bZIP (for example, Fos, Jun, ATF and broadly increased accessibility across the promoter and the entire gene CREB) and Rel/NFκB family members in T cell activation and effector in Nr4a TKO compared to wild-type TILs (Extended Data Fig. 9b). function, and consistent with negative crosstalk between NR4A and We previously used an engineered NFAT protein, CA-RIT-NFAT1, to NFκB27,28. Overall, Nr4a TKO TILs display potent effector function: mimic a dephosphorylated nuclear NFAT that cannot form cooperative decreased inhibitory receptor expression, increased cytokine produc- transcriptional complexes with AP-1 (Fos–Jun)10. CA-RIT-NFAT1- tion and strong enrichment in accessible chromatin for binding motifs transduced cells display a transcriptional program that mimics the early of transcription factors involved in effector function. stages of in vivo exhaustion and/or dysfunction and also express higher We confirmed the binding of haemagglutinin (HA)-tagged NR4A levels of NR4A transcription factors than mock-transduced cells10–12. proteins to selected differentially accessible regions in CD8+ T cells Regions that were more accessible in wild-type versus Nr4a TKO by chromatin immunoprecipitation and quantitative PCR (ChIP– TILs were also more accessible in CA-RIT-NFAT1-expressing versus qPCR). Ccr7, which is highly expressed in naive and memory T cells mock-transduced cells (Extended Data Fig. 9c, top) and in cells trans- and decreased in effector T cells29, is expressed at lower levels in effec- duced with NR4A1, NR4A2 or NR4A3 versus cells transduced with an tor-like Nr4a TKO compared to wild-type TILs (Fig. 4a). The Ccr7 distal empty vector (Extended Data Fig. 9c, middle three panels). Conversely, 5′ region contains two ATAC-seq peaks that are less prominent in Nr4a regions more accessible in Nr4a TKO versus wild-type TILs were more TKO than in wild-type TILs and contain adjacent NFAT- and NR4A- accessible in cells stimulated with PMA and ionomycin compared with binding motifs (Extended Data Fig. 9a, middle panel, peach lines). These resting cells (Extended Data Fig. 9c, bottom). PMA and ionomycin regions bind NR4A (Extended Data Fig. 9a, right, bar plots). In contrast, stimulate bZIP and NFκB transcription factors, indicating that the two ATAC-seq peaks in the Ccr7 proximal promoter and first intron are in vivo effector function of Nr4a TKO compared to wild-type TILs is more prominent in Nr4a TKO compared to wild-type TILs and contain associated with increased activity of these transcription factors.

NAtUre| www.nature.com/nature Letter RESEARCH

Each NR4A protein binds to and is partly responsible for increased 17. Brogdon, J., June, C. H., Loew, A., Maus, M. & Scholler, J. Treatment of cancer using humanized anti-cd19 chimeric antigen receptor. WO patent accessibility of an enhancer located at around 23 kb 5′ of the Pdcd1 WO2014153270A1 (2014). transcription start site (Fig. 4d), noted in all mouse models of exhaus- 18. Im, S. J. et al. Defning CD8+ T cells that provide the proliferative burst after tion or dysfunction investigated so far11–13,15,19. The ATAC-seq peak PD-1 therapy. Nature 537, 417–421 (2016). 19. Pauken, K. E. et al. Epigenetic stability of exhausted T cells limits durability of marking this enhancer is diminished in Nr4a TKO compared to wild- reinvigoration by PD-1 blockade. Science 354, 1160–1165 (2016). type CAR T cells and increased in T cells ectopically expressing NR4A1, 20. Sekiya, T. et al. Nr4a receptors are essential for thymic regulatory T cell NR4A2 or NR4A3 compared to cells transduced with empty vector development and immune homeostasis. Nature Immunol. 14, 230–237 (2013). 21. Leong, Y. A. et al. CXCR5+ follicular cytotoxic T cells control viral infection in alone (Fig. 4d). Deletion of this enhancer results in a decrease in the B cell follicles. Nature Immunol. 17, 1187–1196 (2016). mean fluorescence intensity of PD-1 staining in the EL4 thymoma cell 22. He, R. et al. Follicular CXCR5-expressing CD8+ T cells curtail chronic viral line15. PD-1 blockade19 caused a significant (two-fold) decrease in levels infection. Nature 537, 412–416 (2016). 23. Utzschneider, D. T. et al. T Cell Factor 1-expressing memory-like CD8+ T cells of Nr4a2 mRNA, with a smaller decrease in Nr4a1 and Nr4a3 mRNAs sustain the immune response to chronic viral infections. Immunity 45, (Extended Data Fig. 10a). These data indicate that, together with NFAT, 415–427 (2016). the three NR4A transcription factors are prominent and redundant 24. Subramanian, A. et al. Gene set enrichment analysis: a knowledge-based + approach for interpreting genome-wide expression profles. Proc. Natl Acad. Sci. effectors of the CD8 T cell hyporesponsive program downstream of USA 102, 15545–15550 (2005). NFAT (Fig. 4e). Moreover, because NR4A deficiency results in down- 25. Nowyhed, H. N. et al. The nuclear receptor nr4a1 controls CD8 T cell development regulation of the inhibitory receptors PD-1 and TIM3, the effect of through transcriptional suppression of runx3. Sci. Rep. 5, 9059 (2015). + NR4A deficiency is functionally similar to that of PD-1 blockade19, 26. Milner, J. J. et al. Runx3 programs CD8 T cell residency in non-lymphoid tissues and tumours. Nature 552, 253–257 (2017). but NR4A deficiency affects a wider range of regulatory elements than 27. Harant, H. & Lindley, I. J. D. Negative cross-talk between the human orphan PD-1 blockade alone (Fig. 4e). nuclear receptor Nur77/NAK-1/TR3 and nuclear factor-kappaB. Nucleic Acids Although immune cell therapies offer considerable promise for the Res. 32, 5280–5290 (2004). 28. Saijo, K. et al. A Nurr1/CoREST pathway in microglia and astrocytes protects treatment of cancer, treatment with individual blocking antibodies dopaminergic neurons from infammation-induced death. Cell 137, 47–59 (2009). against targets such as PD-1 and CTLA4 rarely achieve complete cures. 29. Best, J. A. et al. Transcriptional insights into the CD8+ T cell response to We have shown that the NFAT–NR4A axis controls the expression of infection and memory T cell formation. Nature Immunol. 14, 404–412 (2013). 30. Spolski, R. & Leonard, W. J. Interleukin-21: basic biology and implications for multiple inhibitory receptors, and that treatment of tumour-bearing cancer and autoimmunity. Annu. Rev. Immunol. 26, 57–79 (2008). mice with CAR T cells lacking all three NR4A transcription factors 31. Larkin, J. et al. Combined nivolumab and ipilimumab or monotherapy in resulted in tumour regression and prolonged survival. Inhibiting untreated melanoma. N. Engl. J. Med. 373, 23–34 (2015). the function of NR4A family members in tumour-infiltrating T cells Acknowledgements We would like to thank C. Kim, L. Nosworthy, R. Simmons, could be a promising strategy in cancer immunotherapy as it would D. Hinz and C. Dillingham of the LJI Flow Cytometry Core Facility for cell be expected to mimic combination therapies with blocking antibodies sorting; J. Day, S. Wlodychak and C. Kim of the LJI Next Generation Sequencing 31 Facility for next-generation sequencing; S. Schoenberger for B16-OVA cells; against multiple inhibitory receptors . A. W. Goldrath for MC38 cells; V. Wong, S. Trifari and G. Mognol for advice and discussions; B. Peters for statistics discussions; and the Department of Online content Laboratory Animal Care (DLAC) and the animal facility for excellent support. Any methods, additional references, Nature Research reporting summaries, source This work was funded in part by the US National Institutes of Health (NIH) AI109842, AI040127, S10OD016262, S10 RR027366 (A.R.); NIH T32 data, statements of data availability and associated accession codes are available at GM007752 and PhRMA Foundation Paul Calabresi Medical Student Research https://doi.org/10.1038/s41586-019-0985-x. Fellowship (J.C.); UC MEXUS-CONACYT Fellowship (I.F.L.-M.); AACR-Genentech Immuno-oncology Research Fellowship, 18-40-18-SEO (H.S.); Cancer Research Received: 31 December 2017; Accepted: 29 January 2019; Institute (CRI) Irvington Postdoctoral Fellowship (C.-W.J.L.); Fraternal Order of Published online xx xx xxxx. Eagles Fellow of the Damon Runyon Cancer Research Foundation, DRG-2069- 11 (J.P.S.-B.); JSPS KAKENHI Scientific Research (B) 16KT0114 (T.S.); and JSPS KAKENHI (S) JP17H06175, Challenging Research (P) JP18H05376, and 1. Davila, M. L. et al. Efcacy and toxicity management of 19-28z CAR T cell Advanced Research & Development Programs for Medical Innovation (AMED- therapy in B cell acute lymphoblastic leukemia. Sci. Transl. Med. 6, 224ra25 CREST) JP18gm0510019, JP18gm1110009 (A.Y.). (2014). 2. Maude, S. L. et al. Chimeric antigen receptor T cells for sustained remissions in Reviewer information Nature thanks Takeshi Egawa, Golnaz Vahedi and the leukemia. N. Engl. J. Med. 371, 1507–1517 (2014). other anonymous reviewer(s) for their contribution to the peer review of this work. 3. Ahmed, N. et al. Human Epidermal Growth Factor Receptor 2 (HER2)-specifc chimeric antigen receptor-modifed T cells for the immunotherapy of Author Contributions J.C. designed and performed experiments, analysed HER2-positive sarcoma. J. Clin. Oncol. 33, 1688–1696 (2015). data and prepared the sequencing libraries; I.F.L.-M. performed computational 4. O’Rourke, D. M. et al. A single dose of peripherally infused EGFRvIII-directed analyses of the RNA-seq and scRNA-seq data; H.S. assisted with in vivo CAR T cells mediates antigen loss and induces adaptive resistance in patients mouse experiments and in vitro experiments; C.-W.J.L. performed ChIP and with recurrent glioblastoma. Sci. Transl. Med. 9, eaaa0984 (2017). ChIP–qPCR; L.J.H. assisted with in vivo mouse experiments; T.S. and A.Y. 5. Moon, E. K. et al. Multifactorial T-cell hypofunction that is reversible can limit the gave advice and provided the Nr4a-gene-disrupted mice (with permission efcacy of chimeric antigen receptor-transduced human T cells in solid tumors. from P. Chambon); J.P.S.-B. conceived the mouse CAR T cell model, designed Clin. Cancer Res. 20, 4262–4273 (2014). experiments and performed computational analyses of the ATAC-seq data; A.R. 6. Wherry, E. J. T cell exhaustion. Nature Immunol. 12, 492–499 (2011). supervised the project. J.C., J.P.S.-B. and A.R. interpreted data and wrote the 7. Wherry, E. J. et al. Molecular signature of CD8+ T cell exhaustion during chronic manuscript, with all authors contributing to writing and providing feedback. viral infection. Immunity 27, 670–684 (2007). 8. Schietinger, A. et al. Tumor-specifc T cell dysfunction is a dynamic antigen- Competing interests The La Jolla Institute of Immunology has a pending patent, driven diferentiation program initiated early during tumorigenesis. Immunity PCT/US2018/062354, covering the use and production of engineered immune 45, 389–401 (2016). cells to disrupt NFAT-AP1 pathway transcription factors, including the NR4A 9. Schietinger, A. & Greenberg, P. D. Tolerance and exhaustion: defning family members, with J.C., H.S., J.P.S-B. and A.R. listed as inventors. A.R. receives mechanisms of T cell dysfunction. Trends Immunol. 35, 51–60 (2014). funding from Takeda for subsequent research related to this subject matter. 10. Martinez, G. J. et al. The transcription factor NFAT promotes exhaustion of None of the other authors has any competing interests. activated CD8+ T cells. Immunity 42, 265–278 (2015). 11. Scott-Browne, J. P. et al. Dynamic changes in chromatin accessibility occur in Additional information CD8+ T Cells responding to viral infection. Immunity 45, 1327–1340 (2016). Extended data is available for this paper at https://doi.org/10.1038/s41586- 12. Mognol, G. P. et al. Exhaustion-associated regulatory regions in CD8+ 019-0985-x. tumor-infltrating T cells. Proc. Natl Acad. Sci. USA 114, E2776–E2785 (2017). Supplementary information is available for this paper at https://doi.org/ 13. Philip, M. et al. Chromatin states defne tumour-specifc T cell dysfunction and 10.1038/s41586-019-0985-x. reprogramming. Nature 545, 452–456 (2017). Reprints and permissions information is available at http://www.nature.com/ 14. Tirosh, I. et al. Dissecting the multicellular ecosystem of metastatic melanoma reprints. by single-cell RNA-seq. Science 352, 189–196 (2016). Correspondence and requests for materials should be addressed to J.C., 15. Sen, D. R. et al. The epigenetic landscape of T cell exhaustion. Science 354, J.P.S.-B. or A.R. 1165–1169 (2016). Publisher’s note: Springer Nature remains neutral with regard to jurisdictional 16. Nicholson, I. C. et al. Construction and characterisation of a functional CD19 claims in published maps and institutional affiliations. specifc single chain Fv fragment for immunotherapy of B lineage leukaemia and lymphoma. Mol. Immunol. 34, 1157–1165 (1997). © The Author(s), under exclusive licence to Springer Nature Limited 2019

NAtUre| www.nature.com/nature RESEARCH Letter

METHODS colon adenocarcinoma cell line (a gift from A.W. Goldrath, UCSD, La Jolla, CA) Data reporting. No statistical methods were used to predetermine sample size. was originally purchased from Kerafast, Inc. (ENH204). The EL4 cell line stained The investigators were not blinded to allocation during experiments and outcome positive for mouse Thy1.2 and PD-1 and stained negative for hCD19. The B16- assessment. OVA cell line stained negative for hCD19. The MC38 cell line stained negative for Construction of retroviral vector (MSCV-myc-CAR-2A-Thy1.1) containing hCD19. The PlatE and 293T cell lines were not authenticated. Cell lines were not chimeric antigen receptor (CAR). The chimeric antigen receptor was pieced tested for mycoplasma contamination. together using published portions of the clone FMC63 human CD19 single chain Construction of mouse tumour cell lines expressing hCD19. B16-OVA, EL4 and variable fragment16,17, and the published portions of the murine CD28 and CD3ζ MC38 cells were transduced with an amphotropic virus containing human CD19 sequences32. The sequence for the myc tag on the N terminus was obtained from (hCD19) and then sorted for cells expressing high levels of hCD19. published work33. This chimeric antigen construct was then cloned into an MSCV- Preparation of B16-OVA-hCD19 melanoma cells for tumour inoculation. B16- puro (Clontech) murine retroviral vector in place of the PGK-puro. OVA-hCD19 cells were cultured in Dulbecco’s medium (DMEM) with 10% (vol/ Sequence of CAR construct (nt): (1,470 bp). 5′-ATGGCTTTGCCAGTGACA vol) FBS, 1% l-glutamine, 1% penicillin/streptomycin and passaged three times GCTCTTCTCCTTCCACTGGCCCTCCTCCTTCACGCCGCTAGGCCAGAG before inoculation. At the time of injection, cells were trypsinized and resuspended CAGAAACTTATTTCAGAGGAAGACCTGGACATTCAAATGACACAAACT in Hanks balanced salt solution without phenol red at 10 million cells per millilitre. ACTTCTTCTCTCTCCGCCTCACTTGGTGACCGCGTCACTATTAGTTGCC C57BL/6J male or female mice (8–12 weeks old) were injected intradermally with GCGCTAGTCAAGATATTAGTAAGTACCTGAATTGGTATCAACAAAAACC 500,000 B16-OVA-hCD19 cells (50 μl per injection). TGACGGGACTGTAAAGCTGCTTATATATCATACTTCTAGGCTGCATTCTG Preparation of MC38-hCD19 colon adenocarcinoma cells for tumour inocula- GAGTACCTTCACGATTTAGCGGTAGCGGATCCGGCACCGACTACTCCC tion. MC38-hCD19 cells were cultured in Dulbecco’s medium (DMEM) with 10% TCACAATTAGCAATCTGG AGCAAGAGGACATAGCCACCTACTTCTG (vol/vol) FBS, 0.1mM non-essential amino acids, 1 mM sodium pyruvate, 10mM CCAGCAAGGGAATACCTTGCCATACACTTTCGGTGGTGGAACTAAGCTC Hepes, 1% l-glutamine, 1% penicillin/streptomycin and passaged two times before GAAATTACTGGGGGTGGAGGCAGTGGCGGAGGGGGGTCAGGTGGGGG inoculation. At the time of injection, cells were trypsinized and resuspended in AGGTTCAGAAGTCAAACTCCAGGAATCTGGACCTGGACTCGTTGCCCC Hanks balanced salt solution without phenol red at 10 million cells per millilitre. TTCCCAATCCCTTAGTGTTACATGCACTGTATCAGGTGTATCCCTCCCTG C57BL/6J male or female mice (8–12 weeks old) were injected intradermally with ATTACGGTGTCTCCTGGATTCGGCAGCCTCCTCGGAAGGGTCTCGAGTG 500,000 MC38-hCD19 cells (50 μl per injection). GTTGGGAGTGATTTGGGGGTCTGAAACTACTTATTATAACAGTGCCCTT Mice. C57BL/6J, B6.SJL-PtprcaPepcb/BoyJ, Rag1−/− mice were obtained from AAGAGTAGATTGACTATAATTAAGGATAA CAGTAAGTCACAAGTATTCC Jackson Laboratories. Nr4a gene-disrupted strains were obtained from Takashi TCAAAATGA ATTCCTTGCAAACAGACGATACAGCAATATATTACTG Sekiya and Akihiko Yoshimura, with permission from Pierre Chambon. Both male CGCAAAACACTACTACTATGGC GGTAGTTACGCT ATGGACTATTGGG and female mice were used for studies. Mice were age-matched and between 8 GTCAAGGAACCTCTGTC ACAGTTTCTAGCATTGAGTTCATGTATCCCCC and 12 weeks old when used for experiments, and tumour-bearing mice were first ACCTTACTTGGA CAATGAAAGGTCTAATGGGACCATCATACACAT tumour size-matched and then randomly assigned to experimental groups. All TAAAGAGAAACACCTGTGTCATACTCAGAGTTCTCCAAAATTGTTCTGG mice were bred and/or maintained in the animal facility at the La Jolla Institute for GCCTTGGTTGTCGTTGCCGGCGTACTGTTCTGTTACGGTCTCTTGGTT Immunology. All experiments were performed in compliance with the study pro- ACCGTGGCACTTTGTGTTATCTGGACTAATTCCCGGCGGAATCGGGGT tocol approved by the LJI Institutional Animal Care and Use Committee (IACUC) GGACAGAGCGATTACATGAATATGACCCCAAGAAGACCTGGACTGACC regulations. AGGAAACCATATCAACCCTATGCTCCT­ GC­ TC­ GG­ GA­ CT­ TT­ GC­ TG­ CT­ TA­ C­ B16-OVA-hCD19 tumour model. For analysis of CAR CD8+ TILs and endog- C­GC­CC­AC­GCGCAAAGTTTTCTAG­GA­GC­GC­TG­AA­AC­CG­CT­GC­CA­AC­CT­C enous CD8+ TILs: on day 0, 8–12-week-old C57BL/6J mice were injected intra- CAA­ GA­ CC­ CT­ AATCAGCTTTACAATGAATT­ GA­ ACTT­ GG­ GA­ CG­ CC­ GG­ GA­ GG­ dermally with 5 × 105 B16-OVA-hCD19 cells. After tumours became palpable, AG­ TA­ TG­ AC­ GT­ CC­ TT­ GA­ GAAAAAGCGGGCTCGGGATCCAGAAATGGGC­ tumour measurements were recorded with a manual caliper every other day and GGAAAGCAACAGAGGCGAAGAAATCCACAAGAGGGGGTCTATAA­ CG­ C­ tumour area was calculated in centimeters squared (length × width). On day TCT­ TC­ AG­ AA­ AG­ AT­ AA­ AA­ TG­ GC­ TG­ AG­ GCATATAGCGAAATTGGG­ AC­ CA­ A­ 13, 1.5 million CAR-transduced CD45.1+CD8+ T cells were adoptively trans- GGG­ GG­ AG­ AG­ AA­ GA­ AG­ AG­ GC­ AA­ GG­ GA­ CA­ TG­ ACGGGCTTTACCAGGGTT­ ferred into tumour size-matched tumour-bearing mice. On day 21, tumours TGTCTACCGCAACAAAAGACACCTATGATGCTTTGCACATGCAAACACT were collected from mice. For analysis of CAR CD8+ TILs lacking NR4A family GGCTCCTAGA-3′. members: because the Nr4a gene-disrupted mice were originally derived from Sequence of CAR construct (aa): (490 aa). 5′-MALPVTALLLPLALLLHAARPEQ 129/SvJ embroyonic stem cells34, and their genetic background might not have KLISEEDLDIQMTQTTSSLSASLGDRVTISCRASQDISKYLNWYQQKPDGTVK been fully compatible with that of inbred C57BL/6J mice despite stringent back- LLIYHTSRLHSGVPSRFSGSGSGTDYSLTISNLEQEDIATYFCQQGNTLPYTFG crossing, we used Rag1-deficient mice as recipients in most experiments to avoid GGTKLEITGGGGSGGGGSGGGGSEVKLQESGPGLVAPSQSLSVTCTVSGVSL variable rejection. On day 0, 8–12-week-old Rag1−/− mice were injected intra- PDYGVSWIRQPPRKGLEWLGVIWGSETTYYNSALKSRLTIIKDNSKSQVFLK dermally with 5 × 105 B16-OVA-hCD19 cells and tumours were measured every MNSLQTDDTAIYYCAKHYYYGGSYAMDYWGQGTSVTVSSIEFMYPPPYLDN other day after they became palpable. On day 13, 1.5 million CAR- and empty ERSNGTIIHIKEKHLCHTQSSPKLFWALVVVAGVLFCYGLLVTVALCVIWTN vector pMIN-transduced CD8+ Thy1.1+ NGFR+ Nr4a1fl/flNr4a2fl/flNr4a3+/+ SRRNRGGQSDYMNMTPRRPGLTRKPYQPYAPARDFAAYRPRAKFSRSAETAA (wild type) or CAR- and Cre-transduced CD8+ Thy1.1+ NGFR+ Nr4a1fl/fl NLQDPNQLYNELNLGRREEYDVLEKKRARDPEMGGKQQRRRNPQEGVYN Nr4a2fl/flNr4a3−/− (Nr4a TKO) T cells were adoptively transferred into tumour ALQKDKMAEAYSEIGTKGERRRGKGHDGLYQGLSTATKDTYDALHMQTL size-matched tumour-bearing mice. On day 21, tumours were collected from the APR-3′. mice. For monitoring of tumour growth for survival studies after adoptive transfer Construction of retroviral vector containing hCD19. DNA fragment encoding of CAR T cells lacking NR4A family members: again, because the Nr4a-gene- hCD19 was PCR-amplified and cloned into an MSCV-puro (Clontech) murine disrupted mice were originally derived from 129/SvJ embryonic stem cells34, and retroviral vector. their genetic background might not have been fully compatible with that of inbred Construction of retroviral vectors containing Cre (MSCV-Cre-IRES-NGFR), C57BL/6J mice despite stringent backcrossing, we used Rag1-deficient mice as and Nr4a1, Nr4a2, Nr4a3 (MCSV-HA-Nr4a1-IRES-NGFR, MCSV-HA-Nr4a2- recipients in most experiments to avoid variable rejection. On day 0, 8–12-week- IRES-NGFR, MCSV-HA-Nr4a3-IRES-NGFR). DNA fragment encoding Cre was old Rag1−/− mice were injected intradermally with 5 × 105 B16-OVA-hCD19 cells PCR-amplified and cloned into MSCV-IRES-NGFR (Addgene plasmid 27489). and tumours were measured every other day after they became palpable. On day 7, DNA fragment encoding Nr4a1 (a gift from C.-W. J. Lio, La Jolla Institute for 3 million CAR- and empty vector pMIN-transduced or CAR- and Cre-transduced Immunology) was PCR-amplified with 5′ HA-tag and cloned into MSCV-IRES- CD8+ Thy1.1+NGFR+Nr4a-floxed mouse T cells (in combinations to produce NGFR. DNA fragment encoding Nr4a2 (Addgene Plasmid 3500) was PCR- Nr4a1 KO, Nr4a2 KO, Nr4a3 KO, Nr4a TKO, wild type as listed in Extended amplified with 5′ HA-tag and cloned into MSCV-IRES-NGFR. DNA fragment Data Fig. 6a) were adoptively transferred into tumour size-matched Rag1−/− encoding Nr4a3 (DNASU plasmid MmCD00080978) was PCR-amplified with 5′ tumour-bearing mice. For the experiments using immunocompetent recipients: HA-tag and cloned into MSCV-IRES-NGFR. on day 7, 6 million CAR- and empty vector pMIG-transduced or CAR- and Cre- Eukaryotic cell lines. The EL4 mouse thymoma cell line was purchased from the transduced CD8+Thy1.1+GFP+Nr4a-floxed mouse T cells (strains as above to American Type Culture Collection (ATCC): EL4 (ATCC TIB-39, Mus musculus produce Nr4a TKO and wild type) were adoptively transferred into tumour size- T cell lymphoma). The B16-OVA mouse melanoma cell line expressing the oval- matched C57BL/6J tumour-bearing mice. For all survival studies, tumour growth bumin protein (a gift from S. Schoenberger, La Jolla Institute for Immunology) was monitored until experimental endpoint on day 90 after tumour inoculation or was previously described12. The 293T cell line was purchased from ATCC: 293T until IACUC-approved endpoint of a maximal tumour measurement exceeding a (ATCC CRL-3216). The Platinum-E Retroviral Packaging Cell Line, Ecotropic diameter greater than 1.5cm for more than three days without signs of regression. (PlatE) cell line was purchased from Cell BioLabs, Inc.: RV-101. The MC38 mouse In none of the experiments was any of these limits exceeded. Letter RESEARCH

MC38-hCD19 tumour model. The monitoring of tumour growth for survival Staining Kit (eBioscience) according to manufacturer’s instructions. All transcrip- studies after adoptive transfer of CAR T cells lacking NR4A family members into tion factor antibodies were used at 1:200 final concentration. The antibody for Ki67 C57BL/6J mice bearing MC38-hCD19 tumours was performed as described for was used at 1:100 final concentration. the B16-OVA-hCD19 model using immunocompetent recipients. Flow cytometry analysis. All flow cytometry analysis was performed using the Preparation of cells for adoptive transfer. CD8+ T cells were isolated and acti- LSRFortessa (BD Biosciences) or the LSR-II (BD Biosciences). Flow data was ana- vated with 1 μg ml−1 anti-CD3 and 1 μg ml−1 anti-CD28 for 1 day, then removed lysed using FlowJo v.10 (Tree Star, Inc.). Relevant sample gating has been provided from activation and transduced with retrovirus expressing CAR, Cre, empty vector in extended data figures. (corresponding to Cre) or a combination of the above for 1 h at 37 °C and 2,000g. Statistical analyses. Statistical analyses on flow cytometric data and tumour Immediately after the transduction, cells were replaced with media containing growth data were performed using the appropriate statistical comparison, includ- 100 U of IL-2 per ml. One day after the first transduction, a second transduction ing paired or unpaired two-tailed t-tests with Welch’s correction as needed, was performed and immediately after the transduction, cells were replaced with one-way ANOVA with multiple comparisons test (Tukey’s or Dunnett’s), media containing 100 U of IL-2 per ml. On the day of adoptive transfer (either day row-matching (RM) one-way ANOVA with Greenhouse-Geisser correction, or 3 or day 5 post activation), cells were analysed by flow cytometry and cell counts ordinary two-way ANOVA (Prism 7, GraphPad Software). Statistical analyses for were obtained using a haemocytometer. The number of CAR-transduced cells survival curves were performed using the log-rank (Mantel–Cox) test (Prism 7, was obtained using the cell counts from the haemocytometer and the population GraphPad Software). P ≤ 0.05 was considered statistically significant. percentages obtained from flow cytometry. Cells were then collected, washed with In vitro killing assay. Approximately 10,000 B16-OVA-hCD19 cells (target cells) PBS and resuspended at a concentration equivalent of 1.5 million, 3 million or were plated in 100 μl of T cell media (or media only for background) in each well 6 million CAR-transduced cells per 200 μl of PBS. Mice were then adoptively in E-plate 96 (ACEA Biosciences Inc.). Plate was placed in xCELLigence Real- transferred with 200 μl of retro-orbital intravenous injections each. Time Cell Analysis (RTCA) instrument (ACEA Biosciences Inc.) after 30 min and Isolation of TILs for subsequent analyses. Sample preparation for flow cytometry incubated overnight. The following day, the plate was removed from xCELLigence and cell sorting of TILs from CAR and OT-I experiments and for flow cytometry RTCA machine and CD8+ CAR T cells (effector cells) were added in an additional of TILs from Nr4a TKO versus wild-type experiments: on day 21, mice were killed 100 μl of T cell media for 30 min (for lysis positive control, 0.2% TritonX was used, and perfused with PBS before removal of tumour. Tumours were collected, pooled for lysis negative control, only media was used). The plate was then placed back together by group, homogenized, and then dissociated using the MACS Miltenyi into the incubator, and data acquisition began. 5 h after, the cell index (CI) was Mouse Tumour Dissociation kit (Miltenyi Biotec) and the gentleMACs dissocia- obtained from each well. Percentage of specific lysis was calculated for each well tor with Octo Heaters (Miltenyi Biotec) according to manufacturer’s instructions. as follows: percentage of specific lysis = 100 − (CIeach well/(CIpos − CIneg)) × 100. Tumours were then filtered through a 70 μM filter and spun down. Supernatant B16-OVA-hCD19 cells were thawed out 3 days before plating on day 4 (when was aspirated and the tumours were resuspended in the equivalent of 4–5 g of inoculation would usually occur); mouse CD8+ CAR T cells were prepared before tumour per 5 ml of 1%FBS/PBS for CD8 positive isolation using the Dynabeads the experiment to be added to target cells on day 5 post activation of CD8+ T cells. FlowComp Mouse CD8 isolation kit (Invitrogen). After positive isolation, cells Chromatin immunoprecipitation and quantitative PCR (ChIP–qPCR). ChIP were either divided into equal amounts for staining and phenotyping with flow was performed as previously described35. In brief, CD8+ T cells were isolated from cytometry or stained for cell sorting. Sample preparation for cell sorting of TILs C57BL/6J mice as above, activated with plate-bound anti-CD3/CD28, transduced from Nr4a wild-type and Nr4a TKO experiments: on day 21, mice were killed and with either empty vector control or retrovirus expressing NR4A1, NR4A2, or perfused with PBS before removal of tumours. Tumours were collected, pooled NR4A3 with HA-tag on the N terminus. Cells were cultured for a total of 5 days together by group, homogenized, and then dissociated using the MACS Miltenyi post-transduction. For fixation, formaldehyde (16%, ThermoFisher) was added Mouse Tumour Dissociation kit (Miltenyi Biotec) and the gentleMACs dissocia- directly to the cells to a final concentration of 1% and incubated at room temper- tor with Octo Heaters (Miltenyi Biotec) according to manufacturer’s instructions. ature for 10 min with constant agitation. Glycine (final 125mM) was added to Tumours were then filtered through a 70 μM filter and spun down. Supernatant was quench the fixation and the cells were washed twice with ice-cold PBS. Cell pellets aspirated and the tumours were resuspended in 40% Percoll/RPMI and underlaid were snap-frozen with liquid nitrogen and stored at −80 °C until use. For nuclei with 80% Percoll/PBS in 15 ml conical tubes to form an 80%/40% Percoll discon- isolation, cell pellets were thawed on ice and lysed with lysis buffer (50 mM HEPES tinuous density gradient. Samples were spun for 30 min at room temperature at pH 7.5, 140 mM NaCl, 1mM EDTA, 10% glycerol, 0.5% NP40, 0.25% Triton-X100) 1,363g in a large benchtop centrifuge with a swinging bucket. TILs were collected supplemented with 1% Halt protease inhibitor (ThermoFisher) for 10 min at 4 °C from 80%/40% Percoll interface and further purified using CD90.2 Microbeads with constant rotation. Pellets were washed once with washing buffer (10 mM (Miltenyi Biotec) and magnetic separation. After positive isolation, cells were Tris-HCl pH 8.0, 200 mM NaCl, 1 mM EDTA, 0.5 mM EGTA, 1% Halt protease stained for cell sorting. inhibitor) and twice with shearing buffer (10 mM Tris-HCl pH 8.0, 1 mM EDTA, Transfections. Transfections were performed in 10cm dish format, following 0.1% SDS, 1% Halt protease inhibitor). Nuclei were resuspended in 1 ml shearing manufacturer’s instructions for the TransIT-LT1 Transfection Reagent (Mirus Bio buffer, transferred to 1 ml milliTUBE (Covaris, Woburn, MA), and sonicated with LLC) and using the pCL10A1 and pCL-Eco packaging vectors (the former for the Covaris E220 for 18 min (Duty Cycle 5%, intensity 140 Watts, cycles per burst 200). hCD19 virus, and the latter for all other viruses produced). After sonication, insoluble debris was removed by centrifugation at 20,000g for Retroviral transduction. Retroviral transductions were performed in six-well plate 10 min at 4 °C. The concentration of chromatin was quantified using Qubit DNA format, using 3 ml of 0.45 μM filtered virus and 15 μg ml−1 of polybrene per well. BR assay (ThermoFisher). For immunoprecipitation, 25 μg of chromatin was Double transductions were performed using 1.5 ml of each virus for a total of 3 ml. removed and mixed with equal volume of 2× Conversion buffer (10 mM Tris-HCl Cells were spun at 2,000g for 1 h at 37 °C in a pre-warmed centrifuge. Immediately pH 7.5, 280 mM NaCl, 1 mM EDTA, 1 mM EGTA, 0.2% sodium deoxycholate, after the transduction, cells were replaced with media containing 100 U of IL-2 per 0.2% Triton-X100, 1% Halt protease inhibitor) in a 2-ml low-binding tube ml. A second transduction is performed the following day. (Eppendorf). Either 5% or 6% of input chromatin was saved as control. Chromatin Antibodies. Fluorochrome-conjugated antibodies were purchased from Biolegend, was pre-cleared using 30 μl washed protein A magnetic dynabeads (ThermoFisher) BD Sciences, eBioscience, and Cell Signaling Technologies. Primary antibody for 1h at 4 °C with constant rotation. Pre-cleared chromatin was transferred to used for chromatin-immunoprecipitation was purchased from Cell Signaling new tube, added with 10 μg rabbit monoclonal anti-HA (C29F4, Cell Signaling Technologies. Technology) and 30 μl washed protein A magnetic dynabeads, and incubated at Surface marker staining. Cells were spun down and stained with 1:200 final con- 4 °C overnight with constant rotation. Bead-bound chromatin was washed twice centration of antibodies in 50% of 2.4G2 (Fc block) and 50% of FACS Buffer (PBS with RIPA buffer (50 mM Tris-HCl pH 8.0, 250 mM LiCl, 1 mM EDTA, 1% sodium + 1% FBS, 2mM EDTA) for 15 min. deoxycholate, 1% NP-40, 0.1% SDS), once with high salt washing buffer (50 mM Cytokine restimulation and staining. Prior to staining, cells were incubated in Tris-HCl pH 8.0, 500 mM NaCl, 1 mM EDTA, 1% NP-40, 0.1% SDS), once with media containing 10 nM of PMA and 500 nM of ionomycin, and 1 μg ml−1 of Lithium washing buffer (50 mM Tris-HCl pH 8.0, 150 mM NaCl, 1 mM EDTA, brefeldin A at 37 °C for 4 h. After restimulation, cells were then stained for surface 1% sodium deoxycholate, 1% NP-40), and once with TE (10 mM Tris-HCl pH markers and with live/dead dye as described in the surface marker staining protocol 8.0, 1 mM EDTA). All washes were incubated for 5 min at 4 °C with constant above. Cells were then fixed with 4% paraformaldehyde for 30 min, permeabi- rotation. Chromatin was eluted from beads by incubating with elution buffer lized with 1 × BD Perm/Wash (BD Biosciences) for 30 min, and then stained for (100 mM NaHCO3, 1% SDS) at room temperature for 30 min in the presence of cytokines at a final concentration of 1:200 in 1 × BD Perm/Wash buffer. 1 × BD 0.5 mg ml−1 of RNaseA (Qiagen). To de-crosslink protein and DNA, proteinase K Perm/Wash buffer was prepared according to manufacturer’s instructions. All wash (final 0.5 mg ml−1) and NaCl (final 200 mM) were added to the recovered super- steps were performed with FACS buffer (PBS + 1% FBS, 2 mM EDTA). natant and incubated at 65 °C overnight with constant shaking (1,000 r.p.m.) in a Transcription factor staining. Cells were stained for surface markers and with ThermoMixer (Eppendorf). DNA was purified using Zymo ChIP DNA clean and live/dead dye as described in the surface marker staining protocol above. Cells concentration kit (Zymo Research) according to the manual from the manufac- were then fixed, permeabilized, and stained using the Foxp3/Transcription Factor turer. Eluted DNA was analysed by qPCR using Power SYBR Green PCR Master RESEARCH Letter

Mix (Roche) and StepOne Real Time PCR system (ThermoFisher). The signals per kilobase. We used HOMER44 to identify motifs for transcription factor binding from ChIP sample was normalized to those from the input and calculated as ‘per- sites enriched in different groups of peaks. centage of input’. A value of ‘undetected’ was recorded as zero. RNA-seq sample and library preparation. Total RNA was extracted using the ChIP–qPCR primers (all coordinates are for mm10). (1) chrX:7584283–7584409 RNeasy Micro Kit (Qiagen). SMARTseq2 libraries were prepared as described45. 127 bp: Fp3-CNS2-qF (forward) 5′-CCCAACAGACAGTGCAGGAA-3′, (reverse) Briefly, purified RNA was hybridized to polyA to enrich for mRNA, and then Fp3-CNS2-qR 5′-TGGTGTGACTGTGTGATGCA-3′. (2) chr1:94074907– mRNA underwent reverse transcription and template switching before an 18-cycle 94075062 156 bp: Pd1.4A_qF1 (forward) 5′-ACCTTTCCTGTGCCTACGTC-3′, PCR preamplification step. PCR clean-up was then performed using AMPure XP Pd1.4A_qR1 (reverse) 5′-TAAGAGTGGTGGTGGTTGGG-3′. (3) chr11: beads (Beckman Coulter). Quality check of the cDNA library was performed using 99163437–99163632 196 bp: CCR7_E1_F1 (forward) 5′-GGCTCTACTGC an Agilent high-sensitivity DNA chip, and 1 ng of input cDNA was further used for CCTGTTGTC-3′, CCR7_E1_R1 (reverse) 5′-AACACATCATTTTGCCGTGA-3′. library preparation using the Nextera XT LibraryPrep kit (Illumina). Tagmented (4) chr11:99168432–99168614 183 bp: CCR7_E2_F1 (forward) 5′-GGACACAGA DNA was amplified with a 12-cycle PCR and again purified with AMPureXP beads. CGGGTGAGTTT-3′, CCR7_E2_R1 (reverse) 5′-GGCCTGTGTTCAAAT Library size distribution and yield were evaluated using the Agilent high-sensitivity GAGGT-3′. (5) chr17:8196147–8196301 155 bp: CCR6_F1 (forward) 5′-GGCAGG DNA chip. Libraries were pooled at equimolar ratios and sequenced with the rapid ATGTGGCTTTGTAT-3′, CCR6_R1 (reverse) CCTGCATGTAGTGCTGACCA run protocol on a HiSEq2500 (Illumina) with 50-nt single-end cycling. (6) chr10:118460432–118460610 179 bp: IfngE_F1 (forward) 5′-GCGCC RNA-seq analysis. Quality and adaptor trimming was performed on raw RNA- TAGAAGTTCAGTGCT-3′, IfngE_R1 (reverse) 5′-TTTGAGATGCA seq reads using TrimGalore! v0.4.5 (http://www.bioinformatics.babraham.ac.uk/ GCAGTTTGG-3′. projects/trim_galore/) with default parameters, retaining reads with minimal Cell sorting. Cell sorting was performed by the LJI Flow Cytometry Core, using length of 36 bp. Resulting single-end reads were aligned to mouse genome mm10 the FACSAria-I, FACSAria-II, or FACSAria-Fusion (BD Biosciences). For ATAC- using STAR v2.5.3a46 with alignment parameter ‘outFilterMismatchNmax 4’. seq, 50,000 cells were sorted from the isolated CD8+ TILs, with the exception of the Technical replicates were merged. RNA-seq analysis was performed at the gene OT-I samples, for which 15,000–30,000 cells were sorted. In most cases, a second level, employing the transcript annotations of the mouse genome mm10. Reads ATAC-seq technical replicate using 50,000 additional cells was prepared in parallel. aligning to annotated features were counted using the summarizeOverlaps function For RNA-seq, two technical replicates of 10,000 cells each were sorted from the (mode = ‘Union’) of the Bioconductor package GenomicAlignments42 v1.10.1. isolated CD8+ TILs. For the CAR and OT-I experiments, the populations sorted The DESeq2 package47 v1.14.1 was used to normalize the raw counts and identify + + + hi hi + were as follows: CD8 CD45.1 Thy1.1 PD-1 TIM3 CAR (population A), CD8 differentially expressed genes (FDR cutoff of P < 0.1 and fold change (log2 scale) CD45.1+ Thy1.1+ PD-1hiTIM3lo CAR (population B), CD8+CD45.1−Thy1.1−PD- ≥ 1 or ≤ −1 , unless otherwise specified). Genes with less than 10 reads total were 1hiTIM3hi endogenous cells (population C), CD8+CD45.1−Thy1.1−PD-1hiTIM3lo pre-filtered in all comparisons as an initial step. Transformed values (rlog) were endogenous cells (population D) and CD8+CD45.1−Thy1.1−PD-1loTIM3lo calculated within DESeq2 for data visualization. endogenous cells (population E), and CD8+CD45.1+PD-1hiTIM3hi OT-I (pop- Single-cell RNA-seq analysis. Data were obtained from a previously published ulation F). For the NR4A experiments, populations were sorted as follows: study on the cellular ecosystem of human melanoma tumours14. In brief, malignant CD8+Thy1.1+NGFR+ Nr4a wild-type TILs and CD8+Thy1.1+NGFR+ Nr4a TKO and non-malignant cells (including immune, stromal, and endothelial cells) were TILs. For the experiments ectopically expressing NR4A in vitro, populations were profiled by single-cell RNA-seq. Normalized expression values (Ei,j = log2(TP- + sorted on a set level of NGFR expression. Mi,j/10 + 1), where TPMi,j refers to transcripts per million (TPM) for gene i in cell j) ATAC-seq sample and library preparation. ATAC-seq samples were prepared were obtained from Gene Expression Omnibus (GSE72056). For the analysis, we as in a previous study36 with minor modifications. In brief, cells were sorted into only kept genes with non-zero expression values in at least 10 cells. Given the 50%FBS/PBS, spun down, washed once with PBS, and then lysed. Transposition technical noisiness and gene dropout associated with single-cell RNA-seq data, we reaction was performed using Nextera enzyme (Illumina) and purified using the used the MAGIC algorithm48 for imputation in the matrix of normalized expres- MinElute kit (Qiagen) before PCR amplification (KAPA Biosystems) with 10–12 sion values, with diffusion parameter t = 2. An R implementation of the MAGIC cycles using barcoded primers and 2 × 50 cycle paired-end sequencing (Illumina). method was downloaded from (https://www.krishnaswamylab.org/magic-project). ATAC-seq analysis. Sequencing reads in FASTQ format were generated from Tumour-infiltrating T cells were selected based on the inferred cell type annota- Illumina Basespace (for mouse data sets) or were from published data13,15. Reads tion in a previous study14. CD8+ T cells were selected based on the expression of were mapped to mouse (mm10) or human (hg19) genomes using bowtie37 (ver- CD8A (cells with imputed values ≥ 4) and CD4 (cells with imputed values ≤ 1.5). sion 1.0.0 with parameters ‘-p 8 -m 1–best–strata -X 2000 -S–fr–chunkmbs 1024’. Imputed values were used for gene expression visualizations. Unmapped reads were processed with trim_galore using parameters ‘–paired– Gene set enrichment analyses. Gene set enrichment analysis (GSEA) was per- nextera–length 37–stringency 3–three_prime_clip_R1 1–three_prime_clip_R2 1’ formed employing the GSEA Preranked function24, ranking genes by fold change before attempting to map again using the above parameters. These two bam files according to the pertinent comparison, with number of permutations of 10,000 were merged and processed to remove reads mapping to the mitochondrial genome and allowing for gene set size up to 2,000 genes. Gene sets were defined from dif- and duplicate reads (with picard MarkDuplicates). For mouse data sets, technical ferentially expressed genes obtained from pairwise comparisons between effector, replicates were merged together into one single biological replicate at this point. For memory, and exhausted CD8+ T cells from a previously published study11. In this human data sets, samples with low coverage or which did not meet quality control context, differential gene expression was identified employing DESeq2 with FDR metrics were excluded. For the one human sample with two technical replicates, cutoff of P < 0.01 and fold change (log2 scale) cutoff of 1. these matched closely and number 1 was chosen for the analysis. Genomic coverage Reporting summary. Further information on research design is available in the for individual replicates were computed on 10 bp windows with MEDIPS38 using Nature Research Reporting Summary linked to this paper. full fragments captured by ATAC-seq and used to generate average coverage with the Java Genomics Toolkit39 for each group. Data availability To identify peaks, the bam files containing unique, non-chrM reads were pro- All data generated and supporting the findings of this study are available within the cessed with samtools and awk using ‘{if(sqrt($9*$9)<100){print $0}}’ to identify paper. RNA-seq and ATAC-seq data are available in the Gene Expression Omnibus nucleosome free DNA fragments less than 100 nt in length. These subnucleosomal (GEO) database under the SuperSeries reference number GSE123739. Source Data fragments were used to call peak summits for each replicate with MACS2 using for Figs. 2, 4 and Extended Data Figs. 2, 4, 7, 8, 9 are provided in Supplementary parameters ‘--nomodel --nolambda --keep-dup all --call-summits’. For peak calling, Tables 1–5. Additional Source Data are provided in the online version of the paper. we used a q value cutoff of 0.0001 for mouse data sets and 0.01 for human data sets. Additional information and materials will be made available upon request. The summits for each peak from all replicates were expanded to regions with a uniform size of 200 bp for mouse data sets and 300 bp for human data sets. These 32. Kochenderfer, J. N., Yu, Z., Frasheri, D., Restifo, N. P. & Rosenberg, S. A. Adoptive regions from all replicates were merged into one global set of peaks and were transfer of syngeneic T cells transduced with a chimeric antigen receptor that filtered to remove peaks on the Y chromosome or those that overlapped ENCODE recognizes murine CD19 can eradicate lymphoma and normal B cells. Blood 40,41 116, 3875–3886 (2010). blacklisted regions . 33. Roybal, K. T. et al. Precision tumor recognition by T cells with combinatorial We used summarizeOverlaps to compute the number of transposase insertions antigen-sensing circuits. Cell 164, 770–779 (2016). overlapping each peak from all replicates42. For differential coverage, raw ATAC- 34. Kadkhodaei, B. et al. Nurr1 is required for maintenance of maturing and adult seq counts in each peak for all replicates of all samples were normalized between midbrain dopamine neurons. J. Neurosci. 29, 15923–15932 (2009). replicates using voom43. Pairwise contrasts were performed with limma and dif- 35. Lio, C. W. et al. Tet2 and Tet3 cooperate with B-lineage transcription factors to regulate DNA modifcation and chromatin accessibility. eLife 5, e18290 (2016). ferentially accessible regions were filtered based on an fdr adjusted P value of less 36. Buenrostro, J. D., Giresi, P. G., Zaba, L. C., Chang, H. Y. & Greenleaf, W. J. than 0.01 and an estimated fold-change of at least 4. ATAC-seq density (number Transposition of native chromatin for fast and sensitive epigenomic profling of of transposase insertion sites per kilobase per million mapped reads) per peak and open chromatin, DNA-binding proteins and nucleosome position. Nature accessible regions were defined as those with a mean of 5 normalized insertions Methods 10, 1213–1218 (2013). Letter RESEARCH

37. Langmead, B., Trapnell, C., Pop, M. & Salzberg, S. L. Ultrafast and memory- 43. Ritchie, M. E. et al. limma powers diferential expression analyses efcient alignment of short DNA sequences to the human genome. Genome for RNA-sequencing and microarray studies. Nucleic Acids Res. 43, e47 Biol. 10, R25 (2009). (2015). 38. Chavez, L. et al. Computational analysis of genome-wide DNA methylation 44. Heinz, S. et al. Simple combinations of lineage-determining transcription during the diferentiation of human embryonic stem cells along the factors prime cis-regulatory elements required for macrophage and B cell endodermal lineage. Genome Res. 20, 1441–1450 (2010). identities. Mol. Cell 38, 576–589 (2010). 39. Palpant, T. Java Genomics Toolkit. (2016). Available at: https://github.com/ 45. Picelli, S. et al. Full-length RNA-seq from single cells using Smart-seq2. Nat. timpalpant/java-genomics-toolkit. Protocols 9, 171–181 (2014). 40. Consortium, E. P. et al.; ENCODE Project Consortium. An integrated encyclopedia 46. Dobin, A. et al. STAR: ultrafast universal RNA-seq aligner. Bioinformatics 29, of DNA elements in the human genome. Nature 489, 57–74 (2012). 15–21 (2013). 41. Kundaje, A. Blacklisted genomic regions for functional genomics analysis. (2014). 47. Love, M. I., Huber, W. & Anders, S. Moderated estimation of fold change and Available at: https://sites.google.com/site/anshulkundaje/projects/blacklists. dispersion for RNA-seq data with DESeq2. Genome Biol. 15, 550 (2014). 42. Lawrence, M. et al. Software for computing and annotating genomic ranges. 48. van Dijk, D. et al. Recovering gene interactions from single-cell data using data PLOS Comput. Biol. 9, e1003118 (2013). difusion. Cell 174, 716–729.e27 (2018). RESEARCH Letter

Extended Data Fig. 1 | Functional assessment of a hCD19-reactive CD3ζ signalling domains, the 2A self-cleaving peptide and the mouse chimeric antigen receptor (CAR). a, Left three panels, EL4, MC38 and Thy1.1 reporter. d, CAR surface expression monitored by myc epitope- B16-OVA cell lines expressing hCD19. Grey, parental; black, hCD19- tag and Thy1.1 expression. Mock-transduced CD8+ T cells were used expressing cells. Right, B16-OVA-hCD19 cells recovered after growth in as controls. e, Cytokine (TNF, IFNγ) production by CAR CD8+ T cells a C57BL/6J mouse followed by culture for 7 days. Grey, isotype control; after re-stimulation with EL4-hCD19 cells or with PMA and ionomycin. black, anti-hCD19. Data from one biological replicate in each case. f, Quantification of the data shown in e; P values (TNF: ****P < 0.0001, b, Left, growth curves (mean ± s.e.m., 15 mice per group) of 250,000 IFNγ: ***P = 0.0009) were calculated using a two-tailed unpaired t-test. B16-OVA parental or B16-OVA-hCD19 tumour cells in vivo after g, In vitro killing assay (mean ± s.e.m.) of CD8+ CAR and mock- inoculation into C57BL/6J mice. There is no significant difference at any transduced T cells; data from two biologically independent experiments, time point (ordinary two-way ANOVA, P > 0.9999 at day 19). Right, each with three technical replicates. h, Inhibitory surface receptor growth curves (mean ± s.e.m.) of 250,000 (n = 5 mice) or 500,000 (n = 6 expression on CAR- and mock-transduced CD8+ T cells cultured in mice) B16-OVA-hCD19 tumour cells in vivo after inoculation (significant vitro for 5 days; data representative of three biological replicates. Grey difference between the two groups at day 21; ordinary two-way ANOVA; shading, isotype control; black line, mock or CAR. Data in d, e and h *P = 0.0146). c, Diagram of the CAR construct. LS, leader sequence; are representative of 3 independent experiments. *P ≤ 0.05, **P ≤ 0.01, SS, signal sequence; myc, myc epitope-tag; scFv, single chain variable ***P ≤ 0.001, ****P ≤ 0.0001. fragment against human CD19; followed by the mouse (m) CD28 and Letter RESEARCH

Extended Data Fig. 2 | See next page for caption. RESEARCH Letter

Extended Data Fig. 2 | Adoptively transferred CD8+ CAR T cells for PBS compared to CAR (d); P values were calculated using a two- infiltrating B16-OVA-hCD19 tumours exhibit phenotypes and gene tailed unpaired t-test with Welch’s correction, *P ≤ 0.05, **P ≤ 0.01, expression profiles similar to those of OT-I and endogenous CD8+ ***P ≤ 0.001, ****P ≤ 0.0001. e, f, Flow cytometry gating scheme TILs. a, b, Experimental design to assess CD8+CD45.1+ OT-I and for CAR (e) and OT-I (f) CD8+ TILs. g, Mean average plots of genes CD8+CD45.2+ endogenous TILs; 1.5 × 106 OT-I T cells were adoptively differentially expressed in the indicated comparisons. Wald test was transferred into C57BL/6J mice 13 days after tumour inoculation. performed to calculate P values, as implemented in DESeq2; P values were c, Tumour growth curves (mean ± s.e.m.) of mice adoptively transferred adjusted using the Benjamini–Hochberg method. Genes differentially + with CAR or OT-I CD8 T cells; graph is a compilation of 3 independent expressed (adjusted P < 0.1 and fold change (log2 scale) ≥ 1 or ≤ −1) are experiments. At days 7 and 21, mouse numbers were: CAR, n = 24, 17; highlighted. Selected genes are labelled. Top row, comparisons of the CAR OT-I, n = 21, 20. d, Tumour growth curves (mean ± s.e.m.) of mice TIL populations amongst themselves and to endogenous PD-1loTIM3lo adoptively transferred with CAR or PBS; graph is a compilation of 3 TILs; middle row, comparisons within the endogenous TIL populations; independent experiments. At days 7 and 21, mouse numbers were: CAR bottom row, comparisons of CAR and endogenous PD-1hiTIM3hi TILs n = 35, 35; PBS n = 8, 6. c, d, For tumour sizes on day 21 after tumour (left), and CAR and endogenous PD-1hiTIM3lo TILs (right). inoculation, P = 0.3527 for CAR compared to OT-I (c) and P = 0.6240 Letter RESEARCH

ATAC-seq data ATAC-seq data a b Effector_rep1 Effector_rep2 LCMV Arm PD1hiTIM3hi_rep1 PD1hiTIM3hi_rep3 lo LCMV Cl13 lo PD1hiTIM3hi_rep4 Endo. PD1hiTIM3hi_rep2 TIM3 hi hi TIM3

Endo. PD1 TIM3 _rep3 hi Endo. PD1hiTIM3hi_rep6 hi hi hi D1

Endo. PD1 TIM3 _rep7 PD 1 Endo. PD1hiTIM3hi_rep1 Endo. PD1hiTIM3hi_rep8 hi hi

Endo. PD1 TIM3 _rep4 Endo. Endo. PD1hiTIM3hi_rep5 CAR TP CARPD1hiTIM3hi_rep5 CARPD1hiTIM3hi_rep3 CARPD1hiTIM3hi_rep4 Endo. PD1hiTIM3hi CAR TPD1hiTIM3hi CARPD1hiTIM3hi_rep1 B16-OVA- CARPD1hiTIM3hi_rep2 hCD19 hi lo hi

Endo. PD1 TIM3 _rep1 lo Endo. PD1hiTIM3lo_rep2 hi lo CARPD1 TIM3 _rep1 TIM3 hi hi lo TIM3 CARPD1 TIM3 _rep2 lo Endo. PD1hiTIM3lo_rep3 hi hi PD 1 OT-I PD1 TIM3 _rep1 PD 1 OT-I PD1hiTIM3hi_rep2 CARNr4aTKO_rep1 Endo.

CARNr4aTKO_rep2 Endo. CARNr4a3KO_rep1 CARNr4a3KO_rep2 CARNr4a WT_rep1 CARNr4a WT_rep2 Endo PD1hiTIM3lo CAR TPD1hiTIM3hi Naive_rep2 Naive_rep1 Naive_rep3 lo Naive_rep4 lo Memory_rep1 TIM3 Memory_rep2 LCMV Arm TIM3 lo Endo. PD1loTIM3lo_rep4 hi Endo. PD1loTIM3lo_rep2 PD 1 Endo. PD1loTIM3lo_rep3 PD 1 Endo. PD1loTIM3lo_rep1 B16-OVA- Endo. PD1loTIM3lo_rep8 Endo. Endo. PD1loTIM3lo_rep5 hCD19 Endo. Endo. PD1loTIM3lo_rep6 Endo. PD1loTIM3lo_rep7 PMAIono_rep1 Endo.PD1hiTIM3hi CAR TPD1hiTIM3lo PMAIono_rep2 rv Mock_rep1 rv Mock_rep2 hi

Resting_rep1 hi Resting_rep2

rv Empty_rep1 TIM3 hi TIM3

600 rv Empty_rep2 hi rv NR4A2_rep1 PD 1 rv NR4A2_rep2 in vitro PD 1 rv NR4A3_rep1 -I rv NR4A3_rep2 OT

rv NR4A1_rep1 Endo. Distanc e 300 rv NR4A1_rep2 rv NFAT-CA-RIT_rep1

rv NFAT-CA-RIT_rep2 hi hi hi hi CAR_rep2 OT-I PD1 TIM3 CAR TPD1 TIM3 CAR_rep4 CAR_rep3

Euclidean CAR_rep1 0 CAR_rep5

ATAC-seq data c 20 kb 10 kb Peaks hi hi (C) Endo. PD1 TIM3 (F) OT-I PD1hi TIM3hi (A) CAR T PD1hi TIM3hi (D) Endo. PD1hi TIM3lo (B) CAR T PD1hi TIM3lo (E) Endo. PD1lo TIM3lo Naive LCMVArm Effector LCMVArmMemory LCMVCl13 PD1hi TIM3hi Pdcd1 Itgav Extended Data Fig. 3 | Adoptively transferred CD8+ CAR T cells c, Genome browser views of sample loci, Pdcd1 (left), Itgav (right); scale infiltrating B16-OVA-hCD19 tumours exhibit chromatin accessibility range is from 0–600 for all tracks and data are the mean of all replicates. profiles similar to those of endogenous CD8+ TILs. a, Pair-wise CD8+ TIL populations are as indicated and defined in Fig. 1b, Extended hi hi hi lo euclidean distance comparisons of log2 transformed ATAC-seq density Data Fig. 2b: (A) PD-1 TIM3 CAR, (B) PD-1 TIM3 CAR, (C) PD- (Tn5 insertions per kilobase) between all replicates at all peaks accessible 1hiTIM3hi endogenous, (D) PD-1hiTIM3lo endogenous, (E) PD-1loTIM3lo in at least one replicate. b, Scatterplot of pairwise comparison of endogenous, (F) PD-1hiTIM3hi OT-I. ATAC-seq density (Tn5 insertions per kb) between samples indicated. RESEARCH Letter

Extended Data Fig. 4 | Mouse and human CD8+ TILs exhibit increased comparing CAR and endogenous TIL populations (A–E) defined in expression of NR4A1, NR4A2, NR4A3. a, b, Flow cytometry gating Fig. 1b. e–g, Plotting in single cells the expression of PDCD1 and HAVCR2 scheme for CAR (a) and endogenous (b) CD8+ TILs. c, Representative (x and y axis, respectively) and (displayed by the colour scale) the flow cytometry histograms of NR4A proteins in PD-1hiTIM3hi TILs, expression of the following: e, Genes differentially upregulated in PD- PD-1hiTIM3lo TILs, and PD-1loTIM3lo TILs and their corresponding 1hiTIM3hi TILs relative to PD-1loTIM3lo TILs. f, Genes coding for selected fluorescence minus one controls (in off-white). Data are representative of transcription factors showing differential expression in the comparison of 3 independent experiments in which the sample from each independent PD-1hiTIM3hi TILs relative to PD-1loTIM3lo TILs. g, Genes differentially experiment is comprised of TILs pooled together from 9–14 mice. downregulated in PD-1hiTIM3hi TILs relative to PD-1loTIM3lo TILs. Each d, Representative flow cytometry histograms for NR4A protein expression, dot represents a single cell. Human CD8+ TILs data are from ref. 14. Letter RESEARCH

a Control (CD8a stain only) d

Singlets-FSC Inoculate FSC/SSC Adoptive 91.4 Singlets - SSC tumour 64.5 + transfer 96.8 Live CD8 cells d90 or until 95.0 IACUC endpoint C57BL/6J mice

FSC-A FSC-H SSC-H Comp-APC-A :: CD8a d7 x x x x e SSC-A FSC-W SSC-W FSC-A B16-OVA-hCD19 MC38-hCD19 3.0 3.5 ) 2 )

0.19 0.023 2 105 105 2.5 3.0 PBS 2.5 PBS 2.0 104 104 2.0 1.5 1.5 :: CD8 a 1.0 mour size (cm

mour size (cm 1.0 103 :: huNGFR 103 Tu CAR T cells Tu 0.5 0.5 0.43 0.0 0.0 102 102 0102030405060708090 0510 15 20 25 3.0 3.5 Comp-APC-A

0 0 ) ) 2 2 Comp-PE-A 3.0 -102 -102 99.4 0.41 2.5 3 4 5 3 4 5 2.5 0 10 10 10 0 10 10 10 2.0 WT WT 2.0 Comp-FITC-A :: Thy1.1 Comp-FITC-A :: Thy1.1 1.5 1.5 1.0 mour size (cm mour size (cm 1.0 Tu 0.5 Tu b WT 0.5 0.0 0.0 0102030405060708090 0510 15 20 25 Singlets-FSC 3.0 3.5

FSC/SSC ) ) 2 Singlets - SSC 2 58.6 91.3 2.5 3.0 98.8 Live CD8+ cells Nr4a TKO 2.5 Nr4a TKO 95.4 2.0 2.0 1.5 1.5 FSC-A FSC-H SSC-H Comp-APC-A :: CD8a 1.0 1.0 mour size (cm x x x x mour size (cm Tu SSC-A FSC-W SSC-W FSC-A 0.5 Tu 0.5 15.6 83.9 105 105 0.0 0.0 0102030405060708090 0510 15 20 25 Days after tumour inoculation Days after tumour inoculation 104 104 :: CD8a

:: huNGFR B16-OVA-hCD19 tumour sizes MC38-hCD19 tumour sizes 103 103 f CAR T cells on Day 21 after inoculation on Day 19 after inoculation 83.9 p = 0.5010 *** 102 102 p = 0.8180 Comp-APC-A 0 p = 0.3252

Comp-PE-A 0 p = 0.1979 * -102 -102 0.50 0.043 1.5 3.5

3 4 5 3 4 5 ) ) 0 10 10 10 0 10 10 10 2 2 3.0 m m Comp-FITC-A :: Thy1.1 Comp-FITC-A :: Thy1.1 1.0 2.5 2.0 Area (c Area (c 1.5 0.5 1.0 mour

c Nr4a TKO mour Tu

Tu 0.5 FSC/SSC Singlets-FSC 0.0 0.0 91.8 Singlets - SSC 64.8 + PBS WT TKO PBS WT TKO 98.9 Live CD8 cells 95.3 g B16-OVA-hCD19 MC38-hCD19 FSC-A FSC-H SSC-H Comp-APC-A :: CD8a 100 100 x x x x SSC-A FSC-W SSC-W FSC-A 14.8 78.4 75 75 105 105

50 *p = 0.0026 50 *p = 0.0138 104 104 :: CD8a 25 25

3 :: huNGFR Percent survival 10 103 Percent survival CAR T cells 79.9 0 0 2 0102030405060708090 0510 15 20 25 10 102

Comp-APC-A 0 0 Days after tumour inoculation Days after tumour inoculation Comp-PE-A -102 -102 5.28 1.57 0 103 104 105 0 103 104 105 PBS WT Nr4a TKO Comp-FITC-A :: Thy1.1 Comp-FITC-A :: Thy1.1 Extended Data Fig. 5 | Prolonged survival of immunocompetent in individual mice. f, B16-OVA-hCD19 (left) and MC38-hCD19 (right) tumour-bearing mice adoptively transferred with CD8+ Nr4a TKO tumour sizes (mean ± s.d.) at day 21 and 19 post inoculation respectively. CAR T cells compared to mice transferred with CD8+ wild-type P values were calculated using an ordinary one-way ANOVA with Tukey’s CAR T cells. a, CD8a only staining control (previously tested to be multiple comparisons test: B16-OVA-hCD19, no significant difference; + the same as fluorescence minus one controls for CAR expression and MC38-hCD19, PBS versus Nr4a TKO ***P = 0.0001; PBS versus wild + NGFR expression) of CAR T cells before adoptive transfer. b, CAR type, P = 0.3252; wild type versus Nr4a TKO, *P = 0.0120. g, Survival and NGFR expression of CD8+ wild-type CAR T cells before adoptive curves for mice bearing B16-OVA-hCD19 tumours (left) and MC38- transfer. c, CAR and NGFR expression of CD8+ Nr4a TKO CAR T cells hCD19 tumours (right). P values calculated using log-rank (Mantel–Cox) before adoptive transfer. Data in a–c are representative of 4 independent test. For B16-OVA-hCD19, surviving mouse numbers at day 7, day 21, day experiments for adoptive transfer into Rag1−/− recipient mice; preparation 90 were: PBS, n = 13, 11, 0; wild type, n = 15, 11, 0; Nr4a TKO, n = 14, 13, of adoptive transfer into immunocompetent mice was the same except 2; *P = 0.0026. For MC38-hCD19, surviving mouse numbers at day 7 and for the use of GFP-expressing Cre and empty vector. d, 6 × 106 CAR day 19 were: PBS, n = 10, 9; wild type, n = 10, 7; Nr4a TKO, n = 10, 10; all T cells were adoptively transferred into C57BL/6J mice 7 days after mice died by day 23; *P = 0.0138. *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, tumour inoculation. e, Growth of B16-OVA-hCD19 (left; 13–15 mice per ****P ≤ 0.0001. condition) and MC38-hCD19 (right; 10 mice per condition) tumours RESEARCH Letter

Extended Data Fig. 6 | Tumour-bearing mice adoptively transferred a with CD8+ CAR T cells lacking all three NR4A family members exhibit Inoculate Adoptive prolonged survival compared to mice transferred with wild-type tumour transfer CD8+ CAR T cells or CD8+ CAR T cells lacking only one of the three d90 NR4A family members. a, Experimental design; 3 × 106 wild-type, Nr4a Rag1-/- mice TKO, Nr4a1 KO, Nr4a2 KO or Nr4a3 KO CAR T cells were adoptively − − d7 transferred into Rag1 / mice 7 days after tumour inoculation. b, Growth

Donor Genotype Retrovirus 1 Retrovirus 2 of B16-OVA-hCD19 tumours in individual mice, comprised of 17 or PBS n/a none none more mice per condition (these data include the wild type and Nr4a WT Nr4a1 fl/fl Nr4a2 fl/fl Nr4a3 +/+ CAR empty vector TKO data from Fig. 3). c, Graph shows mean ± s.d. and the individual Nr4a TKO Nr4a1 fl/fl Nr4a2 fl/fl Nr4a3 -/- CAR Cre Nr4a1 KO Nr4a1 fl/fl Nr4a2 +/+ Nr4a3 +/+ CAR Cre values of B16-OVA-hCD19 tumour sizes at day 21 after inoculation. Nr4a2 KO Nr4a1 +/+ Nr4a2 fl/fl Nr4a3 +/+ CAR Cre P values were calculated using an ordinary one-way ANOVA with Tukey’s Nr4a3 KO Nr4a1 fl/fl Nr4a2 fl/fl Nr4a3 -/- CAR empty vector multiple comparisons test; PBS versus wild type, *P = 0.0395; wild type versus Nr4a1 KO, P = 0.0511 (not significant); wild type versus Nr4a2 KO, b 3.0 3.0

) = = 2 ) **P 0.002, wild type versus Nr4a3 KO, *P 0.0161; and wild type

2 PBS Nr4a1 KO 2.5 2.5 versus Nr4a TKO, ****P < 0.0001. d, Survival curves. ****P < 0.0001, 2.0 2.0 calculated using log-rank (Mantel–Cox) test. Surviving mouse numbers at Area (cm Area (cm 1.5 1.5 day 7, day 21 and day 90 were n = 31, 14, 0 for PBS; n = 35, 25, 1 for wild our our 1.0 1.0 m m type; n = 17, 12, 0 for Nr4a1 KO; n = 17, 15, 1 for Nr4a2 KO; n = 32, 22, Tu

Tu 0.5 0.5 11 for Nr4a3 KO; and n = 39, 36, 27 for Nr4a TKO. P ≤ 0.05, P ≤ 0.01, 0.0 0.0 * ** 0102030405060708090 0102030405060708090 ***P ≤ 0.001, ****P ≤ 0.0001. 3.0 3.0 ) ) 2 WT 2 Nr4a2 KO 2.5 2.5

2.0 2.0 Area (cm 1.5 Area (cm 1.5

our 1.0 1.0 m mour Tu 0.5 Tu 0.5

0.0 0.0 0102030405060708090 0102030405060708090 3.0 3.0 ) ) 2 2 2.5 Nr4a TKO 2.5 Nr4a3 KO

2.0 2.0 Area (cm Area (cm 1.5 1.5

1.0 our 1.0 m mour Tu

Tu 0.5 0.5

0.0 0.0 0102030405060708090 0102030405060708090 Days after tumour inoculation Days after tumour inoculation

c Tumour sizes on Day 21 after inoculation * * **** p = 0.0511 PBS 3.0 WT ) ** 2

m 2.5 Nr4a TKO 2.0 Nr4a1 KO Area (c 1.5 Nr4a2 KO

mour 1.0 Nr4a3 KO Tu 0.5

0.0 PBSWTTKO 1KO2KO 3KO Days after tumour inoculation

d 100

75

PBS ****p < 0.0001 WT 50 Nr4a TKO

Nr4a1 KO

Percent survival 25 Nr4a2 KO Nr4a3 KO

0 0102030405060708090 Days after tumour inoculation Letter RESEARCH

Extended Data Fig. 7 | See next page for caption. RESEARCH Letter

Extended Data Fig. 7 | Phenotypic and genomic features of mouse ***P ≤ 0.001, ****P ≤ 0.0001. e, PCA plot of RNA-seq data from CD8+ T cells expressing NR4A1, NR4A2 or NR4A3. Mouse CD8+ in vitro resting mouse CD8+ T cells ectopically expressing NR4A1, T cells were isolated, activated, transduced with empty retrovirus or NR4A2, NR4A3 and empty vector control. f, Mean average plots of genes retroviruses encoding HA-tagged Nr4a1, Nr4a2 or Nr4a3 with human differentially expressed in the comparisons of ectopic expression of NGFR reporter, and assayed on day 5 post activation. a, Flow cytometry NR4A1, NR4A2 or NR4A3 against empty vector (top row), and pairwise gating of CD8+NGFR+ empty vector control, NR4A1-, NR4A2- and comparisons between the ectopic expression of various NR4A family NR4A3-expressing cells at a constant expression level of NGFR reporter. members (bottom row). Wald test was performed to calculate P values, b, Quantification of surface receptor expression (data from 3 independent as implemented in DESeq2. P values were adjusted using the Benjamini– replicates), showing geometric MFI normalized across experiments to Hochberg method. Genes differentially expressed (adjusted P < 0.1 and the average of all samples within each experiment. c, Representative fold change (log2 scale) ≥ 1 or ≤ −1) are highlighted using different flow cytometry plots of cytokine production upon re-stimulation with colours as indicated in the PCA plot as in e. Selected genes are labelled. PMA and ionomycin. d, Quantification of the data in c, showing geometric g, Scatterplot of pairwise comparison of ATAC-seq density (Tn5 insertions MFI normalized across experiments to the average of all samples within per kb) between the indicated samples. Data in a–d are from three each experiment. All P values were calculated using an ordinary one-way independent experiments; data in e–g from two independent experiments, ANOVA with Dunnett’s multiple comparisons test; *P ≤ 0.05, **P ≤ 0.01, each with two technical replicates. Letter RESEARCH

Extended Data Fig. 8 | See next page for caption. RESEARCH Letter

Extended Data Fig. 8 | CD8+ Nr4a TKO CAR TILs show increased plots for TIM3 and TCF1 expression in wild-type and Nr4a TKO CAR effector function compared to CD8+ wild-type CAR TILs. a, Tumour TILs (2 independent experiments). Bottom, bar plots (mean ± s.d.) growth curves (mean ± s.e.m.) after adoptive transfer of 1.5 × 106 CAR of transcription factor expression by wild-type and Nr4a TKO CAR T cells into Rag1−/− mice on day 13 after tumour inoculation. Mouse TILs (6 independent experiments). P values were calculated using two- numbers at day 7 and day 21 were: wild type, n = 47, 35; Nr4a TKO, tailed paired t-tests. For all P value calculations, *P ≤ 0.05, **P ≤ 0.01, n = 41, 32. P values were calculated using an ordinary two-way ANOVA ***P ≤ 0.001, ****P ≤ 0.0001. e, PCA plot of RNA-seq data from Nr4a with Tukey’s multiple comparisons test; for wild type versus Nr4a TKO, TKO or wild-type CAR TILs. f, Normalized enrichment scores (NES) of P = 0.5463. b, Flow cytometry gating scheme for surface markers, gene sets defined from pairwise comparisons of effector, memory and cytokines, and transcription factors expressed by wild-type (top) and Nr4a exhausted CD8+ T cells from LCMV-infected mice11. Enrichment score TKO (bottom) TILs. All samples are gated on cells with a set level of CAR was calculated using a Kolmogorov–Smirnov test, as implemented in gene expression (103–104) within the CAR+ NGFR+ population. c, Bar plots set enrichment analysis (GSEA). g, GSEA of RNA-seq data from Nr4a (mean ± s.d.) showing (left) number of wild-type and Nr4a TKO TKO and wild-type CAR TILs displayed as enrichment plots, ranking CAR TILs per g of tumour (5 independent experiments; P value was genes by fold change in expression between those conditions. The false calculated using a two-tailed ratio paired t-test) and (right) mean discovery rate (FDR) for both (f, g) is controlled at a level of 5% by the fluorescence intensity of Ki67 of wild-type and Nr4a TKO CAR TILs Benjamini–Hochberg correction. For e–g, data are from two independent (2 independent experiments). d, Top, representative flow cytometry experiments each consisting of 1–2 technical replicates. Letter RESEARCH

10 kb Ccr7 (1) Ccr7 (2) a empty Peaks 0.03 0.07 Nur77 motifs 0.06

s Nr4a1 NFAT motifs 0.02 0.05 t 0.04 CAR T WT 0.03 Nr4a2 0.01

CAR T Nr4a TKO % input % inpu 0.02 bZIP motifs cell count

normalized 0.01 Nr4a3 NFkB motifs 0.00 0.00 qPCR primers

3 4 5 Ccr7 (1)(2) Nr4a 1 Nr4a 2 Nr4a 3 empt y Nr4a1 Nr4a2 Nr4a3 0 10 10 10 empt y huNGFR: FITC 10 kb Ccr6 Peaks 0.015 Nur77 motifs

NFAT motifs 0.010 CAR T WT t

CAR T Nr4a TKO 0.005 bZIP motifs % inpu NFkB motifs 0.000 qPCR primers Ccr6 Nr4a1 Nr4a2 Nr4a3 empt y

10 kb Ifng 0.05 Peaks Nur77 motifs 0.04

NFAT motifs t 0.03

CAR T WT 0.02

CAR T Nr4a TKO % inpu 0.01 bZIP motifs NFkB motifs 0.00 qPCR primers Ifng Nr4a1 Nr4a2 Nr4a3 empt y

b c Regions where Regions where 5kb Nr4a TKO>WT WT > Nr4a TKO Peaks Nur77 motifs Higher in Higher in NFAT motifs Mock rv NFAT- CAR T WT CA-RIT CAR T Nr4a TKO bZIP motifs Empty rv Nr4a1 NFkB motifs

Il21 Empty rv Nr4a2

2kb Peaks Empty rv Nr4a3 Nur77 motifs NFAT motifs CAR T WT Resting PMA/Iono CAR T Nr4a TKO bZIP motifs NFkB motifs

log2 fold change Tnf

Extended Data Fig. 9 | NR4A family members bind to predicted showing enrichment of NR4A at regions probed; data representative of NR4A-binding motifs that are more accessible in wild-type CAR TILs 2 independent experiments consisting of three technical replicates each. compared to the Nr4a TKO CAR TILs, and regions more accessible in b, Genome browser views of the Il21 (top), Tnf (bottom) loci incorporating wild-type compared to Nr4a TKO CAR TILs are more accessible in CA- wild-type CAR TILs compared to Nr4a TKO CAR TILs, including binding RIT-NFAT1- and NR4A1/2/3-transduced cells. a, Top right, histogram motifs for NFAT, NR4A, bZIP and NFκB. Scale range is 0–600 for Il21 view showing expression of NR4A in cells ectopically expressing HA- and 0–1000 for Tnf; data are mean of two independent experiments. tagged versions of NR4A1, NR4A2, NR4A3; data are representative of 2 c, Top four panels, ATAC-seq data from Nr4a TKO and wild-type CAR independent experiments. Middle, genome browser views of the Ccr7, TILs compared with data from cells ectopically expressing CA-RIT- Ccr6, Ifng loci for wild-type CAR TILs compared to Nr4a TKO CAR TILs, NFAT1, NR4A1, NR4A2 or NR4A3. Bottom panel, ATAC-seq data from including binding motifs for NFAT, NR4A, bZIP, NFκB and the Nr4a TKO and wild-type CAR TILs compared with data from cultured location of the qPCR primers used. Scale range is 0–600 for all tracks cells re-stimulated with PMA and ionomycin. and data are mean of two independent experiments. Right, bar plots RESEARCH Letter

D

3'/!&75/ IROGFKDQJH  ORJ

&75/!3'/

ORJPHDQRIQRUPDOL]HGFRXQWV Extended Data Fig. 10 | Nr4a family members show a moderate decrease in mRNA expression in antigen-specific cells from LCMV- infected mice treated with anti-PDL1 or IgG control. a, Mean average plots of genes differentially expressed in cells treated with anti-PDL1 compared to cells treated with IgG control, highlighting two different categories of differentially expressed genes: those with adjusted P < 0.1 and fold change (log2 scale) ≥ 0.5 or ≤ −0.5 (lighter colours); and those with adjusted P < 0.1 and fold change (log2 scale) ≥ 1 or ≤ −1 (darker colours). Selected genes are labelled. Displayed are the number of genes in each category. The sequencing data in this analysis were obtained from ref. 19. Wald test was performed to calculate P values, as implemented in DESeq2; P values were adjusted using the Benjamini–Hochberg method. nature research | life sciences reporting summary

Corresponding author(s): Rao, Anjana; 2017-12-17355B

Life Sciences Reporting Summary Nature Research wishes to improve the reproducibility of the work that we publish. This form is intended for publication with all accepted life science papers and provides structure for consistency and transparency in reporting. Every life science submission will use this form; some list items might not apply to an individual manuscript, but all fields must be completed for clarity. For further information on the points included in this form, see Reporting Life Sciences Research. For further information on Nature Research policies, including our data availability policy, see Authors & Referees and the Editorial Policy Checklist.

Please do not complete any field with "not applicable" or n/a. Refer to the help text for what text to use if an item is not relevant to your study. For final submission: please carefully check your responses for accuracy; you will not be able to make changes later.

` Experimental design 1. Sample size Describe how sample size was determined. Sample sizes were not predetermined; for the OT-I and CAR mouse experiments, sample sizes were chosen based on a previous study from our lab [ref. 18]. For the Nr4a3-/- and Nr4a TKO experiments in the original submission, power calculations using a one-sided Mann-Whitney- Wilcoxon Test were retroactively performed on the initial experiment, and determined that chosen sample sizes were sufficient. Summary Statement from power calculation as follows: Group sample sizes of 7 and 7 achieve 91% power to show a difference in means when there is a difference of 0.9 between the null hypothesis mean difference of 0.0 and the actual mean difference of -0.9 at the 0.050 significance level (alpha) using a one-sided Mann-Whitney- Wilcoxon Test. These results are based on 2000 Monte Carlo samples from the null distributions: Normal(M0 S) and Normal(M0 S), and the alternative distributions: Normal(M0 S) and Normal(M1 S). Because the initial Nr4a3-/- and Nr4a TKO difference was less significant than that of the WT and Nr4a TKO, we extrapolated that the previous sample sizes would be sufficient for the WT and Nr4a TKO experiments as well. 2. Data exclusions Describe any data exclusions. One replicate of cytokine production collected from in vivo TILs by flow cytometry was excluded due to a machine/cytometer error during data collection.

For human cell ATAC-seq analysis, three samples with less than 10 million unique, non-chrM mapped reads were excluded. Samples with low numbers of unique reads are often indicative of sample viability issues or PCR amplification artifacts. Additionally, samples from one donor had substantial signal at regulatory elements that were not apparent in other samples, and all four samples from this donor (donor 3) were excluded from further comparisons. We have found that ATAC-seq is generally very reproducible between biological replicates and substantial outliers can be indicative of sample preparation issues. These analyses were performed on previously published data from other investigators and thus exclusion criteria were not pre-determined. 3. Replication Describe the measures taken to verify the reproducibility All experimental findings can be and were reliably reproduced. For sequencing and flow of the experimental findings. cytometry, we performed two to six independent biological replicates of each assay and all results were reproducible. For mouse survival studies, altogether we used a minimum of 17 mice and a maximum of 39 mice (independent biological replicates) per transfer group. 4. Randomization Describe how samples/organisms/participants were Tumor-bearing mice were first tumor size-matched and then randomly allocated to groups allocated into experimental groups. for adoptive transfer of CAR or OT-I, or in the later experiments, CAR + empty vector (pMIN) or CAR + Cre consisting of various Nr4a-floxed genotypes. 5. Blinding November 2017 Describe whether the investigators were blinded to Investigators were not blinded to group allocation during data collection and analysis; group allocation during data collection and/or analysis. investigators were aware of the cell type transferred into tumor-bearing mice. As certain experiments already required the simultaneous participation of more than one investigator, we did not have the personnel resources to consistently perform blinding; hence blinding was not used for the course of this study. Note: all in vivo studies must report how sample size was determined and whether blinding and randomization were used.

1 6. Statistical parameters nature research | life sciences reporting summary For all figures and tables that use statistical methods, confirm that the following items are present in relevant figure legends (or in the Methods section if additional space is needed). n/a Confirmed

The exact sample size (n) for each experimental group/condition, given as a discrete number and unit of measurement (animals, litters, cultures, etc.) A description of how samples were collected, noting whether measurements were taken from distinct samples or whether the same sample was measured repeatedly A statement indicating how many times each experiment was replicated The statistical test(s) used and whether they are one- or two-sided Only common tests should be described solely by name; describe more complex techniques in the Methods section. A description of any assumptions or corrections, such as an adjustment for multiple comparisons Test values indicating whether an effect is present Provide confidence intervals or give results of significance tests (e.g. P values) as exact values whenever appropriate and with effect sizes noted. A clear description of statistics including central tendency (e.g. median, mean) and variation (e.g. standard deviation, interquartile range) Clearly defined error bars in all relevant figure captions (with explicit mention of central tendency and variation)

See the web collection on statistics for biologists for further resources and guidance.

` Software Policy information about availability of computer code 7. Software Describe the software used to analyze the data in this ATAC-seq analysis: bowtie 1.0.0, samtools 0.1.8, bedtools v2.16.2, MACS2 v2.1.1.20160309, study. picard tools-1.94, java genomics toolkit 1.1.0, trim_galore 0.3.8, homer v4.10.1, and R v3.3.3 (with packages Biobase v2.34.0, BiocGenerics v0.20.0, Biostrings v2.42.1, data.table v1.11.4, dplyr v0.7.6, GenomeInfoDb v1.10.3, GenomicAlignments v1.10.1, GenomicRanges v1.26.4, ggplot2 v3.0.0, gtools v3.5.0, IRanges v2.8.2, limma v3.30.13, MEDIPS v1.24.0, pheatmap v1.0.8, RColorBrewer v1.1-2, Rsamtools v1.26.2, S4Vectors v0.12.2, SummarizedExperiment v1.4.0, tidyr v0.8.1, XVector v0.14.1)

RNA-seq, scRNA-seq, and GSEAs analysis: TrimGalore v0.4.5, Cutadapt v1.13, STAR v2.5.3a, R v3.3.3, GSEA v3.0; BioConductor packages: (for data analysis), rtracklayer v1.34.2, GenomicAlignments v1.10.1, DESeq2 v1.14.1; BioConductor packages: (for making figures) pheatmap, ggplot2, ggrepel, grid, RColorBrewer, MAGIC (R implementation, Rmagic v1.0.0

Flow cytometry analysis: FlowJo v.10 (Tree Star, Inc), Prism 7 (GraphPad Software) Tumor growth curve / survival curve analysis: Prism 7 (GraphPad Software) Assembly / layout of figures: Adobe Illustrator CS6, Affinity Designer 1.6.1

For manuscripts utilizing custom algorithms or software that are central to the paper but not yet described in the published literature, software must be made available to editors and reviewers upon request. We strongly encourage code deposition in a community repository (e.g. GitHub). Nature Methods guidance for providing algorithms and software for publication provides further information on this topic.

` Materials and reagents Policy information about availability of materials 8. Materials availability Indicate whether there are restrictions on availability of All unique materials will be made available by authors upon request. unique materials or if these materials are only available for distribution by a third party. November 2017

2 9. Antibodies nature research | life sciences reporting summary Describe the antibodies used and how they were validated All antibodies were purchased from Biolegend, eBioscience, BD Bioscience, and Cell Signaling for use in the system under study (i.e. assay and species). Technology with the exception of mouse CD3 which was purified from a monoclonal-antibody producing hybridoma (Clone 145-2C11). All Biolegend, eBioscience, BD Bioscience, and Cell Signaling Technology antibodies provide validation statements/data and relevant citations on the manufacturer’s website, which can be found by searching for the catalog number of the antibody on the corresponding manufacturer's website. For flow cytometry, all antibodies were used at a final concentration of 1:200 with the exception of Ki67, which was used at a final concentration of 1:100. For ChIP, 10ug of the HA-tag antibody was used.

Catalog No. Supplier Name Antibody Clone Name Lot No. 100712 Biolegend APC anti-mouse CD8a 53-6.7 B200238 100706 Biolegend FITC anti-mouse CD8a 53-6.7 B208067, B217242 100708 Biolegend PE anti-mouse CD8a 53-6.7 B134188, B151201 100737 Biolegend BV421 anti-mouse CD8a 53-6.7 B226247, B210399 100734 Biolegend PerCP/Cy5.5 anti-mouse CD8a 53-6.7 B156856 100722 Biolegend PeCy7 anti-mouse CD8a 53-6.7 B190884 110730 Biolegend PeCy7 anti-mouse CD45.1 A20 B188237, B217246 11-0900-85 eBioscience FITC anti mouse/rat CD90.1 (Thy 1.1) HIS51 4310957 554898 BD Biosciences PE mouse anti-rat/mouse CD90.1 OX-7 2317593 202539 Biolegend BV711 anti-rat/mouse CD90.1 Thy1.1 OX-7 B223103 202516 Biolegend PerCP/ Cy5.5 anti-rat/mouse CD90.1 Thy1.1 OX-7 B202057 202519 Biolegend APC-Cy7 Anti-rat/mouse CD 90.1 OX-7 B222663 345106 Biolegend PE anti-human CD271 (NGFR) ME 20.4 B175123 345108 Biolegend APC anti-human CD271 (NGFR) ME 20.4 B204228 345112 Biolegend PerCP/Cy5.5 anti-human CD271 (NGFR) ME 20.4 B218745 135221 Biolegend BV421 anti-mouse CD279 (PD-1) 29F.1A12 B213655 135206 Biolegend PE anti-mouse CD279 (PD-1) 29F.1A12 B142906 135210 Biolegend APC anti-mouse CD279 (PD-1) 29F.1A12 n/a 125210 Biolegend APC anti-mouse CD223 (Lag3) C9B7W B176313 125223 Biolegend PE/Dazzle 594 anti-mouse CD223 (Lag3) C9B7W B224161 12-5870-81 eBioscience PE anti-mouse TIM3 RMT3-23 4301948, 4273433 119723 Biolegend BV421 anti-mouse CD366 (Tim-3) RMT3-23 B235257 119705 Biolegend APC anti-mouse CD366 (Tim-3) RMT3-23 n/a 123907 Biolegend PE anti-mouse CD200R (OX2R) OX-110 B220799 123809 Biolegend APC anti-mouse CD200 (OX2) OX-90 B203310 133507 Biolegend PE anti-mouse CD244.2 (2B4 B6 alloantigen) m2B4 (B6) 458.1 B182843 126309 Biolegend PE anti-mouse CD357 (GITR) DTA-1 B214908 506328 Biolegend BV421 anti-mouse TNF alpha MP6-XT22 B224675 503839 Biolegend PE/Dazzle594 anti-mouse IL-2 JES6-5H4 B211964 12-7021-82 eBioscience PE anti-mouse IL-2 JES6-5H4 E030634 505807 Biolegend PE anti-mouse IFN gamma XMG1.2 B178149 17-7311-82 eBioscience APC anti-mouse IFN gamma XMG1.2 E07379-1633 363004 Biolegend PE anti-human CD19 SJ25C1 B214170 9066S Cell Signaling Technology TCF1/ TCF7 Rabbit mAb (Pacific Blue Conjugate) C63D9 1 12-4875-80 eBioscience PE anti-mouse Eomes Dan11mag 4313231, 4323634 50-5825-82 eBioscience eFluor 660 anti-human/mouse Tbet 4B10 E12136-1632 65-0865-18 eBioscience Fixable Viability Dye eFluor 780 n/a E11447-1674 102112 Biolegend LEAF (TM) Purified CD28 37.51 B229179, B231127, B228119 n/a monoclonal antibody- producing hybridoma purified CD3 145-2C11 n/a 3724S Cell Signaling Technology HA-tag Rabbit mAb C29F4 8 3739S Cell Signaling Technology Myc-tag (9B11) mouse PE mAb 9B11 9 135216 Biolegend PeCy7 anti-mouse CD279 (PD-1) 29F.1A12 B227806 a12-1011-80 Invitrogen PE anti-mouse CD101 Moushi 101 4330771 17-0381-81 eBioscience APC anti-mouse CD38 90 4324890 12-5965-80 eBioscience PE anti-mouse Nur77 (Nr4a1) 12.14 E01954-1636 sc-376984AF647 Santa Cruz Biotechnology AF647 anti-mouse Nr4a2 F-5 G2517 sc-393902PE Santa Cruz Biotechnology PE anti-mouse Nr4a3 H-7 B2818 563786 BD Biosciences BUV395 rat anti-mouse CD8a 53-6.7 8072932, 7096603 652405 Biolegend APC anti-mouse Ki67 16A8 B191905 345104 Biolegend FITC anti-human NGFR ME20.4 B223717 400411 Biolegend APC Rat IgG1 k isotype control RTK2071 B238505 400511 Biolegend APC Rat IgG2a k isotype control RTK2758 n/a November 2017

3 10. Eukaryotic cell lines nature research | life sciences reporting summary a. State the source of each eukaryotic cell line used. The EL4 mouse thymoma cell line was purchased from the American Type Culture Collection (ATCC): EL4 (ATCC® TIB-39™, Mus musculus T cell lymphoma). The B16-OVA mouse melanoma cell line was a kind gift of Dr. Schoenberger (LJI). The 293T cell line was purchased from ATCC: 293T (ATCC® CRL-3216™). The Platinum-E Retroviral Packaging Cell Line, Ecotropic (PlatE) cell line was purchased from Cell BioLabs, Inc: RV-101. The MC-38 mouse colon adenocarcinoma cell line (a kind gift of A.W. Goldrath, UCSD, La Jolla, CA) was originally purchased from Kerafast, Inc (ENH204).

b. Describe the method of cell line authentication used. The EL4 cell line stained positive for mouse Thy1.2 and PD-1; and stained negative for huCD19. The B16-OVA cell line stained negative for huCD19. The MC-38 cell line stained negative for huCD19. The PlatE and 293T cell lines were not authenticated.

c. Report whether the cell lines were tested for Cell lines were not tested for mycoplasma contamination. mycoplasma contamination. d. If any of the cell lines used are listed in the database No commonly misidentified cell lines were used. of commonly misidentified cell lines maintained by ICLAC, provide a scientific rationale for their use.

` Animals and human research participants Policy information about studies involving animals; when reporting animal research, follow the ARRIVE guidelines 11. Description of research animals Provide all relevant details on animals and/or C57BL/6J, B6.SJL-PtprcaPepcb/BoyJ, Rag 1-/- mice were obtained from Jackson Laboratories. animal-derived materials used in the study. Nr4a gene-disrupted strains were obtained from Takashi Sekiya and Akihiko Yoshimura, with permission from Pierre Chambon. Both male and female mice were used for studies. Mice were age-matched and between 8-12 weeks old when used for experiments, and tumor- bearing mice were first tumor size-matched and then randomly assigned to experimental groups. All mice were bred and/or maintained in the animal facility at the La Jolla Institute for Allergy and Immunology. All experiments were performed in compliance with the LJI Institutional Animal Care and Use Committee (IACUC) regulations.

Policy information about studies involving human research participants 12. Description of human research participants Describe the covariate-relevant population There were no human participants in this study. characteristics of the human research participants. November 2017

4 November 2017 nature research | flow cytometry reporting summary 1 s of Rao, Anjana; 2017-12-17355B Rao, Anjana; Corresponding author(s): Corresponding

LSRFortessa, LSR-II, FACSAria-I, FACSAria-II, FACSAria-Fusion (BD Biosciences) FACSDiva8.0 (BD Biosciences), FlowJo v.10 (Tree Star, Inc), Prism 7 (GraphPad Software) Sample preparation for cell sorting of huCD19-expressing cell lines: Sample preparation for cell sorting of huCD19-expressing B16-OVA, or MC-38 cells were spun Bulk populations of huCD19-transduced EL4, down and stained for cell sorting. vitro CD8+ T cells: Sample preparation for flow cytometry of in and stained for phenotyping with Mouse CD8+ T cells in culture were spun down flow cytometry or stained for cell sorting. sorting of TILs from CAR and OT-I Sample preparation for flow cytometry/cell from Nr4aTKO vs WT experiments: experiments, and for flow cytometry of TILs with PBS prior to removal of On Day 21, mice were euthanized and perfused by group, homogenized, and then tumor. Tumors were collected, pooled together Tumor Dissociation kit (Miltenyi dissociated using the MACS Miltenyi Mouse Octo Heaters (Miltenyi Biotec) Biotec) and the gentleMACs dissociator with were then filtered through a according to manufacturer’s instructions. Tumors aspirated and the tumors were 70uM filter and spun down. Supernatant was of tumor per 5mL of 1%FBS/PBS for resuspended in the equivalent of 4-5 grams FlowComp Mouse CD8 isolation kit CD8 positive isolation using the Dynabeads (Invitrogen). After positive isolation, cells were either divided into equal amount for staining and phenotyping with flow cytometry, or stained for cell sorting. Sample preparation for cell sorting of TILs from WT and Nr4a TKO experiments: On Day 21, mice were euthanized and perfused with PBS prior to removal of tumor. Tumors were collected, pooled together by group, homogenized, and then dissociated using the MACS Miltenyi Mouse Tumor Dissociation kit (Miltenyi Biotec) and the gentleMACs dissociator with Octo Heaters (Miltenyi Biotec) according to manufacturer’s instructions. Tumors were then filtered through a 70uM filter and spun down. Supernatant was aspirated and the tumors were resuspended in 40% Percoll/RPMI and underlaid with 80% Percoll/PBS in 15mL conical tubes to form an 80%/40% Percoll discontinuous density gradient. Samples were spun for 30min at room temperature at 1363g in a large benchtop centrifuge with a swinging bucket. TILs were collected from 80%/40% Percoll interface and further purified using CD90.2 Microbeads (Miltenyi Biotec) and magnetic separation. After positive isolation, cells were stained for cell sorting.

identical markers).

the flow cytometry data.

Methodological details Data presentation

4. A numerical value for number of cells or percentage (with statistics) is provided. 4. A numerical value for number of cells 3. All plots are contour plots with outliers or pseudocolor plots. 3. All plots are contour plots with outliers 2. The axis scales are clearly visible. Include numbers along axes only for bottom left plot of group (a 'group' is an analysi axes only for bottom left plot of group clearly visible. Include numbers along 2. The axis scales are 1. The axis labels state the marker and fluorochrome used (e.g. CD4-FITC). the marker and fluorochrome used 1. The axis labels state

7. Describe the software used to collect and analyze

6. Identify the instrument used for data collection. 5. Describe the sample preparation. For all flow cytometry data, confirm that: For all flow cytometry `

`

Form fields will expand as needed. Please do not leave fields blank. as needed. Please do not leave fields Form fields will expand Flow Cytometry Reporting Summary Flow Cytometry

November 2017

nature research | flow cytometry reporting summary 2

No post-sort analysis was done on sorted cell populations from TILs, which were TILs, which from cell populations on sorted was done analysis No post-sort RNA-seq. ATAC-seq or for immediately processed in lines were expanded and MC38-huCD19 cell B16-OVA-huCD19, EL4-huCD19, that huCD19 expression confirmed sorting, and flow cytometry vitro after cell in the sorted population. remained high for cell sorting: Gating strategy -> huCD19+ -> FSC-A/SSC-A -> FSC-W/FSC-H -> SSC-W/SSC-H huCD19 cell line sorts cells) (roughly top 16% of huCD19-expressing -> SSC- cell sorts -> FSC-A/SSC-A -> FSC-W/FSC-H Nr4a1, Nr4a2, Nr4a3-expressing level of NGFR- Live/dead dye neg. -> NGFR+ (set expression W/SSC-H -> (Live) CD8+ Nr4a3 for empty vector (pMIN) or Nr4a1, Nr4a2, expressing CD8+ cells) CD8+ Live/dead -> FSC-W/FSC-H -> SSC-W/SSC-H -> (Live) CAR sorts: FSC-A/SSC-A TIM3lo CAR (CAR) -> PD-1hi TIM3hi CAR, PD-1hi dye neg. -> CD45.1+ Thy1.1+ CD8+ Live/dead -> FSC-W/FSC-H -> SSC-W/SSC-H -> (Live) OT-I sorts: FSC-A/SSC-A (OT-I) -> PD-1hi TIM3hi OT-I dye neg. -> CD45.1+ only SSC-W/SSC-H -> sorts: FSC-A/SSC-A -> FSC-W/FSC-H -> Corresponding Endogenous (Endogenous) -> PD-1hi TIM3hi, (Live) CD8+ Live/dead dye neg. -> CD45.1- Thy1.1- PD-1hi TIM3lo and PD-1lo TIM3lo Endogenous -> SSC-W/SSC-H -> (Live) CD8+ WT or Nr4a TKO sorts: FSC-A/SSC-A -> FSC-W/FSC-H + pMIN empty vector = WT or CAR + Live/dead dye neg. -> Thy1.1+ NGFR+ (CAR Cre = Nr4a TKO) Gating strategy for flow cytometry: similar; gating strategy for all in vivo Gating strategy for all in vitro experiments are below for each set of experiments: experiments are similar. Specifics are detailed B16-OVA-huCD19, MC38-huCD19: Confirming huCD19 expression on EL4-huCD19, gate set on parent population FSC-A/SSC-A -> FSC-W/SSC-H -> huCD19+; negative EL4, B16-OVA, MC38 respectively Assaying CAR expression in vitro: -> FSC-W/SSC-H -> CD8+ CAR in vitro, surface marker expression: FSC-A/SSC-A -> PD-1, TIM3, LAG3 Thy1.1+ (CAR) cells or CD8+ Thy1.1- (mock) -> FSC-W/SSC-H -> CD8+ Thy1.1+ CAR in vitro, cytokine production: FSC-A/SSC-A IFNg; cytokine production negative cells (CAR) or CD8+ Thy1.1- (mock) -> TNF, gates set on mock unstimulated or CAR+ unstimulated Assaying Nr4a1, 2, 3 expression in vitro: Nr4a1, Nr4a2, Nr4a3 in vitro, surface marker and transcription factor expression: FSC-A/SSC-A -> SSC-A/CD8+ -> FSC-W/ live-dead dye -> SSC-W/SSC-A ->NGFR+ cells (pMIN empty vector or Nr4a1, Nr4a2, Nr4a3) -> PD-1, TIM3, LAG3, CD200, GITR, 2B4, CD101, CD38. Nr4a1, Nr4a2, Nr4a3 in vitro, cytokine production: FSC-A/SSC-A -> SSC-A/CD8+ -> FSC-W/ live-dead dye -> SSC-W/SSC-A -> NGFR+ cells (pMIN empty vector or Nr4a1, Nr4a2, Nr4a3) -> TNF, IFNg; cytokine production negative gates set on empty vector (pMIN) unstimulated Assaying TILs (CAR or OT-I): TILs, surface markers or transcription factor expression: FSC-A/SSC-A -> FSC-W/ FSC-H -> SSC-W/SSC-H -> (Live) CD8+ Live/dead dye neg. -> CD45.1+ Thy1.1+ (CAR) or CD45.1+ Thy1.1- (OT-I) -> PD-1, TIM3, LAG3, TCF1, Eomes, T-bet. TILs, cytokine production: FSC-A/SSC-A -> FSC-W/FSC-H -> SSC-W/SSC-H -> (Live) CD8+ Live/dead dye neg. -> CD45.1+ Thy1.1+ (CAR) or CD45.1+ Thy1.1- (OT-I) -> TNF, IFNg, IL-2; cytokine production negative gates set on CAR unstimulated or OT- I unstimulated. Nr4a protein level expression: FSC-A/SSC-A -> FSC-W/FSC-H -> SSC-W/SSC-H -> (Live) CD8+ Live/dead dye neg. -> CD45.1+ Thy1.1+ (CAR) or CD45.1- Thy1.1- endogenous -> PD-1/TIM3 -> Nr4a1, Nr4a2, Nr4a3 Assaying TILs (WT or Nr4a TKO):

populations within post-sort fractions. populations

9. Describe the gating strategy used. the gating strategy 9. Describe 8. Describe the abundance of the relevant cell relevant of the abundance the 8. Describe

November 2017

nature research | flow cytometry reporting summary 3

TILs, surface markers or transcription factor expression: FSC-A/SSC-A -> FSC-W/ -> FSC-A/SSC-A factor expression: or transcription markers TILs, surface + (CAR Thy1.1+ NGFR+ neg. -> dye CD8+ Live/dead -> (Live) -> SSC-W/SSC-H FSC-H Thy1.1 gate on 10^3 – 10^4 + Cre = Nr4a TKO) -> vector = WT, or CAR pMIN empty TCF1, Eomes, T-bet. -> PD-1, TIM3, LAG3, + expression -> SSC-W/SSC-H -> (Live) -> FSC-W/FSC-H production: FSC-A/SSC-A TILs, cytokine or empty vector = WT NGFR+ (CAR + pMIN dye neg. -> Thy1.1+ CD8+ Live/dead set on negative gates IL-2; cytokine production Nr4a TKO) -> TNF, IFNg, CAR + Cre = WT unstimulated. Tick this box to confirm that a figure exemplifying the gating strategy is provided in the Supplementary Information. strategy is provided in the Supplementary that a figure exemplifying the gating Tick this box to confirm