Transcription Factor IRF4 Promotes CD8 T Cell Exhaustion and Limits

Article

Transcription Factor IRF4 Promotes CD8+ T Cell Exhaustion and Limits the Development of Memory- like T Cells during Chronic Infection

Graphical Abstract Authors Kevin Man, Sarah S. Gabriel, Yang Liao, ..., Mark A. Febbraio, Wei Shi, Axel Kallies

Correspondence [email protected]

In Brief During chronic stimulation, CD8+ T cells acquire an exhausted phenotype characterized by expression of inhibitory receptors, loss of effector function, and metabolic impairments. Man et al. have identiﬁed a transcriptional module consisting of the TCR-induced transcription factors IRF4, BATF, and NFATc1 that drives T cell exhaustion and impairs memory T cell development.

Highlights d High IRF4 mediates T cell exhaustion including expression of inhibitory receptors d TCR-responsive transcription factors IRF4, BATF, and NFAT form a transcriptional circuit d Low IRF4 reduces T cell exhaustion and promotes formation of TCF1+ memory-like T cells

Man et al., 2017, Immunity 47, 1129–1141 December 19, 2017 Crown Copyright ª 2017 Published by Elsevier Inc. https://doi.org/10.1016/j.immuni.2017.11.021 Immunity Article

Transcription Factor IRF4 Promotes CD8+ TCell Exhaustion and Limits the Development of Memory-like T Cells during Chronic Infection

Kevin Man,1,2,10,11 Sarah S. Gabriel,1,8,9,10 Yang Liao,1,2 Renee Gloury,1,8,9 Simon Preston,1,2 Darren C. Henstridge,3 Marc Pellegrini,1,2 Dietmar Zehn,4 Friederike Berberich-Siebelt,5,6 Mark A. Febbraio,7 Wei Shi,1,2 and Axel Kallies1,8,9,12,* 1The Walter and Eliza Hall Institute of Medical Research, 1G Royal Parade, Parkville, VIC 3052, Australia 2The Department of Medical Biology, University of Melbourne, Parkville, VIC 3010, Australia 3Baker Heart and Diabetes Institute, Melbourne, VIC 3004, Australia 4Division of Animal Physiology and Immunology, School of Life Sciences Weihenstephan, Technical University of Munich, 85354 Freising, Germany 5Institute of Pathology, University of Wurzburg,€ 97080 Wurzburg,€ Germany 6Comprehensive Cancer Center Mainfranken, University of Wurzburg,€ 97080 Wurzburg,€ Germany 7Cellular and Molecular Metabolism, Garvan Institute, Sydney, NSW 2010, Australia 8Department of Microbiology and Immunology, The University of Melbourne, Parkville, VIC 3010, Australia 9The Peter Doherty Institute for Infection and Immunity, University of Melbourne, Melbourne, VIC 3000, Australia 10These authors contributed equally 11Present address: Cardiovascular Research Institute, University of California, San Francisco, CA 94143, USA 12Lead Author *Correspondence: [email protected] https://doi.org/10.1016/j.immuni.2017.11.021

SUMMARY that is defined by the progressive loss of effector function and impaired memory T cell potential (Virgin et al., 2009; Speiser During chronic stimulation, CD8+ T cells acquire an et al., 2014; Kahan et al., 2015; Wherry and Kurachi, 2015). exhausted phenotype characterized by expression Exhausted T cells successively lose their ability to produce of inhibitory receptors, down-modulation of effector cytokines, including interferon-gamma (IFN-g), tumor necrosis function, and metabolic impairments. T cell exhaus- factor (TNF), and interleukin-2 (IL-2), and exhibit impaired tion protects from excessive immunopathology but survival, eventually leading to the deletion of antigen-specific limits clearance of virus-infected or tumor cells. We cells. They express multiple inhibitory receptors, including programmed cell-death 1 (PD-1), lymphocyte-activation gene transcriptionally profiled antigen-specific T cells 3 (Lag3), TIM-3, CTLA4, TIGIT, and 2B4, associated with func- from mice infected with lymphocytic choriomeningi- tional impairments. Conversely, inhibitory receptors are required tis virus strains that cause acute or chronic disease. to prevent overt immune pathology due to chronic stimulation of T cell exhaustion during chronic infection was driven antigen-specific T cells (Blackburn et al., 2009; Virgin et al., 2009; by high amounts of T cell receptor (TCR)-induced Speiser et al., 2014; Kahan et al., 2015; Wherry and Kurachi, transcription factors IRF4, BATF, and NFATc1. 2015). For example, PD-1 binding to either PD ligand (PD-L)1 These regulators promoted expression of inhibitory or PD-L2 facilitates the dephosphorylation of key proteins receptors, including PD-1, and mediated impaired downstream of the T cell receptor (TCR), leading to the inhibition cellular metabolism. Furthermore, they repressed of TCR-dependent signal transduction (Keir et al., 2008). T cell the expression of TCF1, a transcription factor exhaustion has recently been linked to defects in cellular meta- required for memory T cell differentiation. Reducing bolism (Gubin et al., 2014; Staron et al., 2014; Bengsch et al., 2016; Schurich et al., 2016), which has emerged as a major IRF4 expression restored the functional and checkpoint in adaptive immune responses (Pearce et al., 2013; metabolic properties of antigen-specific T cells and Wang and Green, 2012). Indeed, compromised metabolic promoted memory-like T cell development. These fitness is closely linked to the functional impairments of tumor- findings indicate that IRF4 functions as a central infiltrating T cells that frequently display an exhausted phenotype node in a TCR-responsive transcriptional circuit (Chang et al., 2015; Ho et al., 2015). Recently, PD-1 signaling that establishes and sustains T cell exhaustion was shown to directly contribute to establishing the metabolic during chronic infection. impairments of exhausted T cells (Bengsch et al., 2016), suggesting that exhaustion is underpinned by substantial rewiring of TCR signaling-mediated metabolic processes. INTRODUCTION Several studies have demonstrated that exhausted T cells are characterized by a transcriptional program that is distinct from T cells responding to persistent viral or tumor antigens acquire effector and memory T cells that arise during acute infection an altered state of differentiation referred to as ‘‘exhaustion’’ (Wherry et al., 2007; Doering et al., 2012; Kahan et al., 2015).

Immunity 47, 1129–1141, December 19, 2017 Crown Copyright ª 2017 Published by Elsevier Inc. 1129 A number of key transcription factors were identified that control (TIM-3), Tigit, and others, which were strongly upregulated in ex- the degree of exhaustion during chronic infection, including hausted T cells, as well as transcription factors known to be T-bet (encoded by Tbx21), Blimp-1 (encoded by Prdm1), important for the differentiation of exhausted or memory T cells Eomesodermin (Eomes), von Hippel-Lindau tumor suppressor such as Prdm1 (Blimp1), Eomes, and Tcf7 (TCF1) (Figures 1B (VHL), and Foxo1 (Shin et al., 2009; Kao et al., 2011; Paley and 1C). In addition, our RNA-seq data revealed that exhausted et al., 2012; Doedens et al., 2013; Staron et al., 2014). Recently, T cells in comparison to acutely stimulated T cells expressed the transcriptional regulators TCF1, BCL6, Blimp-1, and Id2 and high amounts of Irf4 mRNA, encoding a transcription factor Id3 (via binding to E2A) were found to control the regenerative that integrates TCR signaling with cellular metabolism and clonal and migratory capacities of exhausted T cells (He et al., 2016; population expansion of effector CD8+ T cells (Man and Kallies, Im et al., 2016; Leong et al., 2016; Utzschneider et al., 2016). 2015; Man et al., 2013). High IRF4 expression in antigen-specific Thus, multiple transcription factors are co-opted in transcrip- CD8+ T cells in mice infected with chronic LCMV Docile as tional modules that differ between acute and chronic infection. compared to mice infected with acute LCMV WE was confirmed However, the molecular network that initiates, drives, and by quantitative PCR and intracellular staining, showing that high sustains the functional impairments of exhausted cells is IRF4 protein expression was established early in chronically incompletely known. stimulated T cells and was maintained up to at least 30 days after To further unravel the transcriptional circuits that underlie the establishment of chronic infection (Figures 1D and 1E). establishment of functional and metabolic exhaustion of Exhausted T cells compared to acutely stimulated T cells also T cells, we here performed transcriptional profiling of antigen- expressed elevated amounts of the AP-1 transcription factor specific CD8+ T cells isolated from mice infected with two BATF (Figures 1F and 1G) that is required for efficient DNA bind- different strains of lymphocytic choriomeningitis virus (LCMV) ing and transcriptional activity of IRF4 in T cells (Glasmacher causing acute and chronic infections, respectively. We demon- et al., 2012; Li et al., 2012). Expression of both IRF4 and BATF strate that exhausted T cells in comparison to fully functional tightly correlated with high expression of PD-1 in P14 T cells (Fig- effector and memory T cells expressed elevated amounts of ures 1H and 1I). Thus, T cell exhaustion is characterized by the the TCR-responsive transcription factors IRF4 and BATF, which elevated expression of the two TCR-responsive transcription promoted the expression of multiple inhibitory receptors factors IRF4 and BATF. including PD-1, impaired effector function, and repressed anabolic metabolism. Furthermore, exhausted T cells expressed High Amounts of IRF4 Drive T Cell Exhaustion high amounts of transcription factor NFATc1, and both NFATc1 To examine the role of IRF4 in T cell exhaustion, we generated and NFATc2 were required for the TCR-dependent induction of mice carrying floxed Irf4 alleles on a Cd4Cre transgenic back- IRF4. Genome-wide occupancy of IRF4, BATF, and NFAT was ground, resulting in conditional ablation of Irf4 in all T cells. As enriched at genes that constituted the exhaustion-specific previously reported in the setting of acute infection (Man et al., gene signature. Reduction of IRF4 expression was sufficient to 2013; Yao et al., 2013), complete loss of IRF4 in T cells in chron- restore anabolic metabolism and favored the formation of ically infected Irf4fl/flCd4Cre mice resulted in severely impaired TCF1-expressing memory-like antigen-specific T cells. Overall population expansion of gp33 and nucleoprotein (NP)396 our data demonstrate that TCR-responsive transcription factors LCMV epitope-specific CD8+ T cells and diminished viral control cooperatively establish T cell exhaustion while repressing (Figures S1A and S1B). To test a potential link between IRF4 memory T cell characteristics during chronic viral infection. dose and exhaustion, we infected mice in which only one copy of Irf4 was deleted in T cells (Irf4fl/+Cd4Cre). Loss of one Irf4 allele RESULTS resulted in decreased numbers of antigen-specific CD8+ T cells compared to controls (Figure S1C). Notably, however, viral con- Exhausted T Cells Express Elevated Amounts of IRF4 trol was not affected (Figure S1D), suggesting that the remaining and BATF antigen-specific T cells were functionally superior compared to To identify molecular mechanisms underlying the distinct their wild-type (WT) counterparts. properties of exhausted T cells, we performed RNA sequencing To examine the potential link between IRF4, inhibitory receptor (RNA-seq) analyses on adoptively transferred P14 TCR expression, and functionality of antigen-specific T cells in more transgenic CD8+ T cells recognizing the LCMV glycoprotein detail, we co-transferred congenically marked WT and Irf4+/ (gp)-derived epitope gp33-41 (gp33) isolated from mice infected P14 T cells into recipient mice, which we infected with chronic with two different strains of LCMV, causing either an acute (WE) LCMV Docile. Co-transfer of the cells ensured that both WT or chronic (Docile) infection. In agreement with earlier reports and Irf4+/ CD8+ T cells were exposed to the same viral load (Doering et al., 2012; Wherry et al., 2007), we observed pro- and inflammation. Irf4+/ P14 T cells expressed reduced nounced differences between the transcriptional profiles of amounts of IRF4 protein (Figure S1E) and were substantially acutely stimulated and exhausted T cells. We identified 945 impaired in their ability to undergo clonal population expansion genes early (day 8) and 1,529 genes late (day 30) specifically (Figure 2A). However, Irf4+/ P14 T cells secreted increased up- or downregulated in chronically stimulated cells in compari- amounts of cytokines IFN-g, TNF, and IL-2 (Figures 2B and 2C) son to acutely stimulated cells (>1.5-fold, <0.1 FDR). This set and displayed reduced expression of inhibitory receptors comprised a total of 2,050 unique ‘‘chronic CD8+ T cell PD-1, TIM-3, Lag3, 2B4, TIGIT, and CTLA4 (Figures 2D and signature’’ genes, of which 826 showed a differential expression 2E) compared to their WT counterparts, which could be reversed of >2-fold (Figure 1A and Table S1). This included several genes by overexpression of IRF4 (Figure S1F). Mice that received encoding inhibitory receptors, including Pdcd1 (PD-1), Havcr2 Irf4+/ P14 T cells showed a more pronounced drop in core

1130 Immunity 47, 1129–1141, December 19, 2017 A B d30 WT Chronic vs Acute Figure 1. Exhausted T Cells Express Chronic signature genes Elevated Levels of the Transcription Factors Lag3 Pdcd1 IRF4 and BATF Id3 Early (d8) Late (d30) Havcr2 Transcriptional proﬁling (RNA sequencing) was Maf Cxcr5 Tigit Irf4 Eomes performed on antigen-speciﬁc P14 T cells isolated Ctla4 Prdm1 Gzmb Batf from mice infected with LCMV strains causing acute 521424 1105 0 (WE) or chronic (Docile) infections. Ccr7 Tcf7 (A) Venn diagram showing overlap between genes

Fold change (log2) Il7r that display signiﬁcant changes (>1.5-fold, < 0.1 FDR) Up in Chronic in expression between P14 T cells isolated from mice Down in Chronic infected with chronic LCMV compared to P14 T cells 0 isolated from mice infected with acute LCMV, early Mean value of log2 RPKM (day 8) or late (day 30) post infection (p.i.). C (B) MA plot showing genes downregulated (blue) or Acute Chronic D gp33+, day 30 E day 30 Acute Acute upregulated (green) in chronically versus acutely naive d8 d30 d8 d30 Chronic Chronic stimulated T cells at day 30 p.i. Key genes of interest Entpd5 14 Dhodh 20 Suclg1 Gpd1l * are indicated. Acyp1 gp33+, day 30 12 Gpx1 Pcca ** (C) Heatmap displaying expression of selected Lef1 Acute Shmt2 15 Chronic * + Slc7a5 10 *** ‘‘chronic CD8 T cell signature’’ genes as in (A). Myc Hprt Isotype Srm Galk1 Shown are genes encoding inhibitory receptors, Sell 8 Pdk1 Cxcr4 100 ND2 10 transcriptional regulators, and chemokine receptors Nr1d1 , normalized to naive) ND4L 80 2 6 ND3 previously implicated in T cell exhaustion (Doering Ears2 Tcf7 60 Pisd 4 et al., 2012; Kahan et al., 2015; Wherry et al., 2007). Il7r 40 Oat mRNA relative to 5 Txnip Naive P14 T cells were included for comparison.

Ndufa6 Irf4 20 Mars2 2 Gstp1

Cox7c % of Max 0 (D) Quantitative RT-PCR analysis of Irf4 mRNA 2 3 4 5 Dmrta1 010 10 10 10 IRF4, GMFI (x10 Uqcrfs1 Satb1 0 IRF4 0 expression in flow cytometry-sorted gp33-specific Uqcc3 Days 5930 Cox16 Idh2 Ndufa9 T cells from acutely and chronically LMCV-infected Acadvl Acaa2 Acadm mice relative to Hprt expression at day 30 p.i. Atp1b3 F gp33+, day 30 G day 30 Mgst3 Sgk1 (E) Histogram (day 30 p.i., left) and geometric mean Guk1 Acute Acute Sema4a Aqp9 Chronic Chronic fluorescence intensity (GMFI) of IRF4 expression in Acss2 Lag3 5 gp33+, day 30 3 *** Cd160 Pdcd1 gp33-specific T cells from mice infected with acute Sh2d2a ** Acute Rgs16 Cd244 Chronic or chronic LCMV, assessed by intracellular staining Ccl4 Casp3 4 Isotype

Ctla4 Hprt at indicated times p.i. (right). GMFI is normalized to Icos Batf 2 Irf4 100 Hif1a 3 IRF4 expression in naive T cells. Zbp1 Tnfrsf9 80 Bcl2l11 (F) Quantitative RT-PCR analysis of Batf mRNA , normalized to naive) Nfatc1 60 3 Ptger2 2 Tox expression in ﬂow cytometry-sorted gp33-speciﬁc Ikzf2 40 1 Bcl2

Cxcr5 mRNA to relative T cells from acutely and chronically LCMV-infected 20 Hk1 Eomes 1 % of Max Batf Batf 0 mice relative to Hprt expression at day 30 p.i. Egr1 Id3 2 3 4 5 Id2 010 10 10 10 Zeb2 BATF Atp2b1 0 0 (G) BATF expression assessed by intracellular

Gzma BATF, GMFI (x10 Tbx21 Itga1 staining in gp-33-speciﬁc T cells 30 days after Il18rap Cd44 Bcl2l1 Tigit H I chronic or acute LCMV infection. GMFI is normalized Ifng Prdm1 Alcam Acute Chronic Acute Chronic to BATF expression in naive T cells (right). Gzmb 0.2 0.03 17.8 78.8 0.660.1 0.13 12.5 85.8 Havcr2 Day 9 Day 9 Slc2a3 (H and I) P14 cells from acutely or chronically LCMV- Klrg1 Prune Atp5h Ly6c1 infected mice stained for PD-1 and IRF4 (H), or PD-1 Prdx4 Il2ra Il1rl1 and BATF (I) and analyzed by ﬂow cytometry at day 9 96.4 3.4 2.0 1.5 97.3 1.9 1.4 0.22 and day 30 p.i. 5 Ro Day 30 105 0.4 0.36 10.5 72.3 Day 30 10 0.1 0.21 17.3 67.6

4 104 10 Error bars denote mean ± SEM. Graphs in (D)–(I) are

3 103 10 representative of 3–4 independent experiments with

2 0 102 10 0 92.3 6.9 5.9 11.3 0 85.2 14.6 11.4 3.81 4–5 mice per group. Statistical analysis was PD-1 PD-1 2 3 4 5 0102 103 104 105 010 10 10 10 IRF4 BATF performed using unpaired two-tailed Student’s t test (*p < 0.05, **p < 0.01, ***p < 0.001). body temperature (Figure 2F) and augmented cellular infiltration also for maintaining their numbers and phenotype during an into multiple non-lymphoid organs as compared to mice ongoing chronic infection (Figure 2G). harboring only WT P14 T cells (Figure S1G); however, viral BATF and IRF4 have been shown to cooperate in the differen- clearance was not improved (Figure S1H). Similar results were tiation of effector CD8+ T cells (Man et al., 2013; Kurachi et al., obtained from chronically infected bone marrow chimeric mice, 2014). Reduction or loss of BATF also resulted in reduced expres- showing impaired population expansion and reduced expression of multiple inhibitory receptors in gp33-specific T cells from sion of inhibitory receptors of endogenous Irf4+/ and Irf4/ chronically LCMV-infected bone marrow chimeric mice (Figures antigen-specific CD8+ T cells in comparison to their WT counter- 2H and S2D). Consistent with this result, Batf+/ P14 T cells iso- parts (Figures S2A–S2C). Tamoxifen-induced deletion of IRF4 lated from chronically LCMV-infected mice showed increased from Irf4fl/flCreERT2 T cells during the chronic phase of the infec- production of IFN-g, but not TNF or IL-2, compared to their tion resulted in dramatic loss of antigen-specific T cells and a wild-type counterparts isolated from the same mice (Figure S2E). reduction in PD-1 expression, indicating that IRF4 was required Overall, these data indicate that high amounts of IRF4 and BATF not only for the initial expansion of antigen-specific T cells but are required for the population expansion and maintenance of

Immunity 47, 1129–1141, December 19, 2017 1131 A B Acute Chronic Chronic Figure 2. IRF4 Drives T Cell Exhaustion and Chronic, co-transfer WT P14 WT P14 Irf4+/- P14 Restrains Effector Function WT P14 67.5 13.6 13.4 2.8 42.1 7.5 Congenically marked wild-type (WT, Ly5.1) and Irf4+/- P14 Irf4+/ (Ly5.2) P14 T cells were co-transferred into recipient mice (Ly5.1+Ly5.2+), which were infected ) 3 5 with acute (WE) or chronic LCMV (Docile) 18.7 0.15 81 2.9 48.4 1.98 subsequently. 2 IL-2 (A) Absolute numbers of adoptively co-transferred +/ 105 16.2 65.5 8.4 2.2 31 13.9 WT and Irf4 P14 T cells over the course of a

1 104 chronic LCMV infection.

103 (B and C) Representative ﬂow cytometric analysis Total # P14 (Spleen, x10

γ (B) and quantiﬁcation (C) of IFN-g, TNF, and IL-2 0 2 10 +/ 0830 0 IFN expression in WT and Irf4 P14 T cells following Days post infection 16.3 2.02 84.9 4.5 52 3.2 0102 103 104 105 ex vivo gp33 peptide restimulation at day 30 post TNF infection (p.i.). (D and E) Geometric mean ﬂuorescence intensity C Chronic, co-transfer, day 30 D Chronic, co-transfer, day 30 (GMFI, D) and representative (E) histograms of WT P14 WT P14 PD-1, TIM-3, Lag3, 2B4, TIGIT, and CTLA4 Irf4+/- P14 Irf4+/- P14 expression on WT and Irf4+/ P14 T cells at day 30 after chronic LCMV infection. Histograms repre- 60 20 15 6 5 15 10 *** 15 *** *** *** senting P14 in acute infection are shown in (E) as a )

) * ) 2 3 50 ** ** 4 2 8 reference. 15 P14 x10 l x10 x10

a 10 4 10

40 t tal P14 10 * (F) Maximum drop of core body temperature of

of totalof P14 3 of totalof P14 6 + + of totalof P14 of to of

+ chronically LCMV-infected mice that received WT or 30 to of 10 (GMFI, + + (GMFI, (GMFI, IL-2 +/ 2 2 TNF 4 + + 20 5 5 2 5 3 Irf4 P14 T cells. Donor cell numbers were IL- %IFN

5 % PD -1 adjusted to achieve similar representation of WT %TNF 1 Lag 2 10 TIM -3 %IFN %IFN and Irf4+/ P14 cells, respectively, at day 13 p.i. 0 0 0 0 0 0 0 0 (G) Absolute numbers (left) and PD-1 expression 15 15 1.5 (right) of gp33-specific CD8+ T cells in Irf4fl/fl mice E Acute, WT P14 ) ) ***

) *** 2 2 carrying a tamoxifen-inducible Cre recombinase Chronic, WT P14 co-transfer 2 *** x10 +/- x10 transgene (ERT2 Cre) after deletion of IRF4 using Chronic, Irf4 P14 x10 10 10 1.0 4-OH tamoxifen (4-OH Tam) or control treatment MFI, G (GMFI,

( with carrier only (peanut oil). Mice were treated at

5 4 0.5 5 day 15 and analyzed at day 30 after chronic LCMV 2B4 TIGIT (GMFI,

CTLA infection. 0 0 0 (H) GMFI of inhibitory receptors PD-1, Lag3, TIM-3, and 2B4 on gp33-speciﬁc CD8 T cells in chronically PD-1 TIM-3 Lag3 LCMV-infected bone marrow chimeric mice 100 H Chronic, gp33+, day 30 reconstituted with WT and Batf+/ or WT and 80 WT (Ly5.1) / 60 Batf hematopoietic cells at day 30 p.i. Batf+/- (Ly5.2) chimeric mice 40 Batf-/- (Ly5.2) Flow cytometry data in (B) and (E) are representative 20 and data in (C) and (D) are pooled from 2 indepen- 3

3 6 4

0 33

% of Max *** 0102 103 104 105 dent experiments with 4–6 mice per group. Data in gp *** 2B4 TIGIT CTLA4 5 gp *** f 3 (F) are representative of 2 independent experi- of *** )o 4 ) 3 3 ments, containing 3–6 mice per group, and data in *** 3 2

(x10 (G) are representative of 2 experiments with 3–4

F G I(x10 FI 2 F N.S. mice each. Data in (H) are pooled from 3 indepen- Chronic, separate transfer Irf4fl/flCreERT2 1 WT P14 carrier control GM 1 dent experiments with 3–5 mice per group. Error 1 Irf4+/- P14 4-OH Tam 0 0 bars denote mean ± SEM and statistical analysis PD- Lag3 GM 15 * 25 was performed using paired (H) or unpaired two- 3

+ * *** 15 *** 3 10 20 tailed (all other) Student’s t test (*p < 0.05, **p < 0.01, *** 33

20 gp **

8 *** gp ***p < 0.001). 10 15 )of of ) of gp33 , Spleen) 2 4 ) 10 4 15 6 2 N.S. 10 (x10 + 10 4 5 5 MFI (x10 ** 5 5 G 2 # gp33 genes (Figures S3A and S3B, Table S2). max % temperature loss PD-1 (GMFI, x10 B4 GMFI (x10 0 0 0 0 0 2

TIM-3 Consistent with this finding, chronically stimulated T cells showed reduced mitochondrial membrane potential and mass, antigen-specific T cells during chronic infection, while promoting exhibited increased oxygen consumption rate (OCR) utilized T cell exhaustion at the same time. for ATP production, and displayed an increased proton leak (Fig- ures 3A and 3B), indicating decreased efficiency of coupled IRF4 Suppresses Anabolic Metabolism and respiration. Chronically stimulated T cells, in particular PD-1hi Mitochondrial Function in Exhausted T Cells cells, showed reduced production of reactive oxygen species Further analysis of the RNA-seq data showed that metabolic (ROS), a byproduct of electron transport chain-dependent respi- pathways were significantly enriched among chronic signature ration (Figures 3C and 3D) that is critical for cellular activation

1132 Immunity 47, 1129–1141, December 19, 2017 A B C Chronic, P14 Figure 3. IRF4 Drives Bioenergetic Insufﬁ- Day 30 Acute, P14 Endogenous ciencies during T Cell Exhaustion Acute P14 Acute P14 Acute P14 Chronic P14 Chronic P14 Chronic P14 (A) Mitochondrial mass and membrane potential Day 8 determined by Mitotracker Green and Mitotracker 1.5 1.5 100 2.0 ) 4 *** * *** * 3 Orange (CMTMRos) labeling, respectively, in P14 ) *** 80 *** T cells isolated from acutely (WE) or chronically Fold *** 1.5 N.S. 3 1.0 I ( 1.0 (Docile) LCMV-infected mice at day 30 post infec- F ** M 60 tion (p.i). Geometric mean ﬂuorescence intensity G 100 S), GMFI (x10 S), GMFI Day 30 ) 1.0 O 2

80 R (GMFI) was normalized to GMFI in P14 cells isolated ( ( 40 *

0.5 ox 0.5 60 from acute LCMV infection. R

Proton Leak (Fold) 0.5 1 20 40 (B) Oxygen consumption rate (OCR) utilized for ATP Cell totracker i 20 M turnover calculated as the difference in basal rate Mitotracker (Mass) GMFI (Fold) % of Max 0.0 0.0 (pMoles/min) ATP OCR utilised for 0 0.0 0 0 Day 8 30 Day 830 0102 103 104 105 Day 8 30 and rate following oligomycin treatment (left), and CellRox (ROS) proton leak (right) calculated as the difference between OCR after oligomycin and after rotenone/ D Chronic, P14, day 30 E Chronic, co-transfer, day 30 F Chronic, co-transfer, day 30 antimycin and expressed as fold change over P14 PD-1 High WT P14 WT P14 PD-1 Int Irf4+/- P14 Irf4+/- P14 T cells in acute infection at days 8 and 30 p.i. 1.5 (C) Representative histograms (left) and correspond- 1.5

1.5

) *** ing GMFI (right) of reactive oxygen species (ROS) d

*** l

o ** (Fold)

100 I (Fold)

F 100 abundance determined by CellRox DeepRed labeling ( F

I 1.0 1.0 F 80 1.0 80 in adoptively transferred P14 T cells during acute and M G

60 ) 60 chronic LCMV infection at days 8 and 30 p.i. (

40 GM (Mass) 40 (D) ROS abundance determined by CellRox 0.5 0.5 0.5 er 20 20 DeepRed labeling in PD-1 high and PD-1 interme- ck a r t % of Max % of Max

0 o 0 diate (int) P14 T cells in chronically LCMV-infected CellRox (ROS) GMFI 0102 103 104 105 Mitotracker 0102 103 104 105 CellRox (ROS) Mit 0.0 0.0 CellRox (ROS) 0.0 mice at day 30 p.i. (E and F) Congenically marked wild-type (WT, Ly5.1) and Irf4+/ (Ly5.2) P14 T cells were co- G Day 30 H Chronic, co-transfer, day 30 transferred into Ly5.1+Ly5.2+ hosts and analyzed at Acute P14 WT P14 Chronic P14 Irf4+/- P14 day 30 after chronic LCMV infection. Mitochondrial mass and membrane potential were determined by 25 2.0 20 2.0 2.0

ls *** ls l Mitotracker Green and Mitotracker Orange * ** ** ce cel 20 *** 6

6 (CMTMRos) labeling (E). ROS abundance was ld) 1.5 0 15 d) 1.5 d) o 1.5 1 l

/ F / 10 determined by CellRox DeepRed labeling (F). M I ( M 15 μ μ

) ) 2

MF Normalized to GMFI of WT P14 from acute LCMV 2 10 1.0 0 1.0 1.0 1

10 x infection at day 30 p.i. DG G 98 GMFI (Fol ate ( ate t

NB (G) Extracellular L-Lactate production from ﬂow tate (x10 0.5 5 0.5 0.5 CD CD71 GMFI (Fo CD71 c 5 2- Lac

- cytometry-sorted P14 cells at day 30 after acute L L-La 0 0.0 0 0.0 0.0 or chronic LCMV infection and stimulated ex vivo with IL-2. I Chronic, day 8 J (H) Uptake of ﬂuorescent glucose analog 2-NBDG, WT extracellular L-Lactate production after ex vivo Irf4+/- Chimeric mice, gp33+, day 30 stimulation with IL-2, and expression of CD98 and 100 2.5 +/- ** ** WT (Ly5.1) : Irf4 (Ly5.2) CD71 (GMFI normalized to WT P14 cells) in WT and +/ ) 80 * 2.0

n Irf4 P14 cells as (E).

i * 105 20.7 14.1 56.1 14.9 m (I) Basal extracellular acidiﬁcation rate (ECAR) / 60 1.5 4 H

(pM/mpH) 10

p measured on bulk effector phenotype

m ( 3 + + 10 40 CAR 1.0 (CD62L CD44 ) CD8 T cells isolated from WT or R A Irf4+/ mice at day 8 after chronic LCMV infection 102

EC 20 0.5

OCR/E 0

CD98 with and without the addition of oligomycin (left) and 2-NBDG 63.6 1.55 28.2 0.78 2 3 4 5 0 0.0 0 10 10 10 10 the OCR/ECAR ratio (right). Oligomycin + Ly5.2 - (J) Representative flow cytometric analysis of fluorescent glucose analog 2-NBDG uptake and expression of CD98 on congenically marked WT (Ly5.1) and Irf4+/ (Ly5.2) gp33-specific CD8+ T cells from mixed bone marrow chimeric mice infected with chronic LCMV and analyzed at day 30 p.i. Data in (A), (C), (E), and (H) are pooled from 2–4 independent experiments with 3–4 mice per group; other data are representative of 2–4 independent experiments with 3–4 mice per group. Error bars denote mean ± SEM and statistical analysis was performed using unpaired two-tailed Student’s t test (*p < 0.05, **p < 0.01, ***p < 0.001). and for the maintenance of cytotoxic T cell proliferation and L-Lactate, an end-product of aerobic glycolysis (Figure 3G). In function (Sena et al., 2013; Weinberg et al., 2015). Reducing contrast, Irf4+/ P14 T cells responding to chronic LCMV infec- the amount of IRF4 in chronically stimulated P14 T cells was suf- tion showed increased uptake of the glucose analog 2-NBDG, ficient to restore mitochondrial size and membrane potential as secreted elevated amounts of L-Lactate, and showed increased well as ROS production (Figures 3E and 3F). Consistent with expression of the amino acid transporter component CD98 and impaired anabolic metabolism, chronically stimulated P14 the transferrin receptor CD71 (Figure 3H), which could be T cells in response to stimulation with IL-2 secreted less reverted by overexpression of IRF4 (Figure S3C). Indeed, as early

Immunity 47, 1129–1141, December 19, 2017 1133 A B C Day 30, P14Day 30 Day 8 IgG Nfatc1 1.0 1.0 1.0 NFATc1 * NFATc2 input) Acute Chronic input)

50 % *** 0.8 *** ( 0.8 0.8 (% exon 1-3 ** y **

NFATc2 c e

* cy 40 exon 2-3 0.6 0.6 0.6 naiv upan NFATc1 cupan ver oc

30 occ

occupancy (% input) **

0.4 0.4 0.4 e 1

20 m 0.2 0.2 o 0.2

** promoter os

nduction o i promoter

IRF4 1 10 cd Irf4 old 0.0 0.0 0.0 f BATF Chrom Pd 0 Actin Acute Acute Acute Chronic Chronic Chronic naive Acute naive Acute Chronic Chronic

D E KO KO CsA: - + controlNfatc1 Nfatc2 DKO

NFATc2 NFATc1

NFATc1 IRF4 IRF4 BATF BATF Actin p50 NF-kB

Figure 4. IRF4 and NFAT Form a Positive Feedback Circuit during T Cell Exhaustion (A) Expression of Nfatc1 isoforms as identified by RNA sequencing. Graph shows expression relative to naive T cells of the short isoform (NFATc1aA, represented by exon 1–3 junctions, skipping exon 2) and the long isoform (represented by exon 2–3 junctions) in P14 T cells isolated from acutely (WE) and chronically (Docile) LCMV-infected mice at day 30 post infection (p.i.). (B) Western blot showing NFATc1, NFATc2, IRF4, BATF, and Actin (loading control) expression in CD8+ T cells flow cytometry sorted as CD62L from acutely and as CD62LPD-1+ from chronically LCMV-infected mice at day 30 p.i. The arrow marks the short isoform of NFATc1. (C) Binding of NFATc1 and NFATc2 to the Irf4 and Pdcd1 (encoding PD-1) promoters demonstrated by chromatin immunoprecipitation using specific antibodies or control IgG and RT-qPCR in total CD44+CD8+ T cells isolated from spleens of acutely and chronically LCMV-infected mice at day 8 p.i. Enrichment is expressed as percentage of total chromatin input and compares isotype control and NFATc1- and NFATc2-specific antibodies. (D) Western blot showing NFATc1, NFATc2, IRF4, BATF, and p50 NF-kB (loading control) expression in polyclonal CD8+ T cells deficient (KO) in NFATc1 (Nfatc1fl/flCd4Cre), NFATc2 (Nfatc2/), or both (DKO) after in vitro activation with anti-CD3 and anti-CD28 for 48 hr. (E) Western blot showing NFATc1, IRF4, BATF, and Actin (loading control) in polyclonal CD8+ T cells activated with anti-CD3, anti-CD28, and IL-2 with or without cyclosporine A (CsA) for 24 hr. Data in (A)–(C) are representative of 2 independent experiments and data in (D) and (E) are representative of 3 independent experiments. Error bars denote mean ± SEM. Statistical analysis was performed using unpaired two-tailed Student’s t test (*p < 0.05, **p < 0.01, ***p < 0.001). as day 8 during chronic infection, bulk Irf4+/ effector CD8+ amounts of Nfatc1 RNA, in particular the short isoform Nfatc1aA T cells showed elevated extracellular acidification rate (ECAR) (Figure 4A), which is transcriptionally induced by TCR signaling and a pronounced shift in the OCR to ECAR ratio in comparison and utilizes exon 1 and an intragenic 30 UTR (Serfling et al., to their WT counterparts (Figure 3I). Similar results were obtained 2012). Accordingly, high amounts of NFATc1 protein were ex- from endogenous Irf4+/ gp33-specific CD8+ T cells from LCMV pressed together with IRF4 and BATF, whereas NFATc2 showed Docile-infected mixed bone marrow chimeric mice (Figures 3J, reduced abundance in effector CD8+ T cells from LCMV Docile in S3D, and S3E). Collectively, these data indicate that increased comparison to WE-infected mice (Figure 4B). Chromatin immu- expression of IRF4 contributes to the suppression of glycolysis noprecipitation (ChIP) and quantitative RT-PCR showed that and OXPHOS in exhausted T cells. NFATc1 and NFATc2 binding was enriched in the promoter regions of the Irf4 and Pdcd1 loci in CD8+ T cells isolated from IRF4 and NFAT Form a Positive Feedback Circuit during acutely or chronically LCMV-infected mice compared to isotype T Cell Exhaustion controls (Figure 4C). Notably, combined ablation of Nfatc1 and Induction of inhibitory receptor expression is mediated by the Nfatc2 resulted in a substantial reduction of IRF4 but not BATF transcription factors NFATc1 and NFATc2 (Lu et al., 2014; in activated CD8+ T cells (Figure 4D). In line with a critical role Martinez et al., 2015; Oestreich et al., 2008). In line with this for NFAT proteins in the induction of IRF4 expression, CD8+ notion, we found that exhausted T cells expressed elevated T cells activated in the presence of cyclosporine A, an inhibitor

1134 Immunity 47, 1129–1141, December 19, 2017 A B C IRF4 bound BATF bound IRF4_BATF_NFAT bound genes genes genes ‘exhaustion’ (d30 acute / chronic) Chronic signature Genes bound by ***p<0.00001 genes IRF4, BATF and NFAT 2 900 1068 262 1 2970

1651 450 1518532 2438 Enrichment 0

5051

p=5.39e-18

NFAT bound Up in acute Up in chronic genes 3.0 1.9 1.2 0.7 0.3 11.0 −0.2 −1.7 −0.8 −3.0 Statistics −20.6 D

NFAT

IRF4 Reads

BATF Reads Input Reads Reads 10 kb Havcr2 Tcf7 10 kb 10 kb Hif1a

E Distance from BATF peak Distance from BATF peak Distance from IRF4 peak to nearest IRF4 peak to nearest NFAT peak to nearest NFAT peak 180 40 45 40 100 35 30 35 80 30 25 25 60 20 20 15

# of peaks 40 15 10 10 20 5 5 0 0 0 -1000-5000 500 1000 -1000-500 0 500 1000 -1000 -500 0 500 1000 Distance (base pairs) Distance (base pairs) Distance (base pairs)

F G Distance IRF4 motif to NFAT motif (anchor) 50 NFAT

40 IRF4 30 Reads

20 BATF Reads

# of sequences 10 0 Input

-100-75-50 -25 0 2550 75 100 Reads Reads Distance (base pairs) 10 kb Nfatc1

H WT Irf4-/- WT Irf4-/- time (h) 24 48 72 24 48 72 time (h) 0 48 0 48

NFATc1 NFATc2

IRF4

IRF4 Actin

Actin

Figure 5. IRF4 and BATF Cooperate with NFAT to Establish T Cell Exhaustion (A) Venn diagram showing overlap between genes bound by IRF4, BATF, and NFAT as identified by chromatin immunoprecipitation (ChIP) sequencing of effector CD8+ T cells (Kurachi et al., 2014; Man et al., 2013; Martinez et al., 2015). (legend continued on next page) Immunity 47, 1129–1141, December 19, 2017 1135 of NFAT activity, expressed very little IRF4 protein and showed IRF4, and BATF cooperate at key genomic loci that control the reduced expression of BATF (Figure 4E). Overall, these data transcriptional signature of exhausted T cells. indicate that NFATs are major inducers of IRF4 expression and show that elevated expression of NFAT, in particular NFATc1, IRF4 Controls the Transcriptional Profile Associated contributes to the increased expression of IRF4 and BATF in with Exhaustion chronically stimulated T cells. To understand how high amounts of IRF4 drive T cell exhaustion, we performed transcriptional profiling of Irf4+/ and WT P14 IRF4 and BATF Cooperate with NFAT to Establish T Cell T cells at d30 after acute or chronic LCMV infection. A decrease Exhaustion in IRF4 dosage resulted in a loss of large parts of the transcrip- To understand how the elevated expression of multiple TCR- tional signature associated with chronic infection and was suffi- responsive transcription factors contributes to the establishment cient for T cells to adopt a transcriptional profile similar to the one of T cell exhaustion, we examined genome-wide DNA binding induced after acute infection (Figures 6A and S6A, Table S4). data for IRF4, BATF, and NFAT (Man et al., 2013; Kurachi Although some memory T cell-associated genes were upregu- et al., 2014; Martinez et al., 2015). IRF4, BATF, and NFAT shared lated in Irf4+/ P14 T cells as early as d8 after infection, the a large proportion of their target genes (Figure 5A). Binding of all majority of these genes were expressed at higher levels only at three factors was significantly enriched among genes encoding later time points (Figure S6A). Irf4+/ P14 T cells showed reduced transcripts that constituted the chronic signature (Figures 5B transcript abundance for genes encoding inhibitory receptors, and 5C) and included genes encoding inhibitory receptors including Havcr2, Tigit, Cd244, Pdcd1, Lag3, and transcriptional (e.g., Pdcd1, Havcr2, Ctla4), molecules involved in cellular meta- regulators usually expressed at high amounts in exhausted cells, bolism (e.g., Hif1a, Slc2a3, Pgm2l1), mitochondrial electron including Prdm1 and Maf (Figure 6A). In contrast, Irf4+/ P14 transport chain complex components (e.g., Atp2c1, Atp7a, T cells upregulated multiple genes associated with memory Atp10a, Ndufv2, Ndufaf2), and effector and memory T cell devel- T cell development, including Tcf7, Ccr7, Il7r, and Sell (Fig- opment (Tcf7, Foxo1, Tbx21, Prdm1)(Figures 5D, S4A, and S4B, ure 6A). Overall, 49% of the chronic signature genes were Table S3). This was in contrast to genes bound by only one or responsive to changes in IRF4 expression (Figure 6B, Table two of the transcription factors, which showed no or little enrich- S5). Gene set enrichment analyses confirmed that expression ment among the chronic signature genes (Figure S4C). of chronic signature genes was reduced, while expression of Genome-wide, BATF and IRF4 binding sites were enriched in genes associated with memory T cell development was enriched close proximity to each other, while NFAT sites were enriched in Irf4+/ compared to WT P14 T cells responding to chronic adjacent to both BATF and IRF4 sites (Figure 5E). Indeed, LCMV Docile (Figures 6C and 6D). Consistent with our observa- the IRF4 motif was found to be enriched in close proximity tion that Irf4 haploinsufficieny restored cellular metabolism to the core NFAT motif, with 50% of the IRF4 motifs located during chronic infection, Irf4+/ P14 T cells showed higher in %23 bp distance to the NFAT motif (Figure 5F). Similar enrich- expression of genes involved in OXPHOS, amino acid meta- ment was observed for the AP-1 motif (recognized by BATF), bolism, fatty acid oxidation, the tricarboxylic acid cycle, and with 50% of the IRF4 and NFAT motifs located in %20 bp mitochondrial respiratory chain complexes (Figures 6E and 6F). distance to the AP-1 motif (Figure S5A), as shown for the Expression of Aqp9, a gene critical for memory T cell develop- composite intronic binding site in Tcf7 (Figure S5B). Overall, ment and strongly downregulated in exhausted T cells, was we detected overlapping ChIP-seq peaks (mean overlap recovered in Irf4+/ P14 T cells (Cui et al., 2015). Collectively, >113 bp) at 2,185 sites, providing evidence for co-binding of these data reveal that high amounts of IRF4 drive the transcrip- all three transcription factors. The AP-1 core motif (TGAnTCA) tional profile associated with CD8+ T cell exhaustion. was the most enriched motif within these regions (Figure S5C). Interestingly, we found that IRF4 together with BATF and High Amounts of IRF4 Repress Memory T Cell NFAT itself bound to the Nfatc1 locus (Figure 5G), suggesting Development a feedback loop regulated by the activity of all three transcrip- Recently, we and others have shown that a population of tion factors. Indeed, Irf4/ CD8+ T cells showed reduced antigen-specific T cells during chronic infection displays a expression of NFATc1, in particular the short isoform, but not memory-like phenotype and maintains the proliferative potential of NFATc2 (Figure 5H). Overall, our data indicate that NFAT, of exhausted T cells (He et al., 2016; Im et al., 2016; Leong et al.,

(B) Venn diagram showing signiﬁcant overlap between ‘‘chronic CD8+ T cell signature’’ genes and genes bound by all three transcription factors, IRF4, BATF, and NFAT (Fisher’s exact test). (C) Gene set enrichment test showing the top 500 genes bound by all three transcription factors, IRF4, BATF, and NFAT, demonstrating enrichment of genes that are associated with ‘‘exhaustion.’’ Signiﬁcance was tested using Gene Set Test. (D) Representative ChIP sequencing tracks showing composite binding of IRF4, BATF, and NFAT to chronic CD8+ T cell signature genes Havcr2 (encoding TIM-3), Tcf7 (encoding TCF1), and Hif1a. (E) Genome wide co-localization of BATF, IRF4, and NFAT. Histograms show normalized average tag density at indicated distance from the nearest neighbor as indicated. (F) Histogram showing the IRF4 motif distribution relative to the NFAT motif (anchor) in BATF/IRF4/NFAT triple binding regions. (G) Representative ChIP sequencing tracks showing composite binding of IRF4, BATF, and NFAT to chronic CD8+ T cell signature gene Nfatc1. (H) Western blot showing NFATc1, NFATc2, IRF4, and Actin (loading control) expression in WT and Irf4/ CD8+ T cells at indicated time points after activation with anti-CD3 and anti-CD28 antibodies. The arrow marks the short isoform of NFATc1. Data in (A)–(G) are combined from two independent experiments. Data in (H) are representative of two independent experiments.

1136 Immunity 47, 1129–1141, December 19, 2017 A Acute, d30 Chronic, d30 B Chronic signature IRF4-responsive Figure 6. IRF4 Establishes the Exhaustion- genes genes Associated Transcriptional Profile +/- +/- WT Irf4 WT Irf4 Transcriptional profiling (RNA sequencing) was Rpsa Rpl18 Rpl32 Rplp0 Rps6 performed on antigen-specific wild-type (WT) and Gm3362 Gm10335 Sidt1 Gm10045 Rpl18a +/ Gm10132 Rpl5 Eef1a1 Irf4 P14 T cells isolated at day 30 post infection Rpl17−ps10 Rps9 10391011 1511 Npm1 Gpr18 Eef1b2 EntrezID 100861774 Rps7 (p.i.) with LCMV strains causing acute (WE) or Gnb2l1 Rpl23 Rps2 Rpl37a Rps19−ps3 Gm9794 Eef1g chronic LCMV (Docile) infections. Myc Rps15 Ccr7 Tbl3 Gm8942 Gm20568 Emb (A) Heatmap showing expression of top 250 Gm11425 A130040M12Rik Rps16 Rps10 Gpr183 + Gm8623 Dse Gm4342 ‘‘chronic CD8 T cell signature’’ genes that are Rpl29 Tnfrsf26 Rps19 Sgsh Fam46a +/- +/ Il4ra Irf4 vs WT (d30 chronic) DE genes tested on Epsti1 differentially expressed (DE) in WT versus Irf4 Rplp1 C Rpl30 Wdr36 Pno1 ‘exhaustion’ genes (acute vs chronic, d30 WT) Gm8692 Bzw2 Rpl5−ps2 P14 T cells in chronic LCMV infection at day 30 p.i. Polr1b Pdlim1 Tpt1 Glo1 Gm6867 Rplp2 Rps28 Rpl37 Expression in acute infection is shown as a Rnf138 Il7r Satb1 Slc16a1 Gramd3 Tspan31 Vkorc1 reference. Genes important for memory T cell Tfap4 Ccr2 05.5 Hs3st3b1 Slc9a9 Enrichment Prmt7 Cd44 Ly6c2 differentiation and function as well as for exhaustion Cd53 Poglut1 Plin3 Klhl6 Klrk1 Arl4c Slco3a1 Gm5424 are highlighted. Pcca Rps26 Tcf7 Up in Rpl3 Cpne3 Rpl35 Down in

Tnfrsf13b exhaustion (B) Venn diagram showing overlap between the Rpl6 exhaustion Kbtbd11 Rps20 Gas5 Mgrn1 + Rpl12 Crtam ‘‘IRF4-responsive’’ and the ‘‘chronic CD8 T cell Pecam1 Psen2 Sell Adk Slamf6 Cd9 Myb ichment Fgl2 r signature’’ gene sets. Abr Prkch

Il21r 5.5 0 Coro2a Cd101 En Rcn1 Mxd1 (C and D) Gene-set enrichment tests showing DE Vps37b Entpd1 Etv3 Gfi1 1.2 1.9 0.7 3.0 0.3 11.0 −0.8 −3.0 −0.2 −1.7

Pear1 −20.6 Ildr1 Havcr2 genes associated with chronic T cell stimulation (C) Cd200r1 Statistics Chst12 Adam19 Glcci1 Enpp5 Plekha1 Ptms and effector to memory conversion (D) to be Gm4956 Ociad2 Wscd2 +/- Slc27a4 Maf Irf4 vs WT (day 30 chronic) DE genes tested on Ccrl2 +/ Cd38 D enriched in Irf4 P14 T cells compared to WT P14 Lgalsl Cd244 Adgrg1 effector (acute WT d8) vs memory (acute WT d30) Glp1r 2900026A02Rik Tigit Gm19765 Itih5 T cells responding to chronic LCMV infection. Red, Nr4a2 Csf1 Pdcd1 Ube2l6 +/ +/ Tox2 Chst2 Lag3 genes up in Irf4 versus WT. Blue: down in Irf4 Gem Chn2 Prr5l Pkdcc Ptger4 Nedd9 Il12rb1 03.8 versus WT. Sh2d2a Gzmb Enrichment Rps6ka1 Nkg7 Sec14l1 Cd200r2 Tnfrsf1b Tnfsf10 (E) Gene-set enrichment analysis of genes DE St3gal3 Fkbp5 Dusp6 Cd200r4 Nfatc1 +/ Osgin1 Cish between WT and Irf4 P14 T cells, indicating Ccl3 Sept4 Up in Prex1 Hic1 memory memory Galnt2 Down in Aplp1 Calhm2 Pias3 signiﬁcant enrichment of genes involved in mito- Nab2 Adrb1 Gpd2 Cela1 0 Sla2 Plekhf1 Timp2 chondrial oxidative phosphorylation (OXPHOS). Casp7 Enpp1

Stk32c ichment Ica1 Htra3 Cd3e Scrib (F) Bar graphs showing expression changes (log fold Csnk1e 2.2

Cd7 Enr Mprip Chsy1 Ptk2b Aplp2 Actr1b Cep170b change RPKM, logFC) of key genes representing Rbm38 Arhgap10 2.5 3.9 1.4 0.6 16.3 −0.8 −0.1 −1.5 −4.1 Kctd9 −2.5 Slc16a6 −20.8 Cited4 + Cmip Rgs3 Statistics metabolic pathways found within the chronic CD8 Gpr65 Elmo2 Dapk2 Slc36a1 Gse1 +/ Baiap2 Perp T cell signature comparing WT and Irf4 P14 Ccl4 Atp6v0a2 Gimap1os Habp4 Ctsc OXPHOS (KEGG) related tested on Metrnl E Trac Cd160 T cells at day 30 p.i. Irak2 +/- Osbpl3 Irf4 vs WT (d30 chronic) Gzma Pacsin2 Trerf1 Baiap3 Padi2 Graphs and data in (A)–(F) are combined of two to Gm7334 Fasl Abi3 Gdf11 Mxi1 Camk2n1 Frmd4b three independent experiments. Cdk2 Prdm1 Sytl2 Plk3 Pofut2 Mettl7a1 1700017B05Rik Slamf1 Soat2 Tmtc4

N4bp3 Enrichment Row Z−Score 01.7 Irf4+/- Irf4+/- −1 0 1 Up in Down in of IRF4 (Figure S5B), we examined the dif- 0.89 2.29 1.41 0.07 0.47 10.94 −0.38 −0.84 −1.42 −2.28 F WT, chronic vs acute (day 30) Statistics −13.68 ferentiation of memory-like T cells during Irf4+/- vs WT, chronic (day 30) chronic infection in more detail. A much I. Glycolysis II. Amino acid metabolism VI. Mitochondrial respiratory chain / OXPHOS larger proportion of adoptively transferred III. Fatty acid beta-oxidation VII. Tricarboxylic acid cycle +/ IV. Purine and pyrimidine metabolism VIII. Mitochondrial associated enzymes Irf4 P14 T cells expressed TCF1 and V. Glutathoine metabolism and redox state IX. Signaling kinases/other CD62L, which correlated inversely with I II III IV V VI VII VIII IX 2.0 TIM-3 expression. In addition, these cells 1.5 also expressed elevated amounts of 1.0 CXCR5 and IL-7R compared to WT cells 0.5 (Figures 7A–7C). Increased proportions of 0.0 memory-like Irf4+/ P14 cells were estab- -0.5 lished early and maintained throughout -1.0 infection (Figure S6B). Although the pro- -1.5 portion of TCF1+ cells was much increased -2 -4 among Irf4+/ P14 T cells, the overall -6 Transcript abundance (Fold change log2)

+ 1 3 5 2 9 p 2 a 1 h m k1 s g s2 numbers of both TCF1 and TCF1 g s p5h Oat uk Hk1 Sr xni Idh rdx4 ND2 ND3 Pisd Pisd

c Pcc Pdk1 S Hif1a +/ Aqp9 G T Gpx1 At ND4L ntpd5 ntpd5 Galk1 Gpd1l P Gstp1 Ears2 Ears2 Prune Mar Shmt2 Acaa2 A Acyp1 Mgst3 Uqcc3 Cox7c Cox16 Sucl Acadvl Slc2a Slc7a Dhod Atp1b3 Atp2b1 Acadm Irf4 P14 T cells were reduced due to E Ndufa Ndufa6 Uqcrfs1 the critical role of IRF4 in T cell population expansion (Figure 7D). However, the 2016; Utzschneider et al., 2016). These cells display increased reduction was much more substantial for TCF1 cells, indicating expression of TCF1, Id3, and chemokine receptor CXCR5 and that IRF4 specifically promotes the differentiation of fully show reduced expression of TIM-3 and Blimp-1, consistent exhausted T cells (Figure 7E). with the transcriptional profile of Irf4+/ P14 T cells. As we To functionally test the recall capacity of chronically stimulated observed that Tcf7, which encodes TCF1, was a direct target antigen-specific T cells expressing different amounts of IRF4, we

Immunity 47, 1129–1141, December 19, 2017 1137 A B Figure 7. IRF4 Represses Memory T Cell 100 *** WT P14 +/- Development Irf4 P14 80 17.2 0.84 85.2 1.49 (A) Representative ﬂow cytometry plots (left) 104

of P14 P14

+ 60 endogenous CD8 showing TCF1 versus TIM-3 and CD62L on wild- WT 40 +/ 103 P14 type (WT) and Irf4 P14 cells isolated at day 13

% TCF1 20 0 post chronic LCMV (Docile) infection and enumer- 0 -103 12.5 69.4 ation (right). 2.16 11.1 +/- 3 3 4 5 -10 0 10 10 10 WT TIM-3 Irf4 (B) Representative histograms showing CXCR5 *** +/ 27.0 7.92 25.4 58.9 80 100 60 105 *** expression (left) on wild-type (WT) or Irf4 P14 Irf4+/- 80 60 P14 cells in chronically LCMV-infected mice at day 13 4 of P14 10 60 of P14

+ 40

+ post infection (p.i.) and enumeration (right). 40 40 103 (C) Expression of memory marker IL-7Ra on WT 20 20 20 +/ TCF1 2 or Irf4 P14 cells in chronically or acutely LCMV- 10 59.0 6.06 11.9 3.81 % CXCR5 % of Max % CD62L 0 0 103 104 105 0 104 105 CD62L 0 CXCR5 0 infected mice as indicated at day 30 p.i. +/- +/- + WT (D) Relative and absolute abundance of TCF1 (left, WT Irf4 Irf4 middle) and TCF1 (right) P14 cells at day 15 after C Acute WT P14 D Chronic, day 15 E chronic LCMV infection. Chronic WT P14 +/ + 3 40 (E) Ratio of Irf4 versus WT P14 TCF1 and TCF1 Chronic Irf4+/- P14 80 10 *** ** *** cells, respectively, as in (D). 100 * , Spleen) , Spleen)

8 vs WT) 6 5 30 (F) Representative ﬂow cytometry plots showing 80 60 2 +/- expression of IRF4/GFP versus PD-1 and TIM-3 60 6 Irf4 40 20

P14 (x10 in CD8 T cells in chronically LCMV-infected IRF4-

40 P14 (x10 of P14 (Spleen) - + 4 + 1 reporter mice at day 15 p.i. 20 20 10 2 hi lo

% of Max (G) Recall capacity of IRF4 or IRF4 gp33-speciﬁc 0 2 3 4 5 010 10 10 10 % TCF1

Fold reduction ( T cells ﬂow cytometry-sorted from IRF4 reporter IL-7R 0 0 Total # TCF1 Total # TCF1 0 0 +/- +/- +/- mice at day 13 after chronic LCMV infection and WT WT WT TCF1- Irf4 Irf4 Irf4 TCF1+ transferred into congenically marked recipient mice infected with acute LCMV (WE) one day earlier. * * F CD8+, Chronic, d15 G + 0.8 10 Proportions (left) and numbers (right) of donor cells 30.4 37.5 22.7 45.8 8 at day 7 after transfer.

4 0.6 , Spleen) 10 5 Data in (A)–(C) are representative of two to four in- 6 of total CD8 3 (x10 10 + 0.4 dependent experiments containing three to six mice + 4 per group. Data in (D), (E), and (G) are representative 2 10 0.2 0 2 of two experiments containing three mice each.

IRF4/GFP 23.3 8.2 23.6 8.4

% donor gp33 0.0 0 Error bars denote mean ± SEM and statistical 3 4 5 # donor gp33 0 10 10 10 PD-1 TIM-3 analysis was performed using unpaired two-tailed IRF4 highIRF4 low IRF4 highIRF4 low Student’s t test (*p < 0.05, **p < 0.01, ***p < 0.001). made use of a newly developed reporter mouse that expresses like T cells that are required for the replenishment of effector green fluorescent protein (GFP) from the Irf4 locus (Figure S6C). T cells in situations of persistent antigen stimulation and mediate Chronically LCMV Docile-infected IRF4-reporter mice expressed the reinvigoration of T cells in response to checkpoint blockade. high amounts of GFP in CD8+ T cells positive for PD-1 and TIM-3 IRF4 is a TCR signaling-sensitive transcription factor, which (Figure 7F). However, a small proportion (15%–20%) of plays essential roles in acute CD8+ T cell responses by linking PD-1+CD8+ T cells expressed low amounts of IRF4/GFP (Fig- anabolic metabolism with clonal population expansion and ure 7F). Notably, IRF4/GFPlo gp33+CD8+ T cells sorted from effector T cell differentiation (Man and Kallies, 2015; Man et al., Docile-infected IRF4-reporter mice and adoptively transferred 2013). We show here that during chronic infection, IRF4 played into acutely LCMV WE-infected host mice displayed greater a more complex role. While IRF4 was required for the initial proliferative potential compared to their IRF4/GFPhi counterparts expansion and subsequent maintenance of antigen-specific (Figure 7G). Overall, these data show that high amounts of T cells during chronic infection, high amounts of IRF4 were central IRF4 limit the memory potential of chronically stimulated anti- to establishing cardinal features of T cell exhaustion, including gen-specific T cells by directly repressing Tcf7. high expression of inhibitory receptors such as PD-1, TIM-3, and Lag3, altered cellular metabolism, and impaired cytokine DISCUSSION secretion. Reducing the dosage of IRF4 resulted in decreased expression of inhibitory receptors, led to an overall increase in The transcriptional program underlying CD8+ Tcellexhaustionin anabolic metabolism, and resulted in partially restored secretion response to chronic infection or tumor growth is incompletely of IFN-g. Expression of high amounts of IRF4 was a physiological known. Here we show that a network of TCR-dependent feature of exhausted T cells that was required to limit immune transcription factors consisting of IRF4, BATF, and NFATc1 was pathology. Thus, IRF4 expression needs to be finely balanced central for establishing T cell exhaustion during chronic infection. to maintain the proliferative capacity of antigen-specific T cells We demonstrate that IRF4 contributed to multiple aspects of T cell on the one side and limit their effector function during conditions exhaustion, including inhibitory receptor expression, impaired of chronic antigen exposure on the other side. cytokine secretion, and the repression of anabolic metabolism. Further analyses showed that IRF4 acted together with AP-1 High amounts of IRF4 also limited the development of memory- factor BATF and NFAT, two transcription factors previously

1138 Immunity 47, 1129–1141, December 19, 2017 implicated in promoting exhaustion and Pdcd1 transcription required to determine how modulation of TCR-responsive (Martinez et al., 2015; Oestreich et al., 2008; Quigley et al., transcription factors, inhibitory receptor signaling, and cellular 2010). Together with IRF4 and BATF, the short isoform of metabolism can be combined to reverse immune dysfunction NFATc1, known as NFATc1/aA, was upregulated in exhausted during chronic infection or cancer. T cells, suggesting that NFATc1 and not NFATc2 is the dominant NFAT family member involved in exhaustion. This finding is in line STAR+METHODS with recent reports (Philip et al., 2017; Tirosh et al., 2016) that found NFATc1 was highly expressed in tumor-infiltrating T cells Detailed methods are provided in the online version of this paper and played a critical role in driving cytotoxic T cell differentiation and include the following: in acute infection (Klein-Hessling et al., 2017). NFATc1 expression was also found to promote IRF4 expression in follicular d KEY RESOURCES TABLE helper T cells (Vaeth et al., 2016), while in vitro experiments d CONTACT FOR REAGENT AND RESOURCE SHARING had shown that NFATc2 can bind to the Irf4 locus (Martinez d EXPERIMENTAL MODEL AND SUBJECT DETAILS et al., 2015). These data are consistent with our results, which B Mice show that both NFATc1 and NFATc2 drive IRF4 expression in B IRF4 reporter mice CD8+ T cells during acute and chronic infection. Notably, both d METHOD DETAILS IRF4 and BATF also contributed to NFATc1 expression. IRF4 B Infections, adoptive cell transfer and viral titer determi- together with BATF bound to an intragenic region of Nfatc1 nation and was required for the efficient expression of NFATc1/aA. B Histology Thus, together with the observation that NFATc1/aA expression B Tetramer, surface and intracellular staining and flow is driven by its own promoter and amplified in an autoregulatory cytometry loop (Serfling et al., 2012), our data indicate that IRF4, BATF, and B Retroviral transduction of IRF4 NFAT participate in a self-reinforcing transcriptional circuit that B Metabolic Assays establishes the transcriptional network promoting T cell exhaus- B Real Time PCR analysis tion during chronic infection. B Chromatin Immunoprecipitation (ChIP) and ChIPseq Collaborative activity and co-binding of NFAT and IRF4 has analysis been reported previously for individual target genes (Farrow B Transcriptome analysis et al., 2011; Hu et al., 2002; Lee et al., 2009; Rengarajan et al., B Gene Ontology Pathway Analysis and Gene Set Tests 2002). Similarly, AP-1 binding sites were found to be enriched B Immunoblotting in an NFAT ChIP-seq analysis (Martinez et al., 2015). We found d QUANTIFICATION AND STATISTICAL ANALYSIS that in CD8+ T cells, IRF4, BATF, and NFAT were recruited to d DATA AND SOFTWARE AVAILABILITY adjacent binding sites, and binding of all three factors was significantly enriched among the core group of genes related to SUPPLEMENTAL INFORMATION exhaustion, including Pdcd1, Lag3, Havcr2, Tigit, and Ctla4. Supplemental Information includes six figures and five tables and can be found Thus, in addition to AP1-IRF4 composite elements (AICE) that with this article online at https://doi.org/10.1016/j.immuni.2017.11.021. play important roles in T cell biology, our data indicate that NFAT_AP-1_IRF4 composite elements (NAICE) are particularly AUTHOR CONTRIBUTIONS enriched among the genes regulating CD8+ T cell exhaustion and may indeed be common among genes that play critical roles K.M. designed the study, performed experiments, analyzed and interpreted in T cell biology. Indeed, our data provide evidence that the IRF4, data, and wrote the manuscript. S.S.G. contributed to the design of the study, BATF, and NFAT transcription factor triplet directly bound and performed experiments, analyzed and interpreted data, and wrote the manu- repressed the expression of Tcf7, encoding TCF1, which plays script. Y.L. and W.S. performed all bioinformatics analysis. R.G., S.P., and D.C.H. performed experiments. M.P., D.Z., F.B.-S., and M.A.F. contributed a critical role in the establishment of long-term memory T cells to the study design and interpretation of results. A.K. designed the study, (Jeannet et al., 2010; Zhou et al., 2010). TCF1 is tightly downre- analyzed and interpreted data, and wrote the manuscript. gulated in terminally differentiated and exhausted T cells; however, its expression is preserved in memory-like T cells dur- ACKNOWLEDGMENTS ing chronic infection and is required to maintain their proliferative potential (He et al., 2016; Im et al., 2016; Leong et al., 2016; We wish to thank Liana Mackiewicz, Carolina Alvarado, and Benjamin Lunz for Utzschneider et al., 2016). Importantly, downregulation of IRF4 expert technical support, and Stephen Nutt, Edgar Serfling, and Stefan Klein- Hessling for scientific discussion. We thank Anjana Rao, Laurie Glimcher, Pa- was sufficient to release Tcf7 from repression, allowing the mela Ohashi, Tak Mak, Ulf Klein, Mark Chong, and Daniel Gray for mice. This development of memory-like T cells despite persistent viremia work was supported by grants and fellowships from the National Health and during chronic infection. Medical Research Council of Australia (project grants 1032850 and 1085151 In summary, our findings reveal that the TCR-responsive tran- to A.K., 1006592, 1045549, and 1065626 to M.P., 1023454 to W.S., Senior scription factors IRF4, BATF, and NFAT converge to establish Research Fellowship 1021168 and Senior Principal Research Fellowship the molecular characteristics of exhaustion, including the 1116936 to M.A.F.), the Walter and Eliza Hall Institute Centenary Fellowship funded by CSL (to W.S.), the Swiss National Science Foundation and the expression of inhibitory receptors. Furthermore, they act to Novartis Foundation for Medical-Biological Research (fellowships to S.S.G.), restrain anabolic and mitochondrial metabolism, limit continued and the Sylvia and Charles Viertel Foundation (Senior Medical Research effector function, and oppose the development of memory-like Fellowships to A.K. and M.P.). Further funding was provided by the Fritz cells during chronic T cell stimulation. Future studies will be Thyssen Stiftung (10.13.2.215) and the Wilhelm Sander-Stiftung (2012.047.1)

Immunity 47, 1129–1141, December 19, 2017 1139 to F.B.-S. This study was made possible through Victorian State Government Hu, C.M., Jang, S.Y., Fanzo, J.C., and Pernis, A.B. (2002). Modulation of T cell Operational Infrastructure Support and Australian Government NHMRC Inde- cytokine production by interferon regulatory factor-4. J. Biol. Chem. 277, pendent Research Institute Infrastructure Support scheme. 49238–49246. Huang, W., Sherman, B.T., and Lempicki, R.A. (2009). Systematic and integra- Received: August 19, 2016 tive analysis of large gene lists using DAVID bioinformatics resources. Revised: August 8, 2017 Nat. Protoc. 4, 44–57. Accepted: November 28, 2017 Im, S.J., Hashimoto, M., Gerner, M.Y., Lee, J., Kissick, H.T., Burger, M.C., Published: December 12, 2017 Shan, Q., Hale, J.S., Lee, J., Nasti, T.H., et al. (2016). Defining CD8+ T cells that provide the proliferative burst after PD-1 therapy. Nature 537, 417–421. REFERENCES Jeannet, G., Boudousquie´ , C., Gardiol, N., Kang, J., Huelsken, J., and Held, W. (2010). Essential role of the Wnt pathway effector Tcf-1 for the establishment of Bengsch, B., Johnson, A.L., Kurachi, M., Odorizzi, P.M., Pauken, K.E., functional CD8 T cell memory. Proc. Natl. Acad. Sci. USA 107, 9777–9782. Attanasio, J., Stelekati, E., McLane, L.M., Paley, M.A., Delgoffe, G.M., and Kahan, S.M., Wherry, E.J., and Zajac, A.J. (2015). T cell exhaustion during Wherry, E.J. (2016). Bioenergetic insufficiencies due to metabolic alterations persistent viral infections. Virology 479-480, 180–193. regulated by the inhibitory receptor PD-1 are an early driver of CD8(+) T cell Kao, C., Oestreich, K.J., Paley, M.A., Crawford, A., Angelosanto, J.M., Ali, exhaustion. Immunity 45, 358–373. M.A., Intlekofer, A.M., Boss, J.M., Reiner, S.L., Weinmann, A.S., and Blackburn, S.D., Shin, H., Haining, W.N., Zou, T., Workman, C.J., Polley, A., Wherry, E.J. (2011). Transcription factor T-bet represses expression of the Betts, M.R., Freeman, G.J., Vignali, D.A., and Wherry, E.J. (2009). inhibitory receptor PD-1 and sustains virus-specific CD8+ T cell responses Coregulation of CD8+ T cell exhaustion by multiple inhibitory receptors during during chronic infection. Nat. Immunol. 12, 663–671. chronic viral infection. Nat. Immunol. 10, 29–37. Keir, M.E., Butte, M.J., Freeman, G.J., and Sharpe, A.H. (2008). PD-1 and its Chang, C.H., Qiu, J., O’Sullivan, D., Buck, M.D., Noguchi, T., Curtis, J.D., ligands in tolerance and immunity. Annu. Rev. Immunol. 26, 677–704. Chen, Q., Gindin, M., Gubin, M.M., van der Windt, G.J., et al. (2015). Klein, U., Casola, S., Cattoretti, G., Shen, Q., Lia, M., Mo, T., Ludwig, T., Metabolic competition in the tumor microenvironment is a driver of cancer pro- Rajewsky, K., and Dalla-Favera, R. (2006). Transcription factor IRF4 controls gression. Cell 162, 1229–1241. plasma cell differentiation and class-switch recombination. Nat. Immunol. 7, Cui, G., Staron, M.M., Gray, S.M., Ho, P.C., Amezquita, R.A., Wu, J., and 773–782. Kaech, S.M. (2015). IL-7-induced glycerol transport and TAG synthesis pro- Klein-Hessling, S., Muhammad, K., Klein, M., Pusch, T., Rudolf, R., Flo¨ ter, J., motes memory CD8+ T cell longevity. Cell 161, 750–761. Qureischi, M., Beilhack, A., Vaeth, M., Kummerow, C., et al. (2017). NFATc1 Doedens, A.L., Phan, A.T., Stradner, M.H., Fujimoto, J.K., Nguyen, J.V., Yang, controls the cytotoxicity of CD8+ T cells. Nat. Commun. 8, 511. E., Johnson, R.S., and Goldrath, A.W. (2013). Hypoxia-inducible factors Kurachi, M., Barnitz, R.A., Yosef, N., Odorizzi, P.M., DiIorio, M.A., Lemieux, enhance the effector responses of CD8(+) T cells to persistent antigen. Nat. M.E., Yates, K., Godec, J., Klatt, M.G., Regev, A., et al. (2014). The transcrip- Immunol. 14, 1173–1182. tion factor BATF operates as an essential differentiation checkpoint in early effector CD8+ T cells. Nat. Immunol. 15, 373–383. Doering, T.A., Crawford, A., Angelosanto, J.M., Paley, M.A., Ziegler, C.G., and Wherry, E.J. (2012). Network analysis reveals centrally connected genes and Law, C.W., Chen, Y., Shi, W., and Smyth, G.K. (2014). voom: Precision weights pathways involved in CD8+ T cell exhaustion versus memory. Immunity 37, unlock linear model analysis tools for RNA-seq read counts. Genome Biol. 1130–1144. 15, R29. Lee, C.G., Kang, K.H., So, J.S., Kwon, H.K., Son, J.S., Song, M.K., Sahoo, A., Farrow, M.A., Kim, E.Y., Wolinsky, S.M., and Sheehy, A.M. (2011). NFAT and Yi, H.J., Hwang, K.C., Matsuyama, T., et al. (2009). A distal cis-regulatory IRF proteins regulate transcription of the anti-HIV gene, APOBEC3G. J. Biol. element, CNS-9, controls NFAT1 and IRF4-mediated IL-10 gene activation Chem. 286, 2567–2577. in T helper cells. Mol. Immunol. 46, 613–621. Glasmacher, E., Agrawal, S., Chang, A.B., Murphy, T.L., Zeng, W., Vander Leong, Y.A., Chen, Y., Ong, H.S., Wu, D., Man, K., Deleage, C., Minnich, M., Lugt, B., Khan, A.A., Ciofani, M., Spooner, C.J., Rutz, S., et al. (2012). A Meckiff, B.J., Wei, Y., Hou, Z., et al. (2016). CXCR5(+) follicular cytotoxic genomic regulatory element that directs assembly and function of immune- T cells control viral infection in B cell follicles. Nat. Immunol. 17, 1187–1196. specific AP-1-IRF complexes. Science 338, 975–980. Li, P., Spolski, R., Liao, W., Wang, L., Murphy, T.L., Murphy, K.M., and Griffon, A., Barbier, Q., Dalino, J., van Helden, J., Spicuglia, S., and Ballester, Leonard, W.J. (2012). BATF-JUN is critical for IRF4-mediated transcription in B. (2015). Integrative analysis of public ChIP-seq experiments reveals a T cells. Nature 490, 543–546. complex multi-cell regulatory landscape. Nucleic Acids Res. 43, e27. Liao, Y., Smyth, G.K., and Shi, W. (2013). The Subread aligner: fast, accurate Gubin, M.M., Zhang, X., Schuster, H., Caron, E., Ward, J.P., Noguchi, T., and scalable read mapping by seed-and-vote. Nucleic Acids Res. 41, e108. Ivanova, Y., Hundal, J., Arthur, C.D., Krebber, W.J., et al. (2014). Checkpoint Liao, Y., Smyth, G.K., and Shi, W. (2014). featureCounts: an efficient general blockade cancer immunotherapy targets tumour-specific mutant antigens. purpose program for assigning sequence reads to genomic features. Nature 515, 577–581. Bioinformatics 30, 923–930. He, R., Hou, S., Liu, C., Zhang, A., Bai, Q., Han, M., Yang, Y., Wei, G., Shen, T., Lu, P., Youngblood, B.A., Austin, J.W., Mohammed, A.U., Butler, R., Ahmed, Yang, X., et al. (2016). Follicular CXCR5- expressing CD8(+) T cells curtail R., and Boss, J.M. (2014). Blimp-1 represses CD8 T cell expression of PD-1 chronic viral infection. Nature 537, 412–428. using a feed-forward transcriptional circuit during acute viral infection. Heinz, S., Benner, C., Spann, N., Bertolino, E., Lin, Y.C., Laslo, P., Cheng, J.X., J. Exp. Med. 211, 515–527. Murre, C., Singh, H., and Glass, C.K. (2010). Simple combinations of lineage- Man, K., and Kallies, A. (2015). Synchronizing transcriptional control of T cell determining transcription factors prime cis-regulatory elements required for metabolism and function. Nat. Rev. Immunol. 15, 574–584. macrophage and B cell identities. Mol. Cell 38, 576–589. Man, K., Miasari, M., Shi, W., Xin, A., Henstridge, D.C., Preston, S., Pellegrini, Ho, P.C., Bihuniak, J.D., Macintyre, A.N., Staron, M., Liu, X., Amezquita, R., M., Belz, G.T., Smyth, G.K., Febbraio, M.A., et al. (2013). The transcription Tsui, Y.C., Cui, G., Micevic, G., Perales, J.C., et al. (2015). factor IRF4 is essential for TCR affinity-mediated metabolic programming Phosphoenolpyruvate is a metabolic checkpoint of anti-tumor T cell and clonal expansion of T cells. Nat. Immunol. 14, 1155–1165. responses. Cell 162, 1217–1228. Martinez, G.J., Pereira, R.M., A¨ ijo¨ , T., Kim, E.Y., Marangoni, F., Pipkin, M.E., Hodge, M.R., Ranger, A.M., Charles de la Brousse, F., Hoey, T., Grusby, M.J., Togher, S., Heissmeyer, V., Zhang, Y.C., Crotty, S., et al. (2015). The transcrip- and Glimcher, L.H. (1996). Hyperproliferation and dysregulation of IL-4 expres- tion factor NFAT promotes exhaustion of activated CD8+ T cells. Immunity 42, sion in NF-ATp-deficient mice. Immunity 4, 397–405. 265–278.

1140 Immunity 47, 1129–1141, December 19, 2017 Mittrucker,€ H.W., Matsuyama, T., Grossman, A., Kundig,€ T.M., Potter, J., Shin, H., Blackburn, S.D., Intlekofer, A.M., Kao, C., Angelosanto, J.M., Reiner, Shahinian, A., Wakeham, A., Patterson, B., Ohashi, P.S., and Mak, T.W. S.L., and Wherry, E.J. (2009). A role for the transcriptional repressor Blimp-1 in (1997). Requirement for the transcription factor LSIRF/IRF4 for mature B and CD8(+) T cell exhaustion during chronic viral infection. Immunity 31, 309–320. T lymphocyte function. Science 275, 540–543. Smyth, G.K. (2004). Linear models and empirical bayes methods for assessing Oestreich, K.J., Yoon, H., Ahmed, R., and Boss, J.M. (2008). NFATc1 regulates differential expression in microarray experiments. Stat. Appl. Genet. Mol. Biol. PD-1 expression upon T cell activation. J. Immunol. 181, 4832–4839. 3,e3. Paley, M.A., Kroy, D.C., Odorizzi, P.M., Johnnidis, J.B., Dolfi, D.V., Barnett, Speiser, D.E., Utzschneider, D.T., Oberle, S.G., Munz,€ C., Romero, P., and B.E., Bikoff, E.K., Robertson, E.J., Lauer, G.M., Reiner, S.L., and Wherry, Zehn, D. (2014). T cell differentiation in chronic infection and cancer: functional E.J. (2012). Progenitor and terminal subsets of CD8+ T cells cooperate to adaptation or exhaustion? Nat. Rev. Immunol. 14, 768–774. contain chronic viral infection. Science 338, 1220–1225. Pearce, E.L., Poffenberger, M.C., Chang, C.H., and Jones, R.G. (2013). Fueling Staron, M.M., Gray, S.M., Marshall, H.D., Parish, I.A., Chen, J.H., Perry, C.J., immunity: insights into metabolism and lymphocyte function. Science 342, Cui, G., Li, M.O., and Kaech, S.M. (2014). The transcription factor FoxO1 1242454. sustains expression of the inhibitory receptor PD-1 and survival of antiviral CD8(+) T cells during chronic infection. Immunity 41, 802–814. Pellegrini, M., Calzascia, T., Toe, J.G., Preston, S.P., Lin, A.E., Elford, A.R., Shahinian, A., Lang, P.A., Lang, K.S., Morre, M., et al. (2011). IL-7 engages Tirosh, I., Izar, B., Prakadan, S.M., Wadsworth, M.H., 2nd, Treacy, D., multiple mechanisms to overcome chronic viral infection and limit organ Trombetta, J.J., Rotem, A., Rodman, C., Lian, C., Murphy, G., et al. (2016). pathology. Cell 144, 601–613. Dissecting the multicellular ecosystem of metastatic melanoma by single- Philip, M., Fairchild, L., Sun, L., Horste, E.L., Camara, S., Shakiba, M., Scott, cell RNA-seq. Science 352, 189–196. A.C., Viale, A., Lauer, P., Merghoub, T., et al. (2017). Chromatin states define Utzschneider, D.T., Charmoy, M., Chennupati, V., Pousse, L., Ferreira, D.P., tumour-specific T cell dysfunction and reprogramming. Nature 545, 452–456. Calderon-Copete, S., Danilo, M., Alfei, F., Hofmann, M., Wieland, D., et al. Pircher, H., Ohashi, P., Miescher, G., Lang, R., Zikopoulos, A., Burki,€ K., Mak, (2016). T cell factor 1-expressing memory-like CD8(+) T cells sustain the T.W., MacDonald, H.R., and Hengartner, H. (1990). T cell receptor (TcR) beta immune response to chronic viral infections. Immunity 45, 415–427. chain transgenic mice: studies on allelic exclusion and on the TcR+ gamma/ Vaeth, M., Schliesser, U., Muller,€ G., Reissig, S., Satoh, K., Tuettenberg, A., delta population. Eur. J. Immunol. 20, 417–424. Jonuleit, H., Waisman, A., Muller,€ M.R., Serfling, E., et al. (2012). Quigley, M., Pereyra, F., Nilsson, B., Porichis, F., Fonseca, C., Eichbaum, Q., Dependence on nuclear factor of activated T-cells (NFAT) levels discriminates Julg, B., Jesneck, J.L., Brosnahan, K., Imam, S., et al. (2010). Transcriptional conventional T cells from Foxp3+ regulatory T cells. Proc. Natl. Acad. Sci. USA analysis of HIV-specific CD8+ T cells shows that PD-1 inhibits T cell function by 109, 16258–16263. upregulating BATF. Nat. Med. 16, 1147–1151. Rengarajan, J., Mowen, K.A., McBride, K.D., Smith, E.D., Singh, H., and Vaeth, M., Eckstein, M., Shaw, P.J., Kozhaya, L., Yang, J., Berberich-Siebelt, Glimcher, L.H. (2002). Interferon regulatory factor 4 (IRF4) interacts with F., Clancy, R., Unutmaz, D., and Feske, S. (2016). Store-operated Ca(2+) entry NFATc2 to modulate interleukin 4 gene expression. J. Exp. Med. 195, in follicular T cells controls humoral immune responses and autoimmunity. 1003–1012. Immunity 44, 1350–1364. Ritchie, M.E., Phipson, B., Wu, D., Hu, Y., Law, C.W., Shi, W., and Smyth, G.K. Virgin, H.W., Wherry, E.J., and Ahmed, R. (2009). Redefining chronic viral (2015). limma powers differential expression analyses for RNA-sequencing infection. Cell 138, 30–50. and microarray studies. Nucleic Acids Res. 43, e47. Wang, R., and Green, D.R. (2012). Metabolic reprogramming and metabolic Sawada, S., Scarborough, J.D., Killeen, N., and Littman, D.R. (1994). A line- dependency in T cells. Immunol. Rev. 249, 14–26. age-specific transcriptional silencer regulates CD4 gene expression during T lymphocyte development. Cell 77, 917–929. Weinberg, S.E., Sena, L.A., and Chandel, N.S. (2015). Mitochondria in the regulation of innate and adaptive immunity. Immunity 42, 406–417. Schraml, B.U., Hildner, K., Ise, W., Lee, W.L., Smith, W.A., Solomon, B., Sahota, G., Sim, J., Mukasa, R., Cemerski, S., et al. (2009). The AP-1 transcrip- Wherry, E.J., and Kurachi, M. (2015). Molecular and cellular insights into T cell tion factor Batf controls T(H)17 differentiation. Nature 460, 405–409. exhaustion. Nat. Rev. Immunol. 15, 486–499. Schurich, A., Pallett, L.J., Jajbhay, D., Wijngaarden, J., Otano, I., Gill, U.S., Wherry, E.J., Ha, S.J., Kaech, S.M., Haining, W.N., Sarkar, S., Kalia, V., Hansi, N., Kennedy, P.T., Nastouli, E., Gilson, R., et al. (2016). Distinct meta- Subramaniam, S., Blattman, J.N., Barber, D.L., and Ahmed, R. (2007). bolic requirements of exhausted and functional virus-specific CD8 T cells in Molecular signature of CD8+ T cell exhaustion during chronic viral infection. the same host. Cell Rep. 16, 1243–1252. Immunity 27, 670–684. Seibler, J., Zevnik, B., Kuter-Luks,€ B., Andreas, S., Kern, H., Hennek, T., Rode, Wu, D., Lim, E., Vaillant, F., Asselin-Labat, M.L., Visvader, J.E., and Smyth, A., Heimann, C., Faust, N., Kauselmann, G., et al. (2003). Rapid generation of G.K. (2010). ROAST: rotation gene set tests for complex microarray experi- inducible mouse mutants. Nucleic Acids Res. 31, e12. ments. Bioinformatics 26, 2176–2182. Sena, L.A., Li, S., Jairaman, A., Prakriya, M., Ezponda, T., Hildeman, D.A., Wang, C.R., Schumacker, P.T., Licht, J.D., Perlman, H., et al. (2013). Yao, S., Buzo, B.F., Pham, D., Jiang, L., Taparowsky, E.J., Kaplan, M.H., and Mitochondria are required for antigen-specific T cell activation through Sun, J. (2013). Interferon regulatory factor 4 sustains CD8(+) T cell expansion reactive oxygen species signaling. Immunity 38, 225–236. and effector differentiation. Immunity 39, 833–845. Serfling, E., Avots, A., Klein-Hessling, S., Rudolf, R., Vaeth, M., and Berberich- Zhou, X., Yu, S., Zhao, D.M., Harty, J.T., Badovinac, V.P., and Xue, H.H. (2010). Siebelt, F. (2012). NFATc1/alphaA: The other face of NFAT factors in lympho- Differentiation and persistence of memory CD8(+) T cells depend on T cell cytes. Cell Commun. Signal. 10,16. factor 1. Immunity 33, 229–240.

Immunity 47, 1129–1141, December 19, 2017 1141