Active STAT5 Regulates T-bet and Eomesodermin Expression in CD8 T Cells and Imprints a T-bet−Dependent Tc1 Program with Repressed IL-6/TGF- β1 This information is current as Signaling of September 25, 2021. Magali Grange, Grégory Verdeil, Fanny Arnoux, Aurélien Griffon, Salvatore Spicuglia, Julien Maurizio, Michel Buferne, Anne-Marie Schmitt-Verhulst and Nathalie

Auphan-Anezin Downloaded from J Immunol published online 4 September 2013 http://www.jimmunol.org/content/early/2013/09/04/jimmun ol.1300319 http://www.jimmunol.org/

Supplementary http://www.jimmunol.org/content/suppl/2013/09/04/jimmunol.130031 Material 9.DC1

Why The JI? Submit online.

• Rapid Reviews! 30 days* from submission to initial decision by guest on September 25, 2021

• No Triage! Every submission reviewed by practicing scientists

• Fast Publication! 4 weeks from acceptance to publication

*average

Subscription Information about subscribing to The Journal of Immunology is online at: http://jimmunol.org/subscription Permissions Submit copyright permission requests at: http://www.aai.org/About/Publications/JI/copyright.html Email Alerts Receive free email-alerts when new articles cite this article. Sign up at: http://jimmunol.org/alerts

The Journal of Immunology is published twice each month by The American Association of Immunologists, Inc., 1451 Rockville Pike, Suite 650, Rockville, MD 20852 Copyright © 2013 by The American Association of Immunologists, Inc. All rights reserved. Print ISSN: 0022-1767 Online ISSN: 1550-6606. Published September 4, 2013, doi:10.4049/jimmunol.1300319 The Journal of Immunology

Active STAT5 Regulates T-bet and Eomesodermin Expression in CD8 T Cells and Imprints a T-bet–Dependent Tc1 Program with Repressed IL-6/TGF-b1 Signaling

Magali Grange,*,†,‡,1 Gre´gory Verdeil,*,†,‡,1 Fanny Arnoux,*,†,‡ Aure´lien Griffon,x Salvatore Spicuglia,x Julien Maurizio,*,†,‡ Michel Buferne,*,†,‡ Anne-Marie Schmitt-Verhulst,*,†,‡ and Nathalie Auphan-Anezin*,†,‡

In adoptive therapy, CD8 T cells expressing active STAT5 (STAT5CA) transcription factors were found to be superior to unma- nipulated counterparts in long-term persistence, capacity to infiltrate autochthonous mouse melanomas, thrive in their microen- vironment, and induce their regression. However, the molecular mechanisms sustaining these properties were undefined. In this study, we report that STAT5CA induced sustained expression of controlling tissue homing, cytolytic granule composition, Downloaded from type 1 CD8 cytotoxic –associated effector molecules granzyme B+, IFN-g+,TNF-a+, and CCL3+, but not IL-2, and transcrip- tion factors T-bet and eomesodermin (Eomes). Chromatin immunoprecipitation sequencing analyses identified the genes possessing regulatory regions to which STAT5 bound in long-term in vivo maintained STAT5CA-expressing CD8 T cells. This analysis identified 34% of the genes differentially expressed between STAT5CA-expressing and nonexpressing effector T cells as direct STAT5CA target genes, including those encoding T-bet, Eomes, and granzyme B. Additionally, genes encoding the IL-6R and TGFbRII subunits were stably repressed, resulting in dampened IL-17–producing CD8 T cell polarization in response to IL-6 and http://www.jimmunol.org/ TGF-b1. The absence of T-bet did not affect STAT5CA-driven accumulation of the T cells in tissue or their granzyme B expression but restored IL-2 secretion and IL-6R and TGFbRII expression and signaling, as illustrated by IL-17 induction. Therefore, concerted STAT5/T-bet/Eomes regulation controls homing, long-term maintenance, recall responses, and resistance to polarization towards IL-17–producing CD8 T cells while maintaining expression of an efficient type 1 CD8 program (granzyme B+, IFN-g+). The Journal of Immunology, 2013, 191: 000–000.

he differentiation of naive CD8 T cells into effector regulators, promoting effector and central T cells, re- T cells (eTCs) and memory T cells is accompanied by spectively (8). by guest on September 25, 2021 T changes in expression of genes associated with their The cytokine receptors sharing the common g-chain mediate migration, survival, self-renewal, and effector functions (1–3). activation of the JAK kinases and participate in activation of the Two subsets of memory T cells, effector and central memory STAT5 TF. IL-7 is involved in the maintenance of naive T cells, T cells, have been characterized (4, 5). These have distinct pro- whereas IL-21, IL-15, and IL-2 have been shown to be important liferative potential, tissue homing, and functions. expression for the development and maintenance of memory stem cells, profiles associated with these distinct functional subsets revealed central memory T cells, and effector memory T cells, respectively the regulation of memory cell fate by several transcription factors (9, 10). The three memory T cell subsets demonstrate decreasing (TFs) (3). Expression of two T-box family members T-bet and multipotency and replicative potential while acquiring increasing eomesodermin (Eomes) has been correlated with differentiation effector functions and the ability to home to nonlymphoid tissues. of short-lived eTCs and long-lived central memory T cells, re- Molecular mechanisms by which STAT5 translates the effects of spectively (6, 7). The antagonistic transcriptional repressors IL-21, IL-15, and IL-2 in defining the replicative potential and Blimp-1 and Bcl6 have also been identified as key transcriptional acquisition of effector functions of memory CD8 T cells remain to

*Centre d’Immunologie de Marseille–Luminy, Aix-Marseille Universite´, Marseille A.-M.S.-V. participated in the design of experiments and writing the paper; and N.A.-A. 13288, France; †INSERM Unite´ Mixte de Recherche 1104, Marseille 13288, France; designed the study, did experiments, analyzed data, and wrote the manuscript. ‡Centre National de la Recherche Scientifique, Unite´ Mixte de Recherche 7280, x The microarray data presented in this article have been submitted to the National Center Marseille 13288, France; and Technological Advances for Genomics and Clinics, for Biotechnology Information Omnibus (http://www.ncbi.nlm.nih. INSERM Unite´ Mixte de Recherche 1090, Marseille 13288, France gov/projects/geo/query/acc.cgi?acc=GSE41819) under accession number GSE41819. 1M.G. and G.V. contributed equally to this work. Address correspondence and reprint requests to Dr. Nathalie Auphan-Anezin, Centre Received for publication February 4, 2013. Accepted for publication July 23, 2013. d’Immunologie de Marseille–Luminy, Parc Scientifique de Luminy, Case 906, 13288 Marseille Cedex 09, France. E-mail address: [email protected] This work was supported by funding from INSERM and the Centre National de la Recherche Scientifique, as well as by grants from the Association pour la Recherche The online version of this article contains supplemental material. sur le (to N.A.-A. and A.-M.S.-V.), the Agence Nationale de la Recherche (to Abbreviations used in this article: ChIP, chromatin immunoprecipitation; ChIP-Seq, G.V.), and the Institut National du Cancer (to A.-M.S.-V.). M.G. was the recipient of chromatin immunoprecipitation sequencing; Eomes, eomesodermin; eTC, effector a doctoral fellowship from Association pour la Recherche sur le Cancer, and G.V. T cell; GSEA, gene set enrichment analysis; GzmB, granzyme B; HPRT, hypoxan- held an Agence Nationale de la Recherche “Retour Postdoctorants” fellowship. thine phosphoribosyltransferase; MFI, mean fluorescence intensity; Tc1, type 1 CD8 M.G. performed all sample preparations for transcriptomic analyses, validated tran- cytotoxic T cell; Tc17, IL-17–producing CD8 T cell; TF, ; TSS, script expression, and performed functional assays with the assistance of F.A. and transcription start site; wt, wild-type. M.B.; G.V. performed the ChIP assays and their validation; S.S. directed the deep Ó sequencing and the related bioinformatics analyses with the help of J.M. and A.G.; Copyright 2013 by The American Association of Immunologists, Inc. 0022-1767/13/$16.00

www.jimmunol.org/cgi/doi/10.4049/jimmunol.1300319 2 STAT5/T-bet IMPRINT Tc1 RESISTANCE TO IL-6/TGF-b be determined. Recent studies reporting antagonistic function in the Centre d’Immunologie de Marseille–Luminy animal facility. Animal of STAT5 and Bcl6 (11) and STAT5-mediated regulation of T- experiments respected French and European directives. bet expression (12) suggest that STAT5 may also be a key TF in- Cell preparation volved in effector and memory T cell fates. 2/2 Adoptive therapies for cancer patients based on the transfer of CD8 T cells were prepared from lymph nodes or spleen of TCRP1A Rag-1 mice according to standard procedures. When prepared from immuno- tumor-specific T cells have shown some success in treatment of competent B10.D2 mice, CD8 T cells were enriched using a mouse CD8 melanoma (13) and lymphoma patients (14). However, those negative selection kit (Dynal/Invitrogen) according to the manufacturer’s approaches are limited by poor survival of the transferred T cells, instructions. For analysis of liver-infiltrating T cells, donor mice were their limited replicative potential and capacity to infiltrate solid anesthetized and perfused with PBS. Livers were dispersed into a single- cell suspension and passed over Ficoll-Paque solution (Amersham Bio- tumors, and their sensitivity to tumor-derived immunosuppression. sciences). Based on the properties of the three memory CD8 T cell subsets described above, several groups have concluded that in vitro–ex- CD8 T cell activation and retroviral infections panded CD8 eTCs are not the best subset for adoptive therapy (15), 27 TCRP1A T cells were activated for 72 h with 10 MP1A35–43 but that less differentiated CD8 T cells might be more appropriate (LPYLGWLVF) peptide. Polyclonal CD8 T cells were activated with (16). Infusions of cytokines such as IL-2 are currently used in patients coated anti-CD3 mAb (145.2C.11, 3 mg/ml) and soluble anti-CD28 mAb to increase survival and cytolytic activity of transferred anti-tumor (37.51, 1 mg/ml). Twenty (for TCR transgenic T cells) or 40 h (for poly- CD8 T cells. However, IL-2 treatments have adverse effects on clonal T cells) after initial stimulation, CD8 T cells were retrovirally transduced as previously described (20). Cultures were then continued for patients, notably through induction of vascular leak syndrome (17). IL- another 48 h. The average transduction efficiency was ∼58% for TCRP1A 2 may also activate suppressive T regulatory lymphocytes. Therefore, eTCs-STA5CA, 31% for TCRP1A eTCs-GFP, and 6% for polyclonal CD8 Downloaded from strategies aimed at mimicking the costimulatory effects of IL-2 while eTCs-STAT5CA or eTCs-GFP. Those eTC populations were either ana- 2/2 bypassing cytokine-mediated untoward effects are of interest. lyzed directly or adoptively transferred in Rag1 syngeneic mice. In some experiments, untransduced eTCs are included as controls instead of Among immunosuppressive factors in the tumor microenvi- eTCs-GFP, as we previously showed that those two T cell subsets had the ronment, IL-10, IL-6, and vascular endothelial growth factor, same behavior upon adoptive transfer into syngeneic mice (20). which activate STAT3, are crucial for both tumor growth and Tumor cells lines immunosuppression or T cell polarization (18), creating a “feed- http://www.jimmunol.org/ forward” increased STAT3 activity both in tumor cells and in P1A+ mastocytoma P815 and its P1A2 variant P1.204 are used as stim- tumor-associated immune cells. In particular, aberrant IL-6/JAK/ ulating cells (20). Fas-mediated cytotoxicity was measured against the STAT3 signaling promotes inhibition of Th1 immune responses leukemia L1210 expressing or not Fas. (19). For these reasons, developing CD8 effector T cells able to In vitro stimulation with immunosuppressive/polarizing counteract the tumor stromal barrier and tumor-derived immuno- cytokines suppression or aberrant polarization is an important challenge. eTCs-STAT5CA or eTCs adoptively transferred in syngeneic Rag-deficient We recently reported that expression of an active form of STAT5 mice were recovered from recipients’ pooled lymph nodes and spleens. (STAT5CA) in CD8 T cells favors their differentiation into long-lived CD8 T cells from B10.D2 mice were included as control. Cells (106) were eTCs, manifesting phenotypic and tissue-homing properties charac- stimulated by coated anti-CD3 mAb (and soluble anti-CD28 mAb for by guest on September 25, 2021 teristic of effector memory T cells while maintaining the high poten- naive CD8 T cells). Cultures were conducted in the absence or presence of tial for Ag recall responses associated with central memory T cells. TGF-b1 (R&D Systems; 5 ng/ml) for 48 h. Supernatants were collected for measurement of IFN-g by ELISA (see below) in the absence or presence of Similar characteristics were acquired whether STAT5CA-express- TGF-b1 (2 ng/ml), IL-6 (5 ng/ml), anti–IL-4 (10 mg/ml), and anti–IFN-g 2/2 ing cells were transferred in immunodeficient (Rag )orimmuno- (10 mg/ml) for 48 h. Supernatants were collected for measurement of IFN- competent hosts (20). Of note, these properties greatly enhanced the g and IL-17A by ELISA (IFN-g by AN18 and biotinylated R46A2 Abs and survival and secondary responses of STAT5CA-expressing eTCs streptavidin-HRP [R&D Systems]; IL-17A by DuoSet [R&D Systems]). when they were transferred into tumor-bearing hosts. As compared Flow cytometry with unmanipulated CD8 eTCs, STAT5CA-expressing eTCs very Abs were from BD Biosciences, except anti–granzyme B (GzmB) mAb efficiently infiltrated autochthonous melanomas, maintained a type 1 (Invitrogen). Cells (106) were analyzed on an LSR2 UV or an LSR2 561 CD8 cytotoxic T cell (Tc1) effector program within the tumor, and cytometer (BD Biosciences). Data were analyzed using FlowJo (Tree Star) induced tumor regression (20). or FACSDiva (BD Biosciences) software. For intracellular cytokine To determine the molecular bases sustaining those functional staining, CD8 T cells were stimulated ex vivo for 4 h in the presence of monensin (4 mM) and permeabilized using a Cytofix/Cytoperm kit (BD properties, in this study we performed the transcriptional analysis of Biosciences). STAT5CA-expressing eTCs. This study reveals a long-lasting increase of both T-bet and Eomes expression. Chromatin immunoprecipitation Transcriptome analyses sequencing (ChIP-Seq) analyses further indicate that T-bet and Naive TCRP1A T cells, untransduced TCRP1A eTCs, and TCRP1A eTCs- Eomes are direct STAT5 target genes. STAT5CA also dampens ex- STAT5CA recovered from spleen and lymph nodes after transfer in con- 2/2 pression of subunits involved in IL-6 and TGF-b1–induced signaling, genic Rag-1 B10.D2 hosts were positively selected using anti-CD8 magnetic beads (Miltenyi Biotec). The same was done for T-bet2/2 conferring resistance to those polarizing/immunosuppressive cyto- eTCs-STAT5CA recovered after transfer in congenic Rag-22/2 B6 hosts. kines. Additionally, we report new insights on the interplay between Cell purity ranged from 95 to 99%. In vitro–activated (72 h) TCRP1A STAT5 and T-bet, delineating the plasticity of effector CD8 T cell eTCs-STAT5CA and eTCs-GFP were FACS-sorted on the basis of GFP programs and their sensitivity to immunosuppression. expression: purity was ∼99%. Two to four independent biological repli- cates have been included in the transcriptomic analysis (see Supplemental Fig. 1). Statistically significant differences in mRNA abundance between Materials and Methods samples were determined using the moderated t test of Bioconductor Mice package linear models for microarray data. We selected probes showing at least a log2FoldChange value .0.5 or ,20.5 with a p value ,0.05. A d Mice heterozygous for the H-2L /P1A35–43–specific TCR transgene Venn diagram (22) showed the distribution of genes significantly differ- (TCRP1A) (21) were kept on the Rag-12/2 B10.D2 background. T-bet2/2 entially expressed between two cell populations. The gene set enrichment mice on a C57BL/6 background (http://jaxmice.jax.org/strain/004648. analysis (GSEA) method from the Massachusetts Institute of Technology html) were initially purchased from The Jackson Laboratory. Rag-12/2 (http://www.broad.mit.edu/gsea) (23) was used to statistically test whether B10.D2 and Rag-22/2B6 mice were also used. All of these mice were bred a set of genes of interest was distributed randomly in the large list of genes The Journal of Immunology 3 sorted on the basis of their expression level for TCRP1A eTCs-STAT5CA Briefly, after 72 h in vitro culture, control untransduced (TCRP1A over TCRP1A eTCs with a p value ,0.05. Both C5 (genes annotated by eTCs) or GFP-tranduced (TCRP1A eTCs-GFP) and STAT5CA- the same terms; Supplemental Table I) and C7 (immunologic transduced (TCRP1A eTCs-STAT5CA) were adoptively trans- signature; see Supplemental Table II) sets of genes of interest were used for 2/2 these analyses. Microarray data have been submitted to National Center for ferred into congenic Rag-1 B10.D2 mice for prolonged periods Biotechnology Information Gene Expression Omnibus database: (http:// to establish a pool of memory T cells. www.ncbi.nlm.nih.gov/projects/geo/query/acc.cgi?acc=GSE41819). Using a whole-genome transcriptomic analysis, in this study we define the distinct gene expression program between untransduced and Quantitative RT-PCR STAT5CA eTCs: those two eTC subsets were analyzed after 3 d of Total RNA was isolated using an RNeasy kit (Qiagen). RNA samples (1 mg) in vitro culture and at later time points (20–68 d) after transfer. were reverse-transcribed using random primers and SuperScript II reverse Compared to naive T cells, adoptively transferred eTCs-STAT5CA transcriptase (Life Technologies, Grand Island, NY). Quantitative real-time and untransduced eTCs showed 1526 and 1001 upregulated genes, PCR was done by nonspecific product detection (SYBR Green; Stratagene) of cDNA. mRNA expression levels were normalized to the housekeeping respectively, with 608 genes being shared between these two hypoxanthine phosphoribosyltransferase (HPRT) gene using the 22DCt method subsets (Fig. 1A). Among this shared group, Tbx21 (encoding T- in which DCt was the difference between the mean Ct value of triplicates of bet) and Eomes were present, although at a higher level in the the sample and of the HPRT control. Values normalized to those of the naive eTCs-STAT5CA (Fig. 1A; genes are marked with an asterisk). CD8 T cell sample are shown.Primer sequences are: IFN-g,forward,59- CAACAGCAAGGCGAAAAAGG-39,reverse,59-GGACCACTCGGATGA- This was also the case for some known T-bet target genes (Gzmb, GCTCA-39; GzmB, forward, 59-TCACAAGGACCAGCTCTGTCCT-39,re- Ifng, Cxcr3, Il2rb) (29, 30) and for Fasl. Additionally, compared verse, 59-GTTGGGTTGTCACAGCATGG-39; T-bet, forward, 59-CAACA- with naive T cells, eTCs-STAT5CA expressed higher levels of

ACCCCTTTGCCAAAG-39,reverse,59-TCCCCCAAGCAGTTGACAGT-39; transcripts encoding effector molecules (Gzma, Gzmk, Gzmm, Downloaded from 9 9 9 Eomes, forward, 5 -CCCCTATGGCTCAAATTCC-3 ,reverse,5-CCAGA- Prf1, Tnf), molecules regulating cytokine signaling (Socs1, Socs2, ACCACTTCCACGAA-39;HPRT,forward,59-AGCCCTCTGTGTGCT- CAAGG-39, reverse, CTGATAAAATCTACAGTCATAGGAATGGA-39. Cish) and costimulatory receptors Tnfrsf18 (GITR) and Tnfrsf9 (4- 1BB). Specific staining corroborated the high expression of pro- ChIP and ChIP-Seq teins GITR and 4-1BB on eTCs-STAT5CA (Fig. 2A). The ob- ChIP of crosslinked chromatin was performed essentially as previously served relative increased expression of a large collection of 6 described with minor modifications (24). Cell extracts from 50 3 10 cells transcripts for killer cell lectin-like receptors on eTCs-STAT5CA http://www.jimmunol.org/ were sonicated 12 times, using the S-4000 Sonifier (Misonix, Farmingdale, was in agreement with a recent whole-genome study illustrating NY) with 30-s pulses, to obtain DNA fragments between 200 and 500 bp in the relationship between NK cells and T cells (31). Expression length. Chromatin extracts were prewashed with 50 ml magnetic beads of some NK cell receptors was confirmed by cell surface staining (Invitrogen) for 1 h at 4˚C in a rotating wheel. Precipitation was then carried out overnight at 4˚C using 10 mg anti-STAT5a Ab (R&D Systems) (not shown), but the functional consequences of this expression on complexed to 50 ml magnetic beads. After reverse crosslinking, enriched CD8 T cells are still unclear (32). DNA fragments were extracted with phenol/chloroform and recovered Comparing adoptively transferred eTCs-STAT5CA and eTCs, using the QIAquick PCR purification kit (Qiagen). For ChIP-Seq, se- GSEA performed with the C5 gene set (Supplemental Table I) quencing of two STAT5a ChIP and input DNA samples was performed on an AB SOLiD v4.0 (Life Technologies) according to the manufacturer’s showed a positive enrichment score for transcripts involved in protocol. Sequence read mapping to version 9 build 37 (mm9) of the cell cycle and DNA replication, consistent with the high potential by guest on September 25, 2021 mouse genome was achieved with BFAST (v0.7.0). Only reads with a for secondary responses of eTCs-STAT5CA (20) (see below) or unique best scoring alignment were retained. To reduce the risk of po- encoding components of intracellular organelles such as micro- tential bias, we discarded all tags with exactly the same coordinates. Reads tubules, cytoskeleton, and secretory vesicles in correlation with were elongated to 204 and 189 bp for replicates 1 and 2, respectively. The average values from the two replicates or for the input sample were the high cytolytic activity of the eTCs-STAT5CA, a characteristic summed up in nonoverlapping 50-bp bins along each and corroborated by a programmed cell death/cell-mediated cytotox- normalized per 10 million reads to generate Wiggle files, using an R icity gene cluster. Interestingly, when GSEAs were done with the custom script. Wiggle files were visualized using the Integrated Gene recently released C7 “immunological signature” gene set (Sup- Browser tool (http://www.affymetrix.com). Significant STAT5-bound regions were retrieved using the HOMER peak-calling tool (v4.1; http:// plemental Table II), a positive enrichment score was found for biowhat.ucsd.edu/homer/) with default settings (false discovery rate, transcripts involved in T effector versus memory CD8 T cells, 0.001; local and input fold enrichment, 4.0). Only peaks found in both naive versus T effector CD8 T cells, and unstimulated versus IL-2 replicates were used for further analyses. Genes associated with Stat5a (or IL-15)–stimulated NK cells. peaks were found using the GREAT Web server tool (http://bejerano. In both microarray and cytometric analyses, eTCs-STAT5CA stanford.edu/great/public/html/) (25). Research of STAT5a core matrices hi (MA00457, MA00460, MA00493, and MA00499 from the TRANSFAC demonstrated a GzmB expression (Fig. 2B–D) and exhibited database) (26) within the 250 bp surrounding each STAT5 peak was a higher cytolytic activity than their control counterparts (20) (Fig. conducted using the regulatory sequence analysis tools matrix scan tool 2E). FasL was not constitutively expressed on eTCs- (27). ChIP-Seq data have been submitted to the National Center for Bio- STAT5CA but could rapidly be induced upon Ag triggering technology Information Gene Expression Omnibus (http://www.ncbi.nlm. nih.gov/projects/geo/query/acc.cgi?acc=GSE41819). (Fig. 2F), due perhaps to the high level of FasL transcript (Fig. 1A), For ChIP, PCR analyses were performed using primers described in contributing to Fas-dependent cytotoxic activity when encounter- Supplemental Table III. ing its target (Fig. 2E). Whether the high level of transcripts encoding IFN-g (Fig. 1A) influenced the kinetics of the secondary response by eTCs-STAT5CAwas evaluated through a kinetic study Results of in vitro restimulation in response to peptide-loaded APCs (Fig. STAT5CA expression in in vivo–transferred CD8 T cells 2G). Although no constitutive cytokine production was detected in induces a sustained Tc1 gene expression program those cells, a larger fraction of eTCs-STAT5CA than control eTCs We derived long-lived eTCs by retroviral expression of a mutated efficiently secreted IFN-g after activation (Fig. 2H). Additionally, STAT5a1*6 (28), referred to as STAT5CA hereafter, in in vitro– eTCs-STAT5CAwere also found to secrete CCL3 and TNF-a but not activated CD8 T cells expressing a transgenic TCR specific for the IL-2 (Fig. 2H). Altogether, STAT5CA regulated a gene expression P1A encoded cancer germline Ag (TCRP1A), as described (20). program that induced very responsive Tc1. 4 STAT5/T-bet IMPRINT Tc1 RESISTANCE TO IL-6/TGF-b Downloaded from http://www.jimmunol.org/ by guest on September 25, 2021

FIGURE 1. Long-term expression of STAT5CA sustains a large gene expression program in CD8 T cells. (A) Venn diagram showing the distribution of genes with significant changes (logFC value .0.5 or ,20.5; p value ,0.05): TCRP1A eTCs-STAT5CA/naive TCRP1A T cells (blue circle) and untransduced TCRP1A eTCs/naive TCRP1A T cells (yellow circle). For each subset the most significant genes (highest ratio) are reported as well as their fold increase in parentheses. For some of the 608 common genes, numbers in parentheses are relative to the fold increase in (Figure legend continues) The Journal of Immunology 5

STAT5CA expression induces either sustained gene regulation Expression of STAT5CA in CD8 eTCs modulates expression of or time-dependent accumulation of transcripts, leading to several cytokine receptors concomitant expression of T-bet and Eomes Functional programs of T cells in vivo are influenced by envi- Among the 100 genes displaying the highest positive enrichment ronmental cues, particularly those delivered by cytokines. As between TCRP1A eTCs-STAT5CA and TCRP1A eTCs after compared with naive CD8 T cells, untransduced TCRP1A eTCs adoptive transfer, we selected those relevant to T cell functions and and eTCs-STAT5CA analyzed 20 d after their adoptive transfer activation as well as those included in GSEA clusters (cell cycle, demonstrated increased il2rb transcript and protein (CD122) DNA replication, microtubules/cytoskeleton, and JAK/STAT sig- expression (Fig. 1A) (20), consistent with their Ag-experienced naling) shown in Supplemental Table I. Two groups of transcripts status. Additionally, eTCs-STAT5CA showed upregulation of il2ra with distinct expression kinetics were identified. In the type 1 transcript and protein (CD25) and downmodulation of il7r tran- group (Fig. 1B), transcripts were already upregulated at 72 h after script and protein (CD127) (Fig. 1A) (20). The transcriptomic activation in both eTCs and eTCs-STAT5CA and their expression analyses of TCRP1A eTCs-STAT5CA also pointed to a strong was either sustained at the same level or slightly increased downmodulation (Fig. 1A) of both il6ra and il6st transcripts, 20 d later in in vivo–transferred eTCs-STAT5CA, while being encoding respectively the IL-6Ra and the gp130 signaling sub- respectively reduced or maintained at the same level in control unit, as well as transcripts for TGFbRII. Staining of eTCs- eTCs. This cluster included transcripts involved in DNA replica- STAT5CA and control eTCs or naive CD8 T cells confirmed the tion, cell cycle, microtubules/cytoskeleton (dotted lines), as well reduced expression of IL-6Ra and gp130 (Fig. 3A) surface pro- as Gzmb, il2ra, Tnfrsf9, Tnfrsf18, and STAT5a. Tbx21 (T-bet) and teins on STAT5CA-expressing T cells. As compared with naive its target genes il2rb and Cxcr3 were not included in this figure CD8 T cells, TGFbRII was also expressed at a lower level at the Downloaded from because they were not ranked within the 100 genes with highest surface of eTCs-STAT5CA (Fig. 3B). enrichment. However, their expression profile was similar to that of the type 1 gene cluster (see Fig. 2B, 2C for T-bet) that also Identification of direct STAT5CA target genes included the T-bet target genes gzmb and il12rb2 (Fig. 1B). This To identify regulatory regions occupied by STAT5 TF in ex vivo– pattern of gene expression is consistent with a role for STAT5 in purified memory TCRP1A eTCs-STAT5CA, we performed ChIP the stabilization of a Tc1 gene expression program initiated by using an anti-STAT5a Ab followed by deep sequencing (ChIP- http://www.jimmunol.org/ TCR stimulation (33, 34). Seq). Two independent preparations of TCRP1A STAT5CA- In the type 2 group (Fig. 1C), transcripts were strongly upreg- eTCs were directly (ex vivo) fixed for ChIP-Seq assays without ulated from 72 h to day 20 after activation in eTCs-STAT5CA, any cytokine treatment. Results from these two independent ex- although this was not (or to a much lower extent) observed in con- periments were merged to define common identified peaks, which trol eTCs. This latter gene cluster included Eomes, Ifng, Tnf, and included 7320 significantly bound DNA regions associated with killer cell lectin-like receptors. We confirmed by qRT-PCR the re- 6969 different genes in the TCRP1A eTCs-STAT5CA (http://www. spective type 1 and type 2 profiles for expression of T-bet and ncbi.nlm.nih.gov/projects/geo/query/acc.cgi?acc=GSE41819). Eomes in eTCs-STAT5CA (Fig. 2B, 2C). Most bound regions were found at .5 kb from the transcription by guest on September 25, 2021 T-bet and Eomes, two homologous T-box TFs expressed in start sites (TSS) (Fig. 4A), suggesting preferential binding to activated CD8 T cells (35), jointly regulate expression of cytotoxic putative distal enhancer regions. The distribution of STAT5a- effector molecules (29). Moreover, T-bet also plays a role in Ag- bound regions was similar to the one previously reported for stimulated IFN-g production by CD8 T cells (36, 37). Data re- 3 d in vitro–preactivated T cells restimulated with IL-2 (12, 40). ported in this study emphasize a STAT5CA-mediated increase in However, as compared with recent cytokine-driven STAT5a acti- immediate effector functions (GzmBhi), a characteristic that has vation, long-lasting expression of STAT5CA might induce its par- been correlated with high T-bet expression, as well as potential for ticipation in transcriptional complexes of a different nature. secondary responses (20), a feature assigned to Eomes expression We compared the 6969 STAT5a-associated genes identified by (38). Thus, long-term expression of STAT5CA in CD8 eTCs led to ChIP-Seq with the 3039 transcripts that were significantly different a stable expression of the T-bet gene and to an accumulation of between TCRP1A STAT5CA-eTCs and TCRP1A eTCs in the Eomes transcripts as well as of the corresponding (Figs. transcriptomic analyses: 1043 (34%) of those transcripts were found 1A, 2B–D). to be STAT5a target genes (Fig. 4B). Thus, a substantial proportion Prdm1 transcripts (encoding Blimp-1) were overexpressed in both of the genes whose expression is modulated in TCRP1A eTCs- eTCs-STAT5CA and untransduced eTCs as compared with naive STAT5CA are likely to be directly controlled by STAT5a. T cells (Fig. 1A). Alternatively, Bcl6 was found expressed only in STAT5 is known to bind to TTCNN(G/A)GAA (where “N” untransduced eTCs, but not in eTCs-STAT5CA (Fig. 1A). This result indicates any nucleotide) IFN-g–activated site motifs. We found is in agreement with the reported repression of Bcl6 expression by that 88.8% of STAT5a-bound regions contained at least one STAT5 (11) and repression by Bcl6 of GzmB expression (39). STAT5 motif. A focused analysis of STAT5-bound regions found Altogether, the particular behavior of eTCs-STAT5CA did not by ChIP-Seq was conducted on the upregulated Tbx21 and Eomes seem to rely on distinct regulation of Blimp-1. Rather, tran- genes and downregulated Tgfbr2 and il6st genes (Fig. 4C, Sup- scriptomic analyses highlighted the combined regulation of T-bet plemental Table III). These analyses showed three bound regions and Eomes in eTCs-STAT5CA that may act in concert with around the Tbx21 (T-bet) gene, at positions 212.3, +35.2, and STAT5CA for large-scale and sustained gene regulation involved in +39.4 kb. The upstream site was recently reported to bind STAT5a long-term survival, cytotoxicity, and secondary responsiveness. in CD4 T cells incubated with IL-2, or STAT4 in CD4 T cells

eTCs-STAT5CA (first) and untransduced eTCs (second), with both being compared with naive T cells. Genes marked with an asterisk are also expressed at a significantly different level between eTCs-STAT5CA and untransduced eTCs. (B and C) For genes expressed at significantly different levels between TCRP1A eTCs-STAT5CA and untransduced eTCs, ratios versus naive T cells are reported for eTCs-STAT5CA (left panels) and untransduced eTCs (right panels) activated for 72 h in vitro or recovered 20 d after their adoptive transfer. Depending on their expression profiles in eTCs-STAT5CA, two gene subsets (shown in (B) for type 1 and in (C) for type 2 gene cluster) were identified. 6 STAT5/T-bet IMPRINT Tc1 RESISTANCE TO IL-6/TGF-b Downloaded from http://www.jimmunol.org/ by guest on September 25, 2021 FIGURE 2. (A) Ex vivo TCRP1A eTCs-STAT5CA show increased expression of activation markers. Rag-12/2 B10.D2 mice were injected with 2 3 105 TCRP1A eTCs-STAT5CA 23 d earlier. Splenocytes were analyzed for expression of the reported activation markers on CD8 GFP+ TCRP1A eTCs- STAT5CA. Naive TCRP1A T cells are included as controls. (B–D) Long-term expression of STAT5CA induces upregulation of T-bet and Eomes TFs. (B and C) Quantitative RT-PCR on RNA samples isolated from the following T cells: purified naive CD8 TC from B10.D2 mice; (B) 72 h in vitro–activated untransduced (control) or STAT5CA-transduced TCRP1A (sorted GFP+ cells); (C) purified TCRP1A eTC-STAT5CA from Rag-12/2 B10.D2 mice injected 20 d earlier. Values (2-DCt: see Materials and Methods) normalized to those of the naive CD8 T cell sample are shown. Results are representative of two or three experiments done in triplicate. Statistical analyses were performed with a Mann–Whitney test using GraphPad and a two-tailed p.*p , 0.05, **p , 0.01, ***p , 0.001. (D) TCRP1A eTCs and TCRP1A eTCs-STAT5CA from Rag-12/2 B10.D2 mice injected 20 d earlier were stained for GzmB, Eomes, and T-bet. On gated CD8+ T cells, their respective expression (with the corresponding MFI) is shown overlayed onto an isotype control (gray). Results are representative of three experiments. (E and F) TCRP1A eTCs-STAT5CA can use both secretory and death pathways for target killing. (E) Splenocytes from Rag-12/2 B10.D2 mice injected with 107 untransduced TCRP1A eTCs (day 53) or 2 3 105 TCRP1A eTCs-STAT5CA (day 46) were used in a cytolytic assay. CD8 T cells were cocultured during 5 h with targets negative (L1210; not shown; lysis never .8%) or positive (L1210-Fas) for Fas surface expression. For the latter, three different conditions were tested: medium, to detect constitutive Fas-mediated killing; peptide P1A, to reinduce FasL d expression, which also triggers GzmB/perforin-mediated lysis because the targets express H-2L ; peptide P1A plus EGTA plus MgCl2, to reinduce FasL expression while inhibiting GzmB/perforin lysis. (F) Naive TCRP1A T cells and TCRP1A eTCs in vitro activated for 72 h and splenocytes of Rag-12/2 B10.D2 injected 46 d earlier with TCRP1A eTCs-STAT5CAwere stimulated for 4 h with either ionomycin/PMA or tumor cells expressing (P1A+, P511) or not (P1A2, P1.204) the P1A Ag and then stained for FasL. The fold increase between FasL surface expression on activated cells relative to cells cultured in medium is reported. (G and H) TCRP1A eTCs-STAT5CA manifest rapid and potent IFN-g secretion upon secondary stimulation; those cells produce IFN-g, TNF-a, and CCL3 but not IL-2. TCRP1A eTCs-STAT5CA and TCRP1A eTCs were adoptively transferred into Rag-12/2 B10.D2 mice. Thirty days later CD8 T cells recovered from pooled lymph nodes and spleens were stimulated (G) for different periods of time with P1A peptide–pulsed APCs and stained for IFN-g (percentage IFN-g+ among gated CD8+ T cells is represented as a function of time; means 6 SEM are shown from three independent experiments; statistical analyses were done with an unpaired t test [GraphPad]; **p , 0.01 by a two-tailed p test); or (H) for 4 h with anti-CD3 bound on a FcRg+ (P1.204) tumor. Percentage cytokine+ cells among gated CD8+ T cells are shown. Data are representative of at least three independent experi- ments. incubated with IL-12 (12), in line with the observation that IL-2/ three regions upstream of il6st were identified in the STAT5a ChIP STAT5 and IL-12/STAT4 may have overlapping actions in CD4 at positions 227, 215.5, and 23.3 kb. No significant STAT5a Th1 differentiation, with both contributing to T-bet expression. We binding was found on the il6ra gene (data not shown). Finally, we also identified two enriched STAT5a-bound regions around the identified two major peaks located at 253 and 274 kb upstream Eomes gene at positions 23.3 and +14.6 kb. For the repressed of the Gzmb TSS. A minor peak (GzmB no. 3) at position 240 bp genes, two peaks were found upstream (23.3 and 23.0 kb) of was found, in agreement with our previous study using classical Tgfbr2 and two in its intronic region (+55 and +66 kb). Likewise, ChIP on in vitro–activated CD8 eTCs (34). We also performed The Journal of Immunology 7

FIGURE 3. Long-term expression of STAT5CA in CD8 eTCs induces downregulation of IL-6R and TGFbRII subunit expression. (A) TCRP1A eTCs- STAT5CA or eTCs were adoptively transferred into Rag-12/2 B10.D2 mice. Thirty days later CD8 T cells recovered from pooled lymph nodes and spleens were stained for gp130 and IL-6Ra surface expressions. B10.D2 naive (CD442) CD8 T cells were included as a control. An overlay with control staining (in gray) is shown for each T cell subset. Fold increase in MFI with respect to control staining is indicated. (B) TCRP1A eTCs-STAT5CA adoptively transferred into Rag-12/2 B10.D2 mice [as in (A)] were stained for TGFbRII surface ex- pression. Naive CD8 T cells are also included in the analysis. The fold increase in MFI with respect to control staining is reported in histograms. Statisti-

cal analyses were done with an unpaired t test Downloaded from (GraphPad) with a two-tailed p.*p , 0.05.

classical ChIP assays for all of the identified peaks using three the corresponding protein. Altogether, this analysis showed that independent TCRP1A STAT5CA-eTCs preparations (Fig. 4D) and STAT5CA expression in CD8 eTCs contributes in a manner in- http://www.jimmunol.org/ thus validated the ChIP-Seq data. dependent of T-bet to their increased accumulation in peripheral Altogether, a direct regulation of both T-bet and Eomes might be tissues and to sustained expression for a cluster of genes. The at the basis of the particular phenotype and functional properties of question may be raised as to whether for some of these increased the eTCs-STAT5CA maintained in vivo, whereas direct regulation transcripts Eomes expression compensates for lack of T-bet. In- of il6st and Tgfbr2 genes may contribute to the dampening of their deed, T-bet2/2 CD8 T cells exposed to lymphocytic choriomen- sensitivity to IL-6 and TGF-b1. ingitis virus in vivo had reduced expression of Cxcr3 and GzmB as compared with wt CD8 T cells, but CD8 T cells lacking both T-bet T-bet deficiency did not compromise STAT5CA-induced and Eomes expression failed to develop into Cxcr3+GzmB+ ef- differentiation of long-lived eTCs and their accumulation in by guest on September 25, 2021 fector cells (29). tissues We next evaluated whether the promotion of long-lived CD8 eTCs in T-bet deficiency in STAT5CA-expressing eTCs leads to tissues was dependent on T-bet. For this purpose, polyclonal CD8 T- a diversification of effector functions bet–deficient T cells were recovered from B6 T-bet2/2 mice, acti- This transcriptomic analysis also pointed toward a subset of genes vated by anti-CD3/CD28, and transduced or not with STAT5CA. that is regulated by STAT5CA in a strictly T-bet–dependent manner. After adoptive transfer in Rag-22/2 B6 hosts, the number of recov- This subset included some of the known T-bet target genes: Klrg1, ered T cells from livers was higher for eTCs-STAT5CA compared Prf1, and other upregulated (Gzmm) and downregulated transcripts with eTCs for both wild-type (wt) and T-bet–deficient cells, whereas (Il7r, IL6ra) (Fig. 6A). Importantly, il2 transcripts appeared to be differences in recovery were not significant in the spleens (Fig. 5A). upregulated in T-bet2/2 eTCs-STAT5CA (Fig. 6A). This observa- This enhanced tissue accumulation of STAT5CA-transduced T cells tion was validated by measuring cytokine production. Indeed, T- may result from a combination of increased tissue-homing, prolif- bet2/2 eTCs-STAT5CA recovered the ability to secrete IL-2 (Fig. eration, and survival in a T-bet–independent fashion. Additionally, 6B–D), as compared with wt eTCs-STAT5CA, which failed to do GzmB was equivalently expressed in T-bet–deficient and in wt eTCs- so. We also observed that STAT5CA expression greatly enhanced STAT5CA recovered from the spleen (Fig. 5B, 5C). In both cases, the the amount of accumulated IFN-g whether the responding T cells level of GzmB was higher than in the corresponding eTCs. Intra- were T-bet sufficient or deficient (Fig. 6E). Untransduced wt CD8 cellular staining for Eomes (Fig. 5B, 5C) showed a STAT5CA- eTCs were also able to secrete IFN-g and IL-2, although the mean dependent increase in expression of Eomes, the magnitude of fluorescence intensity (MFI) value for IFN-g was lower than in cells which was higher in T-bet–deficient as compared with wt eTCs. expressing STAT5CA (Fig. 6B). Finally, untransduced T-bet2/2 This latter observation suggests that increased Eomes expression CD8 eTCs had nearly lost the ability to produce IFN-g,whereas might act as a compensatory mechanism in the absence of T-bet. a small subset of cells had the capacity to secrete IL-2 (Fig. 6B–D). Transcriptomic analysis (Fig. 6A) of T-bet2/2 eTCs-STAT5CA Altogether, these results suggest that T-bet2/2 eTCs-STAT5CA determined that most of the transcripts induced by STAT5CAwere might be less differentiated than wt eTCs-STAT5CA with preserved still upregulated in absence of T-bet at a similar level (Il2ra, Ccl3, IL-2 secretion akin to that seen in central memory CD8 T cells. Tnf, Socs1, Gzmc, Cish, Eomes) or at a lower level (Gzmb, IFNg, Fasl, Socs2, Gzma, Gzmk, Rab27b, Cxcr3). Some of these results T-bet deficiency in STAT5CA-expressing eTCs leads to b have been validated by cytometry (Fig. 5B, 5C, and data not sensitivity to IL-6/TGF- 1 and promotes IL-17 secretion shown). However, we noticed a relatively higher Eomes protein Transcriptomic analysis had indicated a milder downmodulation of (Fig. 5B, 5C) than transcript (Fig. 6A) expression level, suggesting Il6ra, Il6st, and Tgfbr2 in T-bet–deficient as compared with wt that posttranscriptional mechanisms might control expression of eTCs-STAT5CA as compared with naive CD8 T cells (Fig. 6A). 8 STAT5/T-bet IMPRINT Tc1 RESISTANCE TO IL-6/TGF-b Downloaded from http://www.jimmunol.org/ by guest on September 25, 2021

FIGURE 4. Identification of direct STAT5 target genes by ChIP-Seq. (A) Genome-wide distribution of STAT5-bound regions as assessed by ChIP-Seq in ex vivo–purified memory TCRP1A eTCs-STAT5CA. The number of STAT5-bound regions found at the indicated intervals from the TSS is shown at the top of each bar plot. (B) Venn diagram showing the distribution of genes significantly differently expressed between TCRP1A eTCs-STAT5CA and untransduced eTCs (dark gray circle) and genes identified in anti-STAT5a ChIP from TCRP1A eTCs-STAT5CA (light gray circle). We annotated how genes already referenced in Fig. 1 distributed depending on a direct (overlapping) or indirect (not overlapping, dark gray) STAT5a regulation. (C and D) For a set of STAT5CA-dependent upregulated (Tbx21, Eomes, Gzmb) and downregulated (Tgfbr2, Il6st) genes, profiles of ChIP-Seq conducted on anti-STAT5a ChIP-DNA and the corresponding input control are shown (C). Results from PCRs on independent samples (D) validate the position of significantly enriched STAT5-bound regions [corresponding numbers in (C) and (D)].

FACS analysis confirmed a lower expression of the IL-6Ra–chain cells produced high amounts of IL-17, wt eTCs-STAT5CA failed to on wt as compared with T-bet–deficient STAT5CA-expressing cells produce this cytokine. We also observed that wt eTCs were unable to (Fig. 7A). Downregulation of IL-6Ra expression thus appears to produce IL-17 upon IL-6/TGF-b1 treatment. This result suggests that depend on T-bet. T-bet represses differentiation of Tc17. Indeed, T-bet2/2 eTCs were The genetic ablation of T-bet has been shown to strongly favor converted to IL-17 producers under polarizing conditions. IL-17 expression both in CD4 and CD8 T cells (29, 41), with Th17/ It thus appears that T-bet–deficient eTCs, whether they express IL-17–producing CD8 T cells (Tc17) differentiation being induced by STAT5CA or not, conserve some functional plasticity and are IL-6 and TGF-b1. We therefore evaluated levels of IL-17 induced by reprogrammed as Tc17 producers upon IL-6/TGF-b1 treatment. In IL-6/TGF-b1 treatment of the wt and T-bet–deficient eTC-STAT5CA contrast, wt eTCs-STAT5CA appear to be fully differentiated and populations (Fig. 7B). Whereas T-bet–deficient STAT5CA-expressing resistant to polarizing IL-6/TGF-b1 cytokines. The Journal of Immunology 9 Downloaded from http://www.jimmunol.org/

FIGURE 5. STAT5CA expression induces differentiation of long-lived eTCs independently of T-bet. (A–C) Rag-22/2 B6 mice were injected with (2 3 106) T-bet2/2 or wt CD8 eTCs or (106) T-bet2/2 or wt CD8 eTCs-STAT5CA. At day 47, absolute numbers of CD8 T cells recovered in liver and spleen are shown in (A) (mean of four experiments). In (B), expression of GzmB and Eomes on the gated splenic CD8 eTCs are reported together with their MFI, compared with an isotype control (gray). In (C), the fold increase for GzmB and Eomes MFI between T-bet2/2 eTCs-STAT5CA and wt eTCs-STAT5CA is shown for independent donor mice.

Discussion target genes. STAT5 has also been shown to unmask DNA bind-

STAT5 is a TF capable of inducing both positive and negative gene ing sites for T-bet during activation of the Ifng (36). Fur- by guest on September 25, 2021 regulation owing to its capacity to recruit either transcriptional thermore, we show that Cxcr3 and Ifng transcripts were still upreg- coactivators or corepressors (42). STAT5CA might additionally ulated, although at a reduced level on T-bet–deficient as compared modify the chromatin structure for efficient TF recruitment on the with wt eTCs-STAT5CA (Fig. 6A). targeted loci (43). Indeed, STAT5 function is modulated by its Altogether, sustained STAT5 activation correlated with long- interaction with cofactors (44) or with other TFs bound on term effector function, migration, and survival of CD8 eTCs. At neighboring sites (36). In CD8 (33, 34) and CD4 (45) T cells, the transcriptome level, in vivo–maintained CD8 eTCs-STAT5CA sustained activation of STAT5 was shown to be crucial for the demonstrated a long-lasting regulation of expression of several stabilization of gene expression programs. Sustained activation of molecules involved in costimulation (CD25, OX40, GITR, 4- STAT5 was also shown to be important in nonhematopoietic cells 1BB), memory fate (OX40, GITR, 4-1BB), tissue infiltration (46) for chromatin remodeling and transcription of the b-casein (CXCR3hi), and effector functions (GzmB, FasL, IFN-g). Indeed, gene. Transcriptomic analyses of CD8 eTCs expressing STAT5CA although not constitutively activated, CD8 eTCs-STAT5CA were in this study identify a cluster of genes demonstrating an early and capable of enhanced secondary responses. This may be due to sustained expression (cluster 1; Fig. 1B) and also a cluster of genes their increased expression of transcripts encoding effector mole- exhibiting a time-dependent accumulation (cluster 2; Fig. 1C). cules allowing rapid recall responses. Whether a relationship of causality exists between early expression High GzmB expression by CD8 eTCs-STAT5CA may result of some of the genes from type 1 cluster and subsequent expression from direct STAT5-mediated regulation (34) (Fig. 4); indirect of some genes of the type 2 cluster remains to be determined. The regulation through increased T-bet/Eomes expression, as well as pattern of expression of the type 1 gene cluster, including Tbx21/T- decreased Bcl6 expression, as the latter TF is a known repressor bet, is consistent with a role for STAT5CA in the stabilization of of GzmB transcription (39); and/or complementation between the a Tc1 gene expression program initiated by TCR stimulation (33, T-box members T-bet and Eomes (Fig. 5). Our ChIP-Seq data, 34). For Eomes expression, STAT5CA appeared to induce a time- validated by ChIP assays, identified regions around the Eomes (2), dependent accumulation of the transcripts (type 2 cluster). These Tbx21/T-bet (3), and Gzmb (3) genes to which STAT5 was bound patterns of T-bet and Eomes gene expression are consistent with in the long-term in vivo–maintained STAT5CA-expressing eTCs the recent description of an IL-2–dependent expression of Eomes (Fig. 4), indicating a direct implication of STAT5 in the regulation in CD8 T cells several days after removal of the TCR stimulus, of those genes. Differences in the initial control of gene expression whereas T-bet was induced in response to TCR engagement (47). that lead to distinct transcriptional patterns of expression for Of note, persistent expression of T-bet was shown to be necessary Eomes (type 2) and for Tbx21 and Gzmb (type 1) require further for IFN-g production and acquisition of the CD122hiCXCR3hi analyses. phenotype of CD4 T cells (29, 30). Therefore, the sustained T-bet For eTCs, T-bethi/Eomeslow expression patterns have been as- expression triggered by STAT5CA might upregulate those T-bet sociated with their terminal differentiation (6, 7). Of note, this 10 STAT5/T-bet IMPRINT Tc1 RESISTANCE TO IL-6/TGF-b Downloaded from http://www.jimmunol.org/ by guest on September 25, 2021

FIGURE 6. STAT5CA and T-bet cooperate for imprinting a Tc1 program: T-bet–deficient T cells recovered functional plasticity. (A) T-bet2/2 eTCs- STAT5CA were adoptively transferred into Rag-22/2 B6 mice. Twenty days later CD8 T cells recovered from pooled lymph nodes and spleens were included in a transcriptomic analysis. The Venn diagram shows the distribution of genes with significant changes between TCRP1A eTCs-STAT5CAversus naive CD8 T cells (black circle), and T-bet2/2 eTCs-STAT5CA versus naive CD8 T cells (gray circle). For each subset, the most significant genes (highest ratio) are reported as well as their fold increase in parentheses. For some of the 853 common genes, numbers in parentheses are relative to the fold increase in TCRP1A eTCs-STAT5CA (first) and T-bet2/2 eTCs-STAT5CA (second). Genes marked with an asterisk are also expressed at a significantly different level between TCRP1A eTCs-STAT5CA and T-bet2/2 eTCs-STAT5CA. (B–D) Rag-22/2 B6 mice were injected with 2 3 106 T-bet2/2 or wt CD8 eTCs or 106 T-bet2/2 or wt CD8 eTCs-STAT5CA. At day 40 after transfer, CD8 T cells recovered from recipients’ pooled lymph nodes and spleen were stimulated or not with ionomycin/PMA and stained for IFN-g and IL-2. Staining representative of three different experiments with two to three mice per group is shown in (B). Percentage positive cells is reported in each quadrant; MFI for IFN-g+ CD8 T cells is reported in brackets below cell description. Nd, Not determined owing to the paucity of cells. Statistical analyses done with an unpaired t test (GraphPad) for total IFN-g or IL-2–producing T cells are reported in (C) and (D), respectively. *p , 0.05, ***p , 0.001, ****p , 0.0001 by two-tailed p test. (E) T cells (for T-bet2/2 eTCs and eTCs-STAT5CA, same cells as in (B)–(D); for wt eTCs and eTCs-STAT5CA, either polyclonal, as in (B)–(D), or monoclonal [TCRP1A]) were stimulated 48 h with plate-bound anti- CD3. IFN-g present in culture supernatants was quantified by ELISA. correlation was also observed in CD8 effector memory T cells senescence. Altogether, the concomitant T-bethi/Eomeshi expres- developing in the absence of CD4 help (6), suggesting that the sion by CD8 eTCs-STA5TCA may explain their responsive level of T-bet/Eomes not only controls CD8 effector memory memory rather than terminal effector fate, as suggested by others T cell functions (cytolysis, IFN-g) but also, directly or indirectly, (48). Indeed, sustained STAT5, T-bet, and Eomes expression are controls the fate of these cells, possibly through the induction of correlated with a superior efficacy of STAT5CA-expressing eTCs The Journal of Immunology 11

FIGURE 7. T-bet deficiency leads to enhanced response to IL-6 and TGF-b1 exemplified by IL-17 polarization. (A) T-bet2/2 eTCs-STAT5CA (open bars) or wt eTCs-STAT5CA (filled bars) were prepared as in Fig. 6B–D and stained for IL-6Ra surface expression. The fold increase between IL- 6Ra MFI versus control staining is reported for five independent donor mice. (B) T cells (106; same cells as in Fig. 6E) were stimulated 48 h with anti- CD3 in the presence of IL-6 and TGF-b1. IL-17A was quantified in culture supernatants. nd, Not detectable. For (A) and (B), statistical analyses were done with an unpaired t test (GraphPad). *p , 0.05, **p , 0.01, ****p , 0.0001 by two-tailed p test. In (B), comparison of T-bet2/2 eTCs to either wt TCRP1A or CD8 (B10.D2) naive T cells having the same significance are represented with a single line with a dotted extension.

for antitumor adoptive therapies as compared with control eTCs, We show that genetic loss of T-bet abrogates IL-6R and TGFbRII Downloaded from even when the latter is associated with IL-2 complex infusions repression and increases their sensitivity to these cytokines, leading (20), a protocol that might only induce transient STAT5 activation. to the differentiation of Tc17 (Fig. 7). As a consequence, the Comparison of secondary responses of T-bet–deficient and wt concerted action of STAT5CA and T-bet contributes to repression eTCs-STAT5CA further highlights the complex interplay between of IL-6Ra and TGFbRII expression, without compensation by T-bet and STAT5. IFN-g secretion appeared to be T-bet–dependent STAT5CA-induced Eomes expression. A competition between but can be compensated directly by STAT5CA and/or indirectly by binding of STAT5 (repressive) and STAT3 (permissive) on the IL- http://www.jimmunol.org/ STAT5CA-induced Eomes. IL-2 secretion was found inhibited by 17 promoter has recently been reported for CD4 T cells (54). In STAT5CA expression in a T-bet–dependent manner (Fig. 6), in this study, we report the incapacity of STAT5CA expression to agreement with the reported repressive function of T-bet on IL-2 mediate IL-17 repression in absence of T-bet. However, because gene transcription in CD4 Th1 cells (49). Importantly, these the loss of T-bet results in increased IL-6Ra expression, a possible results suggest that T-bet–deficient eTCs might be less differen- increased STAT3 signaling in the absence of T-bet may overcome tiated than wt eTCs-STAT5CA with preserved IL-2 secretion akin the STAT5 repressive effects on IL-17 transcription (54). to that seen in central memory CD8 T cells. Recent studies (10, From the therapeutic point of view, polyfunctional T-bet2/2 16) point to the superior therapeutic potential of less differentiated eTCs-STAT5CA might appear more appealing than fully differ- CD8 T cells in adoptive therapies. The polyfunctional character- entiated Tc1 wt eTCs-STAT5CA. However, the restored sensitiv- by guest on September 25, 2021 istics of T-bet2/2 eTCs-STAT5CA might therefore be considered ity of T-bet2/2 eTCs-STAT5CA to IL-6 and TGF-b1 dampens the useful features for this purpose. However, a main limitation for curative potential of those cells. As a general feature, T-bet might adoptive therapies is tumor-derived immunosuppression that lead to imprinting of a Tc1 functional program and blocking of dampens antitumor responses of both endogenous and transferred subsequent cytokine-induced polarization, whereas STAT5CA might CD8 eTCs. In human melanoma cells with the BRAF(V600E) amplify the effector functions by sustained regulation of the cor- mutation, production of immunosuppressive cytokines IL-6, IL- responding genes (e.g., Ifng, Gzmb) and decreased sensitivity of 10, and vascular endothelial growth factor A depended on acti- CD8 eTCs to IL-6/TGF-b1 through sustained downmodulation vation of both the BRAF/MAPK and STAT3 pathways (50). In of their receptors. A direct regulation by STAT5CA of the ex- a mouse model (TiRP) of autochthonous melanomas (51), ag- pression of those receptors is suggested by our ChIP-Seq/ChIP gressive melanoma development was associated with increased data identifying regions of bound STAT5 around the Il6st and levels of several immune-modulating cytokines in the sera (52), Tgfbr2 genes in long-term in vivo–maintained eTCs-STAT5CA including IL-6, a well-known inducer of STAT3 (19). Indeed, p- (Fig. 4). STAT3 was detected in tumor cells and in some hematopoietic In conclusion, direct gene regulation by STAT5CA induces cells infiltrating the TiRP melanoma tumors (52). Additionally, a combined central memory/effector memory T cell gene tran- those TiRP tumors were infiltrated by T cells with low levels of scription profile with concomitant high expression of T-bet GzmB effector molecules and presenting an “exhausted” pro- and Eomes. These eTCs-STAT5CA are endowed with long- grammed death-1+ phenotype (52). We recently showed (20) that term maintenance, efficient recall responses, and stable down- adoptive transfer of tumor-specific eTCs-STAT5CA was efficient modulation of receptors mediating IL-17 polarization. Alto- at inducing melanoma regression. Within the TiRP tumors, eTCs- gether, our data further indicate the importance of testing both STAT5CA maintained a GzmBhi expression, suggesting acquisi- functional responses and imprinting in T cells for adoptive ther- tion of resistance to the local and systemic immunosuppression or apy, as tumor-derived cytokines might polarize candidate T cells reprogramming (53). Transcriptomic analyses of eTCs-STAT5CA to be used for this purpose. in this study revealed a STAT5-mediated repression of IL-6Ra, IL- 6st, and TGFbRII subunits. Although IL-2 was previously shown Acknowledgments to decrease IL-6R expression on CD4 Th1 T cells (12), in this We acknowledge Pascal Barbry and Chimene Moreihlon from the Plate- study we show that the use of an active form of STAT5 provides forme Transcriptome Nice–Sophia Antipolis, Institut de Pharmacologie long-term IL-6R repression on CD8 eTCs in a T-bet–dependent Mole´culaire et Cellulaire, Sophia Antipolis, France. We acknowledge the manner (Fig. 7). Additionally, STAT5CA also induced stable re- Centre d’Immunologie de Marseille–Luminy bioinformatics platform pression of TGFbRII. (headed by Sebastien Jagger) and personnel funded by Cance´ropoˆle Pro- 12 STAT5/T-bet IMPRINT Tc1 RESISTANCE TO IL-6/TGF-b vence-Alpes-Coˆte d’Azur (Amira Djebbari) for help in analyzing tran- 24. Boyer, L. A., T. I. Lee, M. F. Cole, S. E. Johnstone, S. S. Levine, J. P. Zucker, scriptomic data. We thank Lee Leserman for editing the manuscript, as M. G. Guenther, R. M. Kumar, H. L. Murray, R. G. Jenner, et al. 2005. Core Cell well as the Centre d’Immunologie de Marseille–Luminy imaging and an- transcriptional regulatory circuitry in human embryonic stem cells. 122: 947–956. imal facilities personnel for assistance. 25. McLean, C. Y., D. Bristor, M. Hiller, S. L. Clarke, B. T. Schaar, C. B. Lowe, A. M. Wenger, and G. Bejerano. 2010. GREAT improves functional interpre- tation of cis-regulatory regions. Nat. Biotechnol. 28: 495–501. Disclosures 26. Laz, E. V., A. Sugathan, and D. J. Waxman. 2009. Dynamic in vivo binding of The authors have no financial conflicts of interest. STAT5 to growth hormone-regulated genes in intact rat liver. Sex-specific binding at low- but not high-affinity STAT5 sites. Mol. Endocrinol. 23: 1242– 1254. 27. Turatsinze, J. V., M. Thomas-Chollier, M. Defrance, and J. van Helden. 2008. References Using RSAT to scan genome sequences for transcription factor binding sites and 1. Kaech, S. M., S. Hemby, E. Kersh, and R. Ahmed. 2002. Molecular and func- cis-regulatory modules. Nat. Protoc. 3: 1578–1588. tional profiling of memory CD8 T cell differentiation. Cell 111: 837–851. 28. Onishi, M., T. Nosaka, K. Misawa, A. L. Mui, D. Gorman, M. McMahon, 2. Luckey, C. J., D. Bhattacharya, A. W. Goldrath, I. L. Weissman, C. Benoist, and A. Miyajima, and T. Kitamura. 1998. Identification and characterization of D. Mathis. 2006. Memory T and memory B cells share a transcriptional program a constitutively active STAT5 mutant that promotes cell proliferation. Mol. Cell. of self-renewal with long-term hematopoietic stem cells. Proc. Natl. Acad. Sci. Biol. 18: 3871–3879. USA 103: 3304–3309. 29. Intlekofer, A. M., A. Banerjee, N. Takemoto, S. M. Gordon, C. S. Dejong, 3. Sarkar, S., V. Kalia, W. N. Haining, B. T. Konieczny, S. Subramaniam, and H. Shin, C. A. Hunter, E. J. Wherry, T. Lindsten, and S. L. Reiner. 2008. R. Ahmed. 2008. Functional and genomic profiling of effector CD8 T cell Anomalous type 17 response to viral infection by CD8+ T cells lacking T-bet and subsets with distinct memory fates. J. Exp. Med. 205: 625–640. eomesodermin. Science 321: 408–411. 4. Masopust, D., V. Vezys, A. L. Marzo, and L. Lefranc¸ois. 2001. Preferential lo- 30. Matsuda, J. L., T. C. George, J. Hagman, and L. Gapin. 2007. Temporal dis- calization of effector memory cells in nonlymphoid tissue. Science 291: 2413– section of T-bet functions. J. Immunol. 178: 3457–3465. 2417. 31. Bezman, N. A., C. C. Kim, J. C. Sun, G. Min-Oo, D. W. Hendricks, Downloaded from 5. Mora, J. R., and U. H. von Andrian. 2006. T-cell homing specificity and plas- Y. Kamimura, J. A. Best, A. W. Goldrath, and L. L. Lanier; Immunological ticity: new concepts and future challenges. Trends Immunol. 27: 235–243. Genome Project Consortium. 2012. Molecular definition of the identity and 6. Intlekofer, A. M., N. Takemoto, C. Kao, A. Banerjee, F. Schambach, activation of natural killer cells. Nat. Immunol. 13: 1000–1009. J. K. Northrop, H. Shen, E. J. Wherry, and S. L. Reiner. 2007. Requirement for T- 32. Cush, S. S., and E. Flan˜o. 2011. KLRG1+NKG2A+ CD8 T cells mediate pro- + bet in the aberrant differentiation of unhelped memory CD8 T cells. J. Exp. tection and participate in memory responses during g-herpesvirus infection. J. Med. 204: 2015–2021. Immunol. 186: 4051–4058. 7. Joshi, N. S., W. Cui, A. Chandele, H. K. Lee, D. R. Urso, J. Hagman, L. Gapin, 33. Verdeil, G., J. Chaix, A. M. Schmitt-Verhulst, and N. Auphan-Anezin. 2006. and S. M. Kaech. 2007. Inflammation directs memory precursor and short-lived Temporal cross-talk between TCR and STAT signals for CD8 T cell effector

+ http://www.jimmunol.org/ effector CD8 T cell fates via the graded expression of T-bet transcription factor. differentiation. Eur. J. Immunol. 36: 3090–3100. Immunity 27: 281–295. 34. Verdeil, G., D. Puthier, C. Nguyen, A. M. Schmitt-Verhulst, and N. Auphan- 8. Crotty, S., R. J. Johnston, and S. P. Schoenberger. 2010. Effectors and : Anezin. 2006. STAT5-mediated signals sustain a TCR-initiated gene expression Bcl-6 and Blimp-1 in T and B lymphocyte differentiation. Nat. Immunol. 11: program toward differentiation of CD8 T cell effectors. J. Immunol. 176: 4834– 114–120. 4842. 9. Palmer, D. C., and N. P. Restifo. 2009. Suppressors of cytokine signaling (SOCS) 35. Townsend, M. J., A. S. Weinmann, J. L. Matsuda, R. Salomon, P. J. Farnham, in T cell differentiation, maturation, and function. Trends Immunol. 30: 592–602. C. A. Biron, L. Gapin, and L. H. Glimcher. 2004. T-bet regulates the terminal 10. Gattinoni, L., E. Lugli, Y. Ji, Z. Pos, C. M. Paulos, M. F. Quigley, J. R. Almeida, maturation and homeostasis of NK and Va14i NKT cells. Immunity 20: 477– E. Gostick, Z. Yu, C. Carpenito, et al. 2011. A human memory T cell subset with 494. stem cell-like properties. Nat. Med. 17: 1290–1297. 36. Shi, M., T. H. Lin, K. C. Appell, and L. J. Berg. 2008. Janus-kinase-3-dependent 11. Walker, S. R., E. A. Nelson, and D. A. Frank. 2007. STAT5 represses BCL6 signals induce chromatin remodeling at the Ifng locus during T helper 1 cell expression by binding to a regulatory region frequently mutated in lymphomas.

differentiation. Immunity 28: 763–773. by guest on September 25, 2021 Oncogene 26: 224–233. 37. Sullivan, B. M., A. Juedes, S. J. Szabo, M. von Herrath, and L. H. Glimcher. 12. Liao, W., J. X. Lin, L. Wang, P. Li, and W. J. Leonard. 2011. Modulation of 2003. Antigen-driven effector CD8 T cell function regulated by T-bet. Proc. cytokine receptors by IL-2 broadly regulates differentiation into helper T cell Natl. Acad. Sci. USA 100: 15818–15823. lineages. Nat. Immunol. 12: 551–559. 38. Belz, G. T., and A. Kallies. 2010. Effector and memory CD8+ T cell differen- 13. Dudley, M. E., J. R. Wunderlich, J. C. Yang, R. M. Sherry, S. L. Topalian, tiation: toward a molecular understanding of fate determination. Curr. Opin. N. P. Restifo, R. E. Royal, U. Kammula, D. E. White, S. A. Mavroukakis, et al. Immunol. 22: 279–285. 2005. Adoptive cell transfer therapy following non-myeloablative but lympho- 39.Yoshida,K.,A.Sakamoto,K.Yamashita,E.Arguni,S.Horigome,M.Arima, depleting chemotherapy for the treatment of patients with refractory metastatic M. Hatano, N. Seki, T. Ichikawa, and T. Tokuhisa. 2006. Bcl6 controls melanoma. J. Clin. Oncol. 23: 2346–2357. + 14. Porter, D. L., B. L. Levine, M. Kalos, A. Bagg, and C. H. June. 2011. Chimeric granzyme B expression in effector CD8 Tcells.Eur. J. Immunol. 36: 3146– antigen receptor-modified T cells in chronic lymphoid leukemia. N. Engl. J. 3156. Med. 365: 725–733. 40. Lin, J. X., P. Li, D. Liu, H. T. Jin, J. He, M. Ata Ur Rasheed, Y. Rochman, 15. Gattinoni, L., C. A. Klebanoff, D. C. Palmer, C. Wrzesinski, K. Kerstann, Z. Yu, L. Wang, K. Cui, C. Liu, et al. 2012. Critical Role of STAT5 transcription factor S. E. Finkelstein, M. R. Theoret, S. A. Rosenberg, and N. P. Restifo. 2005. tetramerization for cytokine responses and normal immune function. Immunity Acquisition of full effector function in vitro paradoxically impairs the in vivo 36: 586–599. antitumor efficacy of adoptively transferred CD8+ T cells. J. Clin. Invest. 115: 41. Yang, Y., J. Xu, Y. Niu, J. S. Bromberg, and Y. Ding. 2008. T-bet and eome- 1616–1626. sodermin play critical roles in directing T cell differentiation to Th1 versus 16. Dudley, M. E., C. A. Gross, M. M. Langhan, M. R. Garcia, R. M. Sherry, Th17. J. Immunol. 181: 8700–8710. J. C. Yang, G. Q. Phan, U. S. Kammula, M. S. Hughes, D. E. Citrin, et al. 2010. 42. Moriggl, R., V. Sexl, L. Kenner, C. Duntsch, K. Stangl, S. Gingras, CD8+ enriched “young” tumor infiltrating lymphocytes can mediate regression A. Hoffmeyer, A. Bauer, R. Piekorz, D. Wang, et al. 2005. Stat5 tetramer for- of metastatic melanoma. Clin. Cancer Res. 16: 6122–6131. mation is associated with leukemogenesis. Cancer Cell 7: 87–99. 17. Boyman, O., and J. Sprent. 2012. The role of interleukin-2 during homeostasis 43. Xu, M., L. Nie, S. H. Kim, and X. H. Sun. 2003. STAT5-induced Id-1 tran- and activation of the . Nat. Rev. Immunol. 12: 180–190. scription involves recruitment of HDAC1 and deacetylation of C/EBPb. EMBO 18. Wei, L., A. Laurence, and J. J. O’Shea. 2008. New insights into the roles of J. 22: 893–904. Stat5a/b and Stat3 in T cell development and differentiation. Semin. Cell Dev. 44. Pfitzner, E., R. Ja¨hne, M. Wissler, E. Stoecklin, and B. Groner. 1998. p300/ Biol. 19: 394–400. CREB-binding protein enhances the prolactin-mediated transcriptional induc- 19. Yu, H., D. Pardoll, and R. Jove. 2009. STATs in cancer inflammation and im- tion through direct interaction with the transactivation domain of Stat5, but does munity: a leading role for STAT3. Nat. Rev. Cancer 9: 798–809. not participate in the Stat5-mediated suppression of the glucocorticoid response. 20. Grange, M., M. Buferne, G. Verdeil, L. Leserman, A. M. Schmitt-Verhulst, and Mol. Endocrinol. 12: 1582–1593. N. Auphan-Anezin. 2012. Activated STAT5 promotes long-lived cytotoxic CD8+ 45. Cote-Sierra, J., G. Foucras, L. Guo, L. Chiodetti, H. A. Young, J. Hu-Li, J. Zhu, T cells that induce regression of autochthonous melanoma. Cancer Res. 72: 76–87. and W. E. Paul. 2004. Interleukin 2 plays a central role in Th2 differentiation. 21. Shanker, A., G. Verdeil, M. Buferne, E. M. Inderberg-Suso, D. Puthier, F. Joly, Proc. Natl. Acad. Sci. USA 101: 3880–3885. C. Nguyen, L. Leserman, N. Auphan-Anezin, and A. M. Schmitt-Verhulst. 46. Xu, R., C. M. Nelson, J. L. Muschler, M. Veiseh, B. K. Vonderhaar, and 2007. CD8 T cell help for innate antitumor immunity. J. Immunol. 179: 6651– M. J. Bissell. 2009. Sustained activation of STAT5 is essential for chromatin 6662. remodeling and maintenance of mammary-specific function. J. Cell Biol. 184: 22. Oliveros, J. 2007. VENNY: an interactive tool for comparing lists with Venn 57–66. diagrams. Available at: http://bioinfogp.cnb.csic.es/tools/venny/index.html 47. Pipkin, M. E., J. A. Sacks, F. Cruz-Guilloty, M. G. Lichtenheld, M. J. Bevan, and 23. Subramanian, A., P. Tamayo, V. K. Mootha, S. Mukherjee, B. L. Ebert, A. Rao. 2010. Interleukin-2 and inflammation induce distinct transcriptional M. A. Gillette, A. Paulovich, S. L. Pomeroy, T. R. Golub, E. S. Lander, and programs that promote the differentiation of effector cytolytic T cells. Immunity J. P. Mesirov. 2005. Gene set enrichment analysis: a knowledge-based approach 32: 79–90. for interpreting genome-wide expression profiles. Proc. Natl. Acad. Sci. USA 48. Banerjee, A., S. M. Gordon, A. M. Intlekofer, M. A. Paley, E. C. Mooney, 102: 15545–15550. T. Lindsten, E. J. Wherry, and S. L. Reiner. 2010. Cutting edge: the transcription The Journal of Immunology 13

factor eomesodermin enables CD8+ T cells to compete for the memory cell 52. Soudja, S. M., M. Wehbe, A. Mas, L. Chasson, C. P. de Tenbossche, I. Huijbers, niche. J. Immunol. 185: 4988–4992. B. Van den Eynde, and A. M. Schmitt-Verhulst. 2010. Tumor-initiated inflam- 49. Hwang, E. S., J. H. Hong, and L. H. Glimcher. 2005. IL-2 production in de- mation overrides protective adaptive immunity in an induced melanoma model veloping Th1 cells is regulated by heterodimerization of RelA and T-bet and in mice. Cancer Res. 70: 3515–3525. requires T-bet serine residue 508. J. Exp. Med. 202: 1289–1300. 53. Auphan-Anezin, N., G. Verdeil, M. Grange, S. M. Soudja, M. Wehbe, 50. Sumimoto, H., F. Imabayashi, T. Iwata, and Y. Kawakami. 2006. The BRAF- M. Buferne, A. Mas, and A. M. Schmitt-Verhulst. 2013. Immunosuppression in MAPK signaling pathway is essential for cancer-immune evasion in human inflammatory melanoma: can it be resisted by adoptively transferred T cells? melanoma cells. J. Exp. Med. 203: 1651–1656. Pigment Cell Melanoma Res 26: 167–175. 51. Huijbers, I. J., P. Krimpenfort, P. Chomez, M. A. van der Valk, J. Y. Song, 54. Yang, X. P., K. Ghoreschi, S. M. Steward-Tharp, J. Rodriguez-Canales, J. Zhu, E. M. Inderberg-Suso, A. M. Schmitt-Verhulst, A. Berns, and B. J. Van den J. R. Grainger, K. Hirahara, H. W. Sun, L. Wei, G. Vahedi, et al. 2011. Opposing Eynde. 2006. An inducible mouse model of melanoma expressing a defined regulation of the locus encoding IL-17 through direct, reciprocal actions of tumor antigen. Cancer Res. 66: 3278–3286. STAT3 and STAT5. Nat. Immunol. 12: 247–254. Downloaded from http://www.jimmunol.org/ by guest on September 25, 2021