of June 2, 2018.This information is current as

ResponsesCell-Mediated Antitumor Immune T-bet and Eomesodermin Are Required for T

and Binfeng LuChangyou Li, Penelope Morel, Lin Liu, Xueguang Zhang Yibei Zhu, Songguang Ju, Elizabeth Chen, Shao Dai,

http://www.jimmunol.org/content/185/6/3174doi: 10.4049/jimmunol.1000749August 2010;

2010; 185:3174-3183; Prepublished online 16J Immunol 

MaterialSupplementary

9.DC1http://www.jimmunol.org/content/suppl/2010/08/16/jimmunol.100074

Referenceshttp://www.jimmunol.org/content/185/6/3174.full#ref-list-1

, 20 of which you can access for free at: cites 35 articlesThis article

        average*  

4 weeks from acceptance to publicationFast Publication! •    

Every submission reviewed by practicing scientistsNo Triage! •    

from submission to initial decisionRapid Reviews! 30 days* •    

Submit online. ?The JIWhy

Subscriptionhttp://jimmunol.org/subscription

is online at: The Journal of ImmunologyInformation about subscribing to

Permissionshttp://www.aai.org/About/Publications/JI/copyright.htmlSubmit copyright permission requests at:

Email Alertshttp://jimmunol.org/alertsReceive free email-alerts when new articles cite this article. Sign up at:

Print ISSN: 0022-1767 Online ISSN: 1550-6606. Immunologists, Inc. All rights reserved.Copyright © 2010 by The American Association of1451 Rockville Pike, Suite 650, Rockville, MD 20852The American Association of Immunologists, Inc.,

is published twice each month byThe Journal of Immunology

by guest on June 2, 2018http://w

ww

.jimm

unol.org/D

ownloaded from

by guest on June 2, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

The Journal of Immunology

T-bet and Eomesodermin Are Required for T Cell-MediatedAntitumor Immune Responses

Yibei Zhu,*,†,1 Songguang Ju,†,1 Elizabeth Chen,‡ Shao Dai,* Changyou Li,*

Penelope Morel,* Lin Liu,* Xueguang Zhang,† and Binfeng Lu*,x

Cell-mediated adaptive immunity is very important in tumor immune surveillance and tumor vaccination. However, the genetic pro-

gram underlying an effective adaptive antitumor immunity is elusive. T-bet and Eomesodermin (Eomes) have been suggested to be

master regulators of Th1 cells and CD8+ T cells. However, whether they are important for T cell-mediated antitumor immunity is

controversial. In this paper, we show that the combined germline deletion of T-bet and T cell-specific deletion of Eomes resulted in

profound defects in adaptive antitumor immune responses. T-bet and Eomes drive Tc1 differentiation by preventing alternative CD8+

T cell differentiation to Tc17 or Tc2 cells. Surprisingly, T-bet and Eomes are not critical for the generation of systemic CTL activities

against cancer cells. Instead, T-bet andEomes are crucial for tumor infiltration byCD8+ T cells. This study defines T-bet andEomes as

critical regulators of T cell-mediated immune responses against tumor. The Journal of Immunology, 2010, 185: 3174–3183.

The cells of the immune system are involved in cancer de-velopment. Epidemiological studies have established thatinflammation is associated with cancer progression (1).

Inflammation leads to the production of growth factors, angiogenicfactors, and matrix-degrading proteases by leukocytes, which playan important role in tumor promotion and progression (1). In con-trast, tumor development results in the production of danger signals(2) and the expression of tumor Ags (3), which elicit cell-mediatedimmune responses against cancer cells. This is called tumor im-mune surveillance or immune editing (4). In fact, the presence oftumor-infiltrating T cells is associated with better clinical outcomesfor certain types of cancer (5–7). In particular, IFN-g–producingtype 1 immune cells such as Th1, Tc1, gd T, NKT, and NK cellshave been shown to play critical roles in tumor immune sur-veillance and eradication (7). Currently, many cancer vaccinesare being designed to elicit strong type 1 cell-mediated immuneresponses against cancer (8).The transcription factors T-bet and Eomesodermin (Eomes) are

master transcription factors of the Th1 phenotype. T-bet is highlyexpressed in type 1 immune cells such as Th1, Tc1, NK, NKT,

and gd T cells and is critical for the effector function of Th1 cellsand NK cells but is only partially required for CD8+ Tc1 cells.Eomes was found to be highly expressed in CD8+ T cells but weaklyin CD4+ T cells and is proposed to play a key role in the effectorfunction of CD8+ T cells (9). In addition, T-bet and Eomes have beenimplicated in anticancer responses, and higher expression of T-bet isassociated with a favorable outcome of cancer patients (7). The lackof T-bet leads to increased cancer metastasis (10); however, the roleof T-bet in cancer is attributed to its regulation of innate immunity, inparticular, NK cell function (11). Overall, the role of T-bet andEomes in adaptive immunity against cancer is unclear.To study the role of T-bet and Eomes in tumor vaccination, we

generated mice with the germline deletion of T-bet and T cell-specific deletion of Eomes. Unexpectedly, we have found that T-betand Eomes are involved but not crucial for IFN-g production byCD8+ T cells. In contrast, T-bet and Eomes are necessary for sup-pressing alternative fates of CD8+ T cells such as Tc17 and Tc2.Moresurprisingly, T-bet and Eomes are not required for the generation ofsystemic CTLs against a dominant tumor Ag after tumor vaccination.Nevertheless, T-bet and Eomes are critical for inhibiting tumorgrowth after tumor vaccination. The lack of both T-bet and Eomesresults in a lower expression of CXCR3 in T cells and a drastic de-crease in the number of tumor-infiltrating T cells. Consistent withthis, CXCR3-deficient mice showed compromised antitumor im-mune responses upon tumor vaccination. In addition, T-bet/Eomes(T/E)-deficient tumor-infiltrating CD8+ T cells produced much moreIL-17 and less IFN-g. Therefore, our study establishes T-bet andEomes as key regulators of adaptive cell-mediated immunity againstcancer through promoting migration of CD8+ T cells to the tumortissue, enhancing IFN-g production, and suppressing inflammatoryIL-17 production.

Materials and MethodsMice

The bacteria artificial chromosome containing Eomes was obtained fromP. J. de Jong (Children’s Hospital, Oakland, CA). The targeting vector wasconstructed using the recombineering technique (12). Details of vectorconstruction will be available upon request. This vector was used to inserttwo loxp sites that flank the second exon of Eomes, which encodes a criticalregion for DNA binding. The vector was electroporated into SW9.5 ES cells,a gift from R. Chaillet (University of Pittsburgh, Pittsburgh, PA). Drug-resistant mouse embryonic stem cell clones were screened by Southern blot

*Department of Immunology, University of Pittsburgh School of Medicine, Pitts-burgh, PA 15261; †Department of Immunology, Institute of Medical Biotechnology,Soochow University, Suzhou, People’s Republic of China; ‡Carnegie Mellon Univer-sity, Pittsburgh, PA 15213; and xUniversity of Pittsburgh Cancer Institute, Pittsburgh,PA 15232

1Y.Z. and S.J. contributed equally to this work.

Received for publication March 8, 2010. Accepted for publication July 6, 2010.

This work was supported by a Cancer Institute Investigator Award (to B.L.). L.L. wassupported by National Institutes of Health Training Grant T32 CA082084. This workwas partly supported by Cancer and Aging Program National Institutes of HealthGrant P20 CA103730; National Natural Science Foundation of China Grants30528008 (to B.L. and X.Z.) and 30700728 (to Y.Z.); and National Key Basic Re-search Developmental Program of China Grant 2007CB512402 (to X.Z.).

Address correspondence and reprint requests to Dr. Binfeng Lu, Department of Immu-nology, University of Pittsburgh School of Medicine, 200 Lothrop Street, E1047, EastBiomedical Science Tower, Pittsburgh, PA 15261. E-mail address: Binfeng@pitt.edu

The online version of this article contains supplemental material.

Abbreviations used in this paper: DKO, double knockout; EKO, Eomes–/–; Eomes,Eomesodermin; i.d., intradermally; IP-10, inflammatory protein 10; KO, knockout;T/E, T-bet/Eomesodermin; TIL, tumor-infiltrating lymphocyte; TKO, T-bet–/–; TRP,tyrosinase; VAC, vaccination; WT, wild-type.

Copyright� 2010 by TheAmericanAssociation of Immunologists, Inc. 0022-1767/10/$16.00

www.jimmunol.org/cgi/doi/10.4049/jimmunol.1000749

by guest on June 2, 2018http://w

ww

.jimm

unol.org/D

ownloaded from

analysis of genomic DNA digested with XmnI, with either 39 or 59 probes.Two independent correctly targeted embryonic stem cell clones wereinjected into C57BL/6 blastocysts, and germline transmission was obtainedfrom each clone. The Neo gene was removed by breeding F1 mice witha strain of actin promoter-driven Flipase transgenic mice (JAX). These micewere then bred with T-bet2/2 (TKO) mice in the C57BL/6 background foreight generations to generate TKO Eomes fl/+ mice. Then, TKO Eomes fl/+mice were crossed with CD4-cre on the C57BL/6 background to generateCD4-cre-T-bet+/2 Eomes fl/+. These mice were intercrossed to generateCD4-cre-TKO Eomes fl/fl mice. Age- and gender-matched C57BL/6 mice(JAX) were used as wild-type (WT) controls. All animals were maintainedunder specific pathogen-free conditions. CXCR32/2 mice were provided tous by Dr. F. Lakkis (University of Pittsburgh) with permission from Dr. C.Gerard (Harvard University, Cambridge, MA). All animal work has beenapproved by the Institution Animal Care and Use Committee at the Univer-sity of Pittsburgh. For adoptive transfer experiment, C57BL/6 WT recipientmice were lethally irradiated at 1000 rad and then adoptively transferredwith different donor bone marrow cells (WT, TKO, Eomes2/2 [EKO], ordouble knockout [DKO]) within 24 h of irradiation. The reconstituted micewere used for experiments 2–3 mo later.

Cell lines

B16F0 melanoma cells were provided by Dr. Z. Yin (Yale University,School of Medicine, New Haven, CT). B16-GM-CSF cells were a gift fromDr. W. Storkus. B16F0 and B16-GM-CSF cells were grown in DMEM plus10% FCS.

In vitro CD8+ T cells differentiation

Spleen and lymph node were collected from C57BL/6WT, TKO, EKO, andT/E DKO mice. Naive CD62LhighCD44low CD8+ T cells were purified byFACS and cultured in Tc1, Tc0, Tc2, and Tc17 condition as indicated. Cellswere stimulated with 5 mg/ml plate-bound anti-CD3 and 5 mg/ml plate-bound anti-CD28 mAbs in the presence of anti–IL-2Ra (10 mg/ml) andIL-12 (3.4 ng/ml) plus anti–IL-4 (10 mg/ml, clone 11B11) for Tc1 orIL-2 (20 U/ml) for Tc0 or IL-2 (20 U/ml) and IL-4 (2 ng/ml) plus anti–IFN-g (10 mg/ml, clone XMG) for Tc2 or anti–IL-2Ra (10 mg/ml), IL-23(10 ng/ml), IL-6 (10 ng/ml), TGF-b (1 ng/ml), and anti–IFN-g (10 mg/ml,clone XMG) plus anti–IL-4 (10 mg/ml, clone 11B11) for Tc17. After 48 h,cells were replated to new wells without anti-CD3 and anti-CD28 and withfreshly added IL-2 (20 U/ml) for another 48 h.

Cancer models

Prophylaxis experiment. Depletion of peripheral regulatory T cells wasaccomplished by injection of anti-CD25 (clone PC61, 350 mg, i.p.) on days0 and 7. B16-GM-CSF cells were irradiated at 35 Gy, washed, and resus-pended in 100ml PBS for s.c. administration of 53 105/mouse at days 4 and7. On day 14, mice were challenged with 23 105 B16F0 cells intradermally(i.d.). The size of i.d. growing tumors was measured every 2 d.

Therapeutic experiment. Mice were challenged with 2 3 105 B16F0 cellsi.d. At the same day, treatment was initiated by injecting anti-CD25 (clonePC61, 350 mg, i.p.), and the anti-CD25 injection was repeated 7 d later.Treatment with 5 3 105 irradiated B16-GM-CSF cells was administrateds.c. 4 and 7 d after the inoculation of live B16 cells. Mice were monitoredfor tumor growth every other day.

Histopathology

Tumor samples were removed and embedded in OCT compound on dry ice.For immunofluorescence, 6-mmcryostat sectionswere prepared, air-dried for10 min, and fixed in acetone for 10 min at room temperature. The sectionswere then incubated in PBS with 2%BSA, stained with anti-mouse CD8 andanti-rat Alexo Fluor 488, and subsequently stained with Hoechst dye for 30 s.Images were examined using a fluorescence microscope (Provis AX70;Olympus Optical, Tokyo, Japan).

Gene expression analysis

Naive CD62LhighCD44lowCD8+ T cells were purified from C57BL/6 mice,TKO mice, EKO mice, and T/E DKO mice and cultured with 5 mg/ml plate-bound anti-CD3 and 5 mg/ml plate-bound anti-CD28 Abs in the presenceof IL-2 (20 U/ml) and IL-12 (3.4 ng/ml) plus anti–IL-4 (10 mg/ml, clone11B11). After 48 h, cells were replated to new wells without anti-CD3 andanti-CD28 and with freshly added IL-2 (20 U/ml) for another 48 h. Then,effector CD8+ T cells were restimulated with plate-bound anti-CD3 (5 mg/ml) for 4 h. Total RNA was extracted using the RNeasy mini kit (Qiagen,Valencia, CA) and reverse transcribed using Superscript II (Invitrogen, Carls-bad, CA). The real-time PCR was performed on with SYBR green (AppliedBiosystems, Foster City, CA). Cycling conditions were 10 min at 95˚C,

followed by 40 repeats of 95˚C for 15 s and 60˚C for 60 s. Analysis wasperformed by sequence detection software supplied with the instrument (Ap-plied Biosystems). The primers included perforin sense, 59-TGGTGGG-ACTTCAGCTTTC-39; perforin antisense, 59-TGACCGAGTGGCAGTGT-AG-39; granzyme B sense, 59-CCACTCTCGACCCTACATGG-39; granzymeB antisense, 59-GGCCCCCAAAGTGACATTTATT-39; CXCR3 sense, 59-TCTCGTTTTCCCCATAATCG-39; CXCR3 antisense, 59-AGCCAAGCCAT-GTACCTTGA-39;CCR5sense, 59-AGCCGCAATTTGTTTCACAT-39;CCR5antisense, 59-GAGACATCCGTTCCCCCTAC-39; b-actin sense, 59-TCC TTCGTT GCC GGT CCA C-39; and b-actin antisense, 59-ACC AGC GCA GCGATATCG TC-39.

Flow cytometric analysis

CD4(GK1.5), CD8(53-6.7), CD44(IM7), CD122(5H4), CD62L(MEL-14),CD127(A7R34), CD25(PC61.5), Foxp3(FJK-16s), B220(RA3-6B2),NK1.1(PK136), pan-NK(DX5), and TCRb (H57-597) were all purchasedfrom eBioscience (San Diego, CA). Flow cytometric analysis was per-formed using a FACS flow cytometer (BD Biosciences, San Jose, CA).

For intracellular cytokine staining, naive CD62LhighCD44lowCD8+ T cellspurified from TKO, EKO, T/E DKO, or C57BL/6 WT mice were stimulatedwith plate-bound anti-CD28 (5 mg/ml) plus anti-CD3 (5 mg/ml) in Tc1 orTc17 conditions for 96 h. Cells were stimulated with PMA (10 ng/ml) andionomycin (1 mg/ml) for 4 h and incubated for the last 3 h with brefeldin A(10 mg/ml). Cells were transferred to a V-bottom plate, stained with anti-CD8 in HBSS (containing 1% FCS), then fixed with 2% formaldehyde,which was followed by permeabilization with 0.5% saponin. The cells weresubsequently stained with anti–IL-17 (eBio17B7; eBioscience) and anti–IFN-g Ab (clone XMG1.2; eBioscience). Cells producing IFN-g and IL-17were examined with flow cytometry.

Pentamer staining

Allophycocyanin-labeled Pro5 MHC Class I Pentamer H-2K was obtainedfrom ProImmune (Bradenton, FL). Splenocytes were harvested fromT/E DKO or WT-vaccinated mice and allocated (1, 2) 3106 splenocytesper staining condition. Allophycocyanin-labeled Pentamer (10 ml) wasadded to the cells and incubated at room temperature for 10 min, afterwhich the cells were washed, and incubated with anti-CD8 and anti-B220Abs on ice for 20 min before flow cytometric analysis.

In vivo CTL killing assay

T/E DKO mice or WT mice were vaccinated with anti-CD25 and 5 3 105

irradiated B16-GM-CSF cells as described before. Eight days later, spleno-cytes from a syngeneic mouse were isolated, pulsed with tyrosinase(TRP)-2180–188 (SVYDFFVWL) (10 mg/ml, 2 h), and labeled using a highconcentration of CFSE (5 mM, CFSEhigh); splenocytes from a syngeneicmouse without TRP-2 peptide were labeled using a low concentration ofCFSE (0.5 mM, CFSElow) as an internal control. The mixture of the twoCFSE-labeled splenic populations (5 3 106 cells each) was injected intohost DKO mice or WT mice. After 5 h, the relative abundance of CFSEhigh

and CFSElow cells in peripheral blood was determined by flow cytometry.

Harvest of tumor-infiltrating lymphocytes

Tumor masses were removed, minced, and digested with collagenase andhyaluronidase digestion solution (2.5 mg/ml collagenase I, 1 mg/ml colla-genase IV, 0.25 mg/ml hyaluronidase IV-S, 300 mg/ml DNase I, and 0.01%HEPES in RPMI 1640 medium) at 37˚C for 2 h. The pieces were then gentlypressed between the frosted edges of two sterile glass slides, and the cellsuspension was filtered through a 40-mm cell strainer (BD Biosciences) toremove debris and separate cell clumping. Tumor-infiltrating lymphocytes(TILs) were further purified by using the gradient as per manufacturer pro-tocol, washed, and resuspended in HBSS for analysis.

Chemotaxis assays

Naive CD62LhighCD44low T cells purified from TKO, EKO, T/E DKO, orC57BL/6 WT mice were stimulated in Th1 conditions for 96 h. Thenlive cells were loaded into the top chamber of Transwell inserts (5.0-mmpore size; Costar, Cambridge, MA). RPMI 1640 medium containinginflammatory protein 10 (IP-10) (200 ng/ml) were added to the bottomwell.Plates were incubated at 37˚C for 2 h. A chemotaxis index was calculated bydividing the number of cells migrating in response to chemokines by thenumber of cells migrating in wells with medium alone.

Statistical analysis

We used the Mann-Whitney statistics or two-tailed unpaired Student t testas indicated. We considered p values ,0.05 as significant.

The Journal of Immunology 3175

by guest on June 2, 2018http://w

ww

.jimm

unol.org/D

ownloaded from

ResultsT-bet and Eomes are required for maintaining CD44+ T cells

Both T-bet and Eomes are expressed in CD8+ T cells and seem tohave overlapping functions (9, 13–16). To study the combinedfunction of T-bet and Eomes in CD8+ T cells, we generatedEomes conditional-deficient mice with CD4-cre transgene (Sup-plemental Fig. 1) and crossed them to TKO mice to generateT/E DKO mice. T/E DKO mice showed normal thymocyte de-velopment (Supplemental Fig. 2) as well as normal proportions ofCD4+ and CD8+ T cells in the periphery in young mice. We thenexamined effector/memory T cell markers and found that CD44+

effector/memory T cells are reduced in both CD4+ andCD8+T cellsfrom T/E DKO mice compared with WT mice (Fig. 1A). Interest-ingly, compared with WT T cells (∼47% CD44+), TKO CD4+

T cells showed a decrease in CD44+ cells (29%) similar to T/EDKOCD4+ T cells (29%), and EKO CD4+ T cells showed a modest de-crease in the CD44+ population (∼35%). In contrast, EKO CD8+

T cells showed a great decrease in the CD44+ population (∼16%),but TKOCD8+ T cells showed no decrease in the CD44+ population(∼35%) compared with WT (∼30%) (Fig. 1A, Supplemental Fig.3A). These data suggest a dichotomy in the function of T-bet andEomes in maintaining CD44+ populations in CD4+ and CD8+

T cells in the peripheral lymphoid system. Consistent with a pre-vious publication (13), we also found that CD122 is reduced inT/E DKO CD8+ T cells (Supplemental Fig. 4).

The diminished number of CD44high cells in the T/E DKO miceprompted us to examine whether T-bet and Eomes are required forthe activation of CD8+ T cells. We isolated naive CD8+ T cellsfrom WT and T/E DKO mice and assessed their capacity to pro-liferate upon activation by anti-CD3 using the CFSE assay. Therewas no difference in proliferation observed between WT andT/E DKO CD8+ T cells after 3 d of culture (Supplemental Fig.5). Therefore, T-bet and Eomes do not seem to affect clonal ex-pansion of CD8+ T cells in vitro.We then examined the effector/memory surface markers on WT

and T/E DKO CD8+ T cells during activation. We found a 2- to 3-fold reduction of CD44 expression in T/E DKO T cells comparedwith WT controls (Fig. 1B, Supplemental Fig. 3B). On the basis ofthese results, it appears that T-bet and Eomes are important forregulating the expression of this T cell activation marker withoutaffecting T cell clonal expansion.

T-bet and Eomes are not essential for the IFN-g production byeffector CD8+ T cells and are required for suppressingalternative CD8+ T cell fates

Both T-bet and Eomes have been implicated in regulating the pro-duction of IFN-g. Reports indicate that T-bet is critical for IFN-gproduction by Th1 cells (17). Overexpression of Eomes enhancedIFN-g production in both CD4+ and CD8+ T cells (9). However,T-bet–deficient CD8+ T cells and WT CD8+ T cells producedsimilar levels of IFN-g (17). The expression of Eomes in CD8+

FIGURE 1. T-bet and Eomes are required for expression of T cell activation marker CD44 and inhibition of alternative CD8+ T cell fates. A, Splenocytes

were harvested from TKO, EKO, T/E DKO, or C57BL/6 WT mice and stained for CD4, CD8, CD44, and L-selectin (CD62L). The percentage of each

population within T cell compartments is indicated. B, Naive CD8+ T cells purified from DKO or WT mice were stimulated with plate-bound anti-CD3 plus

anti-CD28 in Tc0 and Tc1 condition as indicated. The expression of CD44 was analyzed by flow cytometry after 24 or 48 h. Cells cultured in Tc1 condition

(C) or Tc17 condition (D) for 96 h were then stimulated with PMA (10 ng/ml) and ionomycin (1 mg/ml) for 4 h and incubated for the last 3 h with brefeldin

A (10 mg/ml). Cells producing IFN-g and IL-17 were examined with intracellular cytokine staining and flow cytometry. E, Naive CD8+ T cells cultured in

Tc1 condition for 96 h were then stimulated with anti-CD3 for 4 h. IFN-g production was determined by ELISA. pp , 0.01; determined by two-tailed

unpaired Student t test. Data are representative of three independent experiments.

3176 T-bet AND Eomes MEDIATE THE ANTITUMOR T CELL RESPONSES

by guest on June 2, 2018http://w

ww

.jimm

unol.org/D

ownloaded from

T cells has been suspected to be responsible for IFN-g productionin T-bet–deficient CD8+ T cells (9). We therefore decided todeterminewhether T-bet and Eomes together are required for IFN-gproduction in CD8+ T cells. T/E DKO CD8+ T cells and WT con-trols were cultured in Th1 conditions for 4 d and subsequentlyrestimulated. The frequency of CD8+ T cells producing IFN-gwas assayed by flow cytometry. To our surprise, similar numbersof IFN-g producers were observed in WT and EKO effector CD8+

T cells (Fig. 1C). TKO CD8+ T cells and T/E DKO CD8+ T cellsshowed a slight reduction (∼10%) in IFN-g+ cells (Fig. 1C). IFN-gproduction was reduced 2- to 3-fold in T/E DKO CD8+ T cellscompared with WT control as measured by ELISA (Fig. 1E).Therefore, T-bet and Eomes are required but not crucial for IFN-gproduction by CD8+ T cells. Runx3 was shown to be regulated byT-bet in Th1 cells (16). However, Runx-1,2,3 were expressed atsimilar levels between T/E DKO CD8+ T cells and WT controls(our unpublished observations). Therefore, the high-level IFN-gproduction in T/E DKO CD8+ T cells might be driven by Runxfactors. It is likely that additional transcription factors such asSTAT-4 are involved in regulating IFN-g production in effectorCD8+ T cells (16).

Perforin and granzymes are key molecules for the cytolyticfunction of CD8+ T cells. We examined the expression of perforinand granzyme B genes in WT and T/E DKO effector CD8+ T cellsby real-time RT-PCR. Deletion of T-bet or Eomes alone resulted in∼2-fold reduction of perforin and granzyme B mRNA (Supple-mental Fig. 6A). Deletion of both T-bet and Eomes led to a smalladditional decrease (Supplemental Fig. 6A). These data suggestthat T-bet and Eomes are involved in regulating the cytolyticfunction of effector CD8+ T cells. However, there are likely ad-ditional factors involved in regulating cytolytic functions.CD8+ T cells also produce large amounts of cytokines and can

be divided into various subsets based on the cytokines theyproduce. T-bet was shown to antagonize Th17 and Th2 differen-tiation in CD4+ T cells. In addition, T/E DKO mice developedwasting diseases during lymphocytic choriomeningitis virus infec-tion because of the generation of a large number of Tc17 cells(15). We therefore investigated whether T-bet and Eomes affectthe differentiation of CD8+ T cells into IL-17 producers. WhenCD8+ T cells were cultured in Th17-polarizing conditions, ∼10%of the WT CD8+ T cells become Tc17 cells (18). In the same cul-ture condition, ∼2- to 3-fold more T-bet–deficient CD8+ T cells

FIGURE 2. T-bet and Eomes mediate adaptive immune responses against cancer. A, TKO, EKO, DKO, or WT mice were vaccinated with anti-CD25

mAb (350 mg, i.p) 14 and 7 d before B16 challenge, and 53 105 irradiated B16-GM-CSF cells were injected s.c.10 and 7 d before B16 inoculation. On day

0, mice were challenged with 23 105 B16F0 cells i.d. B and C, Tumor growth were monitored every other day. D and E, C57BL/6 WT recipient mice were

lethally irradiated at 1000 rad and then adoptively transferred with different donor bone marrow cells (WT, TKO, EKO, or DKO) within 24 h of irradiation.

Two months later, tumor VAC and tumor inoculation were performed as described above. Tumor sizes were measured every 2 d. WT, donor cells were from

WT mice; T-bet KO, donor cells were from T-bet KO mice; Eomes KO, donor cells were from Eomes KO mice; and DKO, donor cells were from T-bet/

Eomes DKO mice. The data are representative of three separate experiments, and at least six mice in each group were used for each experiment. pp, 0.05;

determined by Mann-Whitney statistics. Comparison was made between WT and DKO. VAC, vaccination.

The Journal of Immunology 3177

by guest on June 2, 2018http://w

ww

.jimm

unol.org/D

ownloaded from

became IL-17 producers. Eomes deletion itself did not greatly af-fect Th17 differentiation. In contrast, ∼3-fold more T/E DKO CD8+

T cells became IL-17 producers when compared with WT controls(Fig. 1D). These data demonstrate that T-bet and Eomes synergisti-cally inhibit Tc17differentiation.Wehave found similar resultswhenCD8+ T cells were cultured in the Tc2 condition. Our results indicatethat ∼2- to 3-fold more IL-4 was induced in T/E DKO CD8+ T cellswhen compared with WT cells (Supplemental Fig. 6B), suggestingthat T-bet and Eomes are also required to inhibit Tc2 differentiation.

T-bet and Eomes mediate adaptive immune responses againstcancer

The presence of tumor-infiltrating T cells has been associated withoverall survival of cancer patients (7). In addition, the presence ofhigh levels of infiltrating memory CD45RO+ cells has been corre-lated with the absence of signs that signified early metastatic in-vasion, a less advanced pathological stage, and increased survival(19). This suggests that the quality of T cells present within tumorsis important for immune surveillance. Moreover, T-bet and Eomeswere shown to be expressed in TILs of colorectal cancers, and thehigher expression levels of these factors are correlated with bettersurvival (7, 20). T-bet–deficient mice are susceptible to prostatecancer metastasis (10). The role of T-bet and Eomes in adaptiveantitumor responses is, however, not known. We therefore decidedto investigate whether T-bet and Eomes are required for adaptiveimmunity against cancer. WT, TKO, EKO, and T/E DKOmice werevaccinated with irradiated B16 melanoma cells expressing GM-CSF.To further boost adaptive immune response to tumor cells, we de-pleted peripheral regulatory T cells by anti-CD25 Abs. We thenchallenged these mice with live B16 cells. In unvaccinated WTmice, the challenged B16 cells quickly grew and formed tumornodules. However, no tumor nodules were observed in WT micethat have been vaccinated (Fig. 2B, 2C). Interestingly, no tumornodules formed in vaccinated T-bet–deficient mice, suggestingT-bet is not critical for antitumor immunity in this experimentalsetting. In contrast, ∼30% of Eomes-deficient mice allowed tumorgrowth despite the vaccination (Fig. 2B, 2C). More strikingly,∼60% of the vaccinated T/E DKO mice grew tumors (Fig. 2B,2C). Therefore, T-bet and Eomes are both required to mediateantitumor responses.To further clarify that the antitumor immunity is mainly medi-

ated by lymphocytes rather than tissue cells, we generated bonemarrow chimeric mice.We transferred bonemarrow cells fromWT,Eomes-deficient, T-bet–deficient, and T/E DKO mice to WTC57BL/6 mice, which were lethally irradiated. The transferredstem cells were able to reconstitute all major immune subsetsregardless of their genotypes after 2–3 mo. We then performedvaccination experiments on these mice. The chimeric micereconstituted with WT and T-bet–deficient bone marrow cellsgenerated effective antitumor responses and denied the growth oftransplanted B16 melanoma cells. In contrast, ∼50% of micereconstituted with Eomes-deficient bone marrow cells succumbedto tumor growth. Moreover, 100% of mice, which were adoptivelytransferred with T/E DKO bone marrow cells, developed tumors(Fig. 2D, 2E). These data further demonstrate that T-bet and Eomesare both required for adaptive immunity against tumor growth.

T-bet and Eomes are required for therapeutic cancervaccination

In a clinical setting, it has long been attempted to elicit an adaptiveimmune attack against established tumors (21). Therapeutic vaccineshave achieved some success in animal models as long as the tumorsize is within a certain limit (21). However, the molecular mecha-

nisms mediating antitumor responses remain unclear. We then de-cided to examine whether T-bet and Eomes are important in thecontext of a therapeutic vaccination. Live B16 melanoma cells wereinoculated i.d., and therapeutic vaccination was started at the sameday of tumor cell inoculation (Fig. 3A). Inoculated B16 melanomacells were able to grow and form nodules in WT mice. We observedthat the growth of tumor was stopped in WT mice around day 15,suggesting that the vaccinewas effective (Fig. 3B). In contrast, tumorgrowth was not inhibited by the vaccine in T/E DKOmice (Fig. 3B).

FIGURE 3. T-bet and Eomes are required for therapeutic cancer vaccina-

tion. A, TKO, EKO, DKO, or WT mice were challenged with 23 105 B16F0

cells i.d. At the same day, treatment was initiated by injecting anti-CD25

(350 mg, i.p.), and the anti-CD25 injection was repeated 7 d later. Treatment

with 5 3 105 irradiated B16-GM-CSF cells was administrated s.c. 4 and 7 d

after the inoculation of live B16 cells. B, Mice were monitored for tumor

growth every other day. At least six mice were used in each group, and data

are representative of three independent experiments. pp, 0.05; determined by

Mann-Whitney statistics. Comparison was made between WT and DKO.

FIGURE 4. T-bet and Eomes are not critical for the generation of systemic

CTL activities. T/E DKO mice or WTmice were vaccinated with anti-CD25

and 5 3 105 irradiated B16-GM-CSF cells as described before. A, Eight

days later, splenocytes were harvested and stained with CD8, B220, and

TRP-2180–188/H-2Kb pentamer. B, Eight days later, in vivo CTL assay was

performed. Data are representative of three independent experiments.

3178 T-bet AND Eomes MEDIATE THE ANTITUMOR T CELL RESPONSES

by guest on June 2, 2018http://w

ww

.jimm

unol.org/D

ownloaded from

Tumor growth rate invaccinated T/EDKOmicewas similar to that inunvaccinatedmice. In contrast, tumor growth rate in vaccinated EKOmice was similar to that in WT control mice. More interestingly,tumor growth rate in TKO mice was even slightly slower than thatinWT control mice, although the data did not reach statistical signif-icance in this experiment (Fig. 3B). Taken together, our data indicatesthat T-bet and Eomes are crucial for the efficacy of a therapeutictumor vaccine.

T-bet and Eomes control migration of antitumor T cells to thetumor site through regulating chemokine receptors

Because our results suggest that T-bet and Eomes are important forthe expression of perforin and granzyme B, we decided to examinewhether the lack of antitumor responses was due to a failure to elicitCTL activities against cancer cells.We first examined the frequencyof antitumor T cells by staining spleen cells with TRP-2180–188/H-2Kb pentamer. We found that the frequency of tumor-specificCD8+ T cells was similar between T/E DKOmice and WT controlsat 8 d after tumor vaccination (Fig. 4A). This is consistent with ourin vitro studies, which showed no difference in clonal expansion ofCTLs between these mice (Supplemental Fig. 5). To determinewhether deletion of T-bet and Eomes affects the generation of func-tional CTLs, in vivo CTL assays were performed (Fig. 4B). Sur-prisingly, we did not detect any difference in CTL activitiesbetween vaccinated T/E DKO mice and WT mice. These data sug-gest that T-bet and Eomes are not critical to generate antitumorCTL activities in the peripheral lymphoid system. We also per-

formed the same experiment 30 d after vaccination. The frequencyof pentamer-positive T cells was below the detection limit. The invivo CTL activities were minimal at this later time point, consistentwith contraction of tumor Ag-specific T cells in all mice.We then examined CD8+ T cells in the tumors obtained from the

therapeutic experiments. We found that many CD8+ T cellsinfiltrated the tumor mass in WT mice (Fig. 5A, 5B). We alsofound that the number of CD8+ T cells within tumor parenchymain TKO mice is similar to that of WT mice (Fig. 5A, 5B). Amodest reduction in tumor-infiltrating CD8+ T cells was observedin tumors from EKO mice compared with WT control mice. Incontrast, a drastically reduced number of CD8+ T cells were foundin the tumors from T/E DKO mice. Besides the imaging analysis,we also used flow cytometry to quantify the leukocytes that wereassociated with tumors in our experiments. The percentage ofleukocytes (CD45+) was similar in tumors isolated from mice ofall genotypes (data not shown). There was a slight decrease inCD4+ T cells in T/E DKO mice but not in EKO or TKO micecompared with WT control mice. In contrast, proportions of tumorassociated CD8+ T cells were reduced in the tumors isolated fromEKO mice (13.6%) compared with WT mice (27%) (Fig. 5C,Supplemental Fig. 7). Tumor-infiltrating CD8+ T cells weregreatly reduced in T/E DKO mice (2.6%) compared with WTmice (27%) (Fig. 5C, Supplemental Fig. 7). Besides a reductionin number, DKO CD8+ TILs produced reduced levels of IFN-gand increased levels of IL-17 (Fig. 5D). No such change wasobserved in TKO or EKO CD8+ TILs (Fig. 5D).

FIGURE 5. T-bet and Eomes control migration and cytokine production in TILs. A, TKO, EKO, DKO, or WT mice were vaccinated with anti-CD25 and

5 3 105 irradiated B16-GM-CSF cells as described before. Tumors were resected around day 20. Sections were stained with anti-mouse CD8. Images were

examined using a fluorescence microscope (original magnification 3400). a, WT; b, TKO; c, EKO; d, DKO. B, Quantification of data from A shows the

number of CD8+ T cells per 600 cells in cancer tissues. C, TILs were harvested from tumor-bearing mice. Cells were then stained with CD45, CD4, and

CD8. The expression of CD4 and CD8 were analyzed by flow cytometry. Data are representative of data obtained from three mice of each group. D,

Analysis of cytokine production; TILs were then stimulated with PMA and ionomycin for 4 h and incubated for the last 3 h with brefeldin A. Cells

producing IFN-g and IL-17 were examined. All plots are gated on CD45+CD8+ cells.

The Journal of Immunology 3179

by guest on June 2, 2018http://w

ww

.jimm

unol.org/D

ownloaded from

One potential explanation for the decrease in tumor-infiltratingCD8+ T cells in T/E DKO mice is a lack of migration of antitumorT cells in these mice. We then examined whether T-bet and Eomesare required for expression of the important Th1 chemokine re-ceptor CXCR3, which was shown to be regulated by T-bet inCD4+ T cells (22). We harvested and analyzed the spleen cellsand found that CXCR3 was expressed on ∼6% CD4+ T cells and14% CD8+ T cells in WT mice (Fig. 6A). In contrast, CXCR3 wasexpressed on ∼3% of EKO CD4+ T cells and was absent inTKO and T/E DKO CD4+ T cells. Its level was normal on TKOCD8+ T cells but dropped to ∼5% in EKO CD8+ T cells and ,1%of DKO CD8+ T cells (Fig. 6A, lower panel). We then examinedCXCR3 expression on in vitro-differentiated Th1 and Tc1 cells.More than 70% of WT Th1 and Tc1 cells expressed CXCR3 (Fig.6B). Both TKO and EKO Th1 and Tc1 cells decreased theirCXCR3 expression. Less than 6% DKO cells expressed CXCR3(Fig. 6B). To determine whether the difference in CXCR3 levels isfunctionally significant, we performed chemotaxis assay. Wefound that WT Th1 and Tc1 cell readily migrated toward IP-10(Fig. 6C). T-bet knockout (KO) Th1 and Tc1 cells showed a mod-est decrease in migration. Similarly, Eomes KO Th1 and Tc1 cellsshowed a small decrease in migration. In contrast, DKO Th1 and

Tc1 cells showed total lack of migration toward IP-10 (Fig. 6C).We also examined the expression of CXCR3 on TILs and foundWT but not DKO CD8+ TILs expressed CXCR3 on their surfaces(Supplemental Fig. 8). Therefore, T-bet and Eomes are both re-quired for CXCR3 expression and function in Th1 and Tc1 cells.To examinewhether CXCR3 plays any role in antitumor immune

responses, we vaccinated CXCR32/2 mice along with WT andT/E DKO mice. Tumor growth was significantly faster inCXCR32/2 mice compared with WT control mice, although it wasslightly slower than that in T/E DKO mice (Fig. 6D). Consistentlywith this result, the proportion of CD8+ TILs was reduced ∼3-foldin CXCR32/2 mice compared with WT controls (Fig. 6E). There-fore, the lack of CXCR3 expression on T cells might contribute tothe lack of antitumor responses in T/E DKO mice.

T-bet and Eomes are critical for T cell function duringantitumor immune responses

T-bet deficiency has been associated with increased metastatic dis-ease in a murine prostate cancer model (10). This lack of immunesurveillance of cancer metastasis was attributed to NK cell defectsin T-bet–deficient mice (11). However, T-bet deficiency did notalter adaptive antitumor responses, likely as a result of the ex-

FIGURE 6. T-bet and Eomes are required for migration of antitumor T cells to the tumor site through regulating CXCR3. A, Splenocytes were harvested

from TKO, EKO, DKO, or WT mice and stained for CD4, CD8, and CXCR3. Data are representative of three independent experiments. B, Naive T cells

purified from TKO, EKO, DKO, or WT mice were cultured in Th1 condition for 96 h and stained for CD4, CD8, and CXCR3. CXCR3 expression on CD4+

T cells of each group is 79.7% (WT), 20.4% (TKO), 33.5% (EKO), and 5.5% (DKO), respectively. CXCR3 expression on CD8+ T cells of each group is

78.4% (WT), 20.5% (TKO), 19.7% (EKO), and 3.9% (DKO), respectively. Data are representative of three independent experiments. C, Naive T cells

purified from TKO, EKO, DKO, or WT mice were stimulated with plate-bound anti-CD28 plus anti-CD3 in Th1 condition. Then live cells were placed

in the upper wells of Transwell inserts, and their chemotactic responses to IP-10 (200 ng/ml) were determined. Bars represent means of three samples 6SEM. pp , 0.05; ppp , 0.001; determined by two-tailed unpaired Student t test. Two independent experiments were performed with similar results. D,

CXCR32/2, T/E DKO, or WT mice were challenged and treated as described in Fig. 3. Mice were monitored for tumor growth every other day. pp , 0.05;

determined by Mann-Whitney statistics. Comparison made between WT and CXCR32/2. E, TILs were harvested from tumor-bearing mice. The expression

of CD8 was analyzed by flow cytometry.

3180 T-bet AND Eomes MEDIATE THE ANTITUMOR T CELL RESPONSES

by guest on June 2, 2018http://w

ww

.jimm

unol.org/D

ownloaded from

pression of Eomes in T cells (11). Because T-bet is also expressedin NK cells, DCs and macrophages (11, 23–25), we used an adop-tive transfer system to determine whether T-bet and Eomes expres-sion in T cells is required for adaptive antitumor immune responses.T cells from WT and T/E DKO mice were isolated and mixed withWT T cell-depleted splenocytes. The mixed cells were then trans-ferred to nonlethally irradiated mice. The reconstituted mice had animmune system composed mainly of normal non-T cells and eitherWT or T/E DKO T cells. These mice were then subjected to ther-apeutic vaccination as in Fig. 3. Tumor growth was delayed andeventually controlled by vaccination inWT control mice (Fig. 7). Incontrast, tumor growth in mice reconstituted with T/E DKOT cellsand normal T cell-depleted splenocytes was nonstoppable and wassimilar to that in irradiated mice without any transferred cells (Fig.7). Therefore, the function of T-bet and Eomes is T cell intrinsic.

DiscussionIn this study, we used TKO, EKO, and Tbet/Eomes doubly deficientmice to examine the importance of each of these transcription factorsin T cell-mediated antitumor immune responses. T-bet and Eomessynergistically regulate multiple steps of Th1/Tc1 differentiationfrom activation to tissue trafficking. Taken together, these transcrip-tion factors take a two-handed approach. On the one hand, they arecritical in turning on genes with important functions in cell-mediatedimmune responses such as effector/memory marker CD44, antitu-mor cytokine IFN-g, and chemokine receptors that allow T cellsto migrate to cancer tissues. On the other hand, these transcriptionfactors also prevent CD8+ T cell differentiation to alternative sub-sets such as Tc17 and Tc2. The lack of both T-bet and Eomes inT cells severely perturbed these mechanisms, resulting in profounddefects in antitumor immunity and dysregulated T cell differentia-tion. Moreover, we showed that the regulation of antitumor immuneresponses by T-bet and Eomes is T cell intrinsic. Collectively, wehave identified the Th1-associated transcription factors T-bet andEomes as key regulators of T cell-mediated antitumor immuneresponses.Interestingly, we have found that antitumor CTL activities are not

significantly affected in the peripheral lymphoid organs in T/E DKOmice. These data are consistent with our in vitro finding that levels

of perforin and granzymes were only modestly reduced in T/EDKO T cells compared with WT. The frequency of anticancerCD8+ T cells was also similar between WT and T/E DKO mice,which aligns with our in vitro data that T-bet and Eomes are notrequired for clonal expansion. These data suggest that the antitu-mor CTL activity on a per-cell basis is not significantly differentbetween the vaccinated WT and DKO mice. Other transcriptionfactors such as STAT-4 and Runx-1,2,3 might be involved inregulating CTL activities in effector CD8+ T cells in the periphery(16). In CD8+ T cells, the expression levels of these factors arenot regulated by T-bet and Eomes. Whether or not these transcrip-tion factors synergize with T-bet and Eomes in regulating CTLactivities needs to be further studied. In addition, whether theCTL activities are affected within tumor tissues remains to beexplored.Although T-bet and Eomes were not required for systemic anti-

tumor activities, most vaccinated DKO mice still allowed thegrowth of s.c. challenged tumor cells. The deficiency in antitumorimmune responses seems to be attributed to the lack of tumorinfiltration by T cells. We have further shown that chemokine re-ceptor CXCR3 is critically regulated by T-bet and Eomes. In ad-dition, CXCR32/2 mice showed compromised antitumor immuneresponses. Therefore, T-bet and Eomes seem to be important forT cell migration and retention in the tumor site. This might bepartly mediated by their regulation of CXCR3. This finding issignificant in light of the fact that many tumor vaccines have failedto improve survival despite data showing that they have elicitedimmune responses in cancer patients. This is possibly becauseimmune responses have been measured by CTL and cytokineassays done on peripheral blood cells. Our results challenge theutility of these assays, because the CTL activities measured in theblood do not tell us whether antitumor T cells are able to infiltrateand destroy tumors. Therefore, further study on the T/E-regulatedgenes in tumor Ag-specific T cells might provide useful bio-markers for monitoring tumor vaccine trials.CXCR3 was shown to be regulated by Eomes in thymic CD8+

T cells when IL-4 is overexpressed as a result of deletion of KLF2(26). Whether Eomes regulates CXCR3 in peripheral CD8+ T cellsand whether it is essential for CXCR3 expression are still notknown. T-bet was also shown to regulate T cell migration to heartduring inflammation partly via its regulation of CXCR3 on CD4+

T cells (22). The role of T-bet in CD8+ T cell migration is not clear,although it has been shown that T-bet positively regulates CXCR3in CD8+ T cells (27). In addition, the tumor microenvironment isknown for its unique immune suppressive characteristic, whichdiffers greatly from autoimmune and infection-driven inflamma-tory environment. Therefore, it is important to determine whetherT-bet and Eomes affect T cell migration to tumor sites. Indeed, weshowed that T-bet and Eomes are individually dispensable forT cell migration to tumor. But deletion of both T-bet and Eomesrenders CD8+ T cell unable to infiltrate tumors. This finding cor-related with our data showing that T-bet and Eomes are crucial forCXCR3 expression on the surface of CD8+ T cells. In addition, weshowed that CXCR32/2 mice were partially compromised in theirantitumor immune responses. Besides CXCR3, additional factorsmight further contribute to the lack of CD8+ TILs in T/E DKOmice. For example, it has been reported that T-bet regulates celladhesion in CD4+ T cells through tyrosine sulfation of P-selectinglycoprotein ligand 1. It was further attributed to regulation ofmRNA levels of Tpst2 by T-bet in Th1 cells (22). However, nodifference in the expression of Tpst1 or Tpst2 was detected be-tween WT and T/E DKO CD8+ T cells (data not shown). It is alsopossible that the survival of TILs is affected, because it has beenshown that T-bet make cell survive better because of its regulation

FIGURE 7. Requirement of T-bet and Eomes expression in T cells for

anti-tumor immune responses. B6.SJLWT recipient mice were irradiated at

600 rad. A total of 2 3 107 purified T cells from the spleen of WT or DKO

littermates were mixed with 6 3 107 T cell-depleted splenocytes harvested

from C57BL/6 WT mice, and then the mixed cells were transferred i.v. to

the irradiated recipient mice within 24 h of irradiation. Mice were then

challenged and treated as described in Fig. 3 1 d after the adoptive transfer.

Tumor growth was monitored every other day. Data are the mean 6 SE of

five mice in each group. Data are representative of two separate experi-

ments with similar results. Comparison was made between WT and DKO.

pp , 0.05; determined by Mann-Whitney statistics.

The Journal of Immunology 3181

by guest on June 2, 2018http://w

ww

.jimm

unol.org/D

ownloaded from

of CD122 (13, 28). We have shown that T-bet and Eomes regulateCD44. The lack of CD44 was shown to reduce migration of TILswithin tumor (29). These possibilities and others remain to be tes-ted. Regardless, our data provide a definitive answer that T-bet andEomes are required for the presence of CD8+ T cells in tumor.Despite many data support an anticancer role of T-bet and Eomes,

there is a report that T-bet is involved in the progression of carci-noma in Helicobacter felis-induced gastric cancer (30). This mightbe because H. felis-induced gastric cancer progression is associatedwith Th1-type inflammatory response (30). T-bet likely plays a rolein tissue damages that lead to this cancer. Therefore, different typesof cancer are associated with and promoted by different types ofinflammation. The role of T-bet and Eomes in a cancer is dictatedby the type of inflammation that is associated with that particulartype of cancer. Indeed, T-bet deficiency in dendritic cells leads tocolitis-associated colorectal cancer (25), supporting its role in sup-pressing this cancer. In contrast, T-bet deficiency inhibited Th1-driven inflammatory bowel disease (31), and we would speculatethat T-bet deficiency likely will prevent cancer progression drivenby this Th1-mediated colitis.Previous studies have suggested a minimal role for T-bet in adap-

tive immunity against cancer (11). The lack of impact of T-betdeletion on adaptive immune responses against cancer is likelybecause Eomes is also expressed in CD8+ T cells and Th1 cells.Using T/E DKO mice, we have found that T-bet and Eomes areindeed required for the effector phase of the adaptive immunityagainst cancer. It has also been reported that IL-21 inhibits Eomesexpression and is potent in adoptive T cell therapy against cancer(32). Shutting down Eomes during cell culture is important toachieve survival of transferred T cells, likely because Eomesand T-bet might drive more apoptosis in cultured T cells (our un-published observation). However, our study demonstrates that theexpression of T-bet and Eomes is important to allow antitumorT cells to eventually reach and be retained in the tumor site andfulfill their anticancer mission. IL-12 is a hallmark cytokine forinducing Th1 responses and was shown to change the tumor mi-croenvironment to favor tumor surveillance and eradiation. Para-doxically, IL-12 suppresses Eomes expression in CD8+ T cells(14, 26). Our data clearly established that T-bet and Eomes aresynergistic in eliminating cancer cells. Therefore, Eomes may usesignals other than IL-12, such as IL-2, IL-15, or IL-4 (32), toenhance antitumor responses.We have shown in this study that in optimal Th1 culture condi-

tion, the frequency of IFN-g producers was modestly affectedafter both T-bet and Eomes were deleted in CD8+ T cells. Runx3was shown to be regulated by T-bet in Th1 cells (33). However, inCD8+ T cells, we have found it is not regulated by T-bet or Eomes(data not shown). Therefore, Runx3 and perhaps Stat-4 are allalso involved in regulating Th1 phenotype in CD8+ T cells (11).Regardless, IFN-g production in vivo is clearly reduced in tumorsites. Interestingly, previous studies have shown that during infec-tion, there is a clear lack of development of IFN-g–producingCD8+ T cells (9, 17). The reason for this difference might bebecause during infection, the T/E DKO CD8+ T cells predom-inantly differentiate to Tc17 cells (15). Therefore, T-bet and Eomescontrol balance of inflammation during immune responses.We have shown here that T-bet and Eomes also regulate mem-

ory cells marker CD44. The T/E DKO mice showed a reduction ofCD44+ effector/memory T cells. Therefore, it is possible that T-betand Eomes are important for the generation of memory T cells. This isin line with a prior study, which showed that T-bet and Eomes arerequired for the expression of CD122 (34), an IL-15R that is importantfor the homeostasis of memory T cells. Our study further showed thatCD1222CD44+ memory T cells are also regulated by T/E. Therefore,

T-bet and Eomes might regulate effector/memory pools by CD122-independent mechanisms. It has been proposed that optimal T-betexpression by CTLs promotes the generation of long-term mem-ory T cells, but high levels of T-bet drives T cell differentiation toshort-lived effector cells (35). Our data demonstrate that T-bet isdispensable for the effector function of antitumor immune re-sponses. We further showed that the efficacy of tumor vaccinationis dependent on both T-bet and Eomes. Our data clearly estab-lished that both T-bet and Eomes are required for the effector stageof T cell responses against tumor. Whether T-bet, Eomes, or bothare required for the generation of central and effector memorycells will be the focus for future studies.

AcknowledgmentsWe thank Dr. Jeffrey Kovacs and Christen Shiber for editing the manuscript,

Dr. S. Watkins for fluorescence imaging, and Drs. Pamela Beatty and Oli-

vera Finn for reading the manuscript and providing helpful suggestions.

DisclosuresThe authors have no financial conflicts of interest.

References1. Karin, M. 2006. Nuclear factor-kB in cancer development and progression.

Nature 441: 431–436.2. Girardi, M., D. E. Oppenheim, C. R. Steele, J. M. Lewis, E. Glusac, R. Filler,

P. Hobby, B. Sutton, R. E. Tigelaar, and A. C. Hayday. 2001. Regulation ofcutaneous malignancy by gd T cells. Science 294: 605–609.

3. Finn, O. J. 2003. Cancer vaccines: between the idea and the reality. Nat. Rev.Immunol. 3: 630–641.

4. Dunn, G. P., C. M. Koebel, and R. D. Schreiber. 2006. Interferons, immunity andcancer immunoediting. Nat. Rev. Immunol. 6: 836–848.

5. Vesalainen, S., P. Lipponen, M. Talja, and K. Syrjanen. 1994. Histological grade,perineural infiltration, tumour-infiltrating lymphocytes and apoptosis as determi-nants of long-term prognosis in prostatic adenocarcinoma. Eur. J. Cancer 30A:1797–1803.

6. Schumacher, K., W. Haensch, C. Roefzaad, and P. M. Schlag. 2001. Prognosticsignificance of activated CD8+ T cell infiltrations within esophageal carcinomas.Cancer Res. 61: 3932–3936.

7. Galon, J., A. Costes, F. Sanchez-Cabo, A. Kirilovsky, B. Mlecnik, C. Lagorce-Pages, M. Tosolini, M. Camus, A. Berger, P. Wind, et al. 2006. Type, density, andlocation of immune cells within human colorectal tumors predict clinicaloutcome. Science 313: 1960–1964.

8. Dieu-Nosjean, M. C., M. Antoine, C. Danel, D. Heudes, M. Wislez, V. Poulot,N. Rabbe, L. Laurans, E. Tartour, L. de Chaisemartin, et al. 2008. Long-termsurvival for patients with non-small-cell lung cancer with intratumoral lymphoidstructures. J. Clin. Oncol. 26: 4410–4417.

9. Pearce, E. L., A. C. Mullen, G. A. Martins, C. M. Krawczyk, A. S. Hutchins,V. P. Zediak, M. Banica, C. B. DiCioccio, D. A. Gross, C. A. Mao, et al. 2003.Control of effector CD8+ T cell function by the transcription factor Eomeso-dermin. Science 302: 1041–1043.

10. Peng, S. L., M. J. Townsend, J. L. Hecht, I. A. White, and L. H. Glimcher. 2004.T-bet regulates metastasis rate in a murine model of primary prostate cancer.Cancer Res. 64: 452–455.

11. Werneck, M. B., G. Lugo-Villarino, E. S. Hwang, H. Cantor, and L. H. Glimcher.2008. T-bet plays a key role in NK-mediated control of melanoma metastaticdisease. J. Immunol. 180: 8004–8010.

12. Liu, P., N. A. Jenkins, and N. G. Copeland. 2003. A highly efficient recom-bineering-based method for generating conditional knockout mutations. GenomeRes. 13: 476–484.

13. Intlekofer, A. M., N. Takemoto, E. J. Wherry, S. A. Longworth, J. T. Northrup,V. R. Palanivel, A. C. Mullen, C. R. Gasink, S. M. Kaech, J. D. Miller, et al.2005. Effector and memory CD8+ T cell fate coupled by T-bet and eomesoder-min. Nat. Immunol. 6: 1236–1244.

14. Takemoto, N., A. M. Intlekofer, J. T. Northrup, E. J. Wherry, and S. L. Reiner.2006. Cutting edge: IL-12 inversely regulates T-bet and eomesodermin expres-sion during pathogen-induced CD8+ T cell differentiation. J. Immunol. 177:7515–7519.

15. Intlekofer, A. M., A. Banerjee, N. Takemoto, S. M. Gordon, C. S. Dejong,H. Shin, C. A. Hunter, E. J. Wherry, T. Lindsten, and S. L. Reiner. 2008.Anomalous type 17 response to viral infection by CD8+ T cells lacking T-bet andeomesodermin. Science 321: 408–411.

16. Cruz-Guilloty, F., M. E. Pipkin, I. M. Djuretic, D. Levanon, J. Lotem,M. G. Lichtenheld, Y. Groner, and A. Rao. 2009. Runx3 and T-box proteinscooperate to establish the transcriptional program of effector CTLs. J. Exp. Med.206: 51–59.

17. Szabo, S. J., B. M. Sullivan, C. Stemmann, A. R. Satoskar, B. P. Sleckman, andL. H. Glimcher. 2002. Distinct effects of T-bet in TH1 lineage commitment andIFN-g production in CD4 and CD8 T cells. Science 295: 338–342.

3182 T-bet AND Eomes MEDIATE THE ANTITUMOR T CELL RESPONSES

by guest on June 2, 2018http://w

ww

.jimm

unol.org/D

ownloaded from

18. Liu, S. J., J. P. Tsai, C. R. Shen, Y. P. Sher, C. L. Hsieh, Y. C. Yeh, A. H. Chou,S. R. Chang, K. N. Hsiao, F. W. Yu, and H. W. Chen. 2007. Induction of a dis-tinct CD8 Tnc17 subset by transforming growth factor-b and interleukin-6. J.Leukoc. Biol. 82: 354–360.

19. Pages, F., A. Berger, M. Camus, F. Sanchez-Cabo, A. Costes, R. Molidor,B. Mlecnik, A. Kirilovsky, M. Nilsson, D. Damotte, et al. 2005. Effectormemory T cells, early metastasis, and survival in colorectal cancer. N. Engl. J.Med. 353: 2654–2666.

20. Atreya, I., C. C. Schimanski, C. Becker, S. Wirtz, H. Dornhoff, E. Schnurer,M. R. Berger, P. R. Galle, W. Herr, and M. F. Neurath. 2007. The T-box tran-scription factor eomesodermin controls CD8 T cell activity and lymph nodemetastasis in human colorectal cancer. Gut 56: 1572–1578.

21. Schreiber, K., D. A. Rowley, G. Riethmuller, and H. Schreiber. 2006. Cancerimmunotherapy and preclinical studies: why we are not wasting our time withanimal experiments. Hematol. Oncol. Clin. North Am. 20: 567–584.

22. Lord, G. M., R. M. Rao, H. Choe, B. M. Sullivan, A. H. Lichtman,F. W. Luscinskas, and L. H. Glimcher. 2005. T-bet is required for optimalproinflammatory CD4+ T-cell trafficking. Blood 106: 3432–3439.

23. Wang, J., J. W. Fathman, G. Lugo-Villarino, L. Scimone, U. von Andrian,D. M. Dorfman, and L. H. Glimcher. 2006. Transcription factor T-bet regulatesinflammatory arthritis through its function in dendritic cells. J. Clin. Invest. 116:414–421.

24. Glimcher, L. H. 2007. Trawling for treasure: tales of T-bet. Nat. Immunol. 8:448–450.

25. Garrett, W. S., S. Punit, C. A. Gallini, M. Michaud, D. Zhang, K. S. Sigrist,G. M. Lord, J. N. Glickman, and L. H. Glimcher. 2009. Colitis-associated co-lorectal cancer driven by T-bet deficiency in dendritic cells. Cancer Cell 16:208–219.

26. Weinreich, M. A., K. Takada, C. Skon, S. L. Reiner, S. C. Jameson, andK. A. Hogquist. 2009. KLF2 transcription-factor deficiency in T cells results inunrestrained cytokine production and upregulation of bystander chemokinereceptors. Immunity 31: 122–130.

27. Beima, K. M., M. M. Miazgowicz, M. D. Lewis, P. S. Yan, T. H. Huang, andA. S. Weinmann. 2006. T-bet binding to newly identified target gene promotersis cell type-independent but results in variable context-dependent functionaleffects. J. Biol. Chem. 281: 11992–12000.

28. Li, M. O., S. Sanjabi, and R. A. Flavell. 2006. Transforming growth factor-bcontrols development, homeostasis, and tolerance of T cells by regulatory T cell-dependent and -independent mechanisms. Immunity 25: 455–471.

29. Mrass, P., I. Kinjyo, L. G. Ng, S. L. Reiner, E. Pure, and W. Weninger. 2008.CD44 mediates successful interstitial navigation by killer T cells and enablesefficient antitumor immunity. Immunity 29: 971–985.

30. Stoicov, C., X. Fan, J. H. Liu, G. Bowen, M. Whary, E. Kurt-Jones, andJ. Houghton. 2009. T-bet knockout prevents Helicobacter felis-induced gastriccancer. J. Immunol. 183: 642–649.

31. Neurath, M. F., B. Weigmann, S. Finotto, J. Glickman, E. Nieuwenhuis,H. Iijima, A. Mizoguchi, E. Mizoguchi, J. Mudter, P. R. Galle, et al. 2002. Thetranscription factor T-bet regulates mucosal T cell activation in experimentalcolitis and Crohn’s disease. J. Exp. Med. 195: 1129–1143.

32. Hinrichs, C. S., R. Spolski, C. M. Paulos, L. Gattinoni, K. W. Kerstann,D. C. Palmer, C. A. Klebanoff, S. A. Rosenberg, W. J. Leonard, andN. P. Restifo. 2008. IL-2 and IL-21 confer opposing differentiation programs toCD8+ T cells for adoptive immunotherapy. Blood 111: 5326–5333.

33. Djuretic, I. M., D. Levanon, V. Negreanu, Y. Groner, A. Rao, and K. M. Ansel.2007. Transcription factors T-bet and Runx3 cooperate to activate Ifng andsilence Il4 in T helper type 1 cells. Nat. Immunol. 8: 145–153.

34. Intlekofer, A. M., N. Takemoto, C. Kao, A. Banerjee, F. Schambach,J. K. Northrop, H. Shen, E. J. Wherry, and S. L. Reiner. 2007. Requirement for T-bet in the aberrant differentiation of unhelped memory CD8+ T cells. J. Exp.Med. 204: 2015–2021.

35. Joshi, N. S., W. Cui, A. Chandele, H. K. Lee, D. R. Urso, J. Hagman, L. Gapin,and S. M. Kaech. 2007. Inflammation directs memory precursor and short-livedeffector CD8+ T cell fates via the graded expression of T-bet transcription factor.Immunity 27: 281–295.

The Journal of Immunology 3183

by guest on June 2, 2018http://w

ww

.jimm

unol.org/D

ownloaded from