Embed Size (px)
Citation preview
of April 24, 2018.This information is current as
Cells over a Critical Thresholdupon Recognition of a Number of Target Rapid Deletion and Inactivation of CTLs
VilladangosSandro Prato, Yifan Zhan, Justine D. Mintern and Jose A.
http://www.jimmunol.org/content/191/7/3534doi: 10.4049/jimmunol.1300803September 2013;
2013; 191:3534-3544; Prepublished online 9J Immunol
MaterialSupplementary
3.DC1http://www.jimmunol.org/content/suppl/2013/09/09/jimmunol.130080
Referenceshttp://www.jimmunol.org/content/191/7/3534.full#ref-list-1
, 27 of which you can access for free at: cites 57 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 © 2013 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 April 24, 2018
http://ww
w.jim
munol.org/
Dow
nloaded from
by guest on April 24, 2018
http://ww
w.jim
munol.org/
Dow
nloaded from
The Journal of Immunology
Rapid Deletion and Inactivation of CTLs upon Recognition ofa Number of Target Cells over a Critical Threshold
Sandro Prato,*,†,‡ Yifan Zhan,* Justine D. Mintern,*,x and Jose A. Villadangos*,‡,x
Initiation of CTL responses against foreign pathogens also primes anti-self CTLs. Mechanisms of CTL inactivation inhibit anti-self
CTLs to prevent tissue damage. These mechanisms are exploited by pathogens and tumors to evade the immune response, and
present a major hurdle to adoptive CTL therapies. It is unclear whether CTL inactivation is Ag specific and, if so, which APCs
are involved. Potential candidates include the target cells themselves, dendritic cells, myeloid-derived suppressor cells, and macro-
phages. In this study, we show that lymphoma-specific CTLs are rapidly deleted in an Ag-specific manner after adoptive transfer
into lymphoma-bearing mice, and the surviving CTLs are functionally impaired. The only APCs responsible were the target cells
directly presenting Ag, notwithstanding the presence of myeloid-derived suppressor cells, and CD8+ dendritic cells cross-presenting
tumor Ag. The capacity to inactivate CTLs critically depended on the number of tumor/target cells; small numbers of targets were
readily killed, but a large number caused quick deletion and functional inactivation of the CTLs. Application of mild, nonin-
flammatory, and nonlymphoablative chemotherapy to specifically reduce tumor burden before CTL injection prevented CTL
deletion and inactivation and allowed eradication of tumor. Our results advocate the use of adoptive CTL therapy soon after mild
chemotherapy. They also suggest a simple mechanism for Ag-specific impairment of anti-self CTLs in the face of an active anti-
foreign CTL response. The Journal of Immunology, 2013, 191: 3534–3544.
Priming of naive CD8+ T cells by dendritic cells (DCs)presenting pathogen or tumor Ags in lymphoid organsleads to differentiation of Ag-specific CTLs, which rapidly
expand, access sites of infection or malignancy, and kill cellspresenting their cognate Ags (1). DCs can activate bystander anti-self CD8+ T cells during priming of foreign Ag-specific CTLs.Several processes that inactivate anti-self CTLs have been de-scribed: anergy, exhaustion, and senescence (2–4). Notably, tumorsand certain pathogens exploit CTL inactivation to escape theimmune system (3). Considerable progress has been made in char-acterizing the intrinsic molecular programs resulting in CTL in-activation, in particular those causing exhaustion (2). Receptorshave been identified that, when targeted with Abs, prevent ex-haustion (5–8). A limitation of this strategy is that it may benecessary to target multiple molecules simultaneously. An alter-native approach is to identify the cell(s) that cause CTL inacti-
vation and remove them by selective depletion. However, theidentity and features of the APCs that mediate this process remainpoorly understood. Macrophages, DCs, CD11b+GR1+ myeloid-derived suppressor cells (MDSCs), and the target (pathogen-infected or tumor) cells are all potential candidates (9, 10). In-deed, it is still unclear whether CTL inactivation occurs as a resultof recognition of cognate Ag or is mediated by soluble factorsacting in a non-Ag–specific fashion.Characterizing the cellular players involved in CTL inactivation
would benefit the development of successful adoptive T celltherapies for the treatment of chronic infections or cancer. Thisstrategy is predicated on the basis that pathogen-infected or tumorcells can be specifically targeted in vivo with high-avidity CTLs. Inthe past, the paucity of high-avidity anti-tumor CTLs has limitedthe clinical use of this therapy against cancer, but it is now possibleto obtain massive numbers of high-avidity CTLs in vitro byexpanding tumor-infiltrating lymphocytes or producing T cellsexpressing genetically engineered TCRs (11–14). Recent clinicaltrials demonstrated the potential of this approach to treat solidtumors, lymphoma, and systemic infections (13, 15, 16). Unfor-tunately, transferred CTLs are often inactivated by mechanismsthat resemble those observed during chronic viral infections,failing to expand or survive, with loss of effector function (3, 13).Finding approaches to overcome CTL inactivation may enable thefull potential of adoptive T cell therapy to be realized.In this study, we exploited a mouse model of non-Hodgkin
lymphoma to dissect the role of distinct APCs in CTL inactiva-tion. We identify rapid deletion and inactivation of the CTL asa major mechanism impairing antilymphoma CTLs. Furthermore,direct Ag recognition on a number of target cells over a minimumthreshold, rather than acquisition of inactivating features by thosecells, triggered CTL inactivation. The practical implications wereexplored with mild, noninflammatory, nonlymphoablative che-motherapy that reduced tumor burden. Chemotherapy affordedlittle benefit on its own, but it reduced the number of tumor cellsbelow the CTL inactivation threshold so that injection of CTL1 d after chemotherapy resulted in highly efficacious tumor
*Department of Inflammation, The Walter and Eliza Hall Institute of Medical Re-search, Parkville, Victoria 3052, Australia; †Department of Medical Biology, Univer-sity of Melbourne, Parkville, Victoria 3050, Australia; ‡Department of Microbiologyand Immunology, University of Melbourne, Parkville, Victoria 3010, Australia; andxDepartment of Biochemistry and Molecular Biology, University of Melbourne, Park-ville, Victoria 3010, Australia
Received for publication March 26, 2013. Accepted for publication July 30, 2013.
This work was supported by the National Health and Medical Research Council ofAustralia, the University of Melbourne, and the Leukemia and Lymphoma Society. J.A.V.is a National Health and Medical Research Council of Australia Senior Research Fellow,and J.D.M. is a National Health and Medical Research Council of Australia CareerDevelopment Fellow.
Address correspondence and reprint requests to Dr. Sandro Prato, Dr. Justine Min-tern, and Prof. Jose A. Villadangos, University of Melbourne, Gate 11 Royal Parade,Parkville, Victoria 3010, Australia. E-mail addresses: [email protected] (S.P.),[email protected] (J.D.M.), and [email protected] (J.A.V.)
The online version of this article contains supplemental material.
Abbreviations used in this article: cDC, conventional dendritic cell; CTX, cyclophos-phamide; DC, dendritic cell; MDSC, myeloid-derived suppressor cell; MHC II, MHCclass II; PD-1, programmed cell death 1; pDC, plasmacytoid DC; PD-L1, programmedcell death ligand 1; WT, wild type.
Copyright� 2013 by TheAmericanAssociation of Immunologists, Inc. 0022-1767/13/$16.00
www.jimmunol.org/cgi/doi/10.4049/jimmunol.1300803
by guest on April 24, 2018
http://ww
w.jim
munol.org/
Dow
nloaded from
elimination. Therefore, selective reduction of target cell numbersmay be a relatively simple, but effective, strategy to maintain CTLactivity against tumors and chronic infections.
Materials and MethodsMice
Mice were maintained and bred under specific pathogen–free conditions inthe Walter and Eliza Hall Institute and Bio21 Institute animal facilities.C57BL/6J (Ly5.2), C57BL/6-Pep3b (Ly5.1), F1 (Ly5.1 3 B6.CH-2bm1
[bm1]), MHC I2/2 (17), IFN-g2/2 3 OT-I (18, 19), B6.Batf32/2 (20),Ly5.13 OT-I (19), and gBT-I (21) mice were used between 6 and 12 wk ofage; experimentation was performed in accordance with the InstitutionalAnimal Care and Use Committee guidelines at the Walter and Eliza HallInstitute and the University of Melbourne.
Em-myc lymphomas
Generation, i.v. inoculation, and growth of the Em-myc-GFP and Em-myc-OVA lymphomas have been described (27).
In vitro activation of OT-I and gBT-I
Activated OT-I (wild type [WT] or IFN-g2/2) or gBT-I cells were gen-erated as described (22). Cultures typically contained 94–98% CD8+Va2+
cells.
CTL analysis following transfer into Em-myc lymphoma-bearing mice
Ly5.1 3 gBT-I CTLs (5 3 106) and/or Ly5.1 3 OT-I CTLs (5 3 106) orIFN-g2/2 OT-I CTLs were injected i.v. and spleens or lymph nodes wereharvested at the indicated times for analysis. Organs were disruptedthrough a 70-mm cell strainer (BD Falcon). Cell suspensions were treatedwith red cell removal buffer, washed, and resuspended in an Ab mixturecontaining fluorochrome-bound anti-CD8 (YTS 169.4), B220 (RA3-6B2),and anti-Va2 (B20.1) Abs (all produced in-house), Ly5.1 (A20.1; eBio-science), programmed cell death 1 ([PD-1]; J43; eBioscience), or pro-grammed cell death ligand 1 ([PD-L1]; M1H5; eBioscience). Cells wereanalyzed by flow cytometry. Live PI2Ly5.1+CD8+ cells or PI2FSChighB220+
GFP+ lymphoma cells were analyzed phenotypically and enumerated.
In vivo CTL killing assay
Tumor-bearing mice were injected i.v. with 1 3 107 CFSEhigh-labeledOVA257–264 peptide (Auspep, Melbourne, Australia) -pulsed splenocytesand CFSElow-labeled control unpulsed splenocytes in equal ratios, 2 d afterCTL inoculation. Killing by gBT-I CTLs was assessed with CFSEhigh-la-beled gB498–505 peptide–pulsed splenocytes. Spleens were analyzed byflow cytometry 18–24 h later. Where indicated, splenocytes harvested fromtumor-bearing mice were labeled with 2.5 mM CellTrace Violet (Invi-trogen), and the equivalent of 4 3 106 tumor cells were injected i.v. intonaive mice. One day later, 53 106 OT-I CTLs were inoculated i.v., and thekilling of Violet-labeled target cells was determined 2 d later.
In vitro CTL killing assay
Em-myc-GFP and Em-myc-OVA tumor-bearing mice were injected withLy5.1 3 OT-I CTLs 5 d after tumor inoculation. Two days later, spleensfrom Em-myc-GFP or Em-myc-OVA tumor-bearing mice were pooled, andCD8+ T cells were enriched using a mixture of Abs to deplete cellsexpressing CD19 (1D3), MHC class II (MHC II; M5/114), F4/80, Gr-1(RB6-8C5), and CD4 (GK 1.5), and OT-I CTLs (Ly5.1+CD8+) were pu-rified by flow cytometry. OT-I CTLs (5 3 104) were cocultured with pu-rified Em-myc-GFP (5 3 104) and Em-myc-OVA (5 3 104) target cells,and the killing of Em-myc-OVA cells was analyzed by flow cytometry 22–24 h later.
Intracellular cytokine-staining assay
Splenocytes were stimulated with 1 mg/ml OVA257–264 peptide for 5 h inmedium containing 1 mg/ml GolgiPlug (BD Biosciences). Cells werestained with Abs against CD8, and the percentages of IFN-g– and TNF-a–secreting cells among total CD8+ T cells were measured by flow cytometry(23).
Spleen and lymph node DC phenotype
DCs were enriched from spleens or lymph nodes, as previously described(24, 25). Because of the enlarged spleens of lymphoma-bearing mice, 5 mlNycodenz/spleen was used, followed by two rounds of density-gradient
centrifugation. Cells were washed in BSS-EDTA-FCS 2% and stained withanti-CD11c (N418), anti-CD8 (YTS 169.4), anti-CD205 (NLDC-145) andactivation markers anti-MHC II (M5/114), anti-CD69 (H1.2F3), and anti-CD86 (GL-1) (all produced in-house). DCs were then analyzed by flowcytometry. Where indicated, DCs (0.5 3 106 cells/well) purified fromspleen of naive or mice bearing lymphoma for 7 d were maintainedovernight at 4˚C (fresh) or cultured at 37˚C in the absence or presence ofLPS (0.5 mg/ml). DCs were then washed and stained with anti-CD11c andthe activation markers anti–MHC II or anti-CD86 and analyzed by flowcytometry.
Phenotype of CD11b+Gr-1+ MDSCs
Spleens from lymphoma-bearing Ly5.1 mice were digested in mediumcontaining 7 mg collagenase (Type III; Worthington Biochemicals) and1 mg DNase (Boehringer-Mannheim). Subsequently, splenocytes werewashed and analyzed by flow cytometry using Abs against CD19 (1D3),CD3 (KT3-1.1), Ly5.2 (S450-15.2), CD11b (M1/70), CD11c (N418), Gr-1(RB6-8C5), CD43 (S7), F4/80, Ly6C (5075-3.6), Mac3 (M3/84.6.34),CD115 (AFS98), MHC II (M5/114), and CD86 (GL-1) (all produced in-house). The FL1 fluorescence channel was used as a “dump” channelwhere GFP+CD19+CD3+Ly5.2+ cells were excluded.
Purification of CD11b+Gr-1+ MDSCs
Spleens from lymphoma-bearing mice were digested in medium containing7 mg collagenase (Type III; Worthington Biochemicals) and 1 mg DNase(Boehringer-Mannheim). Spleen cells were depleted of CD19 (1D3) andCD3 (KT3-1.1) cells and selected for CD11b-expressing cells by Ab-mediated magnetic enrichment (Miltenyi Biotec). CD11b+ cells werefurther purified by flow cytometry using Ly5.2, CD11c, CD11b, and CD43.
Differentiation of MDSCs in the presence of GM-CSFor M-CSF cytokines
Purified MDSCs were plated at 105 cells/well in U-bottom 96-well plates(Becton Dickinson) in medium containing 10 ng/ml GM-CSF or 200 U/ml M-CSF or were left untreated. After several days, the cells were har-vested, washed twice, and incubated in the Ab mixture containingfluorochrome-bound anti-CD11c, CD11b, F4/80, and the activation markerMHC II. Where indicated, LPS (0.5 mg/ml) was added for 12 h. Afterincubation, cells were analyzed by flow cytometry.
May–Gr€unwald–Giemsa stains
Purified Em-myc cells or myeloid cells were prepared by cytospin cen-trifugation onto glass slides and stained with May–Gr€unwald–Giemsa.Slides were viewed and photographed using a compound microscope(Zeiss) and a digital camera (Axiocam; Zeiss).
Presentation of OVA Ag in vitro
Purified CD8+ DC, CD82 DC, and MDSC were cultured with the in-dicated concentrations of OVA or OVA257–264 peptide for 40 min. Cellswere washed three times and resuspended in medium containing 5 3104 CFSE-labeled OT-I cells. The number of proliferating OT-I cellswas determined 60–65 h later. Alternatively, the indicated numbers ofpurified and irradiated bm1.Em-myc-OVA cells were cultured with 5 3104 CFSE-labeled naive OT-I cells or cocultured with 2.5 3 104 CD8+
DCs or CD82 DCs (purified from naive mice) and 5 3 104 CFSE-labeled naive OT-I cells. The number of proliferating OT-I cells wasdetermined 60–65 h later. To assess cross-presentation of OVA tumorAg in mice bearing bm1.Em-myc-OVA or control Em-myc-OVA, B6 3bm1 (F1) mice were inoculated with lymphoma cells, and spleens wereharvested 7 d later. CD8+ DCs and CD82 DCs were purified andcocultured at the indicated numbers with 5 3 104 CFSE-labeled OT-Icells. The number of proliferating OT-I cells was determined 60–65 hlater. Each assay was performed in duplicate.
Cytokine-production assay
Sera collected from tumor-bearing mice were investigated for cytokinecontent using the 23-plex assay on the BioPlex 2200 system, according tothe manufacturer’s protocol (Bio-Rad, Hercules, CA) (26).
Presentation of lymphoma-associated OVA in vivo
Mice were injected i.v. with Em-myc-OVA cells, and spleens were har-vested 7 d later. CD8+ DC, CD82 DC, and MDSC populations were pu-rified, as described above, and cocultured at the indicated numbers with
The Journal of Immunology 3535
by guest on April 24, 2018
http://ww
w.jim
munol.org/
Dow
nloaded from
5 3 104 CFSE-labeled OT-I cells. The number of proliferating OT-I cellswas determined 60–65 h later. Each assay was performed in duplicate.
Cyclophosphamide treatment
Tumor-bearing mice were injected i.p. with vehicle (H2O) or cyclophos-phamide (CTX; Sigma). To determine the effect of combined chemo-immunotherapy, tumor-bearing mice were injected i.v. with 5 3 106 OT-ICTLs 1 or 3 d following CTX treatment.
Data analysis
Quantitation of CTLs, tumor cells, or other cell types was performed byaddition of 5 3 104 blank calibration particles (6.0–6.4 mm; BD Bio-sciences) per sample prior to flow cytometry. The p values were calculatedusing a two-tailed unpaired Student t test or a two-tailed Mann–Whitney Utest when data failed normality tests.
ResultsDeletion and inactivation of adoptively transferredantilymphoma CTLs in mice with high tumor burden
We described two new variants of the mouse Em-myc lymphoma,a well-established model of human non-Hodgkin lymphoma. Theyexpress GFP alone (Em-myc-GFP control tumor) or with OVA asa model neoantigen (Em-myc-OVA tumor) (27). The Em-myc-OVA tumor recapitulates several features described for humanlymphomas: it is susceptible to anti-OVA CTL killing but is ig-nored by the immune system of tumor-bearing mice (i.e., it doesnot elicit anti-OVA CD8+ T cell priming spontaneously), and al-though it is possible to prime endogenous anti-OVA CD8+ T cellsin Em-myc-OVA tumor–bearing mice with an anti-OVA vaccine,the T cells do not become effective CTLs (27). In this scenario,adoptive cell therapy with high-avidity anti-OVA CTLs (OT-I)might be a suitable approach to overcome immunosuppressionof the endogenous repertoire.Mice in which Em-myc-GFP or Em-myc-OVA tumors (FSChigh
B220intGFP+) had been growing for 3–5 d were treated with anti-OVA OT-I CTLs generated in vitro (22) (Fig. 1A). Em-myc-OVAlymphoma was almost eradicated within 48 h if OT-I CTLs weretransferred 3 d after tumor inoculation (Fig. 1A, 1B). This was Agspecific because control Em-myc-GFP tumors were not affected.Furthermore, Em-myc-OVA tumor–bearing mice receiving CTLseventually succumbed to tumor growth, but the expanded tumordid not express GFP and was no longer recognized by anti-OVACTLs (Fig. 1C and data not shown). Therefore, OT-I CTLsinjected at day 3 exerted enough pressure on the tumor to selectimmunoedited cells, a clear indication of effective Ag-specificantitumor activity. In contrast, if the OT-I CTLs were injected4 d after tumor inoculation, tumor cell killing was less effective;injection on day 5 had no significant effect (Fig. 1A, 1B). Toaddress whether this was due to loss of CTL activity upon en-counter of large tumors, we compared CTL recovery, phenotype,and function two days after injection of the CTL at the three timepoints.Mice bearing Em-myc-GFP or Em-myc-OVA for 3 d were
injected with OT-I CTLs. Two days later, the number of CTLsin spleen was equivalent (Fig. 2A). These two groups of micecontained more CTLs than did mice with no tumor, probablybecause the larger size of the spleens in the tumor-bearing miceprovided more “space” for CTL accumulation. The number ofCTLs recovered from Em-myc-OVA–bearing mice that receivedOT-I CTLs 4 d after tumor inoculation was more variable (Fig.2B). Few CTLs were recovered from mice treated 5 d after tumorinoculation (Fig. 2C). The remaining CTLs expressed low TCRand high PD-1, signature markers of exhausted T cells (2) (Fig.2D, 2E). The phenotype of the CTLs injected on day 5 (largetumor) contrasted with the phenotype of CTLs injected on day 3
(small tumor). Indeed, virtually all CTLs displayed PD-1 upreg-ulation in both situations, indicative of Ag encounter, but thoseinjected at day 5 expressed higher levels of PD-1 and haddownregulated TCR expression (Fig. 2D, 2E). The killing activityof the OT-I was assessed in vivo by injecting peptide-pulsedsplenocytes 2 d after injecting CTLs (Fig. 2F). Splenocytes werekilled in Em-myc-GFP tumor–bearing mice receiving OT-I CTLs3 or 5 d after tumor inoculation or in mice bearing a low burdenof Em-myc-OVA tumor (day 3) but not in mice with a high Em-myc-OVA tumor burden (day 5, Fig. 2F). This was not due tocompetition between the target splenocytes and the tumor cells. In-deed, the CTLs remaining in mice treated at day 5 were in-trinsically defective, because when we purified them from micetreated at day 5 and assessed their killing activity ex vivo, theydisplayed little lytic activity (Fig. 2G), consistent with their in-capacity to secrete cytokines upon restimulation (Fig. 2H). Thisexperiment also showed that the tumor had not become resistant toCTLs at day 5, because it could be killed ex vivo (Fig. 2H, and seebelow). In summary, in mice bearing large lymphomas most of theinjected CTLs were eliminated, and those remaining had lost ef-fector function so that they were incapable of killing tumor cellsand other cells, in vivo or ex vivo.
CTL inactivation requires direct encounter of cognate Ag
Next, we established to what extent Ag-specific CTL inacti-vation required direct contact with an APC. An alternativepossibility was that Ag recognition by a few CTLs triggered sup-
FIGURE 1. A critical tumor mass impairs tumor killing by adoptively
transferred tumor-specific CTLs. (A) Mice bearing Em-myc-GFP or Em-
myc-OVA (GFP-OVA) tumors were inoculated with OT-I CTLs 3, 4, or 5 d
after tumor inoculation or were left untreated. Two days later, tumor load
in the spleen was analyzed by flow cytometry. Contour plots show FSC/
B220 expression in whole spleens that did (lower plots) or did not (upper
plots) receive CTLs at days 3, 4, or 5. The graphs show GFP expression (a
surrogate of OVA expression) in Em-myc-OVA cells (filled graphs) and
endogenous B cells (open graphs) from mice injected with OT-I CTLs (or
not) at day 5. All contour plots are representative of two independent
experiments. (B) Tumor reduction was determined as the ratio of the number
of tumor cells in mice injected with OT-I CTLs relative to the number of
tumor cells in untreated controls. Data are mean 6 SD, pooled from two
independent experiments (n = 4–10). (C) Mice bearing GFP-OVA tumor
were inoculated with OT-I CTLs 3 d after tumor inoculation. After 16 d,
spleens were analyzed by flow cytometry. Graph shows tumor cell GFP
expression (open graph) compared with tumor cell GFP expression from
mice bearing GFP-OVA tumor for 11 d with no OT-I CTLs transferred (filled
graph). The result is representative of two independent experiments. **p ,0.005, two-tailed unpaired Student t test. n.s., not significant (p . 0.05).
3536 THE APC MEDIATING CTL INACTIVATION
by guest on April 24, 2018
http://ww
w.jim
munol.org/
Dow
nloaded from
pressive cytokine release that indirectly impaired the majority oftransferred CTLs. Distinguishing between these two scenarios isimportant because if most CTLs are inactivated indirectly bycytokines, neutralization of these cytokines might restore activity,but if inactivation is mediated by direct CTL–APC contact,identification of the relevant APC becomes critical.Mice bearing Em-myc-GFP or Em-myc-OVA lymphoma for
5 d were coinjected i.v. with equal numbers of gBT-I CTLs andCFSE-labeled OT-I CTLs (Fig. 3A). The transgenic gBT-I CTLsrecognize a H-2Kb–restricted HSV-1 glycoprotein B epitope(gB498–505) (21), absent in the mice. Three days later, mice bearingcontrol Em-myc-GFP lymphoma contained similar numbers ofgBT-I and OT-I CTLs, with normal TCR and PD-1 levels (Fig.3B). In contrast, in mice containing Em-myc-OVA lymphoma,gBT-I CTLs displayed normal TCR and PD-1 expression, whereasOT-I CTLs exhibited an exhausted phenotype (Fig. 3C). Further-
more, the number of OT-I CTLs in spleens of mice bearing Em-myc-OVA lymphoma was lower than in Em-myc-GFP–bearingcounterparts, whereas the number of gBT-I CTLs was three timeshigher (Fig. 3B, 3C). This suggests that tumor Ag recognition byOT-I CTLs created an environment conducive to nonspecific CTL(gBT-I) expansion, but those CTLs that directly recognized Ag(OT-I) were deleted and inactivated. The killing activity of OT-Iand gBT-I CTLs was determined by measuring lysis of third-partycells in spleens of recipient mice (Fig. 3A). Splenocytes coatedwith either gB498–505 or OVA257–264 peptide were killed in spleensof mice bearing control Em-myc-GFP lymphoma (Fig. 3B). Incontrast, in Em-myc-OVA lymphoma–bearing mice, OT-I CTLkilling activity was reduced ∼50%, whereas gBT-I CTL killingwas not impaired (Fig. 3C). Our results demonstrate that for theCTLs to be inactivated, they are required to recognize their cog-nate Ag on an APC. Which APC?
FIGURE 2. A large fraction of anti-tumor CTLs are not recovered upon transfer into mice bearing Em-myc-OVA, and the remaining CTLs are
functionally impaired. Mice devoid of tumors or injected with Em-myc-GFP or Em-myc-OVA (GFP-OVA) were inoculated with OT-I CTLs 3 d (A),
4 d (B), or 5 d (C) after tumor inoculation. Two days later, OT-I CTLs in the spleen were enumerated. Graphs show data pooled from two to five
independent experiments; each symbol represents an individual mouse (n = 5–10). **p , 0.005, ***p , 0.0001, two-tailed unpaired Student t test.
(D) Representative FACS plots of (TCR) Va2 and PD-1 in OT-I CTLs enumerated in (A) (upper panels) and (C) (bottom panels). The dotted line
represents unstained cells. (E) Percentage of PD-1+ OT-I CTLs in spleens of mice enumerated in (A) and (C). Graph represents data pooled from two
or three independent experiments, with each symbol representing an individual mouse (n = 6–7). **p , 0.005, ***p , 0.0001, two-tailed Mann–
Whitney U test. (F) Tumor-bearing mice were injected with OT-I CTLs 3 or 5 d after lymphoma inoculation. Two days later, anti-OVA killing in the
spleen was determined. Graph represents data pooled from two independent experiments, with each symbol representing an individual mouse (n =
6). **p , 0.005, ***p , 0.0001, two-tailed Mann–Whitney U test. (G) Mice bearing GFP or GFP-OVA tumors were injected with OT-I CTLs. Five
days later, OT-I CTLs were purified from GFP (CTLGFP) or GFP-OVA (CTLGFP-OVA) tumor-bearing mice, and their capacity to kill was measured in
an in vitro killing assay using GFP and GFP-OVA tumor cells (arrows) as targets. Representative FACS plots of the remaining tumor cells are
shown. Data are mean 6 SD and are pooled from two independent experiments (n = 3 or 4 replicates). **p , 0.005, ***p , 0.0001, two-tailed
unpaired Student t test. (H) Mice bearing GFP or GFP-OVA tumor were injected with OT-I CTLs 5 d after tumor inoculation. Two days later,
secretion of IFN-g and TNF-a by splenic CD8+ T cells upon in vitro restimulation was determined. Data are pooled from three independent
experiments, with each symbol representing an individual mouse (n = 10). **p , 0.005, ***p , 0.0001, two-tailed unpaired Student t test. n.s., not
significant (p . 0.05).
The Journal of Immunology 3537
by guest on April 24, 2018
http://ww
w.jim
munol.org/
Dow
nloaded from
CTL inactivation is mediated by direct Ag recognition ontumor cells
Lymphoid organs contain multiple types of DCs that play distinctfunctions in Ag presentation, immunity, and tolerance (28, 29), sowe first addressed whether the number and phenotype of DCswere altered in lymphoma-bearing mice. We did not observesignificant alterations in the proportion and number of any DCpopulation in spleen or lymph nodes of lymphoma-bearing mice(Fig. 4A, 4B). Likewise, expression of the DC activation markersMHC II, CD86, and CD69 was equivalent in DCs from normaland tumor-bearing mice, except for a slight increase in CD86expression on splenic conventional DCs (cDCs) and CD69 onplasmacytoid DCs (pDCs) (Fig. 4C, 4D).To test whether the presence of the lymphoma made the DCs
refractory to stimulation, we purified them from normal or tumor-bearing mice and incubated them in vitro in the presence of LPS.Upregulation of the activation markers MHC II and CD86 showedthat the DCs were capable of undergoing “maturation” in vitro(30–32) (Fig. 4E). This is consistent with our previous observa-tions showing that vaccination of lymphoma-bearing mice withcell-associated OVA and LPS induces expansion of endogenousanti-OVA CD8+ T cells in vivo, a reaction mediated by cross-priming CD8+ DCs (27). We conclude that lymphoma growthdid not cause overt alterations in DC numbers, phenotype, Ag pre-sentation, or capacity to prime T cell responses.Another APC that might inactivate CTLs is MDSCs, which are
reportedly involved in cross-presentation of tumor Ags and sup-pression of cytotoxicity (10). Spleen analysis of mice bearingparental Em-myc tumors, or its derivatives Em-myc-GFP and Em-myc-OVA, showed recruitment, after day 5, of a cell populationthat, according to criteria defined in other studies (10), was iden-
tified as MDSCs (Supplemental Fig. 1). Our characterization iden-tified these cells as mature monocytes (Supplemental Fig. 1).Host APCs can present lymphoma Ags via cross-presentation
(33), an activity performed in vivo primarily by conventionalCD8+ DCs (28, 29, 34). We purified CD8+ DCs, CD82 DCs, andMDSCs from spleens of Em-myc-OVA tumor–bearing mice andincubated them in vitro with naive OT-I to assess cross-presentation of tumor Ag acquired in vivo. Only CD8+ DCs eli-cited OVA-specific OT-I proliferation (Fig. 5A). All three pop-ulations induced proliferation when incubated with OVA257–264
peptide (Supplemental Fig. 2A). It could be argued that MDSCsalso cross-presented tumor Ag, but their suppressive activityprevented OT-I division. Addition of large numbers of MDSCs toCD8+ DCs reduced OT-I proliferation, but this was also causedby the addition of CD82 DCs or MHC I2/2 splenocytes (Sup-plemental Fig. 2B). Therefore, MDSCs appeared unlikely to beresponsible for CTL inactivation. The role of CD8+ DCs wasassessed using mice deficient for the transcription factor Batf3,which lack CD8+ DCs and are deficient in cross-presentation (20).Inactivation of OT-I CTLs also occurred in Batf32/2 tumor-bearing mice (Fig. 5B), excluding a major role for CD8+ DCs.We next assessed whether tumor cells themselves were the APCs
that inactivated CTLs. We generated a new lymphoma, bm1.Em-myc-OVA, which expresses mutant H-2Kbm1 molecules incapableof OVA presentation but is a source of OVA for cross-presentation(Supplemental Fig. 2C, 2D). Assessment of OT-I CTLs transferredinto mice bearing Em-myc-GFP, Em-myc-OVA, or bm1.Em-myc-OVA lymphomas showed that CTL deletion and inactivation re-quired direct Ag presentation by tumor cells (Fig. 5C). Becausethe size of bm1.Em-myc-OVA tumors exceeded that of Em-myc-OVA tumors (Fig. 5C), and CD8+ DC OVA cross-presentation wasmore efficient in bm1.Em-myc-OVA tumor–bearing mice (Sup-plemental Fig. 2E), we conclude that CTL inactivation was me-diated by tumor cells themselves and not by DCs or, indeed, anyother recipient cell that might be capable of cross-presentation (e.g.,MDSCs or macrophages) (35).
Target cells that cause CTL inactivation are not resistant tokilling
When CTLs engage target cells they release IFN-g, which re-portedly induces other cells to acquire an inactivating phenotypecharacterized by the upregulation of inhibitory receptors, such asPD-L1 (36). We observed a small upregulation of this marker inEm-myc-OVA tumor cells, as well as in B cells, cDCs, pDCs, andMDSCs, upon injection of OT-I CTLs in mice bearing the Em-myc-OVA tumor (Fig. 6A). The effect was mediated by IFN-g andrequired Ag recognition because it was not observed in micebearing Em-myc-GFP tumors or if the CTLs lacked IFN-g (Fig.6A). Notably, IFN-g–deficient CTLs were deleted and acquired aninactivated phenotype in Em-myc-OVA tumor–bearing mice (Fig.6B, 6C). This implies that CTL impairment was not caused byacquisition of an IFN-g–induced inactivating phenotype by thetarget cells. We also addressed whether tumor cells became re-sistant to CTL killing at the time that they induced inactivation. Todo this, we isolated spleen cells from mice bearing tumors for5 d and labeled them with a fluorescent violet dye. The cells (ofwhich approximately one quarter were tumor cells) were inocu-lated into naive mice. One day later, one group of mice wasinjected with OT-I CTLs and another was not. After 2 d, the tumorand nontumor cells could be distinguished because the former haddivided and were less bright in the Violet channel (Fig. 6D).Control and CTL-injected mice contained similar numbers ofnontumor (Violethigh) transferred cells, but .75% of the tumorcells had been eliminated in the mice injected with CTLs (Fig.
FIGURE 3. The deletion, phenotypic changes, and reduced killing ca-
pacity of adoptively transferred OT-I CTLs are Ag specific and not due to
bystander effects. (A) Ly5.2 mice bearing Em-myc-GFP or Em-myc-OVA
cells were coinjected with Ly5.13gBT-I CTLs and CFSE-labeled OT-I
CTLs 5 d after tumor inoculation. Three days later, the CTL phenotype,
recovery, and killing activity in the spleen were determined. gBT-I and OT-
I CTLs from spleens of mice bearing Em-myc-GFP (B) or Em-myc-OVA
(C) were analyzed by flow cytometry for the expression of (TCR) Va2 and
PD-1 (left panels) and enumerated (middle panels), and their killing ac-
tivity was assessed in vivo (right panels). Graphs show data pooled from
two independent experiments, with each symbol representing an individual
mouse (n = 8). **p , 0.005, two-tailed Mann–Whitney U test; ***p ,0.0001, two-tailed unpaired Student t test. n.s., not significant (p . 0.05).
3538 THE APC MEDIATING CTL INACTIVATION
by guest on April 24, 2018
http://ww
w.jim
munol.org/
Dow
nloaded from
6D). Therefore, the ability of day-5 tumors to cause CTL inacti-vation was not due to an intrinsic capacity to avoid CTL killing.
Selective reduction of target cell number with mildchemotherapy prevents CTL inactivation
Next, we tested the hypothesis that CTL fate might be dictated bytarget cell density: low in mice bearing the tumor for 3 d and high in
mice bearing the tumor for 5 d. If so, reducing tumor size in thelatter group might prevent inactivation of subsequently injectedCTLs. We titrated the effect of the cytotoxic drug CTX to identifya single low-dose injection (40 mg/kg) sufficient to kill most tumorcells within 24 h (Fig. 7A, day 6) without affecting DCs or MDSCs(data not shown). The tumor rebounded and was detectable2 d later if mice were left untreated but not if they were treated
FIGURE 4. The number and phenotype of cDCs and
pDCs remain largely unchanged in the presence of
lymphoma. Graphs display absolute number of pDCs
and cDCs purified from spleen (A) or lymph nodes (B)
of naive and Em-myc-OVA–bearing (GFP-OVA) mice
after 5 d. Bar graphs display mean6 SD of pooled data
from two independent experiments (n = 2). In each
experiment, four pooled spleens were used for analysis.
DC surface expression of MHC II, CD86, and CD69
was determined by flow cytometry for freshly isolated
pDCs (CD11cintCD45RA+), CD8+ DCs (CD11chigh
CD45RA2CD8+), CD82 DCs (CD11chighCD45RA2
CD82), and migratory (migr) DCs (CD11c+CD45RA2
CD82CD205int/high) purified from spleens (C) or lymph
nodes (D; axillary, brachial, inguinal) of naive mice
(filled graph) or mice bearing the lymphoma (solid
black line). The dotted line represents background
staining. All graphs are representative of two to five
independent experiments, with four mice used in each
analysis. (E) cDCs were purified from naive mice or
mice bearing GFP-OVA tumor for 7 d and maintained
overnight at 4˚C (fresh) or cultured at 37˚C in the
presence of LPS. Expression of MHC II and CD86 was
determined by flow cytometry. Data in (C–E) are rep-
resentative of two independent experiments.
FIGURE 5. Direct Ag presentation by lymphoma cells, but not cross-presenting CD8+ DCs or MDSCs, induces deletion and inactivation of adoptively
transferred lymphoma Ag-specific CTLs. (A) CD8+ DCs, CD82 DCs and MDSCs, were purified from spleens of mice bearing Em-myc-OVA for 7 d and
incubated with naive CFSE-labeled OT-I cells in vitro. The number of proliferating OT-I cells was determined after 60–65 h of culture. Experiments were
performed in duplicate, and data represent the mean 6 SD. The graph is representative of two or three independent experiments. (B) C57BL/6 or Batf32/2
mice were injected with OT-I CTLs 5 d after Em-myc-OVA (GFP-OVA) injection. The number of tumor cells (left panel), OT-I CTLs (middle panel), and
OT-I CTLs that produced IFN-g upon restimulation in vitro (right panel) was determined 2 d later. Graphs represent data pooled from two independent
experiments, with each symbol representing an individual mouse (n = 5–6). (C) B63bm1 (F1) mice inoculated with Em-myc-OVA (GFP-OVA), Em-myc-
GFP, or bm1.Em-myc-OVA (bm1-OVA) tumor were injected with OT-I CTLs 5 d later. After 3 d, the number of OT-I CTLs (left panel) and tumor cells
(middle panel) in the spleens of lymphoma-bearing mice was determined. In addition, OT-I CTL killing in the spleen was measured 2 d after CTL injection
(right panel). Data are pooled from two independent experiments, with each symbol representing an individual mouse (n = 5–11). **p , 0.005, ***p ,0.0001, two-tailed unpaired Student t test, with the exception of the left panel in (C), for which the two-tailed Mann–Whitney U test was used. n.s., not
significant (p . 0.05).
The Journal of Immunology 3539
by guest on April 24, 2018
http://ww
w.jim
munol.org/
Dow
nloaded from
with OT-I CTLs 1 d after CTX (Fig. 7A, day 8). This correlatedwith a lack of OT-I CTL deletion and inactivation in CTX-pretreated mice (Fig. 7B), supporting our hypothesis that botheffects were caused by encounter with a high density of targetcells.We ruled out that CTX was acting to increase CD8+ DC cross-
presentation of tumor Ag released by dying cells, given thatanalysis of Batf32/2 mice showed CD8+ DCs were not involved(Supplemental Fig. 3). Another possibility was that CTX-inducedinflammation, as the result of the release of “danger” signals bydying tumor cells, prevented CTL inactivation (37). We did notobserve upregulation of DC maturation markers (Fig. 7C, Sup-plemental Fig 4) or increased serum inflammatory cytokines(Supplemental Fig. 4) in CTX-treated animals. Furthermore, lowerdoses of CTX (30 mg/ml) could kill 75% of tumor cells (Fig. 7D),which should promote the release of significant danger signals, but
this treatment was insufficient to prevent CTL inactivation andtumor re-expansion (Fig. 7E), suggesting that, with this dose ofCTX, the number of tumor cells had not been reduced below theinactivating threshold.We then examined the fate of CTLs if they were injected 3 d, and
not 1 d, after CTX treatment to allow tumor re-expansion beforeCTL inoculation (Fig. 7A). In this scenario, CTLs were againdeleted and inactivated, with no effect on tumor reduction (Fig.7F). This result supports the notion that the lag between CTXtreatment and CTL injection enabled the tumor burden to againreach an inactivating number threshold.
Efficient Ag-specific elimination of lymphoma in mice treatedwith CTX and CTLs
The effect of combined low-dose CTX therapy and adoptivetransfer of tumor-specific CTLs on the survival of Em-myc-OVA
FIGURE 6. Tumor cells that impair CTL function do not become resistant to killing. (A) Mice inoculated with Em-myc-GFP or Em-myc-OVA (GFP-
OVA) were injected with WT (gray line) or IFN-g2/2 OT-I (dashed line) CTLs 5 d after tumor inoculation or were left untreated (filled graph). One day
after CTL injection, spleens from tumor-bearing mice were analyzed for PD-L1 expression. Graphs show PD-L1 expression for tumor cells, B220+ cells,
CD8+ DCs (CD11chighCD11b2), CD82 DCs (CD11chighCD11b+), pDCs (CD11cintCD11b2), and MDSCs (CD11cintCD11b+). DCs were analyzed by
exclusion of GFP+, CD19+, and CD3+ cells. All graphs are representative of two independent experiments. The dotted line represents unstained controls.
Mice injected with GFP or GFP-OVA tumor cells were inoculated with Ly5.1+ WT (B) or CFSE-labeled IFN-g2/2 (C) OT-I CTLs 5 d after tumor in-
oculation. One day later, WT and IFN-g2/2 OT-I CTLs in the spleen of tumor-bearing mice were analyzed as Ly5.1+Va2+ or CD8+CFSE+Va2+ cells,
respectively. Graphs show PD-1 expression by gated CTLs (dashed circle). Results are representative of two independent experiments. (D) Mice were
injected with Em-myc-OVA tumor, and spleens were harvested after 5 d. Splenocytes were labeled with CellTrace Violet dye, and 20 3 106 cells (con-
taining ∼43 106 tumor cells) were injected into naive mice. After 24 h, mice were inoculated with OT-I CTLs or left untreated; 2 d later, spleens of tumor-
bearing mice were analyzed by flow cytometry. Contour plots (far left panels) show FSC/B220 expression in whole spleens that did (upper panels) or did
not (lower panels) receive CTLs. The tumor cells (FSChighB220+) were identified based on Violet and GFP expression (middle contour plots, rectangle).
The transferred nontumor cells were identified as Violethigh and FSClow (right contour plots). The graph shows the number of tumor cells recovered from
mice that did or did not receive OT-I CTLs (far right panel). Graph shows data pooled from two independent experiments, with each symbol representing an
individual mouse (n = 4–7). **p , 0.005, two-tailed unpaired Student t test.
3540 THE APC MEDIATING CTL INACTIVATION
by guest on April 24, 2018
http://ww
w.jim
munol.org/
Dow
nloaded from
tumor–bearing mice was assessed. Despite the strong tumor re-duction caused by low-dose CTX (Fig. 7A), mice given only thistreatment survived just 2 d longer than untreated animals (Fig.
8A). Injection of CTLs without previous CTX treatment conferredno benefits, as expected (Fig. 8A). However, if mice were injectedwith OT-I CTLs 1 d after CTX, their lifespan nearly doubled (Fig.
FIGURE 7. A low nonlymphoablative, noninflammatory dose of CTX specifically reduces tumor burden and allows tumor-specific CTLs to eradicate
lymphoma cells. (A) Mice bearing Em-myc-OVA tumors were treated with vehicle (H2O) or 40 mg/kg CTX 5 d after tumor inoculation. At day 6, spleens
were analyzed by flow cytometry (left and middle panels) for the presence of tumor cells (arrows). Alternatively, OT-I CTLs were transferred (+CTL) or not
(-CTL) at day 6 into the mice, and spleens were analyzed 2 d later (right panel). The graph shows the number of tumor cells in the spleens harvested at day
8. Data are pooled from two independent experiments, with each symbol representing an individual mouse (n = 8). (B) One day after treatment with vehicle
(H2O) or CTX, as indicated in (A), OT-I CTLs were adoptively transferred into tumor-bearing mice, and their number was determined by flow cytometry
2 d later (left panel). Data are pooled from two independent experiments, with each symbol representing an individual mouse (n = 4–8). FACS plots (right
panel) show TCR (Va2) expression by OT-I CTLs from control Em-myc-GFP (filled graph) and Em-myc-OVA (open graph) tumor-bearing mice given
vehicle (upper panel) or CTX (lower panel) before CTL injection. (C) One day after treatment with vehicle (H2O) or CTX, as indicated in (A), DCs were
purified from spleens of tumor-bearing mice, and the surface expression of MHC II, CD40, CD86, and CD69 was analyzed by FACS. Plots show freshly
isolated CD8+ DCs from spleens of mice treated with vehicle (H2O; solid black line) or CTX (dotted line). The filled histograms show background staining.
All results are representative of two to five independent experiments using two or three mice for analysis. (D) Em-myc-OVA–bearing mice were treated with
vehicle (H2O) or the indicated dose of CTX 5 d following tumor inoculation. One day later, the spleens of mice were analyzed by flow cytometry. Arrows
indicate the tumor cells, and the percentage above each plot indicates their frequency. Mean6 SD of pooled data from two to four independent experiments
are shown (n = 4–10). (E) Em-myc-OVA–bearing mice were treated with vehicle (H2O) or the indicated dose of CTX 5 d following tumor inoculation. One
day later, OT-I CTLs were adoptively transferred into tumor-bearing mice or left untreated, and spleens were analyzed by flow cytometry 3 d later. Tumor
reduction was determined as the ratio of the number of tumor cells in mice injected with CTLs relative to the number in untreated controls. Data are mean
6 SD of one independent experiment (n = 3). (F) OT-I CTLs were adoptively transferred into tumor-bearing mice 3 d after treatment with 40 mg/kg CTX,
and spleens were analyzed 2 d later. The graphs show the number of tumor cells (left panel) and OT-I CTLs (middle panel). Data are pooled from two
independent experiments, with each symbol representing an individual mouse (n = 4–8). Graphs compare TCR (Va2) expression by CTLs from control Em-
myc-GFP (filled graph) and Em-myc-OVA (open graph) tumor-bearing mice treated with vehicle (upper panel) or CTX (lower panel) before CTL injection.
In all panels, statistical analysis was performed with a Mann–Whitney U test, except in (B) (GFP), for which a two-tailed unpaired Student t test was used.
**p , 0.005. n.s., not significant (p . 0.05).
The Journal of Immunology 3541
by guest on April 24, 2018
http://ww
w.jim
munol.org/
Dow
nloaded from
8A). This was Ag-specific because it was not observed in micebearing Em-myc-GFP tumors. Indeed, the tumors that eventuallykilled the mice treated with the combined therapy expressed lowGFP (Fig. 8B), indicating that OVA2/low tumors had been selectedby immunoediting under the pressure of OT-I CTLs (38).
DiscussionThis study was based on analysis of CTL activity against Em-myctumors, a standard model of human non-Hodgkin lymphoma. Em-myc-OVA lymphoma is susceptible to CTL killing but is ignoredby the immune system of naive recipients (27). Injection of high-affinity anti-OVA CTLs generated in vitro should be a suitableimmunotherapy against this tumor; however, as often observedin trials of adoptive T cell therapy (39, 40), high-affinity anti-lymphoma CTLs were deleted. In addition, the surviving CTLsexhibited an inactivated phenotype, suggesting more than onemechanism of T cell impairment. We do not know whether in-activation of surviving CTLs was due to exhaustion, anergy, ora novel mechanism. Exhaustion and anergy often result from long-term (several days to weeks) interaction between T cells and APCs(41, 42), but we observed deletion and functional CTL impairmentwithin 2 d. Although several mechanisms of antitumor T celltolerance have been documented (43), little progress has beenmade toward characterizing APCs that mediate CTL inactivation(2). The purpose of the current study was to establish the Agspecificity and dissect the contribution of different APCs to CTLimpairment. It is unclear to what extent such mechanisms are Agspecific as opposed to resulting from overt immunosuppression.An important conclusion of our study is that deletion and in-
activation only affected CTLs that recognized cognate Ag. By-stander CTLs (gBT-I) coinjected with tumor-specific CTLs did notsuffer the same fate. Our results do not exclude a role for inac-tivating surface receptors or soluble factors in CTL impairment, butwe show that these molecules cannot exert their function withoutTCR engagement. We hypothesize that CTL inactivation is trig-gered by excessive formation of immunological synapses, eitherserially or simultaneously. Although studies of the dynamics ofCTL–APC interactions have defined key attributes of CTL-mediated tumor cell killing, transferred CTLs remained func-tional in all of these models (44–47). To our knowledge, this is thefirst description of CTL inactivation following the engagement oflarge numbers of immunological synapses during adoptive T celltherapy.
The major goal of this study was to identify the APC responsiblefor CTL inactivation. Tumor cells themselves were obvious can-didates, but so were DCs andMDSCs, both of which can reportedlyinduce cross-tolerance (48–50). In contrast to previous reports (49,51, 52), we did not observe tumor Ag cross-presentation byMDSCs. Furthermore, these cells did not inhibit T cell primingby other cells (CD8+ DCs), at least in vitro. The only APCs re-sponsible for CTL inactivation in our system were the tumor cellsthemselves. This was demonstrated using the bm1.Em-myc-OVAlymphoma, which enabled definitive demonstration that DCs (50),MDSCs, or any other host cell capable of cross-presentation (e.g.,macrophages) (35) were not the APCs responsible for CTL in-activation. Furthermore, we could not attribute CTL inactivationto the acquisition of a “suppressive” phenotype by the tumor cells(53).Our observations suggest that target cell density is the critical
parameter that dictates CTL fate. We propose that encounter ofa number of lymphoma cells above a critical threshold causesinactivation of anti-tumor CTLs. In agreement with our study,Budhu et al. (54) support the concept that an appropriate CTL E:Tratio is critical to achieve successful tumor reduction, although inthat report inactivation of CTLs was not examined. In the presentstudy, we show that mild CTX therapy of mice with large tumorsprevents the inactivation of CTLs injected 1 d later. Importantly,although CTX therapy alone drastically reduced Em-myc lym-phoma size, it afforded the mice a relatively small gain in survival(2 d). Of note, the CTX dose that we used did not cause lym-phoablation, so its benefit could not be attributed to providingmore “space” for injected CTLs, nor was it overtly inflammatory.Analysis of Batf3-deficient mice (20) also excluded the possibilitythat CTX-mediated tumor killing enhanced cross-priming byCD8+ DCs, which then prevented CTL inactivation. The mostlogical explanation for the effect of CTX is that it reduced tumorburden below the critical density responsible for CTL inactivation.As a result, CTL injection soon after chemotherapy provokeda dramatic outcome, almost doubling the survival of tumor-bearing mice. The tumor likely would have been completelyeradicated if the Ag targeted by the CTL had been derived froma protein critical to tumor survival and not the model Ag OVA,because the tumor that eventually killed the mice treated withCTX plus CTL was immunoedited and expressed low OVA levels.Our results support the concept of applying adoptive T cell ther-apy soon after mild chemotherapy to treat minimal residual dis-
FIGURE 8. Efficient lymphoma elimination in mice treated with CTX and CTL. (A) Tumor-bearing mice were treated with vehicle (H2O) or 40 mg/kg
CTX 5 d after tumor inoculation. One day later, OT-I CTLs were adoptively transferred. Kaplan–Meier survival curves are shown. The black line with filled
circles includes the following groups: GFP or GFP-OVA tumor–bearing mice treated with H2O (n = 4 mice/group) and GFP or GFP-OVA tumor–bearing
mice treated with H2O + OT-I CTLs (n = 4 mice/group). The dashed line with open circles dots includes the following groups: GFP or GFP-OVA tumor–
bearing mice treated with CTX (n = 8 mice/group) and GFP tumor-bearing mice treated with CTX + OT-I CTLs (n = 8 mice/group). The dotted line with
open triangles includes GFP-OVA tumor–bearing mice treated with CTX + CTL (n = 7 mice/group). The p value (log-rank analysis) for GFP-OVA tumor
(+CTX) versus GFP-OVA tumor (+CTX +OT-I CTLs) is shown. (B) FACS plots of GFP expression (a surrogate of OVA expression) by Em-myc-OVA cells
(FSChighB220+) in tumor-bearing mice that received CTX alone (left panel) or CTX + OT-I CTLs (right panel). GFP expression by endogenous B cells is
indicated as background staining (filled graph). Results are representative of two independent experiments.
3542 THE APC MEDIATING CTL INACTIVATION
by guest on April 24, 2018
http://ww
w.jim
munol.org/
Dow
nloaded from
ease, without waiting for the tumor to become detectable again.Such an approach may also avoid the use of immunostimulatorycytokines or neutralizing Abs together with adoptive cell therapytreatments, which can promote unwanted side effects, such asautoimmunity (55). Elimination of the APCs that mediate CTLinactivation may be more feasible than using blocking Abs in casesin which several receptors (e.g., PD-1, lymphocyte-activation gene3 T cell Ig and mucin domain-containing molecule-3) act simul-taneously to inactivate CTLs (6–8, 56).It is pertinent to ask whether the conclusions of our study are
applicable to other tumor and nontumor settings. A limitation of ourlymphoma model is its rapid growth (27), which may not allow thedevelopment of other tumor-resistance mechanisms. Regardless, itprovides a unique system with which to address, in vivo, the APCcontribution to CTL deletion and inactivation in mice with lowversus large tumor burden. Similar outcomes to those described inthis article were observed in a model of adoptive CTL transfer inwhich the target cells were pancreatic b cells (57). In this case,memory CTLs are inactivated following encounter of Ag pre-sented by a sufficiently large number of different types of APCs(57). Therefore, this suggests that our observations are not uniqueto the lymphoma model that we used.Our study assessed a major hurdle encountered during adoptive
T cell therapy of cancer: the loss and inactivation of infused an-titumor CTLs. Significant research efforts have attempted to definestrategies that improve the expansion and survival of infused CTLs(39). In this study, we identified the in vivo APCs that mediate theloss and/or inactivation of transferred CTLs in mice bearinglymphoma. Furthermore, we propose that CTL inactivation byencounter with a high density of target cells may be a mechanismfor the prevention of immunopathology and autoimmunity. Anti-self CTLs can also be generated during immune responses againstpathogens (e.g., virus) and must be inactivated by a mechanismthat operates only on antiself, but not antivirus, CTLs. A mech-anism that is based on the frequency of target cell encounterswould enable that distinction because, at least during the initialphase of infection, the number of virus-infected cells is smallrelative to the number of cells that express self-Ag. Antiself CTLswould be inactivated, whereas antivirus CTLs would maintaintheir capacity to kill, a scenario that we observed in Em-myc-OVAtumor–bearing mice injected simultaneously with OT-I and gBT-ICTLs. This is a hypothesis that awaits testing in other experi-mental systems.
AcknowledgmentsWe thank all members of the Flow Cytometry and Animal Services Facil-
ities at the Walter and Eliza Hall Institute, and the University of Melbourne
(Department of Immunology and Microbiology and Bio21 Institute), for
technical assistance.
DisclosuresThe authors have no financial conflicts of interest.
References1. Zhang, N., and M. J. Bevan. 2011. CD8(+) T cells: foot soldiers of the immune
system. Immunity 35: 161–168.2. Wherry, E. J. 2011. T cell exhaustion. Nat. Immunol. 12: 492–499.3. Kim, P. S., and R. Ahmed. 2010. Features of responding T cells in cancer and
chronic infection. Curr. Opin. Immunol. 22: 223–230.4. Akbar, A. N., and S. M. Henson. 2011. Are senescence and exhaustion inter-
twined or unrelated processes that compromise immunity? Nat. Rev. Immunol.11: 289–295.
5. Barber, D. L., E. J. Wherry, D. Masopust, B. Zhu, J. P. Allison, A. H. Sharpe,G. J. Freeman, and R. Ahmed. 2006. Restoring function in exhausted CD8T cells during chronic viral infection. Nature 439: 682–687.
6. Fourcade, J., Z. Sun, M. Benallaoua, P. Guillaume, I. F. Luescher, C. Sander,J. M. Kirkwood, V. Kuchroo, and H. M. Zarour. 2010. Upregulation of Tim-3 and
PD-1 expression is associated with tumor antigen-specific CD8+ T cell dys-function in melanoma patients. J. Exp. Med. 207: 2175–2186.
7. Matsuzaki, J., S. Gnjatic, P. Mhawech-Fauceglia, A. Beck, A. Miller, T. Tsuji,C. Eppolito, F. Qian, S. Lele, P. Shrikant, et al. 2010. Tumor-infiltrating NY-ESO-1-specific CD8+ T cells are negatively regulated by LAG-3 and PD-1 inhuman ovarian cancer. Proc. Natl. Acad. Sci. USA 107: 7875–7880.
8. Sakuishi, K., L. Apetoh, J. M. Sullivan, B. R. Blazar, V. K. Kuchroo, andA. C. Anderson. 2010. Targeting Tim-3 and PD-1 pathways to reverse T cellexhaustion and restore anti-tumor immunity. J. Exp. Med. 207: 2187–2194.
9. Melief, C. J. 2008. Cancer immunotherapy by dendritic cells. Immunity 29: 372–383.
10. Gabrilovich, D. I., and S. Nagaraj. 2009. Myeloid-derived suppressor cells asregulators of the immune system. Nat. Rev. Immunol. 9: 162–174.
11. Schumacher, T. N. 2002. T-cell-receptor gene therapy. Nat. Rev. Immunol. 2:512–519.
12. Rosenberg, S. A., N. P. Restifo, J. C. Yang, R. A. Morgan, and M. E. Dudley.2008. Adoptive cell transfer: a clinical path to effective cancer immunotherapy.Nat. Rev. Cancer 8: 299–308.
13. Brenner, M. K., and H. E. Heslop. 2010. Adoptive T cell therapy of cancer. Curr.Opin. Immunol. 22: 251–257.
14. Quintarelli, C., G. Dotti, S. T. Hasan, B. De Angelis, V. Hoyos, S. Errichiello,M. Mims, L. Luciano, J. Shafer, A. M. Leen, et al. 2011. High-avidity cytotoxicT lymphocytes specific for a new PRAME-derived peptide can target leukemicand leukemic-precursor cells. Blood 117: 3353–3362.
15. Rosenberg, S. A. 2011. Cell transfer immunotherapy for metastatic solid cancer—what clinicians need to know. Nat. Rev. Clin. Oncol. 8: 577–585.
16. June, C. H. 2007. Adoptive T cell therapy for cancer in the clinic. J. Clin. Invest.117: 1466–1476.
17. Pascolo, S., N. Bervas, J. M. Ure, A. G. Smith, F. A. Lemonnier, andB. Perarnau. 1997. HLA-A2.1-restricted education and cytolytic activity of CD8(+) T lymphocytes from beta2 microglobulin (beta2m) HLA-A2.1 monochaintransgenic H-2Db beta2m double knockout mice. J. Exp. Med. 185: 2043–2051.
18. Dalton, D. K., S. Pitts-Meek, S. Keshav, I. S. Figari, A. Bradley, andT. A. Stewart. 1993. Multiple defects of immune cell function in mice withdisrupted interferon-gamma genes. Science 259: 1739–1742.
19. Hogquist, K. A., S. C. Jameson, W. R. Heath, J. L. Howard, M. J. Bevan, andF. R. Carbone. 1994. T cell receptor antagonist peptides induce positive selec-tion. Cell 76: 17–27.
20. Hildner, K., B. T. Edelson, W. E. Purtha, M. Diamond, H. Matsushita,M. Kohyama, B. Calderon, B. U. Schraml, E. R. Unanue, M. S. Diamond, et al.2008. Batf3 deficiency reveals a critical role for CD8alpha+ dendritic cells incytotoxic T cell immunity. Science 322: 1097–1100.
21. Mueller, S. N., W. Heath, J. D. McLain, F. R. Carbone, and C. M. Jones. 2002.Characterization of two TCR transgenic mouse lines specific for herpes simplexvirus. Immunol. Cell Biol. 80: 156–163.
22. Parish, I. A., J. Waithman, G. M. Davey, G. T. Belz, J. D. Mintern, C. Kurts,R. M. Sutherland, F. R. Carbone, and W. R. Heath. 2009. Tissue destructioncaused by cytotoxic T lymphocytes induces deletional tolerance. Proc. Natl.Acad. Sci. USA 106: 3901–3906.
23. Guillonneau, C., J. D. Mintern, F. X. Hubert, A. C. Hurt, G. S. Besra, S. Porcelli,I. G. Barr, P. C. Doherty, D. I. Godfrey, and S. J. Turner. 2009. CombinedNKT cell activation and influenza virus vaccination boosts memory CTL gen-eration and protective immunity. Proc. Natl. Acad. Sci. USA 106: 3330–3335.
24. Vremec, D., J. Pooley, H. Hochrein, L. Wu, and K. Shortman. 2000. CD4 andCD8 expression by dendritic cell subtypes in mouse thymus and spleen. J.Immunol. 164: 2978–2986.
25. Henri, S., D. Vremec, A. Kamath, J. Waithman, S. Williams, C. Benoist,K. Burnham, S. Saeland, E. Handman, and K. Shortman. 2001. The dendritic cellpopulations of mouse lymph nodes. J. Immunol. 167: 741–748.
26. Zhan, Y., Y. Zhang, D. Gray, E. M. Carrington, P. Bouillet, H. J. Ko, L. O’Reilly,I. P. Wicks, A. Strasser, and A. M. Lew. 2011. Defects in the Bcl-2-regulatedapoptotic pathway lead to preferential increase of CD25 low Foxp3+ anergicCD4+ T cells. J. Immunol. 187: 1566–1577.
27. Prato, S., J. D. Mintern, M. H. Lahoud, D. C. Huang, and J. A. Villadangos. 2011.Induction of antigen-specific effector-phase tolerance following vaccinationagainst a previously ignored B-cell lymphoma. Immunol. Cell Biol. 89: 595–603.
28. Villadangos, J. A., and P. Schnorrer. 2007. Intrinsic and cooperative antigen-presenting functions of dendritic-cell subsets in vivo. Nat. Rev. Immunol. 7: 543–555.
29. Villadangos, J. A., and L. Young. 2008. Antigen-presentation properties ofplasmacytoid dendritic cells. Immunity 29: 352–361.
30. Inaba, K., M. Witmer-Pack, M. Inaba, K. S. Hathcock, H. Sakuta, M. Azuma,H. Yagita, K. Okumura, P. S. Linsley, S. Ikehara, et al. 1994. The tissue distri-bution of the B7-2 costimulator in mice: abundant expression on dendritic cellsin situ and during maturation in vitro. J. Exp. Med. 180: 1849–1860.
31. Vremec, D., and K. Shortman. 1997. Dendritic cell subtypes in mouse lymphoidorgans: cross-correlation of surface markers, changes with incubation, and dif-ferences among thymus, spleen, and lymph nodes. J. Immunol. 159: 565–573.
32. Wilson, N. S., D. El-Sukkari, G. T. Belz, C. M. Smith, R. J. Steptoe, W. R. Heath,K. Shortman, and J. A. Villadangos. 2003. Most lymphoid organ dendritic celltypes are phenotypically and functionally immature. Blood 102: 2187–2194.
33. Lin, M. L., Y. Zhan, J. A. Villadangos, and A. M. Lew. 2008. The cell biology ofcross-presentation and the role of dendritic cell subsets. Immunol. Cell Biol. 86:353–362.
34. McDonnell, A. M., A. C. Prosser, I. van Bruggen, B. W. Robinson, andA. J. Currie. 2010. CD8alpha+ DC are not the sole subset cross-presenting cell-associated tumor antigens from a solid tumor. Eur. J. Immunol. 40: 1617–1627.
The Journal of Immunology 3543
by guest on April 24, 2018
http://ww
w.jim
munol.org/
Dow
nloaded from
35. Asano, K., A. Nabeyama, Y. Miyake, C. H. Qiu, A. Kurita, M. Tomura,O. Kanagawa, S. Fujii, and M. Tanaka. 2011. CD169-positive macrophagesdominate antitumor immunity by crosspresenting dead cell-associated antigens.Immunity 34: 85–95.
36. Keir, M. E., M. J. Butte, G. J. Freeman, and A. H. Sharpe. 2008. PD-1 and itsligands in tolerance and immunity. Annu. Rev. Immunol. 26: 677–704.
37. Griffith, T. S., and T. A. Ferguson. 2011. Cell death in the maintenance andabrogation of tolerance: the five Ws of dying cells. Immunity 35: 456–466.
38. DuPage, M., C. Mazumdar, L. M. Schmidt, A. F. Cheung, and T. Jacks. 2012.Expression of tumour-specific antigens underlies cancer immunoediting. Nature482: 405–409.
39. Vacchelli, E., A. Eggermont, W. H. Fridman, J. Galon, E. Tartour, L. Zitvogel,G. Kroemer, and L. Galluzzi. 2013. Trial Watch: Adoptive cell transfer for an-ticancer immunotherapy. OncoImmunology 2: e24238.
40. Grupp, S. A., and C. H. June. 2011. Adoptive cellular therapy. Curr. Top.Microbiol. Immunol. 344: 149–172.
41. Mueller, S. N., and R. Ahmed. 2009. High antigen levels are the cause of T cell ex-haustion during chronic viral infection. Proc. Natl. Acad. Sci. USA 106: 8623–8628.
42. Zheng, Y., Y. Zha, G. Driessens, F. Locke, and T. F. Gajewski. 2012. Tran-scriptional regulator early growth response gene 2 (Egr2) is required for T cellanergy in vitro and in vivo. J. Exp. Med. 209: 2157–2163.
43. Lu, B., and O. J. Finn. 2008. T-cell death and cancer immune tolerance. CellDeath Differ. 15: 70–79.
44. Mrass, P., H. Takano, L. G. Ng, S. Daxini, M. O. Lasaro, A. Iparraguirre,L. L. Cavanagh, U. H. von Andrian, H. C. Ertl, P. G. Haydon, and W. Weninger.2006. Random migration precedes stable target cell interactions of tumor-infiltrating T cells. J. Exp. Med. 203: 2749–2761.
45. Boissonnas, A., L. Fetler, I. S. Zeelenberg, S. Hugues, and S. Amigorena. 2007.In vivo imaging of cytotoxic T cell infiltration and elimination of a solid tumor.J. Exp. Med. 204: 345–356.
46. Pittet, M. J., J. Grimm, C. R. Berger, T. Tamura, G. Wojtkiewicz, M. Nahrendorf,P. Romero, F. K. Swirski, and R. Weissleder. 2007. In vivo imaging of T celldelivery to tumors after adoptive transfer therapy. Proc. Natl. Acad. Sci. USA104: 12457–12461.
47. Breart, B., F. Lemaıtre, S. Celli, and P. Bousso. 2008. Two-photon imaging ofintratumoral CD8+ T cell cytotoxic activity during adoptive T cell therapy inmice. J. Clin. Invest. 118: 1390–1397.
48. Serafini, P., S. Mgebroff, K. Noonan, and I. Borrello. 2008. Myeloid-derivedsuppressor cells promote cross-tolerance in B-cell lymphoma by expandingregulatory T cells. Cancer Res. 68: 5439–5449.
49. Nagaraj, S., A. G. Schrum, H. I. Cho, E. Celis, and D. I. Gabrilovich. 2010.Mechanism of T cell tolerance induced by myeloid-derived suppressor cells. J.Immunol. 184: 3106–3116.
50. Engelhardt, J. J., B. Boldajipour, P. Beemiller, P. Pandurangi, C. Sorensen,Z. Werb, M. Egeblad, and M. F. Krummel. 2012. Marginating dendritic cells ofthe tumor microenvironment cross-present tumor antigens and stably engagetumor-specific T cells. Cancer Cell 21: 402–417.
51. Nagaraj, S., K. Gupta, V. Pisarev, L. Kinarsky, S. Sherman, L. Kang,D. L. Herber, J. Schneck, and D. I. Gabrilovich. 2007. Altered recognition ofantigen is a mechanism of CD8+ T cell tolerance in cancer. Nat. Med. 13: 828–835.
52. Movahedi, K., M. Guilliams, J. Van den Bossche, R. Van den Bergh,C. Gysemans, A. Beschin, P. De Baetselier, and J. A. Van Ginderachter. 2008.Identification of discrete tumor-induced myeloid-derived suppressor cell sub-populations with distinct T cell-suppressive activity. Blood 111: 4233–4244.
53. Landsberg, J., J. Kohlmeyer, M. Renn, T. Bald, M. Rogava, M. Cron, M. Fatho,V. Lennerz, T. Wolfel, M. Holzel, and T. T€uting. 2012. Melanomas resist T-celltherapy through inflammation-induced reversible dedifferentiation. Nature 490:412–416.
54. Budhu, S., J. D. Loike, A. Pandolfi, S. Han, G. Catalano, A. Constantinescu,R. Clynes, and S. C. Silverstein. 2010. CD8+ T cell concentration determinestheir efficiency in killing cognate antigen-expressing syngeneic mammalian cellsin vitro and in mouse tissues. J. Exp. Med. 207: 223–235.
55. Amos, S. M., C. P. Duong, J. A. Westwood, D. S. Ritchie, R. P. Junghans,P. K. Darcy, and M. H. Kershaw. 2011. Autoimmunity associated with immu-notherapy of cancer. Blood 118: 499–509.
56. Woo, S. R., M. E. Turnis, M. V. Goldberg, J. Bankoti, M. Selby, C. J. Nirschl,M. L. Bettini, D. M. Gravano, P. Vogel, C. L. Liu, et al. 2012. Immune inhibitorymolecules LAG-3 and PD-1 synergistically regulate T-cell function to promotetumoral immune escape. Cancer Res. 72: 917–927.
57. Kenna, T. J., T. Waldie, A. McNally, M. Thomson, H. Yagita, R. Thomas, andR. J. Steptoe. 2010. Targeting antigen to diverse APCs inactivates memory CD8+T cells without eliciting tissue-destructive effector function. J. Immunol. 184:598–606.
3544 THE APC MEDIATING CTL INACTIVATION
by guest on April 24, 2018
http://ww
w.jim
munol.org/
Dow
nloaded from
