Proc. Natl. Acad. Sci. USAVol. 90, pp. 2774-2778, April 1993Immunology

Mechanisms of rejection induced by tumor cell-targeted genetransfer of interleukin 2, interleukin 4, interleukin 7, tumornecrosis factor, or interferon yHANNO HOCK*, MARION DORSCH, ULRICH KUNZENDORF, ZHIHAI QIN, TIBOR DIAMANTSTEIN,AND THOMAS BLANKENSTEINInstitute of Immunology, Universitatsklinikum Steglitz, Freie UniversitWt Berlin, Hindenburgdamm 27, 1000 Berlin 45, Federal Republic of Germany

Communicated by Avrion Mitchison, December 17, 1992

ABSTRACT Interleukin (IL)-2, IL-4, IL-7, tumor necro-sis factor (TNF), or interferon-y (IFN-y) has been shown to beable to induce tumor rejection if produced locally by the tumorcells after gene transfer. To analyze whether the cellularrejection mechanisms are different or redundant we haveexpressed the cytokines in the same tumor cell line (J558L).Cell depletion experiments revealed that all cytokines requiredCD8+ T cells for complete long-term tumor eradication, al-though effective but transient host-dependent tumor suppres-sion was also observed in the complete absence ofCD8+ T cells.The transient tumor suppression induced by IL-2, IL-4, TNF,or IFN-y was also operative in nude and severe combinedimmunodeficient mice, whereas only tumor suppression in-duced by IL-7 was dependent on the presence of CD4+ T cellsand was not evident in nude mice. The T-cell-independenteffector arm og IL-2 and IFN-- but not IL-4 and TNF wasmediated in part by natural killer cells. The transience oftumorsuppression in the absence of T cells reflected loss of cytokineproduction in the case of TNF, IL-2, and IL-4 but not IFN-y.Immunohistologic analysis revealed all cytokine-producing tu-mors to be heavlly infiltrated by macrophages. IL-4 and IL-7tumors additionally contained eosinophlls. The infiltration byT cells did not necessarily reflect their contribution to tumorrejection. Thus, the different cytokines activate heterogeneoustransient tumor-suppressive mechanisms but always requireCD8+ T cells for complete tumor rejection.

Transfer of cytokine genes into tumor cells has proven avaluable approach for the analysis of cytokine-mediatedeffects on tumor growth [interleukin (IL)-2 (1-4), IL-4 (5-9),IL-7 (10-12), tumor necrosis factor (TNF) (13-17), interferony (IFN-y) (18-20), macrophage colony-stimulating factor(M-CSF) (21), granulocyte-CSF (22), monocyte chemotacticand activating factor (23, 24), IL-1 (25)]. An unexpectednumber of cytokines has been shown to be able to inhibittumor growth by activation of host-dependent anti-tumorresponses. The mechanisms underlying cytokine-inducedtumor rejection are of pivotal interest as a contribution to arational basis for future immunotherapy and may serve tofurther enlarge the still limited understanding of cytokineaction in vivo. However, the immunological mechanisms bywhich tumors are rejected after cytokine gene transfer areinfluenced not only by the cytokine but also by the tumormodel. Unless analyzed in the same model, differences in themechanism of tumor rejection cannot conclusively be attrib-uted to the cytokines. Parameters such as the mouse strain,the level of cytokine expression, or the tumor cell lineinfluence the mode of rejection. For example, tumor cellsmay differ in the susceptibility for host effector mechanismsbecause of different (i) surface molecule expression [e.g.,

major histocompatibility complex (MHC), adhesion mole-cules], (ii) soluble mediators other than the transfectedcytokine gene, (iii) growth kinetics, and (iv) immunogenicity.In an effort to more directly compare the mechanisms oftumor rejection induced by different cytokines, we haveanalyzed five cytokines in parallel in the same tumor cell linewith respect to the cellular requirements for tumor rejection.

MATERIALS AND METHODSAnimals. Six-week-old female BALB/c mice or BALB/c

nu/nu mice, CB17 severe combined immunodeficient (SCID)or SCID/beige mice were obtained from the Zentralinstitutfur Versuchstierzucht (Hannover, F.R.G.) or from the Ge-sellschaft fur Strahlenschutz, Munich. NIH III mice (carry-ing nude, XID, and beige mutations) were obtained fromCharles River Breeding Laboratories.Tumor Cell Line and Genetically Modified Variants. J558L

plasmacytoma cells were cultured as published (21). Theyneither produce IL-2, IL-4, IL-7, TNF, or IFN-y as deter-mined by the assays described below nor IL-1, IL-3, IL-5,IL-6, M-CSF, or lymphotoxin (not shown). None of thecytokine-producing variants differs from parental cells incytokine production except for the transfected cytokine.J558L cells express MHC class I and cellular adhesionmolecule ICAM I but no detectable MHC class II moleculesas determined by fluorescence staining with monoclonalantibodies (mAbs) M142, M5-114, and YN1-1.7 (ATCC) (notshown). None of the transfected variants markedly differsfrom J558L cells with respect to MHC or ICAM expression.The generation of J558-IL-4 (6), J558-TNF (referred to asJB62 in ref. 17), and J558-IL-7 cells (10) has been described.J558-IL-4, which was established by retroviral infection, wasconfirmed to be helper virus free by reverse transcriptionassay after prolonged culture (not shown). For establishingcell lines J558-IL-2 and J558-IFN, the respective mousecDNAs were isolated by PCR using reverse transcribed totalRNA of ConA-activated spleen T-cell blasts according topublished protocols (10, 21) (primers for IL-2: 5'-TGGGATCCTGCAGGCATGTACAGCATGC-3' and 5'-GTGGATCCGGTACATAGTTATTGAGGGCTTG-3';primers for IFN-y. 5'-CCGAGCGGATTCACCATGAACG-CTACACACTGCATCTTGGC-3' and 5'-CGGCCGGATC-CCGAATCAGCAAGCGACTCCTTTTCCGCTTCCT-GAG-3'). DNA fragments were cloned into the Bgl II site ofexpression vector pLTR (10), resulting in plasmids pLTR-IL-2 and pLTR-IFN, which were used to transfect J558Lcells along with plasmid pWLneo as described (10). Trans-fected cells were selected in G418-containing medium (1

Abbreviations: M-CSF, macrophage colony-stimulating factor; IL,interleukin; NK, natural killer; mAb, monoclonal antibody; TNF,tumor necrosis factor; IFN-y, interferon y; SCID, severe combinedimmunodeficient; MHC, major histocompatibility complex.*To whom reprint requests should be addressed.

2774

The publication costs of this article were defrayed in part by page chargepayment. This article must therefore be hereby marked "advertisement"in accordance with 18 U.S.C. §1734 solely to indicate this fact.

Dow

nloa

ded

by g

uest

on

Nov

embe

r 6,

202

0

Proc. Natl. Acad. Sci. USA 90 (1993) 2775

mg/ml) for 2 weeks, cloned, and tested for activity in thesupernatant. The transfectants produced the followingamounts of the transfected cytokine: J558-IL-2, 40 units/ml;J558-IL-4, 75 units/ml; J558-IL-7, 50 units/ml; J558-TNF, 17units/ml; and J558-IFN, 120 ng/ml.

Detection of Cytokine Activities. IL-2 and IL-4 were mea-sured by proliferation assays with CTLL-2 (26) or CT.4S (27)cell lines. IL-7 activity was determined by a thymocyteproliferation assay as described (10). TNF was measured bythe L929 cytotoxicity assay as described (14). Units/milliliterwere defined as the reciprocal of supematant dilution (cellswere cultured for 48 hr at a concentration of 1 x 106 per ml)yielding half-maximal proliferation/cytotoxicity. IFN-y wasmeasured by an ELISA (Holland Biotechnology, Leiden,The Netherlands) and bioactivity was confirmed by fluoro-cytometric analysis of induction of MHC class II expressionon WEHI 3 cells as described (21). Specificity was confirmedby addition of blocking mAbs for each cytokine to thebioassays (20 ,g/ml): anti-IL-2 S4B6 (28), anti-IL-4 llBl1(29), anti-IL-7 mAb (1689-01; Genzyme, Cambridge, MA),anti-IFN-y XMG6 (30), and rabbit anti-mouse TNF serum(IP400; Genzyme).

Analysis of Tumor Growth. Tumor growth was analyzed asdescribed (10, 14, 21). Cells (4 x 106) in 0.2 ml of Dulbecco'sphosphate-buffered saline (D-PBS) were injected s.c. into thebelly region. Tumor size was determined as the mean of thelargest diameter and the diameter at right angle.Antibody Treatments in Vivo. Asialo-GM-1 antiserum

(Wako, Neuss, F.R.G.) was injected into the tail vein of themice at days -2, -1, 1, 4, 7, 10, 13, 16, and 19 of tumor cellinjection (50 ,ul of antiserum diluted to 200 ,ul in PBS perinjection). Two percent heat-inactivated normal rabbit serumdiluted in PBS served as a control. Functional depletion ofnatural killer (NK) cells for 3 weeks was controlled by a 4-hr51Cr release assay with 1 x 104 YAC cells as targets andspleen cells as effector cells at effector-to-target ratios of100:1, 50:1, 25:1, 10:1, and 2:1. mAbs were partially purifiedfrom ascites fluids of nude mice by precipitation with 45%ammonium sulfate and subsequent dialysis against PBS. Theconcentration of rat immunoglobulin was determined byradial immunodiffusion (Serotec; no. AAR02Z). For in vivodepletion experiments, 0.5 ml of a PBS solution containing0.5 mg of the rat mAbs GK 1.5 (CD4), 2.43 (CD8) (ATCC),and 23-378 (isotype-matched control mAb, rat IgG2b; un-published) was injected i.p. starting at day -1 and once perweek thereafter. Throughout the experiments (7 weeks), thistreatment specifically depleted the respective subpopulationas determined by indirect staining and cytofluorometric anal-ysis of spleen and/or lymph node cells with mAbs GK 1.5(CD4) and 53-6.72 (CD8) (ATCC) (data not shown).Immunohistochemistry. Immunohistochemical analysis

was done as described (10) using the following mAbs: anti-CR3 (Mac-1) mAb M1/70 (IgG2b) (31); anti-Mac-3 mAbM3-84 (IgGi) (32); anti-macrophage mAb F4/80 (IgG2b) (33);anti-CD4 mAb RM 4-5 (IgG2a; Paesel and Lorei, Frankfurt);anti-CD8 mAb 53-6.72 (IgG2a; Becton Dickinson); mAbNimp-R6 (IgG2b), which recognizes predominantly eosino-phils (34); and isotype-matched control mAbs.

RESULTSJ558L Cells Transfected To Produce IL-2, IL-4, IL-7, TNF,

or IFN-y Are Rejected in Syngeneic BALB/c Mice. The celllines J558-IL-2, -IL-4, -IL-7, -TNF, and -IFN all derive fromJ558L cells and produce the respective cytokine after genetransfer. These five clones were chosen because each pro-duced a sufficient amount of the respective cytokine tosuppress tumor growth in vivo. Moreover, they had retainedan unchanged doubling time in vitro after gene transfer. Uponinjection of 4 x 106 cells into syngeneic mice, none of the

mice injected with the cytokine producers, but all injectedwith parental cells, developed a tumor within 2 weeks. Tumorsuppression in all cases most likely was mediated by thetransfected cytokine because (i) for all vectors used in thisstudy we have analyzed several mock transfected cloneswhose tumor growth was not suppressed, (ii) we haveexpressed several cytokines in J558L cells that are not tumorsuppressive [e.g., IL-6 (14), M-CSF (21), IL-5, human IL-8;unpublished results], and (iii) tumor suppression could eitherbe abolished by parallel injection of anti-cytokine mAbs(J558-IL-4, -IL-7, -IFN) or correlated with the amount ofcytokine production when tumor growth of different cloneswas analyzed (J558-IL-2, -TNF) (not shown).Tumor Growth of Cytokine-Producing J558L Cells in Syn-

geneic BALB/c Mice Compared to Immunodeficient MouseStrains. The cellular requirements for the tumor rejectioninduced by the different cytokines were assessed by analyz-ing tumor growth in BALB/c compared to immunocompro-mised nude (T-cell deficient), SCID (T- and B-cell deficient),SCID/beige (T-, B-, and NK-cell deficient), and NIH III mice(T, B, and NK deficient). In contrast to BALB/c mice, whichwith few exceptions completely rejected all cytokine-producing J558L variants, almost all nude mice or SCID micethat were injected in parallel eventually developed a tumor(Table 1). However, marked differences existed in the kinet-ics of tumor growth comparing cytokine producers andparental cells in T-cell-deficient mice: J558-IL-2, -IL-4,-TNF, and -IFN cells gave rise to tumors with a delay ofabout 3 weeks compared to parental or J558-IL-7 tumors,indicating these cytokines were able to evoke T-cell-independent mechanisms capable of suppressing tumorgrowth in the early phase after tumor cell injection. Tumorgrowth in SCID/beige and NIH III mice was largely indis-tinguishable from that in nude mice (Table 1), giving nofurther clue about these T-cell-independent mechanisms.To investigate the reason for the late outgrowth of tumors,

a representative number ofJ558-IL2, -IL4, -TNF, and -IFN-ytumors from nude mice, all ofwhich had been injected at least30 days previously, were isolated (as described in ref. 21) andassayed for cytokine activity. All J558-IL-2 (9/9 tumors),-IL-4 (9/9), or -TNF (8/8) cells had ceased or drasticallydecreased their cytokine production. However, when tumorsuppression in the early phase was blocked by neutralizingantibodies against the cytokine (IL-4) (2/2) or anti-asialo-GM-1 antiserum (IL-2) (2/2), the tumor cells that werereisolated proved to have an unchanged cytokine production.The cytokine secretion of J558-IFN cells was not diminishedin 10/12 tumors after in vivo growth in nude mice, althoughthe in vivo growth kinetics were very similar to those ofJ558-IL-2, -IL-4, and -TNF cells. Reinjection of cells isolatedfrom J558-IFN tumors revealed that they grew as expectedfrom their cytokine production: the nonproducers grew likeJ558L tumors; the IFN-y producers were initially suppressedin nude and rejected in BALB/c mice (not shown).Tumor Growth of Cytokine-Producing J558L Cells in

Asialo-GM-1+ Cell-Depleted Nude Mice. Spleen cells derivedfrom SCID/beige or NIH III mice were not completelydeficient of NK cell activity (data not shown). Thus wefurther studied the role ofNK cells in the tumor suppressionof J558-IL-2, -IL-4, -TNF, and -IFN cells in asialo-GM-1+cell-depleted nude mice. Growth of J558-TNF or J558-IL-4tumors was either not (IL-4) or only slightly (TNF) acceler-ated by the anti-asialo-GM-1 serum (not shown). The mar-ginal effect on J558-TNF tumor growth seems negligibleconsidering that the antiserum also accelerates the growth ofparental J558L cells (Fig. 1). However, the asialo-GM-1antiserum clearly interfered with the anti-tumor activities ofIL-2 and IFN-y and restored early tumor growth (Fig. 1). Itshould be noted, however, that these tumors still showed agrowth retardation in comparison to parental cells, suggest-

Immunology: Hock et al.

Dow

nloa

ded

by g

uest

on

Nov

embe

r 6,

202

0

Proc. Natl. Acad. Sci. USA 90 (1993)

Table 1. Analysis of tumor growth of cytokine-producing J558L cells in immunodeficient miceNo. of mice with tumor/no. of injected mice

Cell line Exp. BALB/c Nude SCID SCID/beige NIH III

J558L 1 5/5 5/5 5/5 (10 ± 1) 5/5 ND2 5/5 (10 ± 1) 5/5 (10 ± 1) ND 4/4 (10 ± 1) 4/5 (14 ± 1)

J558-IL-2 1 2/5 4/5 5/5 4/5 ND2 0/5 (50 ± 0) 4/4 (34 ± 5) 4/5 (38 ± 6) 4/5 (33 ± 4) 4/4 (30 ± 2)

J558-IL-4 1 2/5 5/5 4/4 (31 ± 5) 3/3 ND2 1/5 (33 ± 4) 5/5 (27 ± 2) ND 3/3 (33 ± 2) 4/4 (26 ± 3)

J558-IFN 1 0/5 4/5 5/5 (33 ± 3) 5/5 ND2 0/5 3/5 (41 ± 6) ND 2/4 (49 ± 1) 3/6 (41 ± 4)

J558-TNF 1 1/5 5/5 5/5 (31 ± 5) 5/5 ND2 0/5 (28 ± 0) 5/5 (27 ± 1) ND 3/4 (32 ± 5) 5/5 (28 ± 3)

J558-IL-7 1 0/5 4/5 (12 ± 1) ND ND ND

The indicated cells (4 x 106) were injected in parallel in the indicated mouse strains in two independent experiments.Tumor incidence and latency (mean day ofoccurrence oftumors >1 cm ± SD) are given. When two experiments were done,the indicated latency refers to the combined data of both experiments. ND, not done.

ing that asialo-GM-1+ cells contribute to IL-2- and IFN-y-induced tumor suppression, but that further non-T-cell ef-fector mechanisms are likely to be involved.Tumor Growth of Cytokine-Producing J558L Cells in Mice

Depleted of CD4+ or CD8+ T Cells. Since complete tumoreradication did not occur in T-cell-deficient animals, the roleof T-cell subpopulations was further analyzed by mAbsselectively eliminating either CD4+ or CD8+ T cells for >7weeks. As shown in Fig. 2, depletion of CD4+ cells did notaffect tumor growth ofJ558-IL-2, -IL-4, -IFN, or -TNF cells;this is in contrast to J558-IL-7 cells, which grew almost likeparental J558L tumors. Depletion of CD8+ T cells had a verysimilar effect on the growth of all cytokine gene-transfectedtumors: tumor suppression in the early phase was unaffected,but with a 3- to 5-week delay as compared to J558L, tumors

3-

u 2-a)N

0

E 1-

0-

3

u 2

N

0

oE1

J558L +

IJ558-1L2 + Asiolo

J58-1L2

_ _ - J558-IL2 + NRSI-- t 7

0 8 10 12 14 16 18 20 22 24 26 28

J558L + Asialo)[NR

J558L + NRS |J558-IFN Asialo

J558-IFN + NRS

ff.....= J558-IFN

0 8 10 12 14 16 18 20 22 24 26 28days

FIG. 1. Tumor growth of cytokine-producing J558L cells in nudemice depleted for asialo-GM-1+ cells shows the participation of NKcells during IL-2- (Upper) and IFN-y- (Lower) [but not IL-4- andTNF- (not shown)] induced tumor suppression. One of two experi-ments with similar results and five mice per group is shown. Theindicated cells (4 x 106) were injected into nude mice, which weretreated with asialo-GM-1 antiserum (Asialo) or normal rabbit serum(NRS), as indicated.

developed and progressively grew in most cases from all fivegroups. Tumor growth in CD8+ cell-depleted mice closelyresembled that in nude mice except for J558-IL-7 tumors,where tumor growth in nude mice was unimpaired.

Immunohistological Analysis of Cytokine-Producing Tu-mors. Immunohistological analysis of tissue obtained fromthe site of tumor cell injection between days 3 and 9 (in somecases, 13) enabled us to directly compare the consequencesof cytokine production on the infiltrate (Table 2). OnlyJ558-IL-2, -IL-7, and -TNF tumors were infiltrated by T cellsalready at early time points. J558-IL-2 tumors were homog-enously infiltrated by CD8+ T cells, whereas CD4+ cells werescarcely detectable. J558-IL-7 tumors and, even more rap-idly, J558-TNF tumors were densely infiltrated by CD4+ andCD8+ T cells. In some, but not all, J558-IFN tumors a subtleand nonhomogenous infiltration by CD8+ T cells was ob-served. In J558-IL-4 tumors T cells were absent in the firstweek but could be detected (scattered) at later time points.The predominant cells found in the infiltrate of all cytokineproducers were most likely macrophages, identified by theirmorphology and positive staining with anti-CR3 (31), -Mac3(32), and F4/80 (33) mAbs. Whereas in J558-IL-2 and J558-TNF tumors CR3+ cells appeared to be more prominentwithin the tumor, IL-4 and IL-7 also induced a very pro-nounced rim around the tumor (not shown). J558-IFN tumorswere least infiltrated by CR3+ cells. Eosinophils identified bymAb Nimp-R6 (34) and their characteristic ring-like nucleuswere observed to considerably infiltrate J558-IL4 and J558-IL-7 tumors. Notably, we also observed a situation in whichcytokine production by J558L cells did not induce anydetectable tumor infiltrate: in progressively growing J558-IL-6 tumors (14) the tumor infiltrate did not differ fromparental tumors (unpublished observation).

DISCUSSIONThe most striking similarity in the tumor rejection induced byall five cytokines was the requirement of CD8+ T cells forcomplete tumor eradication. However, CD8+ T cells werenot necessary for initial tumor suppression, because tumorsstarted to progressively grow only after a latency period inCD8+ T-cell-depleted mice. Similarly, tumors producingIL-2, IL-4, IFN-y, and TNF were initially suppressed in theirgrowth in T-cell-deficient mouse strains. IL-7 differs from theother cytokines inasmuch as CD4+ T cells were needed forthe activation of the non-T-effector arm (10). The commonnecessity of CD8+ T cells for complete tumor eradication byall five cytokines does not prove that all of them are able toparacrinely stimulate these cells. Considering the observa-tion that tumors started to progressively grow only after a

2776 Immunology: Hock et al.

Dow

nloa

ded

by g

uest

on

Nov

embe

r 6,

202

0

Proc. Natl. Acad. Sci. USA 90 (1993) 2777

0 10 20 30 40 50days 4

I1 m---l0 10 20 30 40 50 60 90

days10 20 30 40 50

days f

FIG. 2. Analysis of CD4+ or CD8+ T-cell contribution to cytokine-induced tumor suppression. Tumor growth of parental J558L cells inBALB/c mice (o) or J558 variants producing the indicated cytokine was followed either in BALB/c mice without treatment (A), mice treatedwith anti-CD4 (*), anti-CD8 (O), or control mAb (o) to deplete the respective T-cell subpopulation for 50 days (arrows), or nude mice (0). Thecombined results of three experiments with four or five mice per group are shown.

latency period in T-cell-deficient or CD8+ T-cell-depletedanimals, it is equally possible that non-T-effector mecha-nisms keep tumor cells from rapidly expanding, which maygive CD8+ T cells the chance to mediate complete tumor celleradication even without being a direct target for the cytokineproduced by the tumor cells. In the case of tumor rejectioninduced by IL-2, IL-4, TNF, and IFN-'y, CD8+ T cells canserve their function also in long-term CD4+ T-cell-depletedanimals (Fig. 2), indicating that CD4+ helper cell activityeither is not needed or can be compensated for by thecytokine production of the tumor cells. The discontinuationof tumor suppression of J558-IL-2, -IL-4, and -TNF in theabsence of T cells was always associated with loss of cyto-kine production. Loss of cytokine production during in vivogrowth has already been observed by others (4, 5, 12, 20) andis probably due to preferential killing of cytokine-producingtumor cells and selection of non-cytokine-producing vari-ants. Our results indicate that CD8+ T cells are of pivotalimportance for prevention of the outgrowth of non-cytokine-producing variants. This system, however, is not infalliblesince a few mice developed late tumors also in the presenceofT cells and, conversely, some mice completely rejected thetumors in the absence of T cells. The growth of J558-IFNtumors in nude mice illustrates that tumor suppression in theabsence of T cells can be transient, although most tumorsretain the capability to produce undiminished amounts ofIFN-y [similarly for J558-IL-7 tumors in CD8+ T-cell-depleted mice (preliminary data)]. In any case, tumor sup-pression in the absence of CD8+ T cells seemed to bedependent on the continuous presence of the cytokine evenif this is not always sufficient for long-term suppression.The histology of T cells in regressing tumors in the early

phase after tumor cell injection allowed no clear prediction oftheir role in the rejection process. CD8+ T cells were abun-dantly present early in J558-IL-2, -IL-7, and -TNF tumors,although they were not needed for tumor suppression duringthis period. Conversely, the role for CD8+ T cells in long-term eradication did not require their early presence sincethey were only scarcely observed early in the infiltrate of

J558-IL-4 or -IFN tumors. CD4+ T cells were found toinfiltrate J558-IL-7 and -TNF tumors but for J558-TNF, incontrast to J558-IL-7, their depletion had no effect.

Analysis of T-cell-independent effectors mediating tumorsuppression suggests that these mechanisms are at least inpart heterogenous. Treatment with an antiserum to asialo-GM-1 could partially restore J558-IL-2 and -IFN, but notJ558-IL-4 and -TNF, tumor growth in nude mice. There is anapparent contradiction between the effect ofthe asialo-GM-1antiserum (Fig. 1) and the lack ofany influence ofthe "beige"mutation (Table 1), but NK cells are not totally absent incrosses such as SCID/beige or NIH III mice and, moreover,cytokines have been shown to be able to amplify "NK"activity in these strains (35). The interpretation that indeedNK cell activity was affected by the anti-asialo-GM-1 serumis supported by the finding that an effect very similar to thatof anti-asialo-GM-1 was observed with the mAb 30-H12 (notshown) directed against Thy-1.2, which is present on 90% ofthe NK cells in nude mice (36). A recent review proposed acrucial role for granulocytes in the tumor inhibition andinduction of T-cell responses by several cytokines (37).However, a final assessment of their role was hampered bythe lack of reagents for their specific depletion or inactiva-tion. Using an antibody against granulocytes, at least theeosinophils observed in IL-4-producing tumors have beenshown to be important for the tumor-suppressive effect (8).We could detect eosinophils not only in J558-IL-4 but also inJ558-IL-7 tumors; their presence in J558-IL-7 tumors is likethe presence of macrophages dependent on CD4+ T cells(data not shown). The macrophages histologically demon-strated in all cytokine-producing tumors under study arefurther potential effector cells. For TNF, previous experi-ments showed that tumor suppression depended on CR3+cells, most probably macrophages, even though the blockingmAb was not completely specific (14). It cannot be excludedthat, in some cases, the macrophages are rather a result thana cause of tumor cell destruction. Notably, expression ofM-CSF in the same tumor cell line led to macrophageinfiltration, but only marginally delayed tumor growth (21). It

'I--E

aF-

days

v 1 ,W

_ A rS ~ 0-

E- 1(cE

a)a)

a)

j-

60 90

Immunology: Hock et al.

Dow

nloa

ded

by g

uest

on

Nov

embe

r 6,

202

0

2778 Immunology: Hock et al.

Table 2. Immunohistochemistry of cytokine-producingJ558L tumors

Tumor infiltrate

CD4+ CD8+ CR3+ Eosino-T cells T cells cells phils

Day Day Day Day Day 3/ Day 3/Tumor 3 9 3 9 day 9 day 9

J558-IL-2 - - + + + -J558-IL-4 - + - + +J558-IL-7 + + + + + +J558-TNF + + + + + -J558-IFN - - - ± + -BALB/c mice were injected s.c. with 4 x 106 cells and tumor tissue

was excised at the indicated time points. -, Histology did not differfrom that in J558L tumors, which were infiltrated by few CR3+ cells(10).

should be noted that some of the cytokine effects may beindirect and due to the release of further cytokines bybystander or effector cells. Our results confirm some previ-ous reports but also reveal discrepancies. Two reports haveattributed the anti-tumor effect of IL-2 after gene transfermainly to cytotoxic T cells (1, 2). However, we observed thatthe T cells are not necessary in the early phase for tumorsuppression. This is in agreement with some tumor-suppressive effects of local IL-2 in nude mice (3) and a similarlag phase observed with other IL-2-producing tumor cellsafter CD8+ T-cell depletion, which also prompted the authorsto hypothesize further effector mechanisms (1). Our resultsindicate that part of these further effectors are NK cells. Theearly tumor-suppressive effect of IL-4 has consistently beenobserved to be mediated by non-T-effector cells such asmacrophages and eosinophils (5, 7, 8, 9). Notwithstanding,late outgrowth oftumors has been described in nude (5, 7) andSCID mice (7). Tepper et al. (5) reported a greater tendencyof late outgrowth of tumors (clone I2B1) that had ceased toproduce IL-4 in nude but not in syngeneic mice, even if thisdid not occur with a clone producing five times more IL-4(clone 13L6), indicating that T cells may be substituted insome instances by a higher cytokine dose. Compatible withour results, the TNF-induced tumor-suppressive effect hasbeen shown to be operative in the absence ofT cells, even iftumors were not completely rejected (13, 15, 17). On theother hand, considering the dense infiltration of J558-TNFtumors with CD4+ and CD8+ T cells, it is understandable thatboth cell populations were necessary for the rejection ofanother tumor cell line (16), reasoning that it was lesssensitive to T-cell-independent mechanisms. A difference inthe susceptibility for host effector mechanisms of the cell linemay also be responsible for the different mode of rejectionreported for an IL-7-transduced murine glioma that wasdependent on CD8+ but not CD4+ T cells (12), especiallysince we could observe both T-cell subpopulations in thetumor infiltrate of J558-IL-7 (10) and have demonstrated thatCD8+ T cells are also needed in the late phase of tumorrejection. McBride et al. (11) reported a cellular infiltrate ina fibrosarcoma after IL-7 gene transfer, which is very similarto that observed by us in J558-IL-7 tumors. In three reportsthat have described tumor-suppressive effects of IFN-y aftergene transfer, the tumors were reported to grow in nude micebut not in syngeneic mice, firmly establishing a role forT cellsin tumor suppression (18-20). However, the relative contri-butions of T- or non-T-effectors appear to be differentdepending on the cell lines, since we observed IFN-y to beoperative in nude mice due to NK cells.Our results and those reported in the literature are com-

patible with the hypothesis that tumor rejection induced byall five cytokines requires CD8+ T cells and further effector

Proc. Natl. Acad. Sci. USA 90 (1993)

cells, which vary depending on the cytokine. The relativecontributions of the different effectors appear to be influ-enced by the particular tumor model.

We express our particular gratitude to G. Schulz and M.-V.Odenwald for their excellent technical assistance. We thank thefollowing colleagues who generously provided us with material: G.Riethm(ller (SCID/beige mice), W. Paul (mAb lBll, CT.4S cells),C. Sanderson (mAb NIMP-R6), R. Coffman (mAbs XMG-6 andS4B6), K. UJberla (J558-IL-2). This work was supported by grantsfrom the Deutsche Forschungsgemeinschaft and the Deutsche Krebs-hilfe, Mildred Scheel Stiftung, e.V.

1. Fearon, E. R., Pardoll, D. M., Itaya, T., Golumbek, P., Levitsky, H. I.,Simons, J. W., Karasuyama, H., Vogelstein, B. & Frost, P. (1990) Cell60, 397-403.

2. Gansbacher, B., Zier, K., Daniels, B., Cronin, K., Bannejy, R. &Gilboa, E. (1990) J. Exp. Med. 172, 1217-1224.

3. Bubenik, J., Voitenok, N. N., Kieler, J., Prassolov, V. S., Chumakov,P. M., Bubenikova, D., Simova, J. & Jandlova, T. (1988) Immunol. Lett.19, 279-282.

4. Russell, S. J., Eccles, S. A., Flemming, C. L., Johnson, C. A. & Collins,M. K. L. (1991) Int. J. Cancer 47, 244-251.

5. Tepper, R. I., Pattengale, P. K. & Leder, P. (1989) Cell 57, 503-512.6. Li, W., Diamantstein, T. & Blankenstein, T. (1990) Mol. Immunol. 27,

1331-1337.7. Golumbek, P. T., Lazenby, A. J., Levitsky, H. I., Jaffee, L. M., Kara-

suyama, H., Baker, M. & Pardoll, D. M. (1991) Science 254, 713-716.8. Tepper, R. I., Coffman, R. L. & Leder, P. (1992) Science 257, 548-551.9. Platzer, C., Richter, G., Uberla, K., Hock, H., Diamantstein, T. &

Blankenstein, T. (1992) Eur. J. Immunol. 22, 1729-1733.10. Hock, H., Dorsch, M., Diamantstein, T. & Blankenstein, T. (1991) J.

Exp. Med. 174, 1291-1298.11. McBride, W. H., Thacker, J. D., Comora, S., Economu, J. S., Kelley,

D., Hogge, D., Dubinnet, S. M. & Dougherty, G. J. (1992) Cancer Res.52, 3931-3937.

12. Aoki, T., Tashiro, K., Miyatake, S.-I., Kinashi, T., Nakano, T., Oda, Y.,Kikuchi, H. & Honji, T. (1992) Proc. Natl. Acad. Sci. USA 89, 3850-3854.

13. Oliff, A., Defeo-Jones, D., Boyer, M., Martinez, D., Kiefer, D., Vuo-colo, G., Wolfe, A. & Socher, S. S. (1987) Cell 50, 555-563.

14. Blankenstein, T., Qin, Z., Oberla, K., MOlier, W., Rosen, H., Volk,H.-D. & Diamantstein, T. (1991) J. Exp. Med. 173, 1047-1052.

15. Teng, M. N., Park, B. H., Koeppen, H. K. W., Tracey, K. J., Fendly,B. M. & Schreiber, H. (1991) Proc. Natl. Acad. Sci. USA 88, 3535-3539.

16. Asher, A. L., Muld, J. J., Kasid, A., Restifo, N. P., Salo, J. C., Re-ichert, C. M., Jaffe, G., Fendly, B. M., Kriegler, M. & Rosenberg, S. A.(1991) J. Immunol. 146, 3227-3234.

17. Qin, Z., Diamantstein, T. & Blankenstein, T. (1993) in Genes in MedicalDiagnosis and Therapy, ed. Kurstak, E. (Dekker, New York), in press.

18. Watanabe, Y., Kuribayashi, K., Miyatake, S., Nishihara, K., Na-kayama,. E.-I., Taniyama, T. & Sakata, T.-A. (1989) Proc. Natl. Acad.Sci. USA 86, 9456-9460.

19. Gansbacher, B., Bannerji, R., Daniels, B., Zier, K., Cronin, K. &Gilboa,E. (1990) Cancer Res. 50, 7820-7825.

20. Esumi, N., Hunt, B., Itaya, T. & Frost, P. (1991) Cancer Res. 51,1185-1189.

21. Dorsch, M., Hock, H., Kunzendorf, U., Diamantstein, T. & Blanken-stein, T. (1993) Eur. J. Immunol. 23, 186-190.

22. Colombo, M. P., Ferrari, G., Stoppacciaro, A., Parenza, M., Rodolfo,M., Mavilio, F. & Parmiani, G. (1991) J. Exp. Med. 173, 889-897.

23. Rollins, B. J. & Sunday, M. E. (1991) Mol. Cell. Biol. 11, 3125-3131.24. Bottazzi, B., Walter, S., Govoni, D., Colotta, F. & Mantovani, A. (1992)

J. Immunol. 148, 1280-1285.25. Douvdevani, A., Huliehel, M., Zoller, M., Segal, S. & Apte, R. N. (1992)

Int. J. Cancer 52, 1-9.26. Gillis, S., Ferm, M. M., Ou, W. & Smith, K. A. (1978) J. Immunol. 120,

2027-2032.27. Hu-Li, J., Ohara, J., Watson, C., Tsang, W. & Paul, W. E. (1989) J.

Immunol. 142, 800-807.28. Zurawski, S. M., Mosmann, T. R., Benedik, M. & Zurawski, G. (1986)

J. Immunol. 137, 3354-3360.29. Ohara, J. & Paul, W. E. (1985) Nature (London) 315, 333-336.30. Cherwinski, H. M., Schumacher, J. H., Brown, K. D. & Mosmann,

T. R. (1987) J. Exp. Med. 166, 1229-1244.31. Ho, M.-K. & Springer, T. A. (1982) J. Immunol. 128, 2281-2286.32. Springer, T. A. (1981) J. Biol. Chem. 256, 3833-3839.33. Gordon, S. (1981) Eur. J. Immunol. 11, 805-815.34. Lopez, A. F., Strath, M. & Sanderson, C. J. (1984) Br. J. Haematol. 57,

489-494.35. Andriole, G. L., Mule, J. J., Hansen, C. T., Linehan, W. M. & Rosen-

berg, S. A. (1985) J. Immunol. 135, 2911-2913.36. Trinchieri, G. (1989) Adv. Immunol. 47, 187-376.37. Colombo, M. P., Modesti, A., Parmiani, G. & Forni, G. (1992) Cancer

Res. 52, 4853-4857.

Dow

nloa

ded

by g

uest

on

Nov

embe

r 6,

202

0