Proc. Natl. Acad. Sci. USAVol. 92, pp. 5540-5544, June 1995Medical Sciences

Priming of tumor-specific T cells in the draining lymph nodesafter immunization with interleukin 2-secreting tumor cells:Three consecutive stages may be required for successfultumor vaccinationGERHARD MAASS*t, WALTER SCHMIDT*, MANFRED BERGER*, FRANz SCHILCHERt, FRIEDER KoSZIK§,ACHIM SCHNEEBERGER§, GEORG STINGL§, MAx L. BIRNSTIEL*¶, AND TAMAS SCHWEIGHOFFER**Research Institute of Molecular Pathology, Dr. Bohr-Gasse 7, A-1030 Vienna, Austria; tInstitute of Pathology and Forensic Veterinary Medicine, University ofVeterinary Medicine of Vienna, Linke Bahngasse 11, 1030 Vienna, Austria; and §Department of Dermatology, University of Vienna, Medical School, WahringerGurtel 18-20, A-1090 Vienna, Austria

Contributed by Max L. Birnstiel, March 6, 1995

ABSTRACT Although both CD4+ and CD8+ T cells areclearly required to generate long-lasting anti-tumor immunityinduced by s.c. vaccination with interleukin 2 (IL-2)-transfected, irradiated M-3 clone murine melanoma cells,some controversy continues about the site and mode of T-cellactivation in this system. Macrophages, granulocytes, andnatural killer cells infiltrate the vaccination site early afterinjection into either syngeneic euthymic DBA/2 mice orathymic nude mice and eliminate the inoculum within 48 hr.We could not find T cells at the vaccination site, which arguesagainst the concept that T-cell priming by the IL-2-secretingcancer cells occurs directly at that location. However, reversetranscription-PCR revealed transcripts indicative of T-cellactivation and expansion in the draining lymph nodes of miceimmunized with the IL-2-secreting vaccine but not in micevaccinated with untransfected, irradiated M-3 cells. We there-fore propose that the antigen-presenting cells, which invadethe vaccination site, process tumor-derived antigens and,subsequently, initiate priming of tumor-specific T lympho-cytes in lymphoid organs. These findings suggest a three-stageprocess for the generation of effector T cells after vaccinationwith IL-2-secreting tumor cells: (i) tumor-antigen uptake andprocessing at the site of injection by antigen-presenting cells,(ii) migration of antigen-presenting cells into the regionaldraining lymph nodes, where T-cell priming occurs, and (iii)circulation of activated T cells that either perform or initiateeffector mechanisms leading to tumor cell destruction.

We have generated a cancer vaccine by transfecting themelanoma Cloudman S91 clone M-3 (referred to as M-3) witha murine interleukin 2 (IL-2) gene driven by a cytomegaloviruspromoter/enhancer and examined its influence on the fate ofsubsequently injected live M-3 cells in syngeneic DBA/2 hosts(1, 2). For gene delivery we used the adenovirus-enhancedreceptor-mediated gene transferrinfection technique (3). Thistechnique is exceptionally efficient in transiently transfectinga large proportion of exposed cells and can be used for a broadrange of cell types. ,IL-2 has been used successfully in othercancer treatment models-administered i.v. or locally (4),secreted by admixed cells (5), and also delivered by transfectedtumor cells as a "vaccine" (6-8). We find that two consecutivevaccinations with i05 transfected, irradiated M-3 cells secret-ing 1-3 x 103 units of IL-2 per 105 cells and 24 hr in vitroprotect animals from tumor development by a challenge with105 or 3 x 105 wild-type M-3 cells 1 week later (9) and thatimmunized animals also reject a tumor challenge after severalmonths (1). As shown by T-cell transfer experiments, this

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

systemic tumor-specific protection critically depends on bothCD4+ and CD8+ T cells. Correspondingly, no long-lastingsystemic protection could be achieved in athymic nude mice.However, T cells are apparently not required for eliminatingviable IL-2-secreting M-3 melanoma cells because these cellswere effectively rejected in athymic nude mice (1).A conceptual dilemma arises when one attempts to apply to

the M-3 system the theory that the mechanism of action ofIL-2-based cancer vaccines is helper T cell-independent-i.e.,through direct activation of CD8+ T cells by IL-2-transfectedtumor cells. For instance, it is unclear how tumor-reactiveCD4+ cells are activated and clonally selected because M-3tumor cells, while weakly reactive with anti-major histocom-patibility complex class I antibodies, are consistently majorhistocompatibility complex class II negative both in vivo and invitro (1). Further, how naive T cells make contact with thetumor antigens of the IL-2-secreting vaccine implanted s.c. isnot understood because naive T cells do not appear in sizablequantities in nonlymphoid tissues (10, 11).

Here, we investigate the fate of the vaccine cells, theirinteraction with the immune system, and the location andcourse of events of T-cell activation and expansion in responseto IL-2-secreting vaccines.

MATERIAL AND METHODSPreparation and Characterization ofTumor Cells Used for

Vaccination. Cloudman S91 melanoma cells, clone M-3 cellswere purchased from American Type Culture Collection. M-3cells were transfected with the IL-2 expression vector pWS2mand irradiated as described (1). Secreted IL-2 was measured bya commercially available ELISA (Becton Dickinson), andunits were calculated for 1 x 105 cells and 24-hr incubationtime. DBA/2 (H-2d) and MF1 nu/nu mice (6 to 8 weeks old)were immunized twice under halothane anesthesia by s.c.injections.PCR Amplification. For competitive PCR amplification,

injection sites were excised and incubated overnight in pro-teinase K buffer, followed by DNA extraction. The plasmidpWS2m-del, which served as an internal control, was con-structed by deleting the IL-2-coding region between positions197 and 377 of pWS2m, which led to the elimination of a180-bp fragment. For reverse transcription (RT)-PCR ampli-fication, RNA was extracted from the injection site, draininglymph nodes or challenge site with Trisolv (Biotecx Labora-

Abbreviations: IL, interleukin; INF, interferon; IL-2Ra, interleukin-2receptor a; GM-CSF, granulocyte/macrophage colony-stimulatingfactor; TGF, transforming growth factor; RT, reverse transcription.tPresent address: Medigen, Lochhammerstrasse 11, D-82152 Martins-ried, Germany.To whom reprint requests should be addressed.

5540

Proc. Natl. Acad. Sci. USA 92 (1995) 5541

tories, Houston), treated with RNase-free DNase I (Boehr-inger Mannheim) to avoid cross-reactivity with genomic DNA,and reverse-transcribed with oligo(dT) primers (12). Cytokinemessages were then amplified by PCR with specific primerspurchased from Clontech, except for the following: CD69,5'-CCTTGGGCTGTGTTAATAGTGG-3' and 5'-AACTT-TCCTGTGGTGCTATGGC-3'; granzyme B, 5'-GGCAAA-TACTCAAACACGCTACAAG-3' and 5'-GCAATCC- TG-GACTCAGCTCTAGG-3'; CTLA-4, 5'-GGTCTCTGC-TGTTTCTTTGAGCAAG-3' and 5'-GCTATATCCTTAT-GCTGCTTCCAAGTG-3'; f3-actin, 5'-GTGGGCCGCTCT-AGGCACCA-3' and 5'-CGGTTG GCCTTAGGGTTC-AGGGGGG-3'. Amplified DNA products were visualized ona 2% agarose gel.

Immunohistology. Sections were prepared as described (1)and treated with one of the following primary antibodies:145-2C1 1-biotin (CD3), RM4-5-biotin (CD4), 53-6.7-biotin(CD8), B220-biotin (CD45-R), M1/70 (Mac-1, CD11b),GR-1 (RB6-8C5) (all from PharMingen), F4/80-biotin (Ser-otec) and anti-asialo-GM1 antiserum (Wako Bioproducts,Richmond, VA). Streptavidin-alkaline phosphatase or goatanti-rabbit alkaline phosphatase was used as a second-stepreagent. Although the complexes used to transfect tumor cellscontain a streptavidin-biotin link, erroneous binding of anti-bodies does not occur at a detectable level, as deduced fromthe following findings: (i) Particularly at early times, thetransfected inoculum is free of all staining for F4/80-biotinantibody and is devoid of staining with 145-2C11-biotin(CD3), RM4-5-biotin (CD4), 53-6.7-biotin (CD8), andCD45R/B220-biotin (B cell) at all times. (ii) For all otherantibodies there is a highly specific topological and temporalsequence of events during the eradication of the transfectedinoculum (see below).

Transfer of Latex Beads from M-3 Cells to Macrophages.M-3 cells were labeled in vitro with 15 gl of FluoSpheres (0.5,um, Molecular Probes) fluorescent latex beads per 106 M-3cells for 24 hr after transfection. Cells were then washed fivetimes with phosphate-buffered saline, and 105 cells wereinjected into animals. Forty-eight hours later, injection siteswere excised and prepared for electron microscopy by standardprocedures.

RESULTSRapid Elimination of Tumor Vaccines and Secondary Cy-

tokine Responses in Vivo. The persistence of the transfectedIL-2 plasmid was used as a measure of survival of the cellvaccine at the vaccination site. Groups of DBA/2 mice (threeanimals each) were injected s.c. with 3 x 105 M-3 cells that hadbeen transfected with the IL-2 expression vector and thenirradiated. Animals were sacrificed 1, 2, or 5 days thereafter.Competitive PCR analysis of the injection sites showed a rapidelimination of the transfected IL-2 plasmid (Fig. 1). Althougheasily detectable on day 1 and in smaller quantities on day 2,the IL-2 sequence could no longer be amplified on day 5. Thisresult contrasts with the situation in vitro where the IL-2transgene can be detected for at least 15 days (Fig. 1).The rapid kinetics of vaccine elimination prompted us to

start analysis of mRNA recovered from vaccination sites 6 hrafter injection. mRNAs of cytokines IL-1, IL-2, IL-4, IL-6,IL-10, tumor necrosis factor a, transforming growth factor(TGF) B, granulocyte/macrophage colony-stimulating factor(GM-CSF), and interferon -y (IFN--y) and of T-cell activationmarkers IL-2 receptor a (IL-2Ra) chain (CD25), CD69,granzyme B (CTLA-1) and CTLA-4 were determined bysemiquantitative RT-PCR. In addition, amplification of f3-ac-tin mRNA was chosen to gauge efficiency in all experiments.IL-2-secreting, irradiated M-3 cells induce enhanced signalsfor IL-1, IL-6, IL-10, and GM-CSF mRNA in vivo. Thus,secretion of IL-2 at appropriate levels initiates a cascade of

Li) -C>e_

~___-

>1 >1 >1-O,vcC2 '

0

C00

a acV S X

- IL-2 plasmid- internal control

in vitro in vivo

FIG. 1. Mobilization of injected M-3 cells. M-3 cells were trans-fected with the IL-2 expression plasmid pWS2m. (i) Cells werecultured in vitro for 15 days after transfection, and at 5-day intervals,aliquots were lysed and total DNA was purified. (ii) Twenty-four hoursafter transfection, 3 x 105 IL-2-transfected M-3 cells were injected s.c.into each of nine DBA/2 mice. After 1, 2, and 5 days, immunizationsites from three mice were excised, and total DNAwas purified. Copies(103) of the internal control plasmid were added for semiquantitativeanalysis for persistence of IL-2 plasmid DNA.

secondary cytokines at the vaccination site (Fig. 2). Nativetumor cells produce mRNAs for TGF-f3 in vitro. At thevaccination site in vivo, irradiated, nontransfected M-3 mela-noma cells (and surrounding tissue) also display a TGF-f3response together with weak IL-10 and GM-CSF signals.However, TGF-,B mRNA was not detected at the immuniza-tion site of the IL-2-secreting vaccine, indicating a down-regulation of this mRNA in vivo probably as a consequence ofIL-2 secretion. Among the activation markers tested, onlygranzyme B mRNA, a transcript also found in natural killercells, was faintly detectable at the injection site of IL-2-secreting vaccines (see Fig. SC) but not at sites of irradiated-only cells. Other T-cell markers such as IL-4, IL-2Ra, orCTLA-4 transcripts were not in evidence; hence there is noindication of T-cell priming at the injection site. At 6 hr afterinjection of IL-2-expressing M-3 cells, the cytokine mRNApattern in nude mice was identical.to that of immunocompe-tent DBA/2 host (Fig. 2). Therefore, the mRNAs that wereinduced or enhanced by the IL-2 vaccination procedure-i.e.,IL-1, IL-6, IL-10, and GM-CSF-are unlikely to be of T-cellorigin but rather reflect the inflammatory reaction at thevaccination site.

Immunohistology of Immunization Sites. Animals wereinjected with 5 x 105 IL-2-transfected and irradiated M-3 cells(expressing a total of 2000 units of IL-2) or with nontrans-fected, irradiated cells as control. Immunization sites andsurrounding tissue were excised at 6 hr, 24 hr, and 48 hr after

v CM X,v-L ! 2 i

I Sn 0 Co.

IIn vItr mlaIa

Odetecable after detedable after0 40 cycles W30cycles

FIG. 2. M-3 cells were irradiated and transfected with the IL-2expression plasmid pWS2m. For in vitro assays (line 1), total mRNAwas purified from 2 x 105 cells 24 hr after transfection. For in vivoassays, DBA/2 mice or nude mice were immunized with 5 x 105 M-3cells. Injection sites (line 2, irradiated only, lines 3 and 4, IL-2transfected cells) were excised at 6 hr after immunization followed bypurification of total mRNA. RNA was reverse-transcribed and am-plified with cytokine mRNA-specific sets of primers.

Medical Sciences: Maass et aL

5542 Medical Sciences: Maass et at

injection and analyzed by immunohistology. Staining withF4/80 antibody revealed that at 6 hr macrophages had invadedthe rim of the IL-2-secreting cell mass (data not shown). At 24hr, F4/80 antibody-positive cells commenced invasion of theinoculum but only if the M-3 cells secreted IL-2 (compare Fig.3 a and d). By 48 hr macrophages had invaded the dispersedand now necrotic inoculum (Fig. 3b) and approximatelyequaled tumor cells in number. Such a redistribution of F4/80antibody-reactive cells was not seen when the vaccine consistedof nontransfected, irradiated M-3 cells, and in this case mosttumor cells were still detectable by 48 hr (compare Fig. 3 d ande). Staining with granulocyte-specific antibody GR-1 revealedaccelerated infiltration by granulocytes at 24 hr as comparedwith F4/80 antibody-stained cells (Fig. 3g). At 48 hr afterimmunization the granulocyte influx was very similar in num-bers to that of macrophage-like cell types (Fig. 3h). Anti-asialo-GM1 antiserum, which identifies natural killer cells,showed less of these cells in tissues surrounding the vaccine at24 hr, and even fewer natural killer cells were seen surroundingnontransfected, irradiated M-3 cells (Fig. 3 c andf). In neithervaccine were T cells (CD3+, CD4+, or CD8+) or B cells (CD45receptor/B220) detectable (data not shown).Athymic nude mice readily reject live IL-2-secreting M-3

cells (1). The composition of the infiltrating cells in nude micequalitatively matched the cells infiltrating the vaccine inimmunocompetent mice; macrophages and granulocytes ac-cumulated rapidly in IL-2-secreting inocula (Fig. 3 k and 1),whereas only very few of these cells were detectable afterinjection of irradiated-only M-3 cells (Fig. 3 n and o). Stainingwith anti-asialo-GM1 antiserum essentially mirrored the situ-

ation seen in DBA/2 mice (Fig. 3 m and p). The majordifference between athymic and euthymic mice was the mag-nitude of the granulocyte response, in that the IL-2-expressinginoculum induced -3-fold influx in nude mice compared withDBA/2 mice at 24 hr after injection (compare Fig. 3 g and 1).As in the immunocompetent host, nude mice eliminatedIL-2-producing tumor masses at 48 hr (data not shown). Thisfact is evidence that IL-2-secreting tumor cell masses can berejected in a system devoid of T cells, presumably by anonspecific immune response based on natural killer cells,macrophages, and granulocytes.

Ingestion of Tumor Cells by Macrophages at the Vaccina-tion Site. The simplest interpretation of Figs. 1 and 3b is thatthe invading macrophage-like cells ingest the tumor cells andtumor cell debris and thus quickly destroy the inoculum ofIL-2-secreting tumor cells in both DBA/2 and nude mice. Thatthis is, indeed, the case can be deduced from experiments inwhich we preloaded IL-2-producing M-3 cells in vitro with latexbeads (0.5 ,tm) and detected the transfer of these beads intomacrophages as shown by electron microscopy (Fig. 4). Thisresult suggests that macrophages also ingest M-3 tumor anti-gens and that these altered macrophages may act as antigen-presenting cells.

Cytokine mRNAs Specific for T-cell Activation Are Ex-pressed in Draining Lymph Nodes After IL-2 Vaccination.Having detected essentially no T cells at the vaccination site,we reasoned that draining lymph nodes were the more likelyanatomical location of T-lymphocyte priming. Groups of 6-10DBA/2 mice were immunized twice with 105 either irradiated-only M-3 cells or IL-2-transfected, irradiated M-3 cells on the

*yS*5.

|ts... # 4 ,a _ _.

.NVs

a1 t~~~~~~~~~~~ipf*.. ;...

V.

.4r ,4& to: ft ,-i -A

W.A..w

4 * 4 ..,. 4% - 4, 4

,) 6- 'kI a

/ 4. 011 'I -t " - a

FIG. 3. Immunohistology of vaccination site. The approximate boundary between tumor mass and surrounding tissue is indicated by the dashedline. inoc, Inoculum. (a-i) DBA/2 mice. (k-p) Nude mice. (a) F4/80 antibody, IL-2 positive, 24 hr. (b) F4/80 antibody, IL-2 positive, 48 hr. (c)GM1 antiserum, IL-2 positive, 24 hr. (d) F4/80 antibody, IL-2 negative, 24 hr. (e) F4/80 antibody, IL-2 negative, 48 hr. (f) GM1 antiserum, IL-2negative, 24 hr. (g) GR-1 antibody, IL-2 positive, 24 hr. (h) GR-1 antibody, IL-2 positive, 48 hr. (i) GR-1 antibody, IL-2 negative, 24 hr. (k) F4/80antibody, IL-2 positive, 24 hr. (1) GR-1 antibody, IL-2 positive, 24 hr. (m) GM1 antiserum, IL-2 positive, 24 hr. (n) F4/80 antibody, IL-2 negative,24 hr. (o) GR-1 antibody, IL-2 negative, 24 hr. (p) GM1 antiserum, IL-2 negative, 24 hr. (X375.)

Proc. Natl. Acad. Sci. USA 92 (1995)

Proc. Natl. Acad. Sci. USA 92 (1995) 5543

FIG. 4. Electronmicrograph of a macrophage (M0) at the vacci-nation site. (x 11,000.) (Inset) The macrophage contains M-3-derivedlatex beads (LB), shown at larger magnification.

left and the right sides of the back. Seven days after the secondimmunization, a point in time at which a challenge with livetumor cells would normally be set (1), draining lymph nodeswere removed, along with lymph nodes from 10 naive DBA/2mice as a control. Lymph nodes of animals from the IL-2-transfected group were clearly enlarged compared with thoseof mice immunized with irradiated M-3 cells or from naivecontrol mice. In IL-2 vaccine-treated mice, the cell number was4.87 ± 1.87 x 106 cells per lymph node as compared with 2.3+ 1.07 x 106 cells in lymph nodes after immunization withirradiated-only tumor cells and with 1.28 ± 0.35 x 106 cells innaive controls. Lymph node cells from individual groups were

pooled, and RT-PCR was done on mRNA derived from 107cells. 13-Actin control transcripts could be amplified up to a

1:1000 dilution in all groups, showing that the efficiency of RTreactions was comparable (Fig. S A and B). Strong signals forIL-2, IL-4, and IFN-y mRNA together with the appearance ofCD69, granzyme B (CTLA-1), and CTLA-4 mRNA, indicatingactivation and expansion of T cells, were detected in miceinoculated with the IL-2-secreting vaccine. Similar patterns ofamplified transcripts were already obtained after a singlevaccination but were enhanced after the second vaccination(Fig. SB). By contrast, in animals immunized with untrans-

m coQ) 'o

- C N ! c

> J * X Z C,) X C,;) X

I.A

IV 0

EoCD >1

QD No~

Bm 0

t ~Ea) >1 CTcc °- 5N

E

,C' ;t Z

CMX FJi > >u L0 - e;

u

FIG. 5. RT-PCR amplification ofmRNAs. Groups of DBA/2 mice(10 animals each) were immunized twice with irradiated or IL-2-transfected M-3 cells. Seven days after the second immunization,draining lymph nodes were removed. Total RNA was prepared from107 pooled cells of single-cell suspensions, reverse-transcribed, andamplified with cytokine mRNA-specific sets of primers in 30 cycles.LM, DNA length standard. The other lanes show the lymph nodemRNA. (A) Mice vaccinated with irradiated-only M-3 cells. (B) Micevaccinated with IL-2-secreting inoculum. (C) RT-PCR of mRNAsobtained from the vaccination site 24 hr after the s.c. injection ofIL-2-secreting M-3 cells.

fected irradiated M-3 cells, only the CD69 amplificationproduct could be detected (Fig. SA), and the same result wasobtained from naive controls (data not shown). Note that theabove transcripts could not be detected at the IL-2-secretingvaccination site, except for the expected strong signal for IL-2and a very weak signal for granzyme B transcripts (Fig. SC),which might be caused by the infiltrating sparse natural killercells detected in Fig. 3c.

DISCUSSIONAn unexpected result of the above immunohistological inves-tigation was the great similarity of cell types infiltrating thevaccination sites of both immunocompetent and nude mice.Interestingly, a similar situation pertains at the challenge sitein euthymic DBA/2 mice, although here participation of Tcells was clearly demonstrated (1). In all three cases F4/80antibody-positive macrophage-like cells (macrophages and/orLangerhans cells of the skin) and granulocytes rapidly pene-trate and clearly predominate in the inocula and may beresponsible for the observed extensive dispersal and elimina-tion of the M-3 cell masses. The qualitative compositions of thecellular infiltrate at the vaccination site in the presence andabsence of IL-2 are similar, but there are significant quanti-tative and topological differences. If nontransfected, irradiatedM-3 cells are used as whole-cell vaccine, granulocytes andmonocyte/macrophage-like cells are observed, but they onlyencircle the M-3 inoculum and do not penetrate the tumor cellmass. The similarity in cell type is not surprising becausevaccination with irradiated-only cells also induces systemicprotection, although to a much smaller degree (2, 9).The amount of IL-2 secreted by the vaccine is critical for the

generation of antitumor immunity, with 1000-3000 units ofIL-2 per mouse yielding the optimal vaccination effect (9),whereas lower as well as higher values are much less effective.We see the contribution of IL-2 not so much in the preferentialrecruitment of specific cell subsets but rather in the augmen-tation of the function of the cell types-particularly macro-phage-like cells, which invade the inoculum. IL-2 can directlyaugment antitumor activity of monocytes (13) and also con-tributes to the induction of other macrophage functions (14).In addition, other secondary mediators, such as GM-CSF (Fig.2), the release of which is triggered by IL-2 may also substan-tially contribute to macrophage activation (15). Granulocytesmost likely represent a population that may not dependcritically on IL-2 per se because (i) IL-2 is not a majorgranulocyte regulator, and (ii) the granulocyte infiltrate hasalso been recorded in granulocyte colony-stimulating factor-(16) and in IL-4-secreting systems (17, 18), although with apredominance of eosinophils in the latter case. We proposethat granulocytes may actually appear not as a cause but ratheras a result of tumor destruction by the macrophage-like cells.For the IL-2-secreting vaccine this view is also supported by thedynamics of appearance of these cells-namely, that the influxpeak of macrophages precedes that of granulocytes (unpub-lished results). However, with the many potent proinflamma-tory, chemotactic factors they release, granulocytes may actu-ally amplify the tumor destruction, promoting the explosivecellular accumulation that we observe.

Massive infiltration with inflammatory cells was previouslyrecorded for inocula of granulocyte colony-stimulating factor-expressing adenocarcinoma C-26 (16) and of IL-2 or INF-y-secreting CMS-5 fibrosarcoma (8), and we here show that thesame situation applies for M-3 inocula. The absence of T cellsat the vaccination site is consistent with the idea that naive cellsare compartmentalized in blood and lymph and that onlymemory cells appear in tissues (10, 11). Absence of T cells inthe M-3 vaccine poses the general question as to how theactivation of CD4+ and CD8+ cells necessary for the estab-lishment of systemic antitumor immunity occurs in these

Medical Sciences: Maass et at

tscr

-2 1.1 C.l 1 z-j :. :. =. ILL

5544 Medical Sciences: Maass et al

systems (1). CD4+ cells clearly cannot be activated directly bythe M-3 cells, which express only low levels of major histo-compatibility complex class I but no class II (1). Theoretically,M-3 cells might react with CD8+ T cells, which, however, areabsent from the inoculum. It may, of course, be argued that thetransition time of CD8+ cells in the cytokine-producing im-plant is very short, making the detection of the cells moredifficult. This argument is, however, offset by the need of agreat many contacts because only a tiny proportion of the naiveCD8+ cells would be expected to react with the tumor antigen,thus requiring a large flux of cells through the inoculum,necessary for activation and clonal selection of CD8+ cells.These considerations suggest an alternative scheme for the

establishment of T-cell activation, which is based on tumorantigens being processed by antigen-presenting cells that trans-fer them to the lymph node, where T-cell priming then occurs.That this hypothesis is a likely alternative is supported by ourfindings that (i) macrophage-like cells ingest tumor cells at thevaccination site, (ii) the draining lymph nodes are extensivelyenlarged, and (iii) mRNAs diagnostic of activated T lympho-cytes (19) are detected in this compartment. Thus, mRNAs forthe IL-2Ra, IL-2, IL-4, and IFN-y, which are indicative ofT-cell activation and expansion, become detectable only afteradministration of an IL-2-secreting vaccine but cannot beamplified after injection of irradiated-only M-3 cells. mRNAsfor granzyme B in combination with mRNA for CTLA-4 arediagnostic for activated T cells with prospective cytotoxicfunction. CD69 is also anticipated soon after T-cell activation,but this marker is expressed by other activated hematopoieticcells (20). We suggest that it is in the lymph node compartmentthat both CD4+ and CD8+ cells are programmed by antigen-presenting cells, which have passed through the vaccinationsite. A similar hypothesis has also been posited from theoret-ical consideration by Gilboa's group (8). The alternative theyconsidered-namely, that IL-2-producing tumor cells mightmigrate into the lymph node and activate T cells at thislocation-is not supported by our experiments in the M-3system. Neither tyrosinase transcripts (Michael Buschle, per-sonal communication), easily detectable in M-3 cells, nor IL-2transgene sequences originating from injected tumor cellscould be amplified by high-sensitivity PCR from draininglymph node cells (21). However, our postulate agrees withrecent work by Huang et al (22), who concluded from elegantlydesigned mouse bone marrow-transplantation experimentsthat bone marrow-derived antigen-presenting cells are instru-mental for the generation of systemic antitumor immunity.

Note Added in Proof. In recent experiments we found that mice thathad their lymph nodes depleted of naive T cells by in vivo antibodytreatment were unable to develop protective anti-tumor immunity.Thus, de novo priming of naive T cells in the lymph nodes isindispensable for a successful immunization.

We thank Dr. Margaret Chipchase for critical reading of themanuscript. We are grateful to Elke Herbst and Helen Kirlappos forexcellent technical assistance and to Hannes Tkadletz and GeorgKafka for artwork. This work was supported in part by the AustrianIndustrial Research Promotion Fund Grant 6/756.

1. Zatloukal, K., Schneeberger, A., Berger, M., Schmidt, W., Ko-szik, F., Kutil, R., Cotten, M., Wagner, E., Buschle, M., Maass,G., Payer, E., Stingl, G. & Birnstiel, M. L. (1995) J. Immunol. 154,3406-3419.

2. Zatloukal, K., Schmidt, W., Cotten, M., Wagner, E., Stingl, G. &Birnstiel, M. L. (1993) Gene 135, 199-207.

3. Cotten, M., Wagner, E., Zatloukal, K., Phillips, S., Curiel, D. T.& Birnstiel, M. L. (1992) Proc. Nati. Acad. Sci. USA 89, 6094-6098.

4. Rosenberg, S. A., Lotze, M. T. & Mule, J. J. (1988) Ann. Intern.Med. 108, 853-864.

5. Bubenik, J., Simova, J. & Jandlova, T. (1990) Immunol. Lett. 23,287-292.

6. Fearon, E. R., Pardoll, D. M., Itaya, T., Golumbek, P., Levitsky,H., Simons, J. W., Karasuyama, H., Vogelstein, B. & Frost, P.(1990) Cell 60, 397-403.

7. Cavallo, F., Giovarelli, M., Gulino, A., Vacca, A., Stoppacciaro,A., Modesti, A. & Forni, G. (1992) J. Immunol. 149, 3627-3635.

8. Bannerji, R., Arroyo, C. D., Cordon-Cardo, C. & Gilboa, E.(1994) J. Immunol. 152, 2324-2332.

9. Schmidt, W., Schweighoffer, T., Herbst, E., Maass, G., Berger,M., Schilcher, F., Schaffner, G. & Birnstiel, M. L. (1995) Proc.Natl. Acad. Sci. USA, in press.

10. Mackay, C. R., Marston, W. L. & Dudler, L. (1990) J. Exp. Med.171, 801-817.

11. Mackay, C. R., Marston, W. & Dudler, L. (1992) Eur. J. Immunol.22, 2205-2210.

12. Lagoo-Deenanayalan, S., Lagoo, A. S., Barber, W. H. & Hardy,K. J. (1993) Lymphokine Cytokine Res. 12, 59-67.

13. Malkovsky, M., Loveland, B., North, M., Asherson, G. L., Gao,L., Ward, P. & Fiers, W. (1987) Nature (London) 325, 262-265.

14. Belosevic, M., Finbloom, D. S., Meltzer, M. S. & Nacy, C. A.(1990) J. Immunol. 145, 831-839.

15. Oppenheim, J. J., Zachariae, C. 0. C., Mukaida, N. & Matsu-shima, K. (1991) Annu. Rev. Immunol. 9, 617-648.

16. Stoppaciaro, A., Melani, C., Parenza, M., Mastrochio, A., Bassi,C., Baroni, C., Parmiani, G. & Colombo, M. P. (1993) J. Exp.Med. 178, 151-161.

17. Tepper, R. I., Pattengale, P. K. & Leder, P. (1989) Cell 57,503-512.

18. Tepper, R. I., Coffman, R. L. & Leder, P. (1992) Science 257,548-551.

19. Crabtree, G. R. (1989) Science 243, 355-361.20. Testi, R., D'Ambrosio, D., De Maria, R. & Santoni, A. (1994)

Immunol. Today 15, 479-483.21. Maass, G., Schweighoffer, T., Berger, M., Schmidt, W., Herbst,

E., Zatloukal, K., Buschle, M. & Birnstiel, M. L. (1995) Int. J.Immunopharmacol., in press.

22. Huang, A. Y. C., Golumbek, P., Ahmadzadeh, M., Jaffee, E.,Pardoll, D. & Levitsky, H. (1994) Science 264, 961-965.

Proc. Natl. Acad. Sci. USA 92 (1995)