MULTIPLE MOLECULAR AND CELLULAR CHANGES ASSOCIATEDWITH TUMOUR STASIS AND REGRESSION DURING IL-12 THERAPYOF A MURINE BREAST CANCER MODELSergio DIAS1*, Hilary THOMAS2 and Frances BALKWILL 1

1Biological Therapies Laboratory, Imperial Cancer Research Fund, London, UK2Department of Clinical Oncology, Hammersmith Hospital, RPMS, London, UK

IL-12 treatment of a murine transplantable breast carci-noma (HTH-K) led to tumour regression and cure which wasrelated to the duration of treatment. We studied the sequen-tial molecular and phenotypic changes in IL-12-treated tu-mours. IFN-g mRNA was detected 8 hr after the firsttreatment. mRNA expression for the IFN-g-inducible genesb2-microglobulin and indoleamine dioxygenase (IDO) wasinduced subsequently, together with the chemokine IP-10.IL-12-treated tumours had an abundant cellular infiltrate,consisting mainly of CD81 T cells. mRNA for granzyme B andperforin also could be detected, suggesting that those cellswere activated. After 7 days of daily therapy, tumours inIL-12-treated mice had a significant reduction in vasculature.Finally, the number of apoptotic tumour cells increasedthroughout IL-12 treatment. We compared the anti-tumoureffects of IL-12 to those induced by IFN-g therapy, whichcaused initial tumour stasis but subsequent tumour progres-sion. IFN-g induced b2-microglobulin and IDO over a 7-dayperiod, but IP-10 was induced only transiently. IFN-g caused alesser cellular infiltrate, a minor anti-angiogenic effect and atransient apoptotic effect. The success of IL-12 may be due toits ability to produce a distinct sequence of molecular andphenotypic changes in tumours, leading to an anti-tumourimmune response, toxicity against tumour cells and an anti-angiogenic effect. Other cytokines, such as IFN-g, inducesome, but not all, of these actions. Comparison of IL-12 andIFN-g suggests that sustained induction of IP-10 and activa-tion of a resulting cellular infiltrate may be key changes inregressing tumours. Int. J. Cancer 75:151–157, 1998.r 1998 Wiley-Liss, Inc.

Interleukin-12 (IL-12) is a heterodimeric cytokine, producedmainly by macrophages/monocytes (D’Andreaet al., 1992), thatregulates a number of biological activities, such as the inhibition ofgrowth factor–induced angiogenesis (Voestet al., 1995) andincreased extra-medullary hematopoiesis (Jacksonet al., 1995).IL-12 also acts at the level of host defense on natural killer (NK)and T cells, promoting cellular immunity and inducing the expres-sion of several cytokines, such as interferon gamma (IFN-g),tumour necrosis factor alpha (TNF-a), IL-2, IL-3, IL-8 and IL-10.IL-12 also stimulates the proliferation and cytotoxicity of NK and Tcells and overall favours the establishment of a T-helper 1(Th1)-type immune response (Trinchieri, 1994). In models ofinfectious diseases (Gazzinelli, 1996) and in several transplantabletumour models (Brundaet al., 1996), IL-12 has revealed clinicalpotential.

One of the most striking features of IL-12 is its ability tostimulate IFN-g synthesis and release by T and NK cells. Further,IFN-g is thought to be important in mediating the effects of IL-12(Brundaet al.,1995). In a number of transplantable tumour models,antibodies against this cytokine almost completely abrogate theanti-tumour effects of IL-12 (Nastalaet al.,1994). IFN-g induces alarge number of proteins (Young and Hardy, 1995), and these mayplay an important role in the IFN-g-mediated anti-tumour effects ofIL-12. Such proteins include the enzyme indoleamine-2,3-dioxygenase (IDO, E.C. 1.13.11.42), the chemokine IP-10 and themajor histocompatibility complex (MHC) antigen classes I and II.These proteins may be involved in the anti-tumour effects of IFN-gby reducing the tumour’s essential nutrient supply (IDO) (Burkeetal., 1995), limiting its blood vessel formation, attracting T lympho-cytes (IP-10) (Sgadariet al., 1996a) and finally allowing the

presentation of tumour antigens to cells of the immune system(MHC class I). The anti-tumour activities of IL-12 cannot,however, all be linked to IFN-g or related gene products sinceIFN-g therapy does not mimic the effects shown by IL-12 (Brundaet al., 1995). The establishment of a cellular infiltrate after IL-12therapy is at least partially independent of IFN-g (Caret al.,1995),but the mechanisms involved in the process have not been studiedin detail.

We have studied sequentially the effects of IL-12 therapy in themicro-environment of a transplantable breast tumour, HTH-K, inBALB-c mice and compared these with IFN-g-treated tumours.The breast tumour originated in a colony of c-neutransgenic micepreviously described (Thomaset al., 1996) and is an invasive,angiogenic, comedo-type, metastatic breast carcinoma. It grows asa solid mass around a central area of necrosis. A highly effectiveanti-tumour response is necessary in order to increase survival inthis rapidly fatal tumour model.

We established a temporal sequence of events that define theactions of IL-12 on the tumour micro-environment. Some of theseactions have been reported individually, but we were able todemonstrate that these occur in a sequential manner and to describenovel actions of this cytokine. We suggest that the ability toproduce a particular sequence and variety of events may explainwhy IL-12, when compared to other cytokines, is effective against awide range of tumour models.

MATERIAL AND METHODS

Mice

BALB/c mice from the Special Pathogen Free (SPF) Unit, ClareHall Laboratories, South Mimms, UK, 5 to 6 weeks of age, wereused in all experiments.

TumourA transplantable breast carcinoma, originated from a colony of

BALB/c c-neutransgenic mice (Thomaset al.,1996) was used as asolid epithelial tumour model. This breast carcinoma was passagedevery 14 to 16 days to BALB/c mice. At the start of eachexperiment, mice were injected s.c. in the right flank with 0.1 ml oftumour homogenate, containing approximately 13 106 cells.Starting 3 days after tumor injection, mice were treated i.p. dailywith IL-12 or IFN-g, for the duration of the experiment. Controlgroups were injected with control diluent (PBS, containing 0.1%murine serum albumin, MSA) throughout the same period until thelimit tumour size (2 cm3) was reached.

CytokinesRecombinant murine IL-12 was a kind gift of Dr. M.J. Brunda

(Hoffman-LaRoche, Nutley, NJ). It had a bioactivity of 2.33 108

Roche U/mg. For each experiment, mice were injected i.p. with 1µg IL-12/day in a volume of 0.1 ml. Rat IFN-g was provided by

*Correspondence to: Biological Therapies Laboratory, Imperial CancerResearch Fund, 44 Lincoln’s Inn Fields, London WC2A 3PX, UK. Fax:(44) 171 269 3094. E-mail: diassjr@icrf.icnet.uk

Received 15 July 1997; Revised 20 August 1997

Int. J. Cancer:75,151–157 (1998)

r 1998 Wiley-Liss, Inc.

Publication of the International Union Against CancerPublication de l’Union Internationale Contre le Cancer

Roussel UCLAF (Romainville, France). It had a specific activity of1 3 107 U/mg. For each experiment, mice were injected i.p. with 1µg IFN-g/day in a volume of 0.1 ml. Both IL-12 and IFN-g werediluted in PBS/0.1% MSA (Sigma, Poole, UK) to a final concentra-tion of 10 µg/ml and stored in 1 ml aliquots at270°C prior to use.

RNA isolation and Northern blottingTotal RNA was isolated from tumours using Tri Reagent

(Molecular Research Center, Cincinnati, OH) following the manu-facturer’s instructions; 15 µg of total RNA were resolved on a 1%denaturing formaldehyde–agarose gel and blotted by capillarytransfer onto a nylon membrane (Hybond N1; Amersham, Ayles-bury, UK). RNA was then cross-linked to the nylon membraneusing a Stratalinker (Stratagene, Cambridge, UK), and blots werepre-hybridised for 2 hr at 42°C and hybridised using 45%formamide (GIBCO BRL, Paisley, UK), 7% SDS (Sigma) inphosphate buffer (pH 7.2). Blots were incubated overnight at 42°Cin hybridisation buffer with a radioactively labelled (32P-dCTP,Amersham) c-DNA probe. Probes forb-actin, b2-microglobulin,crg-2 (murine IP-10) and granzyme B were labelled by randompriming using the Prime-It Labelling kit (Stratagene). After hybridi-sation, blots were rinsed briefly with 23 SSC, 0.1% SDS at roomtemperature (RT) and then washed to high stringency with 0.13SSC, 0.1% SDS twice for 15 min at 60°C. After washing, blotswere exposed to BIOMAX X-ray films (Kodak) at270°C for 16 to72 hr.

cDNA synthesis and RT-PCRTotal RNA was obtained as above and precipitated, and the pellet

was resuspended in diethylpyrocarbonate (DEPC)-treated water.First-strand cDNA synthesis was performed with 5 µg of totalRNA, using the Ready-to-go kit (Pharmacia, Uppsala, Sweden),following the manufacturer’s instructions.

Murine primers were obtained from several sources. IFN-g andglyceraldehyde phosphate dehydrogenase (GAPDH) primers wereobtained from Clontech (Palo Alto, CA). Perforin and granzyme Bprimer sequences were taken from Tannenbaumet al. (1996). Theidentity of these primer sequences was verified by checkingdegrees of homology with all sequences available in GeneBank.

Two microliters of cDNA were used for each amplification, andnegative controls were obtained by using distilled water instead ofcDNA. All PCR reactions were carried out using a Perkin-ElmerGene-Amp kit (Norwalk, CT). All reactions were performed usinga Techne (Cambridge, UK) PCR machine, but a different reactioncycle was used for each set of primers. For Clontech primers, thefollowing protocol was used: 1 cycle, 94°C (5 min), 60°C (1 min),72°C (30 sec); 35 cycles, 94°C (45 sec), 60°C (45 sec), 72°C (2min); 1 cycle 72°C (7 min). For the remaining sets of primers, theprotocol for amplification was developed as described in Tannen-baumet al. (1996). All PCR reactions were resolved on a 1.2%agarose gel, with ethidium bromide (Sigma) staining.

cDNA probesb-actin. This was kindly provided by Dr. L. Kedes (Stanford,

CA). The 0.7 kb Eco R1-Hind III fragment in Bluescript was usedfor Northern analysis.

Murine IP-10 (crg-2).This was kindly provided by Dr. J. Farber(Bethesda, MD). The 0.56 kb Eco R1 fragment in Bluescript wasused for Northern analysis.

Murineb2-microglobulin (MHC-I).This was kindly provided byDr. D. Kioussis (London, UK). The 1.8 kb XbaI cDNA fragment ina Bluescript derivative was used for Northern analysis.

Granzyme B.This was kindly provided by Dr. C. Tannenbaum(Cleveland, OH). The 0.9 kb Eco R1 fragment was used forNorthern analysis.

Immuno-histochemistryAntibodies.All antibodies were purchased from PharMingen

(San Diego, CA) unless otherwise stated. A panel of antibodiesagainst mouse surface molecules was used, to characterise the

cellular infiltrate in this model: anti-CD3 for all T lymphocytes,anti-mac-3 for monocytes/macrophages, anti-CD4 for helper Tlymphocytes, anti-CD8 for cytotoxic T lymphocytes. An anti-CD31 antibody (kindly donated by Dr. R. Bicknell, Oxford, UK)was used as an endothelial cell marker, for the detection andquantification of blood vessels.

Methods.The appropriate antibody concentrations were deter-mined using spleen and thymus sections as a positive control.Negative controls were obtained by omitting the primary antibody.All tissue sections were treated with H2O2 to block endogenousperoxidase prior to staining. Before applying the primary monoclo-nal antibody (MAb), sections were incubated with 1% normalrabbit serum (DAKO, High Wycombe, UK). For the anti-CD3antibody, a 3-step alkaline phosphatase system was used. Briefly,the primary antibody was applied at the appropriate concentration,followed by a biotinylated secondary polyclonal antibody (ratanti-hamster, Jackson Immuno Research, Baltimore, MD) and,finally, by a streptavidin–alkaline phosphatase conjugate (PharMin-gen). Each incubation was performed at RT for 40 min. Betweeneach step, sections were washed twice with PBS/1% BSA for 5min. Staining was developed using the Vector-Red kit (Peterbor-ough, UK), and sections were counterstained with 1% toluidineblue. For the remaining antibodies, a 2-step peroxidase method wasused. Sections were processed as above, but a rabbit anti-ratantibody (DAKO) conjugated with peroxidase was employed todetect the primary antibody. Staining was revealed using thechromogen 3,5-diaminobenzidine (DAB, Sigma) by incubatingsections for 3 to 4 min at RT. Toluidine blue was used forcounterstaining, as above.

Quantitative microscopyCellular infiltrates.Cellular infiltrates were identified by light

microscopy at a magnification of31000, under oil immersion. Foreach time point, 4 sections were counted, 1 section from each of 4animals. Twenty randomly chosen high-power fields (HPF) werecounted for each section, using a 103 10 index grid (Graticules,Tonbridge, UK) and the LUCIA image analysis system (Nikon,Kingston, UK). Only cells with cytoplasmic staining, nuclearcounterstaining and appropriate morphology were included in ourevaluation. Results are shown as the mean infiltrating cells per 80HPF. Each HPF measured 0.012 mm2.

Blood vessels.For the detection and quantification of bloodvessels with anti-CD31, sections were observed at a magnificationof 3400 with the aid of a Chalkley grid (Graticules). For each timepoint, 4 sections were counted, equivalent to 4 animals per timepoint. Blood vessel areas were determined by counting 20 ran-domly chosen HPFs for each section. Results are given as theproportion of blood vessel areas to the remaining areas determinedfor each section (the mean of 4 tumours/time point plus standarddeviations).

In situapoptosis detection assayWe used a variation of the original TUNEL protocol, with an

apoptosis-detection kit (Promega, Madison, WI). Briefly, paraffin-embedded sections were deparaffinised in fresh xylene, rehydratedthrough graded ethanol washes (100%, 95%, 80% and 50%), fixedin 4% paraformaldehyde (5 min), permeabilised with Proteinase K(Sigma) for 8 min and blocked with 0.3% H2O2 in methanol.Sections were washed twice in PBS for 5 min at RT between eachprocedure. Finally, sections were incubated with equilibrationbuffer for 10 min at RT and incubation buffer for 60 min at 37°C.Both buffers were provided in the kit, but we used biotinylatednucleotides (Boehringer Mannheim, Mannheim, Germany) insteadof the FITC-labelled ones suggested in the manufacturer’s instruc-tions. The reaction was terminated by immersing the slides in 300mM sodium chloride, 30 mM sodium citrate for 15 min at RT.Sections were then incubated with a PBS/2% BSA solution for 10min at RT and finally covered with extra-avidin peroxidase (Sigma)for 30 min at 37°C. Colour was developed with the DAB

152 DIAS ET AL.

peroxidase substrate. Slides were counterstained with toluidineblue and examined by optical microscopy.

Quantification of apoptosis.Only brown-stained nuclei (tumourcells) were counted as positive, and a total of 20 HPFs was countedfor each section (4 sections per time point) using a 103 10 indexgrid (Graticules) and the LUCIA image analysis system, as above.Results are given as a proportion of apoptotic cells to the totalnumber of cells counted.

Statistical analysisStatistical evaluation of survival data and tumour volumes was

performed using the non-parametric 1-tailed Mann-Whitney U test.Statistical evaluation of cellular infiltrates and apoptotic cells wasdone using a 2-sample, unpaired Student’st test. Finally, thestatistical evaluation of blood vessel areas was performed using arisk difference test. All tests were used under the advise of the ICRFMedical Statistics Group (Institute for Health Sciences, Oxford, UK).

RESULTS

Assessment of survival

IL-12 had a clear anti-tumour effect in our model. Micereceiving IL-12 survived longer than mice receiving control diluentalone (Fig. 1), and there was a correlation with duration oftreatment. Fifteen days of daily IL-12 therapy produced a 5-foldincrease in survival when compared to mice receiving controldiluent alone. After this time point, continued therapy increased thechances of complete tumour regression (1⁄8 complete regressionwith 15 days of treatment and7⁄8 complete regressions with 45 daysof treatment) and produced a significant difference in survival overcontrol (p 5 0.014 after 45 days of treatment). For the 45 days oftherapy, mice first were injected daily for 30 days and then received1 µg IL-12 twice weekly for the remaining 15 days. With thistreatment schedule, significant reductions in tumour mass wereobserved in IL-12-treated mice compared to controls (p 5 0.01).IFN-g therapy did not cause a significant increase in survival, but adelay in tumour growth was observed (Fig. 2). Initially, IFN-g had

a significant effect on tumour size (p 5 0.05 at the first 7 days), butafter 14 days of treatment this was no longer apparent (Fig. 2).

IFN-g and IFN-g-inducible genesmRNA expression for IFN-g and IFN-g-inducible genes was

studied at 4 time points during treatment with IL-12 (8 and 24 hr, 3and 7 days). In 3 separate experiments, IFN-g mRNA was detectedby RT-PCR after 8 hr of IL-12 therapy and then at all time points upto 7 days (Fig. 3a). IFN-g mRNA was not detected at 8 hr in controltumours but could be detected after 24 hr. IFN-g-treated tumours,however, did not show IFN-g at any time point analysed (Fig. 3a).IFN-g mRNAwas not detectable in any sample by Northern blotting.

Northern analysis showed a strong IP-10 induction in IL-12-treated tumours 24 hr after start of therapy, though it could bedetected 8 hr after the first injection (Fig. 3b). IFN-g-treatedtumours also showed an increase of IP-10 after 24 hr; however, thiswas not sustained, and after 3 days the level of expression of thischemokine was much lower (Fig. 3b). Control tumours had lowlevels of IP-10 expression for all time points analysed.

Using Northern analysis, we next studied MHC class I expres-sion by probing for murineb2-microglobulin. mRNA for thisprotein was detected first on IL-12-treated tumours 8 hr after thestart of treatment. IFN-g-treated tumours showed induction ofb2-microglobulin 24 hr after treatment began. Its expressionpeaked 3 days after the first injection in both groups (Fig. 3b).Control tumours had low levels of expression forb2-microglobu-lin, peaking at 3 days (Fig. 3b).

Expression of T cell effector moleculesmRNAs for perforin and granzyme B were detected first by

RT-PCR after 24 hr of IL-12 therapy (not shown). Further,expression of granzyme B, but not perforin, could be detected byNorthern analysis in IL-12-treated tumours (Fig. 3b), with a stronginduction 3 days after the start of therapy (Fig. 3b). IFN-g-treatedand control tumours showed low levels of expression of both

FIGURE 1 – Effects of IL-12 and IFN-g therapy on survival of micebearing the HTH-K breast tumour model. Forty mice were injectedwith approximately 13 106 tumour cells, as described in ‘‘Materialand Methods’’. Treatment started 3 days after tumour injection. Eightmice were given control diluent (PBS/MSA); 8 were treated i.p. with 1µg IL-12/day for 12, 15, or 45 days (30 daily doses plus twice weeklyfor the remaining 15 days); and the remaining 8 were treated i.p. with 1µg IFN-g/day until they were killed. There was a significant increase insurvival between IL-12-treated tumours (for 45 days) and controls( p 5 0.014).

FIGURE 2 – IL-12 induces tumour growth inhibition and regression.IL-12 (1 µg/mouse/day) was given for 15 or 45 days (30 daily dosesplus twice weekly for the remaining 15 days), and IFN-g (1 µg/mouse/day) was given until tumours reached the legal ethical limit. Tumourgrowth is represented as volume (cm3) and was plotted as the mean ofmice alive at every time point analysed. IL-12 therapy (45 days)resulted in significantly (p 5 0.01 for the duration of treatment)smaller tumour volumes compared to controls. Initially, IFN-g therapyshowed a significant delay in tumour growth (p 5 0.05 for the first 7days) compared to controls, but after 14 days this was no longer visible.

153MOLECULAR AND CELLULAR CHANGES AFTER IL-12 THERAPY

granzyme B and perforin, which could be detected only by RT-PCR(not shown).

Cell countsThe cellular infiltrate in the tumours was composed mainly of

Mac-31 monocytes/macrophages and CD31 T cells (Table I). After24 hr of therapy, IL-12-treated tumours had higher numbers ofCD31 cells than control tumours, and this was maintained for theduration of treatment. However, for the first 7 days of treatment,this difference was not significant. After 7 days, IL-12-treatedtumours had significantly more CD31 cells than controls (Table I,p 5 0.005). This difference was maintained after 14 days (Table I,p 5 0.01). CD81 cytotoxic T cells accounted for a majority of theCD31 cell population (64% at 14 days). CD41 T cells were alsopresent but did not show any significant variation between IL-12-treated and control tumours at any time point analysed. IFN-g-treated tumours had smaller infiltrates than IL-12-treated tumours,but these were more abundant than controls (Table I). Further, inIFN-g-treated tumours there was a smaller increase (not signifi-

cant) in the cellular infiltrate throughout treatment compared withthe IL-12-treated group.

Blood vessel countsCD-31 staining revealed a significant difference between the

proportion of vascularised areas in IL-12-treated and controltumours. After 7 days, control tumours were highly vascularisedcompared to the IL-12-treated group, and this difference was main-tained at 14 days (Fig. 4,p , 0.01 for both time points analysed).After 14 days of therapy, IFN-g-treated tumours were only mar-ginally less vascularized than control tumours (Fig. 4). Differencesbetween the 3 groups were clearly visible macroscopically (notshown). When removed after 7 days of daily therapy, IL-12-treatedtumours were much paler than control or IFN-g-treated tumours.

Levels of apoptosisIn IL-12-treated, IFN-g-treated and control tumours, apoptosis

was detectedin situ using the TUNEL technique. IL-12-treatedtumours had a higher incidence of apoptosis at all time points

FIGURE 3 – (a) Analysis of mRNA expression for IFN-g by RT-PCR, as described in ‘‘Material and Methods’’. Samples were normalised forloading using GAPDH primers, and a positive control (1), provided with the amplimer sets, was used to confirm the identity of the PCR products.mRNA levels are representative of 3 mice/group/time point (all 3 mice yielded identical results).(b) crg-2 (IP-10),b2-microglobulin (MHC classI) and granzyme B mRNA expression was analysed by Northern analysis. To estimate loading of mRNA from untreated controls, IL-12- andIFN-g-treated tumours, blots were exposed to a cDNA probe forb-actin (a house-keeping gene). mRNA levels of a particular gene, for each timepoint, are representative of 3 mice/group/time point (all 3 mice yielded identical results).

TABLE I – PHENOTYPES OF THE CELLULAR INFILTRATE ON IL-12-TREATED, IFN-g-TREATED AND CONTROL TUMOURS

24 hr 7 days 14 days

Control IFN-g IL-12 Control IFN-g IL-12 Control IFN-g IL-12

CD3 126 4.6 146 2.8 23.56 2.1 30.56 2.1 45.36 3.5 926 4.21 37.36 9.5 476 4.6 81.36 10.91

CD4 nd nd nd 14.36 2.1 216 1.8 24.56 1.3 17.86 2.2 20.56 2.1 266 1.8CD8 nd nd nd 15.56 2.6 25.36 1.3 626 2.161 19.86 2.2 256 4.1 52.36 2.91

Mac-3 216 2.8 196 1.4 23.56 2.4 416 1.4 526 3.6 46.56 2.1 40.36 3.8 496 4.2 40.36 1.5

Sections were stained for CD3, CD4, CD8 and Mac-3 and counterstained with toluidine blue. Results are shown as the number of infiltratingcells in 80 high power fields (mean of 4 tumours/time point plus standard deviations). Each high-power field was observed under31,000magnification and measured 0.012 mm2. nd, not done. Due to the small number of infiltrating cells at the 24 hr time point, it was not possible todetermine the number of CD4 and CD8 T cells accurately.–1After 7 days, IL-12-treated tumours had more CD31 (CD81) T cells than controls( p 5 0.005), and this difference was maintained at 14 days (p 5 0.01).

154 DIAS ET AL.

analysed compared to IFN-g-treated and control tumours (Fig. 5,p , 0.001 at 7 days compared to control tumours). IFN-g-treatedtumours also showed significantly higher levels of apoptosis thancontrols at 24 hr and 7 days (Fig. 5,p , 0.01) but not after 14 daysof treatment. In control tumours, the number of apoptotic cells didnot show significant variation at all time points analysed (Fig. 5).

DISCUSSION

In experimental cancer models, IL-12 produces more effectiveanti-tumour responses than most other cytokines. Although some

of the actions described here have been reported in other models(Tannenbaumet al.,1996), we have defined a number of reportedand novel actions of IL-12 in a new murine breast carcinoma modeland have determined the temporal sequence of these events. Wealso compared the molecular and cellular changes induced intumours by IL-12 with those induced by IFN-g. A dose of 1µg/mouse/day was used for each cytokine. As IFN-g (m.w. of 15.9)is biologically active as a homodimer and IL-12 is a heterodimermolecule (m.w. of 62.6), these doses were reasonably comparable.However, the half-life of IL-12 is considerably longer than that ofIFN-g, and this should be taken into account when assessing theresults. The dose of IFN-g used in this study is equivalent to aclinically achievable dose in tumour patients and gives a similarpharmacokinetic profile (Maliket al.,1991).

We have shown in this model that the actions of IL-12 in thetumour micro-environment involve a variety of events exerted atdifferent levels: the establishment of an anti-tumour immuneresponse, direct toxicity against tumour cells and anti-angiogeniceffects. The observation that after IL-12 therapy tumour cellapoptosis occurs early on and is sustained throughout treatment hasnot been reported before, and the range of molecular and cellularchanges have not been studied in a single model. Data from thisreport and many other experimental tumour results suggest thatcytokines, such as IFN-g, produce some, but not all, of the effectsinduced by IL-12 and, thus, achieve only partial anti-tumourresponses. Finally, our results suggest that induction of thechemokine IP-10 and activation of the infiltrating cells are some ofthe key changes seen after IL-12 therapy. A diagramme of thesequence of cellular and molecular events that occurs in IL-12-treated tumours is summarised in Figure 6.

Our data and other published work (Tannenbaumet al., 1996)suggest that one of the first actions of IL-12 against a developingtumour results in the establishment of an immune response. Thisinvolves the production of IFN-g by infiltrating cells (Nastalaet al.,1994). mRNA for IFN-g was detected at very low levels inIL-12-treated HTH-K tumours as early as 8 hr after the start oftreatment. Following this early presence of IFN-g, several geneswere induced, such as IP-10. This protein is an IFN-g-induciblechemokine, an attractant of T lymphocytes with anti-angiogenicproperties (Sgadariet al., 1996b). mRNA for IP-10 was detectedafter 8 hr of IL-12 therapy. In IFN-g-treated tumours, IP-10 wasdetected 24 hr after the first treatment, and its induction was notsustained. There was a quantitative difference between the infil-trates in IL-12/IFN-g-treated tumours and controls after 24 hr,which suggests that induction of IP-10 may have contributed to therecruitment of cellular infiltrates into the tumours. The sustainedinduction of this chemokine in IL-12-treated tumours may accountfor the higher level of infiltrating CD81 T cells seen in these, butnot in IFN-g-treated or control tumours at later time points.

The presence of local or systemic IFN-g led also to the inductionof other genes, likeb2-microglobulin, which was detected inIL-12-treated tumours 8 hr after the first injection.b2-microglobu-lin continued to be expressed at high levels after 7 days of therapyin both IL-12- and IFN-g-treated tumours. This suggests thatantigen presentationvia MHC class I is an important step in theestablishment of an immune response by IL-12, mediated byIFN-g. Further, we could find evidence of T-cell activation only inIL-12-treated tumours. mRNA for granzyme B and perforin wasdetected 24 hr after the start of treatment in IL-12-treated tumoursbut not in IFN-g-treated or control tumours. mRNA levels forgranzyme B in IL-12-treated tumours were higher than those forperforin, confirming previous studies suggesting that IL-12 is astrong inducer of granzyme B (Salcedoet al.,1993). The detectionof their specific mRNA confirmed the presence of activatedcytotoxic T cells after IL-12 therapy, as determined by immuno-histochemistry. Further, immuno-histochemical detection of thevery early activation antigen CD-69 also indicated the presence ofactivated T cells in IL-12-treated tumours but not in IFN-g orcontrol tumours (data not shown). These observations confirm

FIGURE 4 – Vascularisation within the tumours is reduced afterIL-12 therapy but not after IFN-g therapy. Tumour sections werestained for CD-31, as described in ‘‘Material and Methods,’’ and bloodvessel areas determined using the Chalkley array. Sections wereobserved at3400 magnification. Results are shown as the proportionof blood vessel areas to the remaining areas in the tumour system (themean of 4 tumours/time point plus standard deviations). **IL-12-treated tumours have significantly smaller vascularised areas thancontrols (p , 0.01).

FIGURE 5 – IL-12 therapy induces tumour cell killing by apoptosis.Sections were stained using the TUNEL technique, as described in‘‘Material and Methods’’. Time points after treatment were 24 hr, 7days and 14 days. Twenty high-power fields were counted for eachsection (4 tumours per time point). Each high-power field was observedunder 31,000 magnification and measured 0.012 mm2. Results areshown as the proportion of apoptotic cells to the total number of cells(mean of 4 tumours/time point plus standard deviations). *Significantlyhigher than controls, p, 0.01; **p , 0.005; ***p , 0.001; ns, notsignificant.

155MOLECULAR AND CELLULAR CHANGES AFTER IL-12 THERAPY

other reports (Nastalaet al., 1994) suggesting that an essentialmechanism developed during the anti-tumour actions of IL-12 isthe establishment of an immune response. The involvement ofCD81 cytotoxic T cells has been suggested previously (Brundaetal., 1993), but our results suggest that antigen presentation andsubsequent T-cell activation occurvia MHC class I after IL-12therapy.

Following early IFN-g production and the establishment of animmune response, a subsequent phase in the actions of IL-12involves toxicity against tumour cells. This may be mediated byother IFN-g-inducible genes, such as the enzyme IDO. This proteinis involved in the degradation of tryptophan, which may be one ofthe mechanisms by which IFN-g produces its anti-proliferativeeffects (Burkeet al., 1995). IDO was detected in IL-12- andIFN-g-treated tumours, but not in controls, after 3 days of dailytherapy and could be detected still after 7 days (data not shown). Byreducing the tryptophan content of tumour cells, IDO may reducetheir viability and ability to grow. The detection of mRNA for IDOin both IL-12- and IFN-g-treated tumours suggests that toxicityagainst tumour cells, through the induction of IDO, may benecessary for the development of an anti-tumour response, but it isnot sufficient, as shown by Burkeet al.(1995).

Other effects developed during the anti-tumour actions of IL-12involve an anti-angiogenic response. The anti-angiogenic effectsof IL-12 have been reported in other tumour model systems (Voestet al.,1995). After 7 days of daily treatment, there was a reductionin the vascularity of IL-12-treated tumours when compared tocontrols, and this effect was maintained after 14 days of therapy.IFN-g caused a lesser and transient anti-angiogenic effect through-out the course of treatment. These observations could be explainedby the levels of expression of the chemokine IP-10. This chemo-kine has anti-angiogenic properties (Sgadariet al., 1996b) and is

involved in the anti-angiogenic effects of IL-12 in other models(Sgadari et al., 1996b). Our data suggest that its sustainedinduction at the tumour site is necessary for a full anti-angiogeniceffect to develop. However, we do not discount the possibility ofother cytokines, chemokines or cytokine antagonists being in-volved in the process.

All of the events described above may contribute to the reductionin tumour volume and mass observed after IL-12 therapy. IL-12 isnot capable of acting directly against tumour cellsin vitro (Brundaet al., 1993), but some of its secondary products may affect theirgrowth and survivalin vivo. TNF-a was not induced in IL-12-treated tumours (data not shown), suggesting that the anti-tumoureffects of IL-12, at least in this model, do not require this cytokine.However, IFN-g (Burke et al., 1997), CD81 cytotoxic T cellsviathe perforin granzymes (Berke, 1995) and inhibition of angiogen-esis (Hanahan and Folkman, 1996) all induce cell death byapoptosis, and each one of these mechanisms was detected afterIL-12 therapy. IL-12-treated tumours showed a higher incidence ofapoptosis compared to IFN-g-treated or control tumours immedi-ately after 24 hr of treatment. This may be the result of the earlydevelopment of an anti-tumour immune response and toxicityagainst tumour cells, possibly through the induction of nitric oxidesynthase (Wiggintonet al.,1996). Levels of apoptosis are sustainedthroughout treatment as a result of the remaining actions of IL-12,induction of other toxic factors such as IDO, maintenance of ananti-tumour immune response and anti-angiogenic mechanisms.The combined action of these effects led to eradication of theestablished tumour.

Individual cytokine therapy of human tumours has been oflimited success (Gutterman, 1994). An understanding of thesequence of molecular and cellular events in an experimentaltumour successfully treated with a cytokine may lead to the design

FIGURE 6 – Schematic representation of the temporal sequence of events produced by IL-12. Scale starts 3 days after tumour injection (day 0 oftreatment). Molecular changes were studied over 7 days and phenotypic changes over 14 days.

156 DIAS ET AL.

of sequential or combination cytokine/biological therapies that canachieve similar results with acceptable toxicity.

ACKNOWLEDGEMENTS

The authors are grateful to Mrs. H. Holdsworth for experttechnical assistance; Dr. R. Negus for help with the image analysis

system and valuable discussions; Mr. G. Elia for histologicalprocessing of the tumours; Dr. R. Bicknell for the CD-31 antibody;Mr. M. Bradburn (ICRF Medical Statistics Group) for statisticalevaluation of the data; and Drs. M. Owen, J. Taylor-Papadimitriouand M.J. Brunda for critically reviewing this manuscript. S.D. isthe recipient of a JNICT (Portugal) fellowship (PFMRH/BD/1546/94).

REFERENCES

BERKE, G., The CTL’s kiss of death.Cell, 81,9–12 (1995).

BRUNDA, M.J., LUISTRO, L., HENDRZAK, J.A., FONTOULAKIS, M., GAROTTA,G. and GATELY, M.K., Role of interferon-gamma in mediating the antitumorefficacy of interleukin-12.J. Immunother.,17,71–77 (1995).

BRUNDA, M.J., LUISTRO, L., RUMENNIK, L., WRIGHT, R.B., DVOROZNIAK, M.,ALIGONE, A., WIGGINTON, J.M., WILTROUT, R.H., HENDRZAK, J.A. andPALLERONI, A.V., Antitumor activity of interleukin 12 in preclinical models.Cancer Chemother. Pharmacol. (Suppl.),38,S16–S21 (1996).

BRUNDA, M.J., LUISTRO, L., WARRIER, R.R., WRIGHT, R.B., HUBBARD, B.R.,MURPHY, M., WOLF, S.F. and GATELY, M.K., Antitumor and antimetastaticactivity of interleukin 12 against murine tumors.J. exp. Med.,178,1223–1230 (1993).

BURKE, F., EAST, N., UPTON, C., PATEL, K. and BALKWILL , F., Interferon-ginduces cell cycle arrest and apoptosis in a model of ovarian cancer.Europ.J. Cancer,33,1114–1121 (1997).

BURKE, F., KNOWLES, R.G., EAST, N. and BALKWILL , F.R., The role ofindoleamine 2,3-dioxygenase in the anti-tumour activity of human interferon-gammain vivo. Int. J. Cancer,60,115–122 (1995).

CAR, B.D., ENG, V.M., SCHNYDER, B., LEHIR, M., SHAKHOV, A.N., WOERLY,G., HUANG, S., AGUET, M., ANDERSON, T.D. and RYFFEL, B., Role ofinterferon-gamma in interleukin 12-induced pathology in mice.Amer. J.Pathol.,147,1693–1707 (1995).

D’A NDREA, A. and 12OTHERS, Production of natural killer cell stimulatoryfactor (interleukin 12) by peripheral blood mononuclear cells.J. exp. Med.,176,1387–1398 (1992).

GAZZINELLI , R.T., Molecular and cellular basis of interleukin 12 activity inprophylaxis and therapy against infectious diseases.Mol. Med. Today,2,258–267 (1996).

GUTTERMAN, J.U., Cytokine therapeutics: lessons from interferona. Proc.nat. Acad. Sci. (Wash.),91,1198–1205 (1994).

HANAHAN , D. and FOLKMAN , J., Patterns and emerging mechanisms of theangiogenic switch during tumorigenesis.Cell, 86,353–364 (1996).

JACKSON, J.D., YAN, Y., BRUNDA, M.J., KELSEY, L.S. and TALMADGE, J.E.,Interleukin-12 enhances peripheral hematopoiesisin-vivo. Blood, 85,2371–2376 (1995).

MALIK , S.T.A., KNOWLES, R.G., EAST, N., LANDO, D., STAMP, G. andBALKWILL , F.R., Antitumour activity of gamma interferon in ascitic and

solid tumour models of human ovarian cancer.Cancer Res.,51,6643–6649(1991).NASTALA, C.L., EDINGTON, H.D., MCKINNEY, T.G., TAHARA, H., NALESNIK,M.A., BRUNDA, M.J., GATELY, M.K., WOLF, S.F., SCHREIBER, R.D., STORKUS,W.J. and LOTZE, M.T., Recombinant IL-12 administration induces tumorregression in association with IFN-gamma production.J. Immunol.,153,1697–1706 (1994).SALCEDO, T.W., AZZONI, L., WOLF, S.F. and PERUSSIA, B., Modulation ofperforin and granzyme messenger RNA expression in human natural killercells.J. Immunol.,151,2511–2520 (1993).SGADARI, C., ANGIOLILLO, A.L., CHERNEY, B.W., PIKE, S.E., FARBER, J.M.,KONIARIS, L.G., VANGURI, P., BURD, P.R., SHEIKH, N., GUPTA, G., TERUYA,F.J. and TOSATO, G., Interferon-inducible protein-10 identified as a mediatorof tumor necrosisin vivo. Proc. nat. Acad. Sci. (Wash.),93, 13791–13796(1996a).SGADARI, C., ANGIOLILLO, A.L. and TOSATO, G., Inhibition of angiogenesisby interleukin-12 is mediated by the interferon-inducible protein 10.Blood,87,3877–3882 (1996b).TANNENBAUM, C.S., WICKER, N., ARMSTRONG, D., TUBBS, R., FINKE, J.,BUKOWSKI, R.M. and HAMILTON , T.A., Cytokine and chemokine expressionin tumors of mice receiving systemic therapy with IL-12.J. Immunol.,156,693–699 (1996).THOMAS, H., HANBY, A.M., SMITH, R.A., HAGGER, P., PATEL, K., RAIKUNDA -LIA , B., CAMPLEJOHN, R.S. and BALKWILL , F.R., An inbred colony ofoncogene transgenic mice: diversity of tumours and potential as a therapeu-tic model.Brit. J. Cancer,73,65–72 (1996).TRINCHIERI, G., Interleukin-12: a cytokine produced by antigen-presentingcells with immunoregulatory functions in the generation of T-helper cellstype 1 and cytotoxic lymphocytes.Blood,84,4008–4027 (1994).VOEST, E.E., KENYON, B.M., O’REILLY, M.S., TRUITT, G., D’AMATO, R.J.and FOLKMAN , J., Inhibition of angiogenesisin vivoby interleukin 12.J. nat.Cancer Inst.,87,581–586 (1995).WIGGINTON, J.M., KUHNS, D.B., BACK, T.C., BRUNDA, M.J., WILTROUT, R.H.and COX, G.W., Interleukin-12 primes macrophages for nitric-oxide produc-tion in-vivoand restores depressed nitric-oxide production by macrophagesfrom tumor-bearing mice—implications for the antitumor-activity of inter-leukin-12 and/or interleukin-2.Cancer Res.,56,1131–1136 (1996).YOUNG, H.A. and HARDY, K.J., Role of interferon-gamma in immune cellregulation.J. Leukocyte Biol.,58,373–381 (1995).

157MOLECULAR AND CELLULAR CHANGES AFTER IL-12 THERAPY