Interleukin 12-activated Lymphocytes Influence Tumor Genetic Programs¹ Free

Cancer results from the accumulation of a series of genetic events that convert a normal cell into a transformed cell able to proliferate without restraint, attract its own blood vessels, spread around the body, and evade immune reactivity (1, 2). Numerous ways by which tumors sneak through immune reactions have been elucidated. In many cases, cells release factors that favor tumor growth by inhibiting gene expression in T-LYs⁴ (3), impairing antigen presentation (4) or activating suppressor functions in macrophages (5). On the other hand, the immune system hampers tumor growth through the activity of CTLs (6), natural killer cells (7), macrophages (5), and granulocytes (8) that specifically and nonspecifically kill tumor cells. Lymphoid helper cells also may activate and guide an inflammatory-like reaction at the tumor site (9) by releasing cytokines and other factors.

Exogenous proinflammatory cytokines also trigger the immune system to develop strong antitumor reactions. Systemic (10) and local (11) administration of IL-12 is particularly effective because it activates a complex reaction that halts the growth of a large number of transplantable (10, 11) as well as chemically (12) and oncogene-induced (13) primary mouse tumors. Tumor inhibition appears to result from the activation of endothelial cells and their ability to recruit lymphoid cells to the tumor site. IL-12-activated lymphoid cells kill tumor cells, damage tumor vessels, release secondary messengers, such as IFN-γ and TNF-α, and trigger the release of third-level chemokines, such as IP-10, MIG, and other factors, that counteract tumor angiogenesis (11, 14).

Here we show that by directly influencing tumor cells through the cytokines they release, lymphoid cells activated in the presence of IL-12 also modulate the growth of a tumor in a more subtle way by changing its genetic programs, so that it becomes a party to its own inhibition.

Seven-week-old female BALB/cAnCr mice and BALB/c-GKO mice from Charles River Laboratories (Calco, Italy) were treated in accordance with Italian and European Union guidelines.

TSA is an aggressive and poorly immunogenic cell line established from the first in vivo transplant of a moderately differentiated mammary adenocarcinoma that arose spontaneously in a BALB/c female mouse (15). TUBO cells are a p185neu⁺ cell line established from a lobular carcinoma that spontaneously arose in a BALB/c female mouse transgenic for the transforming rat Her-2/neu oncogene driven by the mouse mammary tumor virus promoter (13). TSA and TUBO cells express MHC class I but not MHC class II glycoproteins and spontaneously secrete granulocyte colony-stimulating factor, GM-CSF, TGF-β1, VEGF, and basic FGF (11). TSA cells were maintained in RPMI 1640 (Bio*Whittaker Europe, Verviers, Belgium) with 50 μg/ml gentamicin (Bio*Whittaker Europe), 2.5 × 10⁻⁵ m 2β-mercaptoethanol (Flow Laboratories, Opera, Italy), and 10% FCS (Life Technologies, Inc., San Giuliano Milanese, Italy; RPMI complete medium). TUBO cells were maintained in DMEM (Bio*Whittaker Europe) supplemented with 50 μg/ml gentamicin (Bio*Whittaker Europe), 2.5 × 10⁻⁵ m 2β-mercaptoethanol (Flow Laboratories), and 20% FCS (Life Technologies, Inc.; DMEM complete medium). All tumor and lymphoid cell cultures were performed in a humidified 5% CO₂ atmosphere at 37°C.

Total Spcs (2 × 10⁶)/ml from normal and GKO mice were stimulated for 18 h with 2 μg/ml anti-CD3 and anti-CD28 mAbs (PharMingen, San Diego, CA) in RPMI complete medium in the presence or absence of 10 ng/ml recombinant IL-12 (Dr. Michael Brunda; Hoffmann-La Roche, Nutley, NJ). After incubation, activated LYs were washed and placed in the transwell inserts (Falcon; Becton Dickinson Labware Europe, Milan, Italy) or used in other tests. Supernatants were collected to assess cytokine contents.

A total of 200 μl of a suspension of 2 × 10⁶ activated LYs/ml were cultured in complete RPMI 1640 in triplicate in round-bottomed 96-well plates (Falcon) and pulsed with 1 μCi of [³H]thymidine (Amersham, Milan, Italy). After 4 h, the cells were harvested on a glass fiber filter, and [³H]thymidine uptake was evaluated with a Matrix-96 beta counter (Camberra-Packard, Milan, Italy). The results were expressed as the arithmetic mean ± SD of total cpm.

Tumor cells (4 × 10⁵) from confluent monolayers were cultured at the bottom of the wells of 6-well plates (Falcon) in DMEM complete medium. After 18 h, adherent tumor cells were washed with PBS, and 3 ml of RPMI complete medium were added to each well together with a cell culture insert of 0.4 μm pore size. The insert was filled with 3 ml of complete RPMI 1640 containing 2 × 10⁶ Spcs/ml with or without 10 ng/ml IL-12. The plates were then incubated for 96 h.

Rat monoclonal IgG (An18) neutralizing IFN-γ but not IFN-α/β was produced as described elsewhere (16). Ten μg/ml high-performance liquid chromatography-fractionated An18 mAb was added to transwell cocultures. A similar preparation of rat IgG of unknown specificity was used as the negative control (16).

Supernatants from activated LYs cultured in the transwells alone or in the presence of tumor cells were tested for the presence of IFN-γ, TNF-α, IL-2, IL-4, IL-6, IL-10, and GM-CSF using sandwich ELISA kits (PharMingen).

Tumor cells recovered by trypsinization from the transwells after 96 h of coculture with activated LYs were stained using a standard direct immunofluorescence procedure with FITC-conjugated antimouse mAb (all from PharMingen) against: (a) MHC class I (anti-H-2D^d, clone 34-2-12; anti-H-2K^d, clone SF1-1.1); (b) MHC class II (anti-I-Ad, clone 39-10-8); and (c) intracellular adhesion molecule 1 (anti-CD54, clone 3E2). Cells were then suspended in PBS containing 10 μg/ml propidium iodide to gate out dead cells and analyzed with a FACScan flow cytometer (Becton Dickinson, Mountain View, CA). After overnight activation, LYs were analyzed for the expression of surface markers using the following mAbs (PharMingen): (a) anti-CD4; (b) anti-CD8; (c) anti-CD25; (d) anti-CD54; (e) anti-CD69; (f) anti-CD80; (g) anti-CD86; (h) anti-CD154; (i) anti-H-2D^d; (j) anti H-2K^d; (k) anti I-A^d; and (l) anti-B220.2. For the intracellular detection of phosphorylated STAT1 and STAT3, both tumor and lymphoid cells were treated as described previously by Fleisher et al. (17). Briefly, after the fixation and permeabilization steps, 1 × 10⁶ cells were incubated with 0.1 μg of rabbit IgG (Caltag Laboratories, Burlingame, CA) and 1 μg of rabbit antiphosphorylated STAT1 or rabbit antiphosphorylated STAT3 polyclonal antibodies (New England Biolabs, Beverly, MA) at 4°C for 30 min. Cells were washed twice and stained with 1 μg of FITC-conjugated F(ab′)₂ swine antirabbit IgG (Dako, Glostrup, Denmark) and incubated for 30 min at 4°C. After a final washing step, the cells were suspended in 200 μl of PBS and analyzed with a FACScan flow cytometer (Becton Dickinson). Each plot represents the results from 10,000 events.

To avoid contaminations of genomic DNA, 40 μg of total RNA were incubated with 4 units of DNase I (Clontech, Palo Alto, CA) for 30 min at 37°C. The digestion was stopped with 20 μl of 10× termination mix [0.1 m EDTA (pH 8.0) and 1 mg/ml glycogen], and the RNA was extracted once with 300 μl of 2:1 (v:v) phenol/chloroform solution [equilibrated with 0.1 m sodium citrate (pH 4.5) and 1 mm EDTA] and then with 200 μl of chloroform. The aqueous phase was precipitated by adding 0.1 volume of 2 m NaOAc (pH 4.5) and 2.5 volumes of 95% ethanol (−20°C, 20 min). After centrifugation (14,000 rpm at 4°C for 15 min), the pellet was washed with 80% ethanol and air dried. RNA was dissolved in 10 μl of RNase-free water. Its concentration was determined at A_{260 nm}, and its quality was evaluated by denaturing agarose gel analysis.

RNA labeling was performed with Superscript II Reverse Transcriptase (Life Technologies, Inc., Grand Island, NY) as follows: 500 ng to 1 μg of total RNA and 1 μl of 10× CDS primer mix (Atlas cDNA Expression Arrays; Clontech) in a final volume of 10 μl were incubated at 70°C for 10 min and then chilled on ice. The primed RNA was incubated at 37°C for 90 min with 6 μl of 5× First Strand Buffer (Life Technologies, Inc.), 1 μl of 0.1 m DTT (Sigma), 1.5 μl of a deoxynucleotide triphosphate mixture (20 mm dCTP, 20 mm dGTP, and 20 mm dTTP), 1.5 μl of Superscript II Reverse Transcriptase (Life Technologies, Inc.), and 10 μl of [α-³³P]dATP (10 mCi/ml; 3000 Ci/mmol specific activity; Amersham International, Bucks, United Kingdom). Labeled cDNA was purified by column chromatography with the Chroma Spin 200 DEPC-H₂O columns (Atlas cDNA Expression Arrays; Clontech).

cDNA probes (∼200 μl) were first denatured with 22 μl of 10× denaturing solution (1 m NaOH and 10 mm EDTA) for 20 min at 68°C. The probe solution was neutralized with 225 μl of 2× neutralization solution [1 m NaH₂PO₄ (pH 7.0)] and incubated at 68°C for 10 min with 5 μl of C_ot DNA (Clontech). Atlas Mouse cDNA Expression Array I (Clontech) is a positively charged nylon membrane (8 × 12 cm) spotted in duplicate with cDNA fragments representing 588 known genes and 21 housekeeping genes or control sequences. These arrays were prehybridized at 68°C for 30 min with 5 ml of ExpressHyb (Clontech) and 500 μg of heat-denatured sheared salmon testes DNA. Hybridization was performed for 18 h at 68°C in roller bottles, with continuous agitation. Arrays were washed four times in 150 ml of 2× SSC and 1% SDS and once in 150 ml of 0.1× SSC and 0.5% SDS at 68°C for 30 min with continuous agitation. One final wash in 150 ml of 2× SSC was performed at room temperature for 5 min. Hybridized arrays were exposed in a phosphorimaging cassette for 20 h.

The numerical data corresponding to the integrated radioactive intensity of each DNA array spot (radioactivity volume) were generated by phosphorimager analysis. They were analyzed with the DNA_MAP program. DNA_MAP analysis approach information can be obtained on the internet.⁵

Total cellular RNA was extracted from TUBO and TSA tumor cells cocultured in the transwell with activated LYs as described by Chomczynski and Sacchi (18) by using the RNAble solution (Eurobio, Les Ulis Cedex, France). cDNA was prepared by reverse transcription at 42°C for 30 min in a 50-μl reaction mixture containing 4 μg of total RNA, 0.5 μg of oligo(dT), 1 mm 2′-deoxynucleotide-5′-triphosphate, 5 μl of 10× RT buffer [100 mm Tris-HCl (pH 8.8), 500 mm KCl, and 1% Triton X-100], 5 mm MgCl₂, 40 units of recombinant RNasin RNase inhibitor, and 25 units of avian Moloney virus reverse transcriptase. All reagents for cDNA synthesis were from Promega Corp. (Madison, WI). cDNA (2.5 μl) from each sample (generated using 0.2 μg of total RNA) was amplified using a 9600 Thermal Cycle (PE Biosystem, Norwalk, CT) in a final volume of 25 μl with individual PCR cycle conditions for each set of primers. Specific primers for mouse G3PDH and iNOS were obtained from Clontech and used in the following conditions: 1× PCR Buffer, 0.2 mm deoxynucleotide triphosphates, 1 mm MgCl₂, 0.2 μm each primer, and 1 unit of Taq polymerase (all from Polymed, Florence, Italy). The other primers used in this work were designed on the basis of the gene sequences: (a) LMP2, 5′-AGGAACAGCAGTGGTGAACC-3′ and 5′-TGTAGGAGCTTCCAGAACCG-3′ (amplified fragment of 334 bp); (b) LMP7, 5′-CACACTCGCCTTCAAGTTCC-3′ and 5′-AACCGTCTTCCTTCATGTGG-3′ (amplified fragment of 554 bp); (c) PA28, 5′-AGGAGGCTGATGACTTCCTC-3′ and 5′-TCCAGACTTCTGGCTTAACC-3′ (amplified fragment of 260 bp); (d) IP-10, 5′-GCGTTAACCTCCCCATCAGCACCATGAAC-3′ and 5′-CCGCTCGAGGTGGCTTCTCTCCAGTTAAGGA-3′ (amplified fragment of 300 bp); (e) MIG1, 5′-TCCGCTGTTCTTTTCCTTTTGG-3′ and 5′-TTGAACGACGACGACTTTGGGG-3′ (amplified fragment of 361 bp); (f) MCP1, 5′-GCTCTCTCTTCCTCCACCAC-3′ and 5′-CGGGTCAACTTCACATTCAA-3′ (amplified fragment of 383 bp); (g) VEGF, 5′-CACAGCCAATGTGAATGCA-3′ and 5′-ACGTAGATCTTCACTTTCGCGGCTTCCG-3′ (two amplified fragments of 324 and 223 bp corresponding to two different isoforms of VEGF); (h) Ang1, 5′-GAAGATATAACCGGATTCAAC-3′ and 5′-TGACAAGGTTATGAACTGTGT-3′ (amplified fragment of 698 bp); and (i) Ang2, 5′-ACTGACTGATGTGGAAGC-3′ and 5′-CTCTCAGTGCCTTGGAGTTAA-3′ (amplified fragment of 1131 bp).

Ten μl of each PCR product were electrophoresed in a 1.5% agarose gel in Tris/boric-acid/EDTA buffer, and then specific bands were analyzed using a program for densitometry kindly provided by P. L. Lollini (University of Bologna, Bologna, Italy).

Tumor cells grown on slides during cocultures were fixed in acetone for 10 min. They were then washed with PBS and incubated for 1 h at room temperature with anti-MIG (R&D Systems Inc., Minneapolis, MI); anti-GM-CSF (Genzyme, Milan, Italy); anti-iNOS (Tranduction Laboratories, Lexington, KY); anti-VEGF, anti-basic FGF, and anti-TGF-β1 (Santa Cruz Biotechnology, Santa Cruz, CA); anti-MIP-2 (Serotec Ltd., Oxford, United Kingdom); and anti-IP-10 (Peprotech Ec Ltd., London, United Kingdom). Hydrogen peroxide, normal goat blocking serum, biotinylated immunoglobulins, avidin-biotin complex, and fuchsin (Fuchsin Substrate Chromogen; Dako Spa, Milan, Italy) solutions were used according to the manufacturer’s instructions (ABC ELITE detection system; Vector Laboratories, Burlingame, CA). Cells were lightly counterstained with Mayer’s hematoxylin and mounted with Crystal/Mount (Biomeda, Foster City, CA). The positivity of the reactions was assessed independently in a blind fashion by two pathologists on three samples from each experiment, and 100 consecutive cells were evaluated in three or more fields. The expression of cytokines was defined as absent (−) or scarcely (±), moderately (+), frequently (++), or strongly (+++) present on samples tested with the corresponding antibodies.

To evaluate the effect of IL-12 on the activation of T-LYs, Spcs from normal BALB/c and BALB/c-GKO mice were stimulated with anti-CD3 and anti-CD28 mAbs for 18 h in a complete culture medium supplemented or not supplemented with 10 ng/ml mouse recombinant IL-12. Whereas no significant variations in the cell surface differentiation and activation markers were found at the end of stimulation with or without IL-12, both normal and GKO LYs recovered from cultures with IL-12 displayed a 30% lower [³H]thymidine uptake (Table 1). Moreover, normal IL-12-activated LYs produced 5-fold more IFN-γ, 20% less IL-2, and equal amounts of IL-4. LYs from GKO mice activated with or without IL-12 did not produce IFN-γ, whereas they released 3-fold more GM-CSF than normal LYs, as recently observed in vivo (19). The amounts of IL-10 and TNF-α produced by all these LYs were below the detection threshold (data not shown).

The ability of activated LYs cocultured in transwell inserts to modulate the gene expression of two distinct BALB/c mammary carcinoma cell lines [TSA (Ref. 11) and TUBO (Ref. 20)] growing at the bottom of wells was then evaluated. Cocultures with LYs activated in medium only were established in complete culture medium without IL-12, whereas those with LYs activated in the presence of IL-12 were established in medium supplemented with 5 ng/ml IL-12. After 96 h, modulation of gene expression was first assessed with the commercially available Atlas Mouse cDNA Expression Array from Clontech. With 2-fold expression as the cutoff threshold, TUBO cells cocultured with activated LYs from both normal and GKO mice up-modulated expression of the gene coding for the LPSR (CD14; LPSR, SwissProt accession number P10810; Ref. 21; Fig. 1, TUBO CELLS). By contrast, TUBO cells cocultured with normal IL-12-activated LYs up-modulated expression of LPSR and STAT1 (SwissProt accession number U06924), IRF-1 (SwissProt accession number M21065), and IP-1 (SwissProt accession number U19119). Expression of these three genes is regulated by IFN-γ secreted by IL-12-activated lymphoid cells (22). IFNγ absence when TUBO cells were cocultured with IL-12-activated GKO LYs further points to the central role of downstream IFN-γ in their up-modulation (Fig. 1, TUBO CELLS). By contrast, up-regulation of LPSR gene expression appears to be independent of both the downstream secretion of IFN-γ and the presence of IL-12. When the ability of activated LYs and IL-12-activated LYs to modulate gene expression in TSA cells was evaluated, LPSR gene up-modulation was no longer evident (Fig. 1, TSA CELLS). In the presence of IL-12-activated LYs, IP-1 and IRF-1 gene expression was up-regulated as seen in TUBO cells, although to a lesser extent. In this case as well, no up-regulated IRF-1 and IP-1 gene expression was found when TSA cells were cocultured with GKO IL-12-activated LYs (Fig. 1, TSA CELLS). The expression of transcription termination factor 1 (SwissProt accession number Q62187) was up-regulated when TSA cells were cocultured with both normal and GKO IL-12-activated LYs (Fig. 1, TSA CELLS). Interestingly, the absence of IL-12 and the absence of IFN-γ led activated LYs from GKO mice to down-regulate expression of the laminin receptor 1 gene (LAMR1, SwissProt accession number P14206), which is differently expressed during tumor invasion and metastasis (23), and the glucose-6-phosphate isomerase gene (GPI, SwissProt accession number P06745), which promotes the survival in culture of spinal neurons (24). No modulation of STAT1 gene expression was detected.

These data suggest that the factors released by activated and IL-12-activated LYs change the gene expression pattern in tumor cells. To overcome the rigidity imposed by the fixed cutoff chosen for evaluation of the macroarray data, the modulation of a few genes regulating key features of the tumor-host immune relationship was probed by semiquantitative RT-PCR, and the results were expressed as densitometric values (Table 2, A and B). A marked overexpression of the LMP2 and LMP7 genes and an increased expression of PA28 were found in TUBO cells cocultured with activated LYs from normal mice. The expression of these three immunoproteasome genes was further up-modulated when TUBO cells were cocultured with normal IL-12-activated LYs, whereas no up-modulation was found when they were cocultured with activated LYs and IL-12-activated LYs from GKO mice. A similar inhibition of LMP2, LMP7, and PA28 was observed when anti-IFN-γ mAb An18 was added at the beginning of the cocultures (data not shown). Whereas immunoproteasome genes are naturally expressed by TSA cells, their further up-regulation was found in the presence of normal IL-12-activated LYs but not in the presence of those from GKO mice. In both TUBO and TSA cells, the up-regulation of immunoproteasome genes increases with the increasing amount of IFN-γ in the culture medium and the membrane expression of MHC class I glycoproteins on tumor cells (data not shown). Expression of IP-10 and MIG, another two IFN-γ-inducible genes (25), is much more markedly up-modulated when TUBO (Table 2A) and TSA cells (Table 2B) are cocultured in the presence of IL-12-activated LYs than with activated LYs from normal mice. Surprisingly, the up-modulated expression of IP-10 in these tumor cell lines cocultured with activated LYs and IL-12-activated LYs from GKO mice was not inhibited, nor was it inhibited by the addition of An18 mAb (data not shown). These data suggest that factors other than IFN-γ released by activated lymphoid cells regulate IP-10 gene expression. A similar IFN-γ-independent regulation takes place with the iNOS gene. It was up-regulated after tumor cell coculture with activated LYs and further up-regulated by IL-12-activated LYs from normal mice. However, its up-regulation was also evident when TUBO cells were cocultured with activated LYs and IL-12-activated LYs from GKO mice (Table 2A). Expression of MCP1 is markedly up-modulated in TUBO and TSA cells cocultured with normal IL-12-activated LYs and not in those cocultured with GKO IL-12-activated LYs. It is also up-modulated in TSA cells cocultured with activated LYs (Table 2B). Expression of VEGF, a factor of crucial importance for tumor angiogenesis (26), is markedly down-modulated in tumor cell lines cocultured with IL-12-activated LYs and, to a lesser extent, with activated LYs. Lastly, in TUBO cells cocultured with IL-12-activated LYs only, expression of Ang2 (26, 27) became evident. No down-modulation of VEGF gene expression or up-modulation of Ang2 gene expression was found when tumor cells were cocultured with activated LYs and IL-12-activated LYs from GKO mice (Table 2, A and B). In both cases, Ang1 expression remained undetectable (data not shown).

In a blind fashion, two trained pathologists also evaluated whether the up-modulated expression of a few genes correlated with the overexpression of the proteins they encode, as assessed by immunocytochemistry. On culturing TUBO and TSA cells on microscope slides placed at the bottom of the transwells in cocultures with activated LYs and IL-12-activated LYs, it was found that the up-modulation of MIG and IP-10 gene expression detected by semiquantitative RT-PCR fit in well with the intensity of protein expression (Fig. 2; Table 2, A and B). Immunocytochemistry data also endorse the indication of an IFN-γ-independent up-regulation of IP-10 but not of MIG expression when tumor cells are cocultured with activated LYs and IL-12-activated LYs from GKO mice. A similar activated and IL-12-activated LY-dependent but IFN-γ-independent up-regulation is also suggested for MIP-2, TGF, and FGF protein expression. In the absence of IFN-γ, MIP-2 is much more strongly expressed by TSA cells (Table 2B). Lastly, VEGF is poorly expressed in tumor cells cultured with IL-12-activated LYs.

The data reported here show that these two distinct lines of mammary carcinoma cells change gene and protein expression when cocultured in the presence of activated T-LYs. This finding suggests a new way in which the immune system affects the growth of a tumor so that it becomes a party to its own inhibition.

The amount of IFN-γ released by T-LYs appears to be of major importance in the gene and protein modulation in tumor cells. This finding fits in well with the strong influence that IFN-γ has on the expression of a great number of genes. Indeed, the greater ability of IL-12-activated LYs to modulate the genetic program of tumor cells appears to rest on IL-12’s ability to induce the prolonged release of large amounts of IFN-γ in normal T-LYs. However, the present data lead to the rather obvious conclusion that induction of a high and prolonged IFN-γ release results in marked gene modulation in tumor cells and disclose a rather complex pattern of gene modulation by T-LYs. Parallel cocultures of tumor cells with activated and IL-12-activated LYs from normal and GKO mice revealed distinct categories of genes. The first group embraces genes (STAT1, IRF-1, LMP2, LMP7, MIG, MCP1, and Ang2) that are mostly up-modulated by high amounts of IL-12-induced IFN-γ. These genes are up-modulated in tumor cells from cocultures with normal activated LYs, more so in cells from cocultures with normal IL-12-activated LYs and not at all in cocultures with both activated and IL-12-activated LYs from GKO mice and with normal IL-12-activated LYs in the presence of anti-IFN-γ An18 mAb. Other genes (PA28, IP-10, iNOS, and MIP-2) are overexpressed in the presence of IFN-γ even if their expression is still enhanced in cocultures with activated and IL-12-activated LYs from GKO mice and in those in the presence of anti-IFN-γ An18 mAb. This suggests that IFN-γ is not the only factor released by activated LYs that is implicated in the up-regulation of their expression (19). IP-10 can also be modulated by IFN-α and IFN-β (28), which are not neutralized by An18 mAb. The third group embraces genes (LPSR, transcription termination factor 1, TGF, and FGF) whose expression is promoted by factors released by activated LYs apart from the presence or absence of IFN-γ. In contrast, in the absence of factors released by normal activated LYs, the expression of the LAMR1 and GPI genes is inhibited. Distinct factors, such as those released by GKO IL-12-activated LYs, may restore their expression to normal levels.

Because immunocytochemistry detects only relatively high protein levels, it is conceivable that any modulation in protein expression detected may have a functional consequence. The modulation of a few genes and/or the overexpression of their protein product take place in both TUBO and TSA carcinoma cells (IP-10, MIG, iNOS, MCP1, MIP-2, VEGF, and FGF). LPSR and STAT1 genes are modulated in TUBO cells, but not in TSA cells. Modulation of LAMR1 and GP1 occurs the other way around, pointing to the idiosyncratic features of each tumor cell line. The activation of LMP1, LMP7, and PA28 in TUBO cells cocultured with IL-12-activated LYs implies their critical passage from normal to immunoproteasome (29). This passage is not evident in TSA cells because they constitutively express the immunoproteasome.

The gene modulations described seem to be a consequence of the factors released by activated T-LYs. The repertoire, the amount, and the persistence with which these factors are released are influenced by the presence or absence of cytokines during LY activation. The sole presence of IL-12 never affected gene and protein expression in TUBO and TSA cells, nor was gene modulation found in tumor cells recovered after 4 and 96 h of culture in medium only (data not shown). In contrast, it markedly increased the release of IFN-γ in normal activated LYs and that of GM-CSF in GKO activated LYs. The presence of the whole Spc population during T-LY activation allows these IL-12-induced cytokines to promote the release of several additional downstream factors that are not necessarily directly released by T-LYs. In effect, the reaction activated in vivo by IL-12 involves many types of cells and factors, in both the presence (11, 14) and the absence (19) of secondary IFN-γ.

Among the events activated by IL-12, the modulation of gene expression by tumor cells may play a significant role (14). Whereas the purpose of this study was to point out this new way in which lymphoid cells may interfere with tumor growth, the pattern of down-modulation of VEGF observed in TUBO and TSA cells and the up-modulation of Ang2, MIG, IP-10 and iNOS fit in well with the inhibition of neoangiogenesis and the damage of neoformed vessels that characterize the antitumor reaction activated by IL-12 leading to ischemic necrotic tumor rejection (11, 12). High Ang2 expression in tumor vessels causes their destabilization and regression, perhaps as part of the host reaction (27). On the other hand, tumor-derived VEGF represses these Ang2 regression signals (30). In our cocultures, induction of Ang2 expression coincided with VEGF down-modulation, thus resulting in a strong antiangiogenic setting. Up-modulation of MIG1, IP-10, and iNOS stresses this antiangiogenic scenario even further.

It is not surprising that the number of genes modulated by IL-12-activated LYs is low because it has been recently shown that independent samples taken from the same tumor after surgery, chemotherapy, and metastasis retain the same gene expression pattern (31). Whereas the consequences of other gene modulations in shaping tumor growth and inhibition and tumor interaction with immune cells are still speculative, a few changes in the tumor genetic program induced by IL-12-activated LYs may be truly important because they enroll tumor cells themselves in IL-12-activated tumor inhibition. IL-12-induced reaction very effectively inhibits TSA cell growth (11) as well as that of many other tumors (10). It is less effective on TUBO cells (data not shown), whereas it marginally affects the growth of a few other tumors (10, 11). These outcomes may be influenced by differences in the pattern of tumor gene modulation that may well be induced by IL-12-activated LYs on individual tumors.

We thank P-L. Lollini and F. Bussolino for helpful discussion and John Iliffe for critical comments.