Immune Events Associated with the Cure of Established Tumors and Spontaneous Metastases by Local and Systemic Interleukin 12¹

The antitumor activity of IL-12³ is documented by a large set of data from mouse models in which murine rIL-12 was administered both locally around the tumor and systemically by i.p. administration (1, 2, 3, 4, 5, 6, 7, 8, 9, 10). In other studies, tumor cells engineered to release IL-12 have also been used (10, 11, 12, 13, 14, 15, 16). Nevertheless, the mechanisms by which IL-12 impairs tumor growth are still not fully defined. There are contrasting data about the host lymphoid cell populations mainly involved in tumor rejection and how IL-12 triggers their activities (1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20). What critically influences the antitumor efficacy of IL-12 in each model is still poorly determined.

To gain insight into a few of these issues, we compared the effects of local and systemic rIL-12 administration in mice harboring an aggressive, metastasizing, and moderately differentiated mammary adenocarcinoma (TSA; Ref. 21). It was found that rIL-12 therapeutic activity ranged from very high to moderate according to the administration route alone. The immune reactions elicited via the two routes seem to be the same. The variation in the efficacy of rIL-12 seems to be determined by the differences in action kinetics and intensity.

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 mouse (21). TSA cells express MHC class I but not MHC class II glycoproteins (21). They secrete granulocyte colony-stimulating factors and granulocyte macrophage colony-stimulating factors, like most human and mouse mammary adenocarcinomas (22), TGF-β1 (21), vascular endothelial growth factor, and basic fibroblast growth factor (23); they do not stimulate a syngeneic antitumor response in vivo or in vitro (21, 24). F1F is a transformed BALB/c mouse fibroblast cell line that does not immunologically cross-react with TSA-pc (21). Confluent monolayers of tumor cells treated with a 0.25% solution of trypsin (Sigma, St. Louis, MO) in HBSS were used. In a few experiments, subconfluent monolayers of TSA cells were cultured for 24 h with 100 units/ml mouse rIFN-γ (PharMingen, San Diego, CA) and then dissolved in Ultraspect (Biotecx Laboratories, Inc., Houston, TX) for total mRNA extraction.

Female 7-week-old BALB/c mice (Charles River Laboratories) were treated in accordance with the European Community guidelines. When required, mice were selectively immunodepressed by in vivo treatment with antibodies specific for leukocyte subpopulations, as reported previously in detail (10).

Mice received two courses of five daily injections (with a 2-day interval) of mouse rIL-12 (Dr. Michael Brunda, Hoffmann-La Roche, Nutley, NJ) diluted in HBSS supplemented with 100 μg/ml MSA (Sigma). rIL-12 (100 ng/day) was injected either locally s.c. in the tumor area or systemically i.p. (10). Control mice received HBSS with MSA only (MSA controls).

At day 0, mice were challenged s.c. in the left flank with 0.2 ml of a single cell suspension containing 4 × 10⁴ TSA cells. This is about the minimal 100% tumor-inducing dose of TSA cells in BALB/c mice (21). The cages were then coded, and tumor incidence and growth were evaluated as reported previously (10).

A few mice bearing s.c. TSA tumor masses with a mean diameter of 8–10 mm were anesthetized with ketamine hydrochloride (Ketavet; Farmaceutici Gellini, Aprilia, Italy), and the tumor was removed through a 6-mm-long skin cut and careful dissection. The few mice that eventually displayed local tumor relapse were excluded from the experiment. Spontaneous metastases grew in the lungs of mice that were successfully treated surgically. These mice died about 50 days after surgery from respiratory failure due to outgrowth of the metastases.

This was evaluated after three and eight injections of rIL-12, i.e., 10 and 17 days after tumor challenge. The CTL activity of SPCs that were restimulated for 6 days with mitomycin C (Sigma) in the presence of 20 units/ml rIL-2 (Eurocetus, Milan, Italy) against [³H]thymidine-labeled target cells was determined as described previously (21), and the values were expressed as the percentage of lysis and lytic unit/1 × 10⁷ effector cells calculated according to the equation of Pross (25).

Supernatants from SPCs that were restimulated in vitro with soluble anti-CD3 and soluble anti-CD28 (2 μg/ml each; PharMingen) were collected after an 18-h culture and tested for IL-4 and IFN-γ titer using ELISA kits (PharMingen).

For histological evaluation, tissue samples from groups of five mice killed 10 or 17 days after the TSA challenge were fixed, embedded in paraffin, sectioned at 4 μm, and stained with H&E or Giemsa. For immunohistochemistry, acetone-fixed cryostat sections were incubated for 30 min with the following antibodies: anti-MAC-1, anti-MAC-3, anti-I-A/I-E, and anti-IL-6 (PharMingen); anti-PMNs [RB6-8C5 hybridoma; provided by Dr. R. L. Coffman (DNAX, Inc., Palo Alto, CA)]; anti-L3T4 (CD4) and anti-Lyt-2 (CD8; Sera-Lab, Sussex, United Kingdom); anti-Asialo GM1 (Wako Chemicals Gm&H, Dusseldorf, Germany); anti-IL-1β (Genzyme, Cambridge, MA); anti-TNF-α (Immuno Kontact, Frankfurt, Germany); anti-IFN-γ [provided by Dr. S. Landolfo (University of Turin, Turin, Italy)]; anti-CD31 (MEC-13.3) and anti-ELAM-1 (E-selectin; both provided by Dr. A. Vecchi (Istituto M. Negri, Milan, Italy)]; anti-ICAM-1 (CD54) and anti-VCAM-1 (PharMingen); and anti-iNOS (Transduction Laboratories, Lexington, KY). After washing, they were overlaid with biotinylated goat antirat, antihamster, and antirabbit and horse and goat immunoglobulin (Vector Laboratories, Burlingame, CA) for 30 min. Unbound immunoglobulin was removed by washing, and the slides were incubated with avidin-biotin complex/alkaline phosphatase (DAKO, Glostrup, Denmark). Quantitative studies of the immunohistochemically stained sections were performed by three pathologists in a blind fashion on three or more samples from distinct mice by evaluating 10 randomly chosen fields in each sample. Individual cells were counted under a ×400 microscopic field (×40 objective and ×10 ocular lens; 0.180 mm²/field). The expression of cytokines and adhesion molecules was defined as absent (−) or scarcely (±), moderately (+), or frequently (++) present on the cryostat sections tested with the corresponding antibody.

Total RNA was prepared from fresh tumor masses from three mice bearing 10- or 17-day-old tumors that were treated or not treated with rIL-12. Total RNA was extracted using Ultraspect (Biotecx Laboratories, Inc.). RNA (2 μg) was reverse transcribed by using Moloney murine leukemia virus reverse transcriptase (200 units) in 50 ml of reaction mixture with oligodeoxythymidylic acid and deoxynucleotide triphosphate (all from Life Technologies, Inc., Paisley, United Kingdom). Each cDNA (5 ml) was diluted in a 50-ml reaction mixture (GeneAmp Kit; Perkin-Elmer Corp., Norwalk, CT) and amplified by 30 PCR cycles to identify the presence of murine glyceraldehyde-3-phosphate dehydrogenase, IL-1β, IL-4, IL-6, IL-10, IFN-γ, TGF-β1, TNF-α, and iNOS gene sequences by using specific primer pairs (Clontech, Palo Alto, CA). PCR primers were used for murine IFN IP-10 (5′-GCGTTAACCTCCCCATCAGCACCATGAAC and 3′-CCGCTCGAGGTGGCTTCTCTCCAGTTAAGGA), MIG (5′-TCCGCTGTTCTTTTCCTTTTGG and 3′-TTGAACGACGACGACTTTGGGG), ICAM-1 (5′-GACTCTGTGTCAGCCACTGCCTT and 3′-CTCCTCCTGAGCCTTCTGTAACT), VCAM (5′-GCTGCGAGTCACCCATTGTTCT and 3′-CCAGATGGTCGAGGACT CTAA), and ELAM (5′-TACTACAATGCCTCCAGTGAG and 3′-ACATCTCTCGTCATTCCACAT). The bands were defined as almost absent (±), faint (+), intense (++), and very marked (+++).

Three pools of sera from groups of five mice each were collected 10 and 17 days after tumor challenge. Normal sera were three pools from five age-matched, untreated female mice. The total amount of IgM, IgA, IgG1, IgG2a, IgG2b, and IgG3 was evaluated by radial immunodiffusion using the mouse immunoglobulin NL RID kits (The Binding Site, Ltd., Birmingham, United Kingdom). The specific TSA binding potential of the sera was evaluated by flow cytometry after indirect immunofluorescence, as described previously in detail (21).

All in vivo experiments were performed two or three times with groups of five to eight mice, and the results were cumulated as they gave homogeneous results. The significance of differences in tumor takes was determined by Pearson’s χ² test.

As we have described previously, at day 7 after challenge mice display a solid mass of about 2.5 mm in mean diameter. It consists of well-vascularized and invasive tumor tissue with many mitotic figures and only a marginal peripheral reactive cell infiltrate (25). Starting on day 7, these mice received MSA only or rIL-12 for 2 weeks. All MSA controls developed tumors. Local rIL-12 and i.p. rIL-12 cured 20 and 80% of mice, respectively (Fig. 1).

CD4⁺ lymphocytes and NK cells do not seem to be involved in this local cure (Fig. 2). CD4⁺ cells were certainly not involved in the rejection provoked by i.p. rIL-12. The major roles were played by CD8⁺ cells and PMNs and, to a lesser extent, by NK cells.

Histological examination of s.c. tumors after three rIL-12 i.p. administrations (day 10 after challenge) showed that the tumor was partly destroyed by ischemic/coagulative and hemorrhagic necrosis (Fig. 3). Vessel alteration and leukocyte infiltration were evident. At day 17, after eight i.p. rIL-12 administrations, the necrotic areas were large and numerous, and the tumor mass was often completely destroyed. Intratumoral and peritumoral vessels were severely damaged and frequently obstructed by thrombi. These features were much less pronounced after three or eight local administrations.

At day 10, the number of infiltrating macrophages, PMNs, CD8⁺, and CD4⁺ lymphocytes and NK cells was significantly higher in mice that received rIL-12 i.p. as compared to MSA controls (Table 1). In mice that received local rIL-12, only the number of CD8⁺ lymphocytes and NK cells was significantly higher. At day 17, tumor-infiltrating CD8⁺ and CD4⁺ lymphocytes were significantly more numerous after both treatments than they were in the controls. However, whereas in mice receiving rIL-12 i.p., the number of macrophages was also significantly higher, in those receiving local rIL-12, the number of NK cells was significantly higher. These increments in infiltrating leukocytes were paralleled by an increased expression of VCAM-1 by tumor vessels. The expression of IL-1β, TNF-α, IFN-γ, and iNOS was more marked in both i.p. and s.c. treated mice than in MSA controls. TNF-α and IFN-γ were more strongly expressed in the i.p. treated group.

At day 10, a marked increase in expression of mRNA for IP-10, MIG, iNOS, IFN-γ, TNF-α, IL-1β, ICAM-1, VCAM-1, ELAM-1 (Table 1), and IL-4 (data not shown) was evident in the tumor growth area of mice that had received three local or i.p. administrations of rIL-12 compared to the controls. As suggested by the histological data, most of the tumor is necrotic by day 17. At this time, the mRNA for IL-6 and IL-10 becomes evident (data not shown), whereas that for IFN-γ, IP-10, MIG (Table 1), and TGF-β1 (data not shown) decreases.

The rIL-12-provoked cytotoxicity against TSA cells was evaluated in SPCs that were restimulated in vitro with TSA cells in the presence of 20 units of IL-2 and assessed for CTL activity. SPCs obtained at day 10 after three local or i.p. rIL-12 administrations displayed a significantly higher TSA-specific CTL activity than MSA controls (Table 2). Moreover, the CTL activity of mice receiving rIL-12 i.p. was significantly higher than it was in those treated locally. No increase was found in normal mice receiving rIL-12 either locally or i.p (Table 2).

After eight rIL-12 administrations (day 17 after challenge), SPCs from TSA-bearing mice that received rIL-12 either locally or i.p. displayed a similarly high CTL activity (Table 2). A slight increase was also evident in normal mice that received rIL-12 either locally or i.p. No increase was found against antigenically unrelated F1F target cells at any time (data not shown).

SPCs from five mice were pooled and restimulated in vitro for 18 h with soluble anti-CD3 and anti-CD28 monoclonal antibody to evaluate their ability to produce IFN-γ and IL-4. IFN-γ production decreased from about 15 ng/ml before tumor challenge to 11.6 (Fig. 4, top) and 8.3 (Fig. 4, bottom) ng/ml 10 and 17 days after tumor challenge, respectively. Both s.c. and i.p. rIL-12 administration to tumor-bearing mice maintained IFN-γ production at normal levels. In normal mice, rIL-12 administration enhanced IFN-γ production.

Three s.c and i.p. administrations caused a decrease in IL-4 production that was more pronounced in normal mice than in tumor-bearing mice. No modulation of IL-4 release was evident in normal mice that received eight s.c. administrations of rIL-12 (89 pg/ml) compared to the 92 pg/ml released by normal mice. In contrast, eight i.p. administrations enhanced the IL-4 production to 157 pg/ml. Mice bearing 17-day-old tumors displayed a marked decrease in IL-4 production (12 pg/ml), whereas eight rIL-12 administrations enhanced it, especially in those receiving rIL-12 locally (234 pg/ml). These results are representative of three independent tests.

After three rIL-12 administrations (day 10), only tumor-bearing mice receiving rIL-12 i.p. displayed a significant increase in anti-TSA antibodies, which was mostly due to higher levels of IgG1 and IgG3 (Table 3). The amount of IgM and IgA did not change, whereas IgG2a and IgG2b remained undetectable. No significant isotype variation (data not shown) or increase in anti-TSA antibodies (Table 3) was found after local rIL-12. Normal mice treated with local or i.p. rIL-12 or MSA only did not display isotype variations (data not shown) or anti-TSA antibodies (Table 3).

Sera were also collected after eight rIL-12 administrations (day 17). As compared to sera from MSA controls, all those from rIL-12-treated mice displayed a significant increase in the titer of TSA-binding antibody (Table 3), which was associated with a generalized increase in the IgG titer (data not shown). However, the increase was more evident in mice receiving rIL-12 s.c. and was associated with an increase in IgA concentration to 1250 mg/liter from 1100 mg/liter in tumor-bearing mice treated or not treated with rIL-12 i.p. The increase in anti-TSA antibody was also found in normal mice receiving i.p. rIL-12 (Table 3). An increase in antibodies binding the antigenically unrelated F1F cells paralleled the increases of anti-TSA antibodies. However, these increases were lower and were either barely significant or not significant (data not shown).

Multiple spontaneous lung metastases become evident during the growth of s.c. TSA tumors (Fig. 5,B). To evaluate whether the anatomical site of the tumor also influences the therapeutic potential of rIL-12, mice from which s.c. TSA tumors that were 10 mm in mean diameter had been surgically removed (Fig. 5,A) were left to recover from the anesthesia and surgical shock for 2 weeks and then received three i.p. 5-day/week courses of rIL-12 or MSA only. MSA controls died within 40 days after surgery due to respiratory failure associated with the overgrowth of metastases. In contrast, the survival of rIL-12-treated mice was significantly enhanced (Fig. 5,C). At the end of the experiment (day 100 after surgery), 50% were still alive. Whereas PMNs seem to play no role, the cure was abrogated by the removal of CD8⁺ lymphocytes and was significantly weakened by the removal of NK cells (Fig. 2).

These data show that a low rIL-12 dose can cure established solid tumors and spontaneous lung metastases in the mouse, and that its efficacy is highly dependent on the administration route. The markedly higher efficacy of systemic rIL-12 seems to correlate with a quicker recruitment of the effector mechanisms elicited less efficiently by local rIL-12. In contrast with most cytokines, a low dose of rIL-12 at a tumor site does not directly activate the significant local inflammatory reaction on which the induction of a specific immune response depends (24). Rather, the antitumor efficacy of both local and systemic rIL-12 seems to rest on its ability to recruit a systemic reaction. This is activated more quickly and intensely by systemic administration. Leukocyte infiltration of the tumor rejection site, production of mRNA for secondary cytokines and chemokines, spleen cytotoxicity, IFN-γ and IL-4 secretion, and the appearance of serum antibodies, in fact, are all earlier and more effective. In this case, NK cells are also an additional component of the antitumor reaction. This greater speed and intensity is associated with the more effective curing of large s.c. tumors. The same dose of rIL-12 administered i.p. or locally has a different effect. Whereas i.p. rIL-12 rapidly diffuses in the body fluids, reaching low concentrations and promptly triggering an antitumor reaction, the local persistence of high doses of rIL-12 may favor the induction of immunosuppressive effects (17).

In these studies, the therapeutic efficacy of local and systemic rIL-12 was tested against a large, highly invasive, 7-day-old TSA tumor. The tumor mass formed at this time is well vascularized by self-induced neoformed vessels (23). Investigation of the events associated with the partial or total destruction of these relatively large and actively proliferating tumor masses provides an indication of the antitumor mechanisms of rIL-12 (Fig. 6). However, the activation of a given immune activity suggests that it plays a role. It also serves to illustrate the distinctive features of the reaction but clearly does not directly define the weight of its role in tumor inhibition.

Both local and i.p. rIL-12 act mainly by interacting with the IL-12 receptor expressed by activated T lymphocytes and NK cells. The release of IFN-γ and TNF-α is one of the most important events triggered by rIL-12 in these cells (26). These cytokines synergize during monocyte and macrophage activation. Macrophages are usually associated with TSA (24) as well as many mouse and human tumors (27). After their activation by IFN-γ and TNF-α, they become a source of tertiary cytokines and mediators secreted in the tumor environment. rIL-12 administration, in fact, was followed by local expression of the mRNA of IL-1β, IFN-γ, TNF-α, IP-10, MIG, and iNOS in the tumor microenvironment. Immunohistochemically, IL-1β, IFN-γ, TNF-α, and iNOS expression was localized in cells that were morphologically identifiable as macrophages.

Besides their action on immune cells, secondary cytokines seem to affect TSA cells directly. rIL-12-elicited IFN-γ induces overexpression of MHC glycoproteins on the TSA cell surface (28). It also induces TSA cells to generate antiangiogenic activity, as observed with other tumors (7, 20, 29). TSA cells cultured in vitro express a slight amount of IP-10 mRNA but do not express mRNA for MIG. After 48 h of incubation with rIFN-γ, TSA cells displayed both MIG mRNA and an increased expression of IP-10 mRNA (data not shown). This IFN-γ-dependent induction of antiangiogenic activity can explain the necrotic areas observed during the delayed progression of TSA cells engineered to produce IFN-γ (24, 28). This modulation of antiangiogenic factors fits in well with the data showing that their expression by normal and diseased tissues is associated with tissue necrosis and vascular damage (30).

IFN-γ and TNF-α, along with third-level antiangiogenic chemokines and factors, are able to preferentially activate neoformed endothelium and induce the expression of adhesion molecules, as we have observed in peritumoral and intratumoral vessels of rIL-12-treated mice (9, 10). Third-level chemokines (IP-10 and MIG; Ref. 6 and 31) and iNOS (32, 33) as well as secondary IFN-γ and TNF-α (34) may contribute to the inhibition of neoangiogenesis and the damaging of neoformed vessels. Because they are expressed at high levels in tumors from rIL-12-treated mice, all of these factors may be directly responsible for the ischemic necrotic process that characterizes the rejection of established TSA tumors. Indeed, in many tumor models, the inhibition of tumor angiogenesis is an important part of the systemic antitumor effect produced by rIL-12 (9, 10, 35).

Endothelial adhesion molecules, and VCAM-1 in particular, are directly involved in tumor leukocyte recruitment (36, 37). Massive recruitment of CD8⁺ lymphocytes was clearly shown by morphological examination of tumors from mice treated three or eight times with rIL-12 and may depend on IP-10 and MIG selective chemotaxis for CD8⁺ cells (38). Their key role in the antitumor reaction elicited by rIL-12 is endorsed by the results obtained in many other systems (1, 10, 16, 17) as well as by the present findings in selectively immunodepleted mice. The removal of CD8 lymphocytes almost completely abolished the ability of i.p. rIL-12 to cure s.c. tumors and lung metastases. The less important role of CD8⁺ lymphocytes when rIL-12 is injected s.c. may be no more than apparent due to the poor therapeutic efficacy of s.c. rIL-12. Furthermore, systemic rIL-12 results in a more rapid induction of specific anti-TSA CTLs in the spleen of tumor-bearing mice. At day 17, there seems to be no difference in the CTL response induced by systemic or local IL-12. At this late time, the differences in the intensity of CTL activity are no longer evident.

PMNs seem to have no role in the inhibition of lung metastases, whereas they are of crucial importance against s.c. tumors, because their selective depletion abolishes the effect of both s.c. and i.p. rIL-12. Differences in the accessibility of the tumor microenvironment, blood supply, and resident lymphoid cells, especially macrophages, between the lung and the s.c. tissue may account for their distinct weight (39). The number of tumor-infiltrating PMNs was significantly enhanced after three i.p. administrations of rIL-12. This rapid influx of PMNs endowed with a high destructive potential is probably involved in the vascular damage (40) and subsequent extensive ischemic/hemorrhagic necrosis observed after three and eight i.p. administrations. Their early recruitment may be supported by the marked expression of ELAM-1 by the tumor vessels, because this was almost absent in locally treated mice and the controls.

As has been shown with many tumors (41), SPCs released fewer cytokines as the TSA tumor mass increased. Both s.c. and i.p. rIL-12 restored the IFN-γ production of SPCs from tumor-bearing mice to normal levels and enhanced that of normal mice, as observed by Nastala et al. (2) and Fujiwara et al. (19, 20). Three s.c. or i.p. administrations inhibited the production of IL-4, a typical class II lymphokine, in both normal and (to a lesser extent) tumor-bearing mice, whereas eight administrations increased it.

The role of antitumor antibodies is still unclear (7, 24, 42). Even so, their production is stimulated by both local and systemic rIL-12. Anti-TSA antibodies appeared after only three systemic administrations, and their level was unchanged after eight administrations, whereas their production was not induced by three local administrations and only became evident and very marked after eight administrations. The kinetics of this production was correlated with the production of IL-4 by SPCs.

In conclusion, our data indicate that rIL-12 is effective against a large and highly vascularized tumor. Its systemic administration is more powerful than local administration. Tumor destruction is dependent on a prompt antitumor response that must result from the cooperation of several subsets of reactive cells. Three events seem to be of major importance: (a) the destruction of tumor vessels by PMNs; (b) the indirect inhibition of angiogenesis by secondary IFN-γ and TNF-α and third-level chemokines (IP-10 and MIG); and (c) the activation of leukocyte subsets capable of producing proinflammatory cytokines, CTLs, and antitumor antibodies.

After eight rIL-12 administrations (day 17 after tumor challenge), when tumor rejection comes to an end, the local immune response probably starts to wane, as suggested by the appearance of mRNA for IL-6 and IL-10 and a decrease in the production of that for IFN-γ, IP-10, MIG, and vascular endothelial growth factor β (Table 2).

We are grateful to Drs. R. L. Coffman, M. Brunda, S. Landolfo, and A. Vecchi for providing reagents. We thank Dr. John Iliffe for critical review of the manuscript.