Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

Malignant mesothelioma (MM) is an aggressive tumor of the mesothelium that arises in the serosal cavities, usually in individuals who have been exposed to asbestos. The latency period of this disease is between 15 and 40 yr, and the frequency of MM is expected to continue to increase in the near future (1, 2), with the total number of new cases expected to peak by the year 2010 (3). Conventional therapies such as surgical excision, radiotherapy, and chemotherapy have not altered the prognosis of this malignancy, the mean survival time from diagnosis being 9 mo (4). However, there are experimental (5, 6) and clinical (7) data supporting the notion that MM is susceptible to immunologic intervention.

To study the biology of and explore potential treatments for MM, we have developed an animal model of this disease (8). This represents one of the few models in which the cancer-inducing agent asbestos is exactly the same in the mouse and human. The murine model of MM mimics the human disease very closely in terms of histopathology, cell biology, and immunobiologic features such as major histocompatibility complex (MHC) expression (8, 9).

Interleukin-12 (IL-12) has been shown to have potent immunoregulatory properties. These include augmentation of the proliferation of activated natural killer cells (NK) and T cells, enhancement of NK cell activity and lysis by cytotoxic T-lymphocytes (CTL) (10), and promotion of the development of T-helper type 1 (Th1) cells from naive T cells (13). IL-12 is also particularly potent at inducing the production of interferon- (IFN-), and to a lesser extent of tumor necrosis factor- (TNF-), by NK and T cells (14, 15).

Additionally, IL-12 has been shown to induce an antitumor immune response against several murine tumors (16- 22), and clinical trials of this effect have commenced (23). The potential role of IL-12 as an immunotherapeutic agent has not been evaluated in MM. Because MM is refractory to conventional treatments, the present study was designed to assess the potential of IL-12 to induce an immune response against a nonimmunogenic, aggressive murine MM tumor. The data show that systemic administration of rIL-12 prevents tumor growth, and that this inhibitory effect requires both CD4⁺ and CD8⁺ but not NK cells. Importantly, intratumor IL-12 given by injection or gene transfer in established tumors induces tumor regression or growth inhibition that is associated with an increase in the number of CD4⁺ and CD8⁺ cells infiltrating the tumor mass. These results suggest that IL-12 may have a role in the immunotherapy of MM.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Mice

BALB/c and BALB/c nu/nu mice were obtained from the Animal Resource Center, Perth, Western Australia, and maintained under standard conditions at the University of Western Australia, Department of Medicine animal facility.

Cell Lines

The establishment of the murine MM cell line AB1 has been described previously (8). Cell lines were maintained in RPMI 1640 medium (GIBCO, Glen Waverly, Australia) supplemented with 5% fetal calf serum (FCS) (Life Technologies, Inc., Melbourne, Australia), 15 mM N-2-hydroxyethylpiperazine-N'-ethanesulfonic acid (Hepes) (Sigma, Castle Hill, Australia), 200 mM L-glutamine (GIBCO, Grand Island, NY), 0.05 mM mercaptoethanol, and 50 µg/ml gentamicin (Glaxo Australia, Boronia, Australia). Additionally, cell lines transfected with the bacterial neomycin phosphotransferase gene were maintained in media containing G418 (400 µg/ml) (Geneticin; GIBCO). All cell cultures were grown at 37°C in a 5% CO₂ humidified atmosphere.

Tumor Challenge

Exponentially growing AB1 cells were harvested by brief trypsinization, washed, and resuspended in phosphate-buffered saline (PBS). BALB/c mice were injected with syngeneic AB1 tumor cells in the left flank. In experiments in which murine recombinant IL-12 (rIL-12) (a generous gift from Dr. Stanley Wolf of the Genetics Institute, Cambridge, MA) was administered, the test groups received various doses of rIL-12 diluted in saline solution containing 1% BALB/c serum, whereas mice in the sham group were injected with saline solution containing 1% BALB/c serum only. The initiation of rIL-12 treatment and the doses given, injection sites, and tumor inoculum varied with the experiment and are specified in the figure legends. Tumor size was measured every 2 to 4 d with calipers, and the resulting data are presented as a mean tumor area, which is the product of the two largest perpendicular diameters.

Depletion Studies

Depleting monoclonal antibodies (mAbs) were derived from the hybridomas YTS 191.1 (anti-CD4) and YTS 169.4 (anti-CD8) (24) by generation of ascites in nude mice. One-in-two dilutions of the ascites containing anti-CD4 and/or anti-CD8 antibody were given to mice on Days 6, 4, and 0, and weekly thereafter. Animals were inoculated with tumor on Day 0. Subset depletion was verified using fluorescence-activated cell sorting (FACS) analysis, and this protocol resulted in less than 2% CD4⁺ or CD8⁺ cells being detectable in the spleens and lymph nodes of treated mice. NK cells were depleted by intraperitoneal injections of 20 µl of antiasialo-ganglioside GM1 (Nova Chem, Melbourne, Australia), reconstituted as advised by the manufacturer and injected every 5 d from Day 5. To confirm that NK cells were depleted, mice were stimulated with polyinosinic:polycytidylic acid (poly I:C) 24 h before preparation of red-blood-cell (RBC)-free splenocytes. The ability of splenocytes to lyse NK-sensitive target YAC cells was determined in a 4-h ⁵¹Cr release assay as previously described (25). Splenocytes from mice that were depleted of NK cells caused less than 2% lysis of YAC cells at an 100:1 effector-to-target-cell ratio, as compared with 30 to 40% lysis caused by splenocytes from undepleted mice.

Immunohistologic Staining

Tissues from various sites were removed, placed in compound-embedding medium (ornithyl carbamyltransferase [OCT]; Miles, Inc., Elkhart, IN), and snap frozen by placing the mold on dry ice. Ten-micrometer-thick sections were cut, collected on poly-L-lysine-coated slides, and allowed to air-dry for 1 h. Slides were stored at 4°C (for a maximum of 2 d) prior to staining. Routine hematoxylin-eosin staining was performed on the sections. Prior to immunostaining, sections were fixed for 5 min with 1% paraformaldehyde and blocked with 10% FCS/1% bovine serum albumin (BSA) for 15 min. Sections were then incubated with the primary immunohistologic Abs or with rat anti-CD4 (GK1.5), rat anti-CD8 (53.6.72) for 1 h. This was followed by secondary Ab (biotinylated rabbit antirat Ig; Jackson Immunoresearch Labs, West Grove, PA) for 30 min, streptavidin/horseradish peroxidase (Dako, Glostrup, Denmark) for 30 min, and diaminobenzidine/H₂O₂ (Sigma) for 5 to 10 min. Slides were washed for 5 min in PBS between each incubation step.

DNA Constructs and Transfection Procedures

Murine complementary DNA (cDNA) encoding the IL-12 subunits p35 and p40 in the bluescript SK⁺ plasmid was kindly provided by Dr. Ueli Gubler (Hoffman LaRoche, Nutley, NJ). The p40 and p35 subunits were subcloned into the expression vectors pHbApr-1-neo and kCMVintPolyli (Vical, San Diego, CA), respectively. The tumor cell line AB1 was cotransfected with both expression vectors, using cationic lipid N-[1-(2,3-dioloyloxy)propyl]-N,N,N-triethylammonium methylsulfate (DOTAP) (Boehringer, Mannheim, Germany) as previously described (25). Briefly, 3 × 10⁵ tumor cells were seeded into six-well plates and allowed to recover for 24 h. DNA (10 µg of p35 and 10 µg of p40) was mixed with 80 µg of DOTAP, and tumor cells were incubated with the DNA/transfection mixture for 24 h in OPTI-MEM medium (GIBCO). Cells were trypsinized, transferred into 80-cm² flasks, and maintained in 5% FCS/ RPMI medium for 48 h before commencing selection with G418 (Geneticin; GIBCO) at 400 µg/ml for 14 to 21 d. Resistant colonies were pooled and cloned by limiting dilution. Control AB1 transfectants (AB1-neo) were transfected with the expression vector pHbApr-1-neo encoding the bacterial neomycin phosphotransferase gene only.

Enzyme-Linked Immunosorbent Assay for IL-12 and IFN-

Production of IL-12 was quantified with a sandwich-type enzyme-linked immunosorbent assay (ELISA). Supernatants from 10⁶ cells were collected after a 48-h incubation period and stored at 20°C until assayed. Briefly, plates were coated with 8 µg/ml of anti-IL-12 Ab (Red-T/G297-289; Pharmingen, La Jolla, CA) at 4°C overnight and blocked with 10% FCS/PBS for 1 h at 37°C, after which murine rIL-12 standards and supernatants were added for an overnight incubation at 4°C. Anti-IL-12 Ab C17.8 (kindly made available by Dr. Giorgio Trinchieri, Wistar Institute, Philadelphia, PA) was biotinylated and used at 4 µg/ml (1 h at room temperature), followed by an incubation with streptavidin-alkaline phosphatase (1:1,000; Dako) for 0.5 h at room temperature. Between each incubation step, plates were washed several times in 0.5% Tween/PBS. After addition of p-nitrophenylphosphatase, plates were developed for 15 to 45 min and absorbance read at 405 nm. The IL-12 transfectant (AB1-[IL-12]-2A) used in these experiments produced approximately 10 ng/10⁶ cells/48 h of IL-12, and failed to produce tumors in normal mice (I. Caminschi and colleagues, manuscript in preparation). The control transfectant AB1-neo formed tumors at the same rate as did parental AB1 (data not shown). The serum levels of IFN- in RBC-free blood samples were measured with a commercially available IFN- ELISA kit (Pharmingen) according to the manufacturer's protocol. Briefly, flexible, 96-well plates with U-shaped wells were incubated overnight at 4°C with capturing antibody (1 µg/ml in 0.1 M NaHCO₃, pH = 8.5; 50 µl/well). Nonspecific binding was blocked by an incubation for 1 h with 10% FCS/PBS (200 µl/well). Triplicate samples of several dilutions of the test samples and IFN- standards were incubated for 2 h to allow binding. The incubation with detecting antibody (100 µl/well; 1 µg/ml) was done for 1 h. Subsequent incubation with streptavidin-alkaline phosphatase and p-nitrophenylphosphatase were conducted as described for the IL-12 ELISA.

Statistical Analysis

Data presented as percentage of tumor-free animals were analyzed with Kaplan-Meier survival curves, using the log rank test to determine statistical significance. Tumor growth was compared through repeated-measures analysis of variance (ANOVA), and significance was determined by multivariant (mixed model) analysis. Differences were considered significant when the value of P was less than 0.05. Statistics were calculated with the SPSS for Windows program (SPSS, Inc., Chicago, IL).

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

Effects of Systemic Administration of rIL-12 on Tumor Growth

To evaluate the effects of rIL-12 on the in vivo growth of the MM cell line AB1 by injecting BALB/c mice subcutaneously with AB1 (10⁶ cells) and commencing rIL-12 treatment at the time of tumor inoculation, we injected 10 mice intraperitoneally with 0.5 µg of rIL-12 daily for 5 d per week over a period of 3 wk. In two separate experiments, in which tumor growth was monitored for 3 mo, systemic rIL-12 prevented the growth of AB1 tumors in four of 10 (data not shown) and seven of 10 mice, respectively (Figure 1a). All mice in the control groups receiving saline injections developed tumors (Figure 1a). Although not all rIL-12-treated animals remained tumor free, those tumors that did develop were delayed in onset (Figure 1b). However, once tumors emerged in the rIL-12-treated mice, their growth rate was comparable to that seen in control animals.

View larger version (18K): [in this window] [in a new window]   Figure 1.   Systemic administration of rIL-12 prevents and delays tumor emergence. BALB/c mice were injected subcutaneously with syngeneic mesothelioma tumor AB1 (106 cells). Starting on Day 0, 10 mice were injected daily intraperitoneally with 0.5 µg of murine recombinant rIL-12 (n = 10) or saline (n = 10), five times per week for 3 wk. (A) Tumor take and latency. (B) In vivo growth rate of AB1 in mice that developed tumor (n = 3). Data are presented as means ± SE, and represent one of two experiments. *Treatment with rIL-12 significantly inhibited tumor development (P < 0.0001).

It is well documented that IL-12 is a potent inducer of IFN-, a cytokine that has been shown to inhibit directly the growth of various murine tumors including AB1. Therefore, we determined whether the dose of rIL-12 that had antitumor effects caused IFN- production. When blood serum was analyzed 24 h after one intraperitoneal injection of 0.5 µg of rIL-12, it was found to have increased levels of IFN- (Table 1).

To determine whether the initial protection conferred by the systemic administration of rIL-12 induced long-term protection against AB1 tumors, mice that had been treated with rIL-12 and that did not develop tumors were rechallenged with AB1 (10⁶ cells). Four of seven mice did not develop tumor when subsequently rechallenged with AB1. In naive recipients, tumor incidence was 100% (Figure 2).

View larger version (19K): [in this window] [in a new window]   Figure 2.   Mice that had previously rejected AB1 tumors while being given rIL-12 were partly protected against rechallenge with AB1. BALB/c mice that had failed to develop tumors when challenged subcutaneously with 106 AB1 cells while receiving rIL-12 (Figure 1) were rechallenged in the opposite flank with 106 AB1 cells 3 mo later. *Mice previously primed with AB1 in the presence of rIL-12 (n = 7) had reduced tumor emergence compared with mice (n = 10) that were challenged with AB1 for the first time (P = 0.005). Data presented are representative of two such experiments.

Having established that systemic administration of rIL-12 prevents or delays the onset of AB1 tumors, we next determined which cells were involved in this rIL-12-induced antitumor effect. Both CD4⁺ and CD8⁺ cells were required for the rIL-12-induced immune response against AB1 tumor, because the depletion of either cell subset resulted in an increased incidence of tumor development (Figure 3a). Neither subset had a predominant effect, because the incidence of tumor growth was not significantly different between groups treated with the different antibody regimens. NK cells did not appear to be involved in the rIL-12-induced antitumor immunity, because depletion of NK cells did not increase the incidence of AB1 tumor in mice given rIL-12 (Figure 3b). Additionally, in the group of mice depleted of NK cells, CD4⁺ cells, and CD8⁺ cells, the rate of tumor emergence was the same as in the group of mice depleted of CD4⁺ and CD8⁺ cells only, indicating that the depletion of NK cells did not increase the rate or incidence of tumor formation.

View larger version (29K): [in this window] [in a new window]   View larger version (30K): [in this window] [in a new window]   Figure 3.   CD4+ and CD8+ cells, but not NK cells, are required in the antitumor immune response induced by rIL-12. BALB/c mice were injected subcutaneously with syngeneic AB1 tumor (106 cells). Starting on Day 0, mice received daily intraperitoneal injections of 0.5 µg of murine rIL-12 or saline three times per week. A total of 11 injections were given, and are indicated by the inserted triangles at the base of the graph. (A) Mice undepleted (n = 5) or depleted of CD4+ cells (n = 5), CD8+ cells (n = 5), and CD4+ and CD8+ cells (n = 5) were inoculated with tumor and given IL-12. The control group (n = 7) was given tumor but no IL-12 or depleting antibodies. (B) Mice undepleted (n = 5) or depleted of CD4+ and CD8+ cells (n = 5), or of CD4+ and CD8+ cells and NK cells (n = 5), were inoculated with tumor and given rIL-12. The control group (n = 7) was given tumor but no IL-12 or depleting antibodies. Data presented are representative of two such experiments. *Tumor development was significantly inhibited only in mice depleted of NK cells during rIL-12 treatment, and only in mice given rIL-12 (P < 0.02).

Effects of Intratumoral Administration of rIL-12 on Tumor Growth

It is quite feasible that prevention of tumor growth at the onset of disease may be more readily achievable than the induction of regression of an already established tumor. To model this situation, we investigated the capacity of rIL-12 to generate an immune response against established AB1 tumor. Twenty-three BALB/c mice were inoculated subcutaneously in the left flank with 10⁶ AB1 cells, and by Day 16 all of the animals had developed palpable tumors. Mice were then subdivided into two groups in which the individual tumor sizes were matched. Mice in the test group were given intratumoral injections of 0.5 µg of rIL-12 (50 µl volume) every second day, three times per week, for 2 wk, whereas mice in the control group were given saline injections. Overall, intratumoral treatment with rIL-12 inhibited tumor growth (Figure 4). Interestingly, individual mice exhibited a range of responses. These could be broadly categorized into three types of responses. Progressive tumor growth (continued tumor growth that was not significantly different from that of control animals) was observed in two of 13 mice. Growth arrest (inhibition of growth during and after cessation of rIL-12 treatment, but with subsequent outgrowth) was seen in six of 13 animals. Tumor regression (tumors diminished in size until no longer palpable) occurred in only two of 13 mice. Of the five mice that had tumor regressions, four had reemerging tumor within 3 mo and the other remained tumor free for 6 mo. The important conclusion from these data is that the effect of rIL-12 on established tumors was generally not maintained after cessation of treatment.

View larger version (17K): [in this window] [in a new window]   Figure 4.   Intratumoral administration of rIL-12 inhibits tumor growth. BALB/c mice were inoculated subcutaneously with 106 AB1 tumor cells. Starting on Day 16, mice were given intratumoral injection of 0.5 µg of rIL-12 (n = 13) or saline (n = 10) three times per week for 2 wk (injection times are indicated by inserted triangles). Data are presented as means ± SE, and represent one of three experiments. *Animals treated with rIL-12 had significantly inhibited tumor growth (P < 0.0001).

Because systemic administration of rIL-12 required both CD4⁺ and CD8⁺ cells to induce protective immunity, we determined whether the CD4⁺/CD8⁺ cellular infiltrate of tumors paralleled the various stages (regressing, arrested growth, and progressing) after intratumoral administration of rIL-12. Animals that exhibited progressive tumor growth had very few infiltrating CD4⁺ and CD8⁺ cells, as did the saline-treated controls (Figures 5C, 5D, 5E, and 5H). In contrast, mice that showed inhibition of tumor growth or tumor regression had increased numbers of both CD4⁺ and CD8⁺ infiltrating cells (Figures 5A and 5B). Additionally, regressing tumors had more CD4⁺ and CD8⁺ infiltrating cells than did growth-arrested tumors. Also, in mice in which tumors had regressed, reemerging tumors had very few CD4⁺ or CD8⁺ infiltrating cells (Figures 5E and 5F).

View larger version (151K): [in this window] [in a new window]   Figure 5.   Intratumoral injections of rIL-12 result in tumor regression, which correlates with massive CD4+ and CD8+ cell infiltration into the remaining tumor mass. Tumor growth of mice treated intratumorally with rIL-12 or saline was monitored every 2 or 3 d. Three to four days after the last injection, tumor growth was categorized as regressive (A and B) if the tumor area continued to shrink during treatment, and as progressive (C and D) if the tumor area continued to expand similarly to the tumor area of control mice. Tumors that regressed completely during and after rIL-12 treatment, but that developed again within 1 mo thereafter, are referred to as reemerging tumors (E and F). G and H represent control mice receiving saline injections. Sections were stained anti-CD4 (A, C, E, G) and anti-CD8 (B, D, F, H) mAbs. Sections depicted are representative of several such samples.

Effect of IL-12 Transfectants on Parental Tumor Growth in Mixing Experiments

Although local intratumoral administration of IL-12 could potentially avoid the toxicity seen with systemic administration of IL-12, our demonstration that continuous exposure to IL-12 was required to induce an effective antitumor response against established tumor indicated that local injections of IL-12, if intermittent, might not be effective. We were therefore prompted to investigate the capacity of IL-12 gene-transfer approaches as a means of generating prolonged, local increases in IL-12 in tumors. Groups of mice were inoculated subcutaneously with a total of 10⁶ tumor cells containing a mixture of IL-12-transfectants and parental tumor cells (Figure 6) at ratios of transfectant to parental cells of 1:4 and 4:1 (Figure 6). Only mice inoculated with the highest number of transfectants (4:1) exhibited significant protection, with 60% of mice remaining tumor-free compared with 10% in control groups. This result closely paralleled that obtained with systemic rIL-12 administration.

View larger version (19K): [in this window] [in a new window]   Figure 6.   Coinjection of AB1-(IL-12) protected against tumor development by parental AB1 cells. BALB/c mice were inoculated subcutaneously with a total of 106 cells of AB1 containing 8 × 105 cells of the IL-12 transfectant, AB1-(IL-12). Mice in the control groups were inoculated only with the corresponding number of parental AB1 tumor cells. All groups consisted of 10 animals, and the data presented are representative of two experiments. *Animals inoculated with a mixture of AB1 and AB1-(IL-12) had significantly inhibited tumor development compared with mice challenged with AB1 only (P = 0.001).

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

The unresponsiveness of MM to conventional therapy, its potential susceptibility to immunomodulatory approaches, and the powerful antitumor effects of IL-12 in other studies prompted us to evaluate IL-12 in our MM model. Our data show that the systemic administration of rIL-12 prevents or delays onset of the MM murine tumor AB1. At least three features demonstrate that this is an immunologically mediated phenomenon. These are: (1) the need for both CD4⁺ and CD8⁺ cells for this effect; (2) the association of arrested tumor growth with increased lymphocytic infiltration; and (3) the generation of immunologic memory. Additionally, this phenomenon is independent of NK cells. The absolute requirement for both CD4⁺ and CD8⁺ T-cell subsets to generate a protective immune response induced by systemic rIL-12 differs from findings in other tumor models, in which either CD8⁺ cells alone were essential (16) or either subset alone was sufficient to confer IL-12-induced antitumor immunity (17, 19).

The reason for the lack of protection in some animals is not clear. Even animals that originally rejected the tumor cells with IL-12 treatment exhibited only a 50% protection rate on secondary challenge. Thus, there is at least one additional factor, apart from the presence of effector (or memory) cells, that can limit the efficacy of the response to IL-12 in vivo. It is possible that during induction of the immune response there is a crucial phase at which effector cells and tumor cells must coexist at a critical ratio in order to eradicate fully all tumor cells. It is also likely that once a tumor is seeded, physical or immunosuppressive factors are generated within the tumor milieu that limit the cytotoxic capacity of the effector cells. Transforming growth factor- (TGF-), a potent immune modulator, is produced by MM cells, including the line used in the present study (26), and could represent such a barrier. It might be possible to overcome this variation in individual response by decreasing the number of tumor cells inoculated or by increasing the dose of IL-12, although larger doses of rIL-12 were given in preliminary experiments but were found to affect the health of the animals (data not shown). IL-12 toxicity has also been reported in human studies (27).

The relative effectiveness of the three therapeutic regimens used in the present study is worth noting. Systemic administration and coinoculation of rIL-12 resulted in similar outcomes: substantial protection against tumor growth. Both methods involve introducing IL-12 before tumors have been established, maximizing immune-system access to the tumor and allowing complete tumor eradication in most animals. By contrast, the established tumor provides a hostile environment toward potential effector cells in terms of the production of matrix molecules and immunosuppressive factors (28), and limits their access to target cells (29). Under such conditions, IL-12 effects diminish after cessation of treatment.

The cells infiltrating tumors regressing in response to IL-12 vary in reported tumor models. As with our results, others have found that tumor regression induced by rIL-12 is accompanied by a large infiltration of CD8⁺ cells (17) or CD8⁺ cells and macrophages (30, 31), or of CD4⁺ and CD8⁺ cells and macrophages (19, 32). Importantly, tumors that failed to regress had very little cellular infiltrate (32), providing circumstantial evidence that the infiltrating cells are mediating a specific antitumor immune response.

The requirement for both CD4⁺- and CD8⁺-cell subsets implies that the antitumor effect of CD8⁺ cells depends on CD4⁺ cell help. It is possible that this dependency is due to the nature of the tumor antigen, the capacity for this antigen to be presented, the repertoire of T cells that can respond to the antigen, or all of these factors. It is generally considered that most tumor antigens are self-antigens (33), the repertoire of which is diminished by the process of negative selection during T-cell development (34). Potentially self-reactive cells that survive this process are thought to be of low avidity (35), for which reason potential tumor-reactive T cells are also likely to be of low avidity. One of the factors that determines the dependence of CD8⁺ cells on help from CD4⁺ cells is the avidity of the CD8⁺ cells for a particular tumor antigen (36). High-avidity CD8⁺ cells produce their own IL-2 and are independent of CD4⁺ cells. Low-avidity CD8⁺ cells will require help in the form of IL-2 (36). Such interdependence could occur within our model.

The data obtained from the experiments in which rIL-12 was given intratumorally indicate that tumors may become targets for effector cells, yet may fail to sustain the initiated immune response without additional stimulation. Thus, tumor cells that were not eradicated during the enhanced immune response in the present study resulted in reemergent tumors. Our observation that tumor cells alone were insufficient to sustain the initiated immune response mirrors findings in models of autoimmune disease (37), in which CD8⁺ cells autoreactive toward the beta cells of the islets of Langerhans require multiple primings to induce autoimmunity, indicating that the beta cells are transient targets of the CD8⁺ cells but cannot sustain the stimulation of primed CD8⁺ cells (37).

The termination of treatment with IL-12 could alter the balance between the development of Th1 and Th2 effectors or the production of suppressive factors by a tumor. One of the most consistent effects of IL-12 is the induction of IFN- production by T cells and NK cells (14, 15). IL-12 has also been shown to enhance NK and cytotoxic T-lymphocyte (CTL) activity (11, 12), and to facilitate the development of a Th1-type immune response (13). In our tumor model, the doses of rIL-12 used also resulted in high levels of IFN- in the serum within 24 h of rIL-12 injection (Table 1). The suggestion that production of IFN- within the tumor milieu may be important is supported by the report by Zou and colleagues (19) that IL-12 restored suppressed IFN- production within the tumor milieu and subsequently resulted in tumor rejection. In our model of MM, tumor-infiltrating lymphocytes (TILs) produce messenger RNA (mRNA) for IFN- in the early stages of tumor development, but this declines as the tumor progresses (38), implying that the restoration of IFN- production within the tumor microenvironment may be an important step in initiating and maintaining effective tumor immunity.

MM tumors exhibit large numbers of infiltrating macrophages, which may play a role in amplifying the immune response. IL-12 treatment, through the induction of IFN-, could affect macrophages in numerous ways that may lead to an enhanced antitumor response. IFN- induces the production of TNF by macrophages (39) and the upregulation of MHC Class I and II antigen expression on antigen-presenting cells, thus potentially improving antigen presentation capacity (40). Additionally, IFN- is one of the components involved in inducing nitric oxide (NO) production by macrophages (41). NO is a multifunctional molecule that has, among other properties, cytostatic/cytotoxic activity against tumor cells (42, 43), and it has recently been shown both that IL-12 increases serum levels of the NO end products nitrite and nitrate (17), and that the antitumor activity of NO may depend on the production of NO (44). IL-12 also restores depressed NO production by macrophages from tumor-bearing mice via indirect mechanisms that we think may at least in part be IFN- dependent (45). Additionally, IFN- induces expression of interferon-inducible protein 10 (IP-10) (46), a chemokine that has been reported to be chemotactic for monocytes and activated T cells (47). Chemokines such as IP-10 may be partly responsible for the increased number of TILs observed in tumors undergoing regression. In addition, IP-10 is also a potent inhibitor of angiogenesis (48), and has been shown to be the mediator of the antiangiogenic effects of IL-12 (49). Therefore, because neovascularization is an important factor determining tumor growth, IL-12-induced expression of IP-10 (30) could additionally inhibit tumor development.

In other systems, it has been reported that IFN- can act directly on tumors to inhibit their growth (50, 51). The AB1 MM cell line used in the present study is not inhibited by IFN- in in vitro growth analysis (Marzo and coworkers, unpublished results). Nor does IL-12 inhibit the growth of AB1 in vitro (data not shown). Consequently, it is unlikely that direct growth inhibition of the tumor cells occurs in vivo. Upregulation of MHC Class I antigen expression induced by IFN- can result in increased tumor-antigen presentation by the tumor, making it a better target for recognition. Certainly AB1 cells respond to IFN- by upregulating Class I antigen expression in vitro (data not shown).

On the basis of our observations, approaches to treating MM with IL-12 may prove useful; even a rapidly growing, nonimmunogenic MM tumor can regress with IL-12 therapy. However, several factors need to be considered when planning clinical trials of IL-12. First, the tumor needs to be continuously exposed to IL-12 for the effect of this cytokine to be seen, suggesting that continuous IL-12 infusion or IL-12 gene transfer will be required. Second, responses may be variable, even in otherwise immunologically sensitive tumor clones, suggesting that additive immunomodulatory agents may need to be used in many patients. Third, when IL-12 is used in gene-therapeutic approaches, a high efficiency of transfer will be required for maximal antitumor efficacy.