Tumor Cells Engineered with IL-12 and IL-15 Genes Induce Protective Antibody Responses in Nude Mice¹

Several lines of evidence indicate that MHC class I-restricted CTLs play an important role as effector cells in the immune response against tumors. In addition, different CTL-defined tumor-associated Ags have been molecularly characterized, thus representing potential tools for tumor immunotherapy (1, 2, 3). However, in a fraction of human tumors the lack or loss of MHC class I Ag expression is a major “tumor escape” mechanism that hampers the development of CTL-based immunotherapy (4, 5, 6, 7, 8). In tumors lacking MHC class I Ags, anti-tumor functions may be exerted by effectors of innate immunity (9) or by Abs (10), which can be regarded as tools for immunotherapy strategies. In addition, several mAbs directed against surface tumor-associated Ags have been shown to display anti-tumor activity in vivo in both preclinical and clinical studies (11).

A suitable approach to trigger different anti-tumor immune responses is based on the gene transfer of immune stimulatory cytokines into tumor cells (12, 13). In most preclinical models the use of cytokine gene-modified cells as a vaccine resulted in the induction of long-lasting protective immunity against the wild-type tumor, related, in most instances, to the development of tumor-specific T cell responses (14, 15), while the induction of Ab responses was less frequently involved (16).

IL-15 is a four α-helix bundle cytokine displaying IL-2 like functions, such as the induction of T and NK cell proliferation (17, 18). IL-15 plays an essential role in the control of NK cell differentiation (19) and is also capable of stimulating NK cell cytolytic functions (20, 21) and chemotaxis (22). Moreover, IL-15 can provide a costimulus to induce B cell proliferation and Ig secretion (23). Therefore, IL-15 has been proposed as a possible candidate for the development of cancer immunotherapy (24) or gene therapy approaches (25, 26). We have previously shown that a human MHC class I-negative tumor cell line (N592), genetically engineered to secrete IL-15, displayed a reduced growth when xenotransplanted in nude mice, as an effect of NK cell recruitment and activation (27). Since the effect of transduced IL-15 was only partial, in this model we combined IL-15 with another NK-stimulating factor(s), such as IL-12.

IL-12 is a heterodimeric cytokine, acting as a potent inducer of Th1 responses and as a stimulator of NK cell proliferation, cytotoxic activity and IFN-γ production (28). In addition, IL-12 has been shown to drive the differentiation of naive B cells into Ab-secreting plasma cells and to induce plasma cells proliferation and Ig class switch to IgG2 through IFN-γ induction (16, 29, 30). Several studies have demonstrated a potent anti-tumor activity of IL-12 either as a recombinant cytokine or in gene transfer approaches in different syngeneic mice models (16, 31, 32). Synergistic anti-tumor effects were reported by the combined use of IL-12 and IL-15 in syngeneic mice models in immunocompetent (33) and IFN-γ-deficient mice (34). In addition, the simultaneous IL-12 and IL-15 gene transfer in the MHC class I-negative N592 tumor cells synergistically triggered potent tumor rejection responses in nude mice (35).

Here we show that most nude mice rejecting N592 cells engineered with IL-12 and IL-15 genes develop resistance to N592 parental cells (N592pc)³ rechallenge in relationship to the induction of protective IgG2b Ab responses.

The human N592pc small cell lung cancer cell line was provided by Dr. J. Minna (National Cancer Institute, Bethesda, MD). Cells were cultured in endotoxin-free RPMI 1640 medium (endotoxin content, <0.005 EU/ml) supplemented with l-glutamine and antibiotics (all from Cambrex BioWhittaker, Milan, Italy) and 10% heat-inactivated FCS (Seromed Biochrom, Berlin, Germany; endotoxin content, 20 EU/ml). N592 cell transfectants engineered with a modified cDNA encoding a secreted form of IL-15 (36) or with a dicistronic insert encoding the two IL-12 chains (37) were previously described (35). Transfectants were cultured in the presence of G418 (500 μg/ml) and/or hygromycin (250 μg/ml) and were periodically tested for IL-15 and/or IL-12 production.

Female athymic (nu/nu, CD1) and SCID (scid/scid) mice, 6–8 wk old, were obtained from Charles River Laboratories (Lecco, Italy). Homozygous nonobese diabetic (NOD)-SCID mice were originally obtained from The Jackson Laboratory (Bar Harbor, ME) and bred in-house. All mice were housed under pathogen-free conditions and received autoclaved food and water.

Animals (six to eight mice for each group) were injected s.c. with 2 × 10⁷ N592pc or N592/IL-12/IL-15 tumor cells/mouse. Cells were mycoplasma-free, as assessed by ELISA (Roche, Milan, Italy). Cells were washed three times in endotoxin-free RPMI 1640 medium without FCS and once in endotoxin-free PBS before injection. The larger and smaller diameters of the s.c. tumors were measured using a caliper at weekly intervals; these two diameters were multiplied to obtain an estimate of the tumor area. The data are displayed as the mean ± SD of the tumor areas for each group of animals at a given time point. Statistical analysis was performed using the Mann-Whitney test; p < 0.05 was considered significant. Rechallenge was performed 4 wk following the inoculation of N592/IL-12/IL-15 cells by s.c. injection of 2 × 10⁷ N592pc/mouse in rejecting nude and SCID mice.

The Ab titer of sera was analyzed by indirect immunofluorescence and cytofluorimetric analysis. N592pc were stained with serial 2-fold dilutions (ranging from 1/12.5 to 1/1600) of sera. The titer was defined as the last dilution producing mean fluorescence intensity (MFI) values 2 times higher than background levels obtained by staining N592pc only with secondary Ab. PE-conjugated goat anti-mouse Ig Abs or isotype-specific Abs were used as second-step reagents. Samples were analyzed with a FACScan analyzer (BD Biosciences, San Jose, CA).

N592-rejecting nude mice were sacrificed on day 8 after rechallenge, and immune splenocytes were fused with SP2/0-Ag14 myeloma cells according to a standardized polyethylene glycol fusion protocol. The hybridoma supernatant screening was performed on N592pc by immunofluorescence or ELISA. For in vivo experiments hybridoma cells were cultured in serum-free DMEM/F-12 medium supplemented with BIOGRO-1 (Biological Industries, Kibbutz Beit Haemek, Israel), and Abs were purified by protein A chromatography (Amersham International, Milan, Italy)

Lymph nodes and spleens from untreated mice and from mice previously rejecting N592/IL-12/IL-15 cells were examined 4 wk after rechallenge with N592pc. For histologic evaluation, tissues were fixed in 10% neutral buffered formalin, embedded in paraffin, sectioned at 4 μm, and stained with H&E.

For immunohistochemistry, acetone-fixed cryostat sections (or formalin-fixed paraffin-embedded sections for proliferating cell nuclear Ag (PCNA) (immunostaining) were incubated for 30 min with one of the following Abs: anti-CD3, anti-CD4, and anti-CD8a (Sera-Lab, Crawley Down, U.K.); anti-Mac-1 (anti-CD11b/CD18), anti-Mac-3, and anti-Ia (Roche, Milan, Italy); anti-asialo GM1 (NK cells; Wako Chemicals, Dusseldorf, Germany); anti-PCNA (Ylem, Rome, Italy); anti-IFN-γ and anti-CD40 ligand (anti-CD40L; Santa Cruz, Biotechnology, Santa Cruz, CA). After washing, sections were overlaid with biotinylated goat anti-rat, anti-hamster, and anti-rabbit Ig and horse anti-goat Ig (Vector Laboratories, Burlingame, CA) for 30 min at 37°C. Unbound Ig was removed by washing, and the slides were incubated with avidin-biotin-peroxidase complex/AP (DAKO, Glostrup, Denmark). For double-immunofluorescence analysis the reaction was revealed using FITC-conjugated goat anti-rabbit, FITC-conjugated goat anti-rat (BD PharMingen, San Diego, CA) and rhodamine isothiocyanate-conjugated chicken anti-goat (Rockland, Gilbertsville, PA) Abs.

Mononuclear cells were isolated from the spleens of untreated nude mice and used as effector cells in a 6-h ⁵¹Cr release assay at different E:T cell ratios (from 200:1 to 25:1). ⁵¹Cr-labeled N592pc were pretreated with mAbs at 5 μg/ml or with a 1/20 dilution of pooled mouse serum for 30 min at room temperature before the addition of effector cells. Supernatants were collected for the evaluation of ⁵¹Cr release after 6 h of incubation at 37°C, and the percent lysis was then calculated.

The effect of cytokine engineering on tumorigenicity was evaluated by heterotopic (s.c.) implant in nude mice, which display functional natural immunity and T cell-independent B cell responses. In agreement with our previous report (35), N592 cells transfected with empty vectors (N592 mock) and N592pc showed rapid growth kinetics in nude mice, and N592/IL-12 cells displayed only a slightly decreased growth pattern. In contrast, N592/IL-15 cells showed a clearly reduced tumor growth rate, and N592/IL-12/IL-15 cells were completely rejected by all animals (Fig. 1 A), thus indicating a strong synergistic action of coexpressed IL-12 and IL-15.

The tumorigenicity of N592/IL-12/IL-15 cells was then tested in SCID mice, which have a more profound immune deficiency, lacking both T and B cells. N592/IL-12/IL-15 cells were rejected by 100% of SCID mice, while injection of N592pc produced rapid tumor growth (Fig. 1 B). Thus, tumor rejection responses induced by IL-12 and IL-15 gene cotransfer occur in both nude and SCID mice.

To investigate whether the rejection of cytokine-engineered cells could also result in a protective effect against wild-type tumor cells, we rechallenged N592/IL-12/IL-15 (N592 cells genetically modified with IL-12 and IL-15)-rejecting nude and SCID mice with N592pc. As shown in Fig. 2,A, rechallenge by N592pc induced transient tumor growth, followed by complete rejection in ∼60% of N592/IL-12/IL-15-primed nude mice (Fig. 2,B). The rejection process began in most animals at 2 wk after rechallenge, but in some instances it was completed only after 4 wk. The N592pc tumor growth rate in the remaining N592/IL-12/IL-15-immunized animals was similar to that in control mice (Fig. 2,A). In contrast to what we observed in nude mice, N592pc showed rapid growth kinetics in 100% of SCID mice that previously rejected N592/IL-12/IL-15 cells (Fig. 2, C and D), suggesting a role for B cells in the resistance to N592pc.

Sera from N592/IL-12/IL-15-injected nude mice were obtained 4 wk after rechallenge with N592pc and were tested by indirect immunofluorescence on N592pc using either anti-total Ig or subclass-specific secondary reagents. A noticeable individual variability in the anti-N592pc total Ig (not shown) or isotype-specific reactivity was found in sera from animals that rejected N592/IL-12/IL-15 cells; however, these sera showed, on the average, a stronger reactivity (as MFI) than sera from mice injected only with N592pc (Fig. 3). The predominant anti-N592pc Ig isotypes in N592/IL-12/IL-15-vaccinated mice were IgM and IgG1, followed by IgG2b and IgG2a (Fig. 3). If only N592/IL-12/IL-15-vaccinated mice, which were immune to N592pc rechallenge, were considered, their sera displayed significantly higher IgG2b titers (ranging from 1/100 to 1/800; mean, 1/300; p < 0.05) than sera of vaccinated mice, which failed to reject N592pc rechallenge (≤1/50). A similar, but less striking, difference was observed for anti-N592 IgG2a Ab titers (p = NS), while no differences between the two groups of mice were observed for anti-N592 total Ig, IgM, and IgG1 titers. Fluorescence intensity data at the representative 1/100 dilutions are shown in Fig. 3.

We did not find any difference in residual T, B, or NK cell counts in the spleens of mice protected against N592 rechallenge compared with those in mice that were not protected. All animals were female and of similar age and were negative for common pathogen screening. Since the only difference was the anti-N592 IgG2b titer in the serum, this suggested that cytotoxic IgG2 plays a protective role in this model.

To gain more information on the B cell response in nude mice we studied the histopathologic changes in draining lymph nodes of the tumor cell injection site. Lymph nodes of the untreated athymic nude mice were characterized by a marked depletion of lymphocytes in the thymus-dependent areas and an almost complete absence of GCs. In some mice, sporadic, poorly developed GCs appeared, which in addition to B cells contained very few T lymphocytes, rare apoptotic and mitotic figures, and a thin rime of lymphocytes.

In three nude mice rejecting N592/IL-12/IL-15 cells, which showed no resistance to rechallenge with N592pc, the draining lymph nodes showed evident histiocytosis. Histiocytes were found within central and peripheral sinuses. Scarce GCs were present (Fig. 4, A and C) with centroblasts in the dark zone and centrocytes in the light zone, which was almost devoid of T cells. A distinct mantle zone was observed.

Mice that developed resistance to N592pc rechallenge also showed a marked histiocytosis in the draining lymph node sinuses. However, by analyzing several lymph node sections of three different mice that were protected against N592pc rechallenge, we observed that GCs were significantly more numerous (6.3 ± 2) than in unprotected mice (2.5 ± 1; p ≤ 0.01). The more numerous (Fig. 4,B) and large GCs revealed a distinct dark zone constituted by several PCNA-positive centroblasts (Fig. 4 D), indicating that an active B cell response is taking place. In the light zone some T lymphocytes appeared intermingled with centrocytes. Few asialo-GM-1⁺ NK cells were observed in the GCs, while they were numerous in the T cell-depopulated paracortical area (data not shown).

Histological aspects of spleens obtained from the different mouse groups (untreated and treated mice, resistant and not resistant to rechallenge with N592pc cells) closely mimic those found in lymph nodes of the respective animals.

Among the cytokines driving B cell differentiation, IFN-γ is known to mediate IgG2 production. Indeed, IFN-γ was produced by several cells located in the paracortical areas of the draining lymph nodes (Fig. 4,f) and spleens of mice that developed resistance to N592pc. By double immunostaining the production of IFN-γ mostly colocalized with asialo-GM-1⁺ NK cells (Fig. 5, a and b) and with a few residual CD8⁺ cells (Fig. 5, c and d). In the lymph nodes and spleens of these mice, distinct expression of the costimulatory molecule ligand CD40L, involved in Ig class switch, was also observed (Fig. 4 h). The expression of CD40L was also confirmed by immunofluorescence and FACS analysis on cell suspensions obtained from draining lymph nodes of rejecting animals. About 3–6% of CD40L-positive cells were detected by FACS analysis, and about half of these cells appeared positive for the DX-5 NK cell marker (data not shown).

In contrast, a scarce to absent IFN-γ production was detected in the lymph nodes and spleens from mice rejecting N592/IL-12/IL-15 cells but not resistant to rechallenge (Fig. 4,E). In these mice the expression of CD40L was barely detectable (Fig. 4 G). IFN-γ was completely absent in lymph nodes and spleens from untreated mice, in which no evident expression of CD40L was found.

To further study the Ab response in mice developing N592pc resistance we selected mAbs by cell fusion of splenocytes from a nude mouse that rejected N592pc rechallenge with SP2/0-Ag14 myeloma cells. The reactivity of hybridoma supernatants with N592pc was screened by immunofluorescence or ELISA. Among different mAbs obtained, we focused our interest on two mAbs, ASA21 (IgM) and ASA52 (IgG2b), which showed similar surface reactivity against N592pc (Fig. 6). These mAbs appeared to react with the same or a closely related epitope(s) on the basis of mAb binding cross-competition experiments and showed no inhibitory activity on N592pc growth and viability in vitro (data not shown). We then analyzed whether these mAbs could exert a protective effect against N592pc in vivo. To this end, a tumorigenic dose of N592pc was injected s.c. in different groups of nude mice, and then animals were treated (5 μg/mouse) with ASA21 or ASA52 mAb or PBS by i.p. injection every third day. As shown in Fig. 6, B and C, ∼60% mice treated with ASA52 mAb showed complete tumor rejection, and the remaining mice displayed reduced tumor growth kinetics. Under the same experimental conditions, ASA21 mAb or an unrelated IgG2b Ab displayed no inhibitory effect on take rate and tumor growth kinetics. At doses ≥20 μg every third day ASA52 mAb induced tumor rejection in 100% (10 of 10) treated mice (Fig. 6,C), while IgM ASA21 had still no effect (data not shown). Treatment with ASA52 (20 μg/mouse) was significantly less effective (p < 0.05) in NOD-SCID mice bearing N592pc implants (Fig. 6 D) than in nude or SCID mice (not shown).

Therefore, these findings indicate that IgG2b Abs display a potent anti-tumor activity against N592pc tumor cells in vivo, whereas IgM Abs are ineffective. The sensitivity of N592pc to ADCC was then analyzed in a ⁵¹Cr release assay in the presence of nude mouse splenocytes as effector cells and mAbs at a concentration of 5 μg/ml. ASA52 clearly triggered ADCC activity against N592pc, whereas ASA21 or an irrelevant IgG2b had no effect (Fig. 7,A). Also, pooled sera from N592/IL-12/IL-15-primed mice that showed resistance to rechallenge with N592pc induced low, but detectable, ADCC activity, while sera from mice that did not develop resistance were ineffective (Fig. 7 B).

Human MHC class I-negative N592 cells genetically modified to coexpress IL-12 and IL-15 were rejected when implanted in nude or SCID mice, while cells expressing single cytokines displayed only a partial growth inhibition, in agreement with our previous results (35). More interestingly, we show that ∼60% of nude mice that rejected N592/IL-12/IL-15 cells became resistant to rechallenge with unmodified N592 cells, while no protection was found in SCID mice. These findings together with morphological evidence of distinct GC formation in nude mice showing resistance to rechallenge suggested a possible role of Abs in an adaptive response. Nonetheless, the primary rejection of N592/IL-12/IL-15 cells did not require such an adaptive Ab response and appeared related to the activation of innate immunity, since it occurred in SCID mice. The role of Abs in the adaptive response against N592pc was further indicated by the correlation between serum anti-N592pc IgG2, particularly IgG2b, titers and resistance to N592pc rechallenge in nude mice. Other previous reports stressed the role of anti-tumor Ag Abs in anti-tumor responses through mechanisms involving receptor blockade (38), complement fixation (39), or ADCC, mediated primarily by NK cells through FcR (40, 41). Here we also show that an IgG2b mAb derived from immune nude mice splenocytes showed strong protective effects against N592pc in unprimed nude mice while an IgM mAb derived from the same fusion was ineffective. These mAbs showed similar reactivity against N592pc and displayed no inhibitory activity on N592pc growth in vitro. Since both IgM and IgG2b are known to mediate complement fixation, the protective effect of Abs in this model may instead depend on the ability of IgG2b to mediate ADCC, a function that is not shared by IgM. Indeed, we show that N592pc were sensitive to ADCC activity mediated by fresh murine NK cells through the IgG2b mAb and also that sera from mice that developed resistance to N592pc rechallenge induced ADCC. In addition, in NOD-SCID mice, a strain that has reduced NK cells and macrophages, ASA52 IgG2b mAb showed only a partial inhibitory effect on N592pc growth. Together these observations support a role for ADCC mechanisms in N592 tumor cell rejection mediated by IgG2b Abs.

In previous reports IL-12 engineered tumor cells (16) or rIL-12 administered together with tumor cells have been shown to trigger not only cellular, but also Ab responses in immunocompetent syngeneic mice models (32). Thus, IL-12 can act as an adjuvant for humoral immunity, increasing predominantly IgG2a and IgG2b responses and suppressing IgG1 and IgE production (42). Part of these effects could be mediated, in immunocompetent hosts, through Th1 polarization and IFN-γ production (43).

Induction of Ig class switch in athymic nude mice may depend on the synergistic induction of IFN-γ secretion by IL-12 and IL-15 (35, 44) and/or the induction of an innate immunity/B cell cross-talk. Indeed, several lines of evidence indicate that NK cells may be capable of providing efficient help to B cells (45, 46) and may also express CD40L (47), another factor that is essential in providing costimulatory signals to Ag-primed B cells for survival, Ig class switch, and generation of memory B cells (48, 49). In our nude mouse model we could demonstrate the expression of both IFN-γ and CD40L by NK cells present in draining lymph nodes of N592/IL-12/IL-15-primed animals, which were capable of rejecting N592pc upon rechallenge. Although a contribution of extrathymic residual T cells cannot be excluded, these findings strongly suggest that in nude mice the help to B cells may be supplied by NK cells through CD40L and IFN-γ release. In addition to these two factors it is conceivable that IL-12 and/or IL-15, produced by transfectants during the primary rejection, may directly act on B cells. IL-15 has indeed been shown to stimulate proliferation and Ig production by B cells (23).

Several lines of evidence indicate that human tumors may frequently display a loss of MHC class I molecules, thus escaping from control of the immune system (4, 5, 6). Although it has been proposed that in some tumors the loss of specific alleles may be a consequence of in vivo selective pressure by CTLs (7), other tumors, such as neuroblastomas (50) and small cell lung carcinomas (46), display a lineage-dependent down-regulation of MHC class I-related genes at the transcriptional level. The stimulation of Ab responses against MHC class I-negative tumors may be of major interest in the development of a cancer vaccine in humans. In addition, the possibility of by-passing the need of Th cells, which may be functionally suppressed in their responses in tumor-bearing patients, by stimulating natural immunity/B cell cross-talk is an attractive possibility. Finally, the selection of hybridomas from nude mice, which display protective anti-tumor Ab responses, may represent a suitable tool in the search for novel mAbs endowed with anti-tumor activity in vivo.