Systemic interleukin 12 displays anti-tumour activity in the mouse central nervous system.

In various systemic cancers, interleukin 12 (IL-12) induces anti-tumour immunity mediated by T lymphocytes and natural killer cells. To determine whether IL-12 has anti-tumour activity against malignant gliomas in the central nervous system (CNS), which is considered to be an immunologically privileged site, we treated mice with meningeal gliomatosis by intraperitoneal (i.p.) or intrathecal (i.t.) administration of recombinant murine IL-12. Although untreated mice revealed symptoms, such as body weight loss or paraplegia as a result of the meningeal gliomatosis within 8 days after tumour inoculation, 80% of the mice treated with IL-12 at 0.5 microg i.p. were cured. Many lymphocytes, mostly CD4+ and CD8+ cells, infiltrated to the tumours of IL-12-treated mice. The numbers of these cells increased in the cervical lymph nodes, into which the cerebrospinal fluid drains, and there they secreted a considerable amount of interferon-gamma. Mice cured by IL-12 rejected subcutaneous or i.t. rechallenge with their original glioma cells, but the same mice were not able to reject other syngeneic tumour cells. These results indicate that the immune system recognizes malignant glioma cells in the subarachnoid space of the CNS and that systemic IL-12 may produce effective anti-tumour activity and long-lasting tumour-specific immunity.

The central nervous system (CNS) has long been considered to be an immunologically privileged site (de Micco, 1989), as the blood-brain barrier (BBB) had been thought to prevent components of the immune system from entering. Recently, several immunological reactions have been reported to take place in the CNS. For instance, T lymphocytes infiltrate the CNS in several disease models , some glial cells express major histocompatibility complex (MHC) antigens (Frank et al, 1986;Fontana et al, 1987), microglia are derived from bone marrow (Hickey et al, 1992), and cytokine networks exist in the CNS (Fontana et al, 1984). Among gliomas, which account for 35% of all primary CNS tumours (Annegers et al, 1981), glioblastomas are the most frequent and intractable tumours (Nazzaro et al, 1990;Yoshida et al, 1994). Less than 10% of glioblastoma patients survive longer than 5 years despite radical surgeries and various adjuvant therapies (Nazzaro et al, 1990;Yoshida et al, 1994). The rest of the patients usually suffer from progressive tumour growth and, less frequently, meningeal dissemination. Various types of immunotherapy, such as cytokine therapy, antibodies and adoptive immunotherapy, have been tried in malignant glioma patients (Shimizu et al, 1987;Barba et al, 1989;Nitta et al, 1990;Lillehei et al, 1991;Farkkila et al, 1994), sometimes showing a therapeutic effect without significantly prolonging survival.
In advanced phases of human glioblastoma, some tumours undergo cerebrospinal fluid (CSF) pathway dissemination or meningeal gliomatosis (MG) (Arita et al, 1994). The MG model used in the present experiments effectively mimics the terminal state of such patients. Using this model, we demonstrated that systemically administered IL-12 produces a potent anti-tumour effect and long-lasting immunity in the CNS by promoting the proliferation of TIL and by stimulating local IFN-y production.
Anti-tumour activity of IL-12 in the CNS 447 A )6 B Figure 1 Tumours in the brain of meningeal gliomatosis (MG) mice. RSV-M glioma cells (5 x 106) were inoculated intrathecally into C3H/HeN mice. Tumour formation was observed in the basal cistern. A, Three days after tumour inoculation; B; 8 days after tumour inoculation. Haematoxylin and eosin (H and E) staining. Magnification x 20 Glioma models and in vivo IL-12 treatment RSV-M glioma cells were harvested with 0.25% trypsin and washed three times with PBS (137 mm sodium chloride, 2.7 mM potassium chloride, 8.1 mM disodium hydrogen phosphate and 1.5 mm potassium dihydrogen phosphate). The cells were suspended in PBS at a concentration of 108 cells ml. An MG model was produced in 7-week-old female C3H/HeN mice weighing 18-21 g by inoculating 50 gl of the cell suspension into the cistema magna using a 26-gauge needle. An MG model was also produced in 7-week-old female BALB/c-nu/nu mice by inoculating 5 x 106 cells as described above. Recombinant murine IL-12 (rmIL-12) (Hoffmann-La Roche, Nutley, NJ, USA)  was diluted in PBS and injected intraperitoneal (i.p.) (0.5 or 0.1 gg 200 il-') or intrathecal (i.t.) (0.5, 0.1 or 0.02 gg 50 gt1') into the C3H/HeN mice and i.p. (0.5 gg 200 ,ul-') into the BALB/c-nu/nu mice once daily from days 3 to 7 after tumour inoculation. When MG mice revealed physical symptoms, such as 20% body weight loss or paraplegia (MG symptoms), we killed them to confirm the tumour formation histologically. Each group consisted of five mice, and all experiments were repeated more than twice. Differences in days without MG symptoms between the treated groups and the control group were determined using Student's t-test. Animals were handled in accordance with the guidelines of the Animal Committee of Osaka University Medical School and UKCCR guidelines for the welfare of animals in experimental neoplasia (UKCCR, 1988).
Histology and immunohistochemistry MG mice which had been treated with rmIL-12 at 0.5 jg i.p. and untreated controls were killed on days 4, 6 and 8 after tumour inoculation. The brain was removed, immediately frozen and embedded in OCT compound (Tissue Tek, Miles, Elkhart, IN, USA). Coronal sections were cut at a thickness of 7 gm using a cryostat, fixed in acetone for 15 min and then stained with biotinylated rat monoclonal antibody (MAb) against T-lymphocyte antigens CD4 and CD8 (PharMingen, San Diego, CA, USA). Other sections were stained with haematoxylin and eosin (H and E). The spleen and cervical lymph node (CLN) of MG mice which had been treated with rmIL-12 at 0.5 ,ug i.p. were immediately removed after the animals were killed. Untreated MG mice were used as controls. The CLN and spleen were suspended mechanically in PBS. Cells remaining after the lysing of red blood cells were washed twice and counted (Hashimoto et al, 1995). Spleen cells (106) were incubated with FITCor PE-conjugated MAb against mouse lymphocyte antigens CD3, CD4 and CD8 (PharMingen), and were analysed by FACScan (Becton Dickinson, Mountain View, CA).

RESULTS
Tumour formation was observed in the CNS of MG mice Three days after i.t. tumour inoculation ( Figure IA), glioma cells were frequently observed in the basal cisterns and ventricles and on the brain surface. Mass formation was observed in the basal cistern after 8 days ( Figure 1B). Hydrocephalus was seen in several mice. Untreated mice lost body weight, developed paraplegia and were expected to die within 2 weeks after tumour inoculation.
Intraperitoneal administration of IL-12 was effective in immunocompetent MG mice but not in T-cell-deficient MG mice All untreated MG mice showed MG symptoms described in Materials and methods within 8 days (Figure 2A, thin solid line). The symptom-free time was prolonged in MG mice treated with IL-12 at 0.1 tg i.p., and 40% of these mice did not reveal any symptoms (P = 0.053; Figure 2A, dashed line) for longer than 30 days. Eighty per cent of MG mice treated with IL-12 at 0.5 jg i.p. had been without MG symptoms for longer than 30 days (P < 0.01; Figure 2A, bold solid line), and they were judged to be cured. The median symptomfree time was not prolonged in any of the groups treated i.t. in comparison with the control group (P > 0.07; Figure 2B), although a few mice had been without MG symptoms for longer than 30 days in both the IL-12 at 0.1 and the IL-12 at 0.05-,ug i.t. groups. These results showed that the local administration of IL-12 into the CNS was less effective in vivo than the systemic administration.
In BALB/c-nu/nu mice, in which NK activity is considered to be strong, all MG mice showed MG symptoms within 9 days both in the untreated group and in the IL-12 i.p. group (P = 0.17; Figure  3). These findings revealed that i.p. IL-12 was effective in immunocompetent MG mice but not effective in T-cell-deficient MG mice. Furthermore, NK cells alone, without T cells, may not be capable of rejecting tumour cells transplanted into the CNS. CD4+ and CD8+ lymphocytes primarily infiltrated the tumour-bearing areas of MG mice after IL-12 treatment Tumour formation and hydrocephalus were observed in brain sections of untreated mice, with tumour masses daily increasing in size (Figure 1). Few lymphocytes were present in or around the tumours of untreated mice ( Figure 4A). In the MG mice treated with IL-12 i.p., tumours gradually increased in size from day 4 to day 6, but then became smaller on day 8, and no tumours were found on day 16 (data not shown). TIL and some necrotic regions were observed on day 6 in tumour areas of the MG mice treated with IL-12 i.p. ( Figure 4B). These lymphocytes consisted mostly of CD4+ and CD8+ cells ( Figure 4C and D), in a respective ratio of approximately 0.4-0.6. These results indicated that i.p. administration of IL-12 stimulates CD4+ and CD8+ lymphocytes to infiltrate into and around the tumours of MG mice. Numbers of CD3-, CD4and CD8-positive T lymphocytes in CLN significantly increased in IL-12-treated MG mice IL-12 did not have a significant effect on the cell distribution (CD3, CD4 and CD8) in the spleen and CLN of MG mice (data not shown). However, because the spleen and CLN were enlarged in i.p. IL-12-treated MG mice, the total number of T lymphocytes also increased as a natural consequence. Although the total numbers of CD3 ( Figure SA), CD4 ( Figure 5B) and CD8 cells ( Figure 5C) in the spleen of i.p. IL-1 2-treated MG mice (grey bars) were larger than in untreated mice (black bars), the increase was not statistically significant. T-cell population in the CLN of i.p. IL-12-treated MG mice significantly increased in comparison with untreated MG mice ( Figures 5D-F). These MG models were then administered IL-12 i.p. from days 3 to 13 to compare the results with those in the mice treated for 5 days. Prolonged IL-12 treatment did not further affect the T-cell markers ( Figure 5, white bars). Each T-cell population showed a maximum on day 8 in the spleen and on day 4 in the CLN.

IFN-y production by lymphocytes in the CLN was markedly increased in IL-12-treated MG mice
We examined IFN-y and IL-10 production by lymphocytes in the CLN and spleen in MG mice with or without IL-12 treatment. The splenocytes obtained from i.p. IL-12-treated MG mice secreted the same level of IFN-y as those from untreated mice ( Figure 6A). On the other hand, a significant increase in IFN-y was consistently observed in the lymphocytes derived from the CLN of i.p. IL-12treated MG mice ( Figure 6C). Although the splenocytes from treated MG mice secreted less IL-10 on days 6 and 8 than those from untreated mice ( Figure 6B), the lymphocytes in the CLN obtained from treated MG mice secreted more IL-10 on days 4 and 6 than those from untreated mice ( Figure 6D). In treated MG mice, lymphocytes from the CLN maintained a high level of IFN-y secretion until day 6 and then production decreased on day 8, but IL-10 production was already decreased on day 6 (Figs 6C and D grey bars).
MG mice cured by IL-12 rejected rechallenge with glioma cells MG mice cured by IL-12 and surviving more than 8 weeks after the first inoculation of glioma cells were reinoculated with 5 x 106 glioma cells injected into the cisterna magna. These survivors had shown no MG symptoms without any treatment for more than 8 weeks after reinoculation with glioma cells (P < 0.01; Figure 7A). Apparent tumours were observed s.c. after 1 week when 1 x 107 glioma cells were inoculated s.c. into untreated mice ( Figure 7B, black squares), but MG mice cured by IL-12 rejected these s.c. inoculated glioma cells completely (P < 0.01; Figure 7B, black circles). Syngeneic C3H/MCA clone 15 fibrosarcoma cells, however, grew s.c. larger in cured MG mice ( Figure 7B

DISCUSSION
In this report, we applied the MG model to investigate the immunological response against tumour cells in the CNS. Although the pia mater separates subarachnoid space from brain parenchyma, some intrathecally injected glioma cells show parenchymal invasion (Kitamura et al, 1996;unpublished data); in addition, glioma cells undergo CSF pathway dissemination in animal models and in patients with advanced glioma (Arita et al, 1994;Yamada et al, 1994). On the other hand, the BBB or the blood-CSF barrier around the tumour mass in the CNS was thought to be destroyed, and thus tumour antigens gained access to the systemic immune system via tumour vessels or Virchow-Robin space, which continues to subarachnoid space. Further, it was easy to carry out the MG model on mice, and the appearance of MG symptoms showed good correlation with the numbers of intrathecally injected tumour cells. In this way, we used the MG model to investigate the anti-tumour efficacy of IL-12 in the CNS. In this paper, we have demonstrated that systemic administration of IL-12 was surprisingly effective against malignant tumours in the subarachnoid space of the CNS (MG mice). As previously reported , IL-12 does not directly affect the growth of RSV-M glioma cells in vitro (data not shown). However, systemic IL-12 prolonged the symptom-free time in the present study (P < 0.01; Figure 2A). Not only systemic administration but also local or peritumoral injection of IL-12 has been reported to induce the regression of some subcutaneous tumours (Brunda et al, 1993;Tahara et al, 1994). Although i.t. IL-12 treatment was less effective when judged on the basis of the median symptom-free time (P > 0.07; Figure 2B), some MG mice were nevertheless cured by i.t. administration. A large i.t. dose of IL-12 may have some toxicity because some normal mice had lost body weight during and after i.t. administration of 0.5 jg of IL-12 (data not shown). As treatment with IL-12 at 0.5 ,ug i.t. induced splenomegaly, as did i.p. treatment (data not shown), i.t. IL-12 would affect systemic immunity. These findings suggest that i.t. injection of IL-12 has some anti-tumour effect, but that its toxicity offsets its efficacy. Thus, systemic IL-12 may be the preferable choice even for the treatment of CNS tumours.
Under normal conditions, no lymphocytes are present in the brain parenchyma. TIL, however, are found in and around the tumours in 35-80% of glioma patients (de Micco, 1989). B lymphocytes and NK cells do not play major roles in suppressing the growth of human gliomas (Sawamura et al, 1988). TIL are activated by   (Andrews et al, 1993), and IL-12 can induce the production of CD8+ CTL in vitro (Bloom et al, 1994). Large numbers of TIL were observed in immunocompetent MG mice treated with IL-12 (Figure 4), and IL-12 did not show antitumour activity in T-cell-deficient mice (P = 0.17; Figure 3). RSV-M glioma cells were rejected when they were reinoculated i.t. or s.c. into MG mice cured by IL-12 (P < 0.01; Figure 7), but syngeneic C3H/MCA clone 15 fibrosarcoma cells were not rejected (Figure 7). These findings indicate that tumour-specific immunity is induced systemically and/or in the CNS of MG mice cured by IL-12, and that T lymphocytes produced as a result of activation by IL-12 act as major anti-tumour effectors in MG mice.
Many tumour antigens in the CNS are mainly phagocytized by microglia (Shrikant et al, 1996), and a significant percentage of the CSF drains into the CLN (Kida et al, 1993(Kida et al, , 1995Cserr et al, 1992). In MG models, tumour antigens would reach the CLN via this route. In the CLN, APC activate Th cells and CTL, and they secrete a moderate amount of IFN-y. Systemic administration of IL-12 also activates Thl cells and CTL (Seder et al, 1993;Kennedy et al, 1994;Nastala et al, 1994). In the CLN from IL-12treated MG mice, a large number of T lymphocytes and a high level of IFN-y production were detected (Figures 5 and 6). IFN-y production by lymphocytes in the CLN from IL-12-treated MG mice was markedly increased (P < 0.01; Figure 6C higher than that in the CLN from normal mice treated with IL-12 (data not shown). These findings suggest that both systemic IL-12 and tumour antigens are necessary for a high level of IFN-,y secretion. The expression of MHC antigens can be induced in both glial and glioma cells in the CNS by IFN-y (Wong et al, 1984;de Micco, 1989;Tamura et al, 1989). Activated glial cells produce granulocyte-macrophage colony-stimulating factor (Ohno et al, 1990;Lee et al, 1994), and this factor activates microglia, which are one type of APC in the CNS Matsumoto et al, 1992;Shrikant et al, 1996). In this way, although it has been thought that the presence of the BBB and the absence of lymphatic tissues segregate the CNS from systemic immunity, the immune system would appear to recognize the tumour antigens in the subarachnoid space of the CNS, and IL-12 may activate immunological reactions against tumours in the CNS. Furthermore, Th cells and CTL also efficiently infiltrate tumours in the CNS.
Although it is well known that IL-12 stimulates T and NK cells to secrete IFN-y, there is relatively little information on the effect of IL-12 on T cells for production of IL-10. It has been reported that in vivo treatment of IL-12 increases in both IFN-y and IL-1O mRNA and that the presence of IL-12 results in an efficient priming of the clones for high production of both IFN-y and IL-10 (Finkelman et al, 1994;Gerosa et al, 1996). In the present study, both IFN-y and IL-10 production elevated on days 4 and 6 in the CLN from IL-12 treated MG mice ( Figure 6D). Elevation of IL-10 is supposed to affect an anti-inflammatory response involving down-regulation of Thl (Gerosa et al, 1996). Some investigators have, however, reported that production of IL-10 in bulk T-cell cultures stimulated in vitro with allergen is inhibited if IL-12 is added to the cultures (Marshall et al, 1995). The other investigators have reported that the IL-10-producing clone is primed only if IL-12 is added to the culture in the first few days (Gerosa et al, 1996). In this study, IL-10 levels of the CLN decreased by day 6 and were not detected on day 8, while IFN-y production was still maintained at a high level on day 6 and present on day 8 ( Figures  6C and D). The release of the hyperimmune state and/or the prolonged stimulation of IL-12 might result in the decrease of IL-10 level preceding that of IFN-y level.
The results of this study indicate that systemic IL-12 administration induces T-lymphocyte-mediated immune responses in the CNS, and that CTL appear to play an important role in the rejection of glioma cells in the CNS. Furthermore, IL-10 secretion may suppress such inflammatory functions in these models.