An Interleukin (IL)-10/IL-12 Immunoregulatory Circuit Controls Susceptibility to Autoimmune Disease

SJL/J, C57BL/6, C57BL/10, BALB/c, BALB/c nu/ nu, and C.B-17 SCID mice were all obtained from the National Cancer Institute (Frederick, MD). Breeding pairs of C57BL/6 IL-12−/− (N6) and C57BL/6 IFN-γ−/− (N7) mice were originally provided by J. Magram (Hoffman LaRoche, Nutley, NJ) and Dyana Dalton and T. Stewart (Genentech Inc., South San Francisco, CA), respectively. Breeding pairs of IL-4−/− and IL-10−/− mice were originally obtained from R. Kuhn and W. Muller (University of Koln, Koln, Germany) and backcrossed in our facilities onto the C57BL/6 (N13) and C57BL/10 (N7) backgrounds, respectively. All mice were housed under specific pathogen-free conditions. They were exclusively female and between 8 and 12 wk of age when experiments were initiated.

Bovine MBP was obtained from Sigma Chemical Co. (St. Louis, MO). MBP_87–106, corresponding to residues 87–106 of murine MBP (VVHFFKNIVTPRTPPPQGK), was synthesized by the Laboratory of Molecular Structure, Peptide Synthesis Laboratory (NIAID, NIH, Bethesda, MD) and analyzed and purified by high pressure liquid chromatography. For induction of EAE by active immunization, mice were injected with bovine MBP (400 μg) emulsified in CFA (1:1) by the subcutaneous route over the left flank on day 0 and over the right flank on day 7. Mice were examined daily and rated for severity of neurological impairment as previously described (62, 63). For disease induction by adoptive transfer, donor mice were immunized with MBP_87–106 (100 μg) or bovine MBP (400 μg) emulsified in CFA (1:1) by subcutaneous injection at four sites over the flanks. 10–14 d later, draining LN cells (axillary and inguinal) were removed and processed as previously described (62). Cells were cultured for 96 h with MBP_87–106 (50 μg/ml) or bovine MBP (50 μg/ml) in RPMI 1640 containing 10% FCS and standard supplements (TCM). Recovered cells (5 × 10⁷) were injected intraperitoneally into naive syngeneic recipients that were examined daily for signs of EAE. In certain studies, the mice were injected intraperitoneally with polyclonal goat anti–mouse IL-12 (0.5 μg per injection; a gift of Drs. D. Presky and M. Gately, Hoffman-La Roche), normal goat IgG (0.5 mg per injection; Sigma Chemical Co.), rat anti–mouse IL-10 (1 mg each of mAbs SXC-1 and SXC-2 per injection; reference 66), or control rat IgG (2 mg; Sigma Chemical Co.).

Single-cell suspensions of spleen or LN tissue were prepared by passage through wire mesh and red blood cells lysed with ACK buffer (NIH Media Unit, Bethesda, MD). For detection of cytokine production during primary culture, cells (4 × 10⁶/ml) were cultured in TCM in 24-well plates (Costar Corp., Cambridge, MA) for 96–144 h. In some experiments, spleens were depleted of various subpopulations before culture using sheep antifluorescein biomag beads (Perspective Bioresearch Products, Cambridge, MA) together with FITC-conjugated mAbs specific for CD4, CD8, B220, and Mac-1 (PharMingen, San Diego, CA). In other experiments, purified CD4⁺ T cells and B220⁺ B cells were cocultured (2 × 10⁵ of each in 200 μl) in 96-well microtiter plates (Costar Corp.). The CD4⁺ T cells were isolated using CD4 subset enrichment columns (R & D Systems, Inc., Minneapolis, MN) with 90–95% purity achieved. B cells were positively selected to 93–98% purity by incubation with anti-B220 microbeads followed by magnetic separation using VS+ columns (Miltenyi Biotec, Inc., Auburn, CA).

For measurement of cytokine production by MBP-reactive LN cells during secondary cultures, the cells were harvested and washed extensively after 96 h of primary in vitro stimulation. Then they were resuspended in fresh media (1 × 10⁶ cells/ml) and supplemented with irradiated syngeneic splenocytes (3,000 rads; 4 × 10⁶ cells/ml) with or without MBP_87–106 (50 μg/ml). Supernatants were assayed 48 h later.

IFN-γ, IL-4, and IL-10 were quantified using a sandwich ELISA technique based on noncompeting pairs of antibodies as previously described (62). The lower limit of detection for each assay was as follows: IFN-γ, 12–24 pg/ml; IL-4, 8–16 pg/ml; and IL-10, 16–32 pg/ml.

Previous studies have attempted to address the contribution of individual cytokines to the pathogenesis of EAE by examining the susceptibility of cytokine-deficient mice to disease induction. Unfortunately, a comparison of these studies was difficult as different myelin-derived antigens were used and the genetically deficient mice were only backcrossed onto “susceptible” backgrounds for two to four generations. To avoid these problems, we used a single antigen to examine the susceptibility of wild-type mice and several different cytokine-deficient mice that had been backcrossed more completely to either the C57BL/6 or the C57BL/10 backgrounds. C57BL/6 wild-type mice were relatively resistant to EAE, with only 31% of actively immunized individuals exhibiting clinical symptoms (Fig. 1). Among the affected cohort the mean peak clinical score was 2.67 ± 0.8 and the average day of onset was 8.2 ± 0.4. On the other hand, IL-12−/− mice were completely resistant to disease whereas IFN-γ−/− animals exhibited severe EAE at 100% incidence (mean peak clinical score, 4.5 ± 0.68; average day of onset, 9.1 ± 1.6). Surprisingly, the induction of EAE in the IFN-γ−/− animals was completely prevented by the administration of neutralizing antibody to IL-12, but not isotype control antibody, during the course of active immunization (Fig. 1, top). This result suggested that IL-12 promotes the development of EAE by an IFN-γ–independent mechanism. The failure of IL-12−/− and anti–IL-12–treated IFN-γ−/− mice to develop disease was not secondary to the induction of a Th2 response as neither IL-4 nor IL-10 was detected after stimulation of LN cells from these animals with MBP in vitro (data not shown). Interestingly, the incidence and severity of disease in IL-4−/− mice was comparable to that of the C57BL/6 wild-type controls (Fig. 1, bottom; mean peak clinical score, 1.95 ± 0.7) with slightly delayed onset (13 d ± 3). By contrast IL-10−/− mice exhibited enhanced disease incidence and severity in comparison to C57B/10 wild-type controls (Fig. 1, bottom; mean peak clinical scores, 2.4 ± 0.6 vs. 1 ± 0.1 in IL-10−/− and wild-type mice, respectively) with similar kinetics (average day of onset, 11.5 ± 0.5 and 11 ± 1.3, respectively). LN cells from both the IL-4−/− and IL-10−/− strains produced elevated levels of IFN-γ after stimulation with MBP in vitro compared to LN cells from wild-type mice (data not shown). A deficiency of MBP-specific Th2 cells could, theoretically, be responsible for the increased incidence of EAE, as well as the propensity for Th1 cytokine production, in IL-10−/− mice. However, this explanation is unlikely since the IL-4−/− mice did not differ significantly from controls with regard to those characteristics and IL-4 is required for Th2 development (67).

To define IL-12–dependent stage(s) in the development of EAE effectors, we neutralized IL-12 during individual steps of an adoptive transfer protocol. Susceptible female SJL mice were immunized once with MBP_87–106 peptide (the immunodominant epitope for H-2^s strains) in CFA; 10 d later, draining LN cells were stimulated with antigen in vitro for 4 d and then transferred into naive syngeneic recipients. As both the in vivo immunization and the in vitro boost are required for the generation of encephalitogenic effectors, we treated mice with a polyclonal anti–IL-12 antiserum during the course of priming and/or added the antiserum to the in vitro cultures. MBP-reactive LN cells exposed to anti–IL-12 only during culture were mildly inhibited in their ability to produce IFN-γ on subsequent antigenic challenge in vitro (Fig. 2,A), but were unimpaired in their capacity to transfer disease (Fig. 2,B). In contrast, LN cells from donor mice treated with anti–IL-12 in vivo exclusively were severely compromised in their ability to produce antigen-dependent IFN-γ on secondary challenge in vitro (Fig. 2,A) and their encephalitogenicity was markedly decreased on adoptive transfer (Fig. 2,B). Administration of anti–IL-12 both in vivo and in vitro abrogated IFN-γ production in secondary cultures, but did not further impair the ability of cells to transfer disease when compared to cells from animals that were only exposed to anti–IL-12 in vivo (Fig. 2, A and B).

To investigate whether the MBP-reactive cells from the susceptible SJL strain had assumed a Th2 phenotype after treatment with anti–IL-12 in vivo and/or in vitro, we measured IL-4 in the supernatants of the primary and secondary cultures of the four experimental groups studied in Fig. 2. We were unable to detect IL-4 in any of the supernatants tested. On the other hand, both LN cells and splenocytes from anti–IL-12 treated donors, but not control IgG–treated donors, produced significant quantities of IL-10 during the primary culture (Fig. 3,A). This result raised the question of whether the suppression of the encephalitogenic phenotype by anti–IL-12 treatment was secondary, in part, to the production of IL-10 during the activation of MBP-specific T cells. Indeed, neutralization of IL-10 using a mixture of mAbs during both immunization and the in vitro culture largely reversed the protection from disease afforded by the treatment of donor mice with anti–IL-12 (Fig. 3 B).

Initially, we postulated that the source of IL-10 was an MBP-specific T cell population induced during priming in the presence of anti–IL-12. However, similar amounts of IL-10 were produced by cultures in the presence or absence of MBP_87–106 (data not shown). In fact, IL-10 production was not at all dependent on immunization with MBP in CFA as splenocytes from naive SJL mice previously treated with anti–IL-12 produced as much IL-10 as splenocytes from identically treated animals immunized with MBP (Fig. 4 A). An anti–IL-12 antiserum from a separate source (sheep anti–mouse IL-12; gift of Genetics Institute, Cambridge, MA) was equally effective (not shown).

Since mRNA encoding IL-12 p40 and p35 has been found in the spleen and lymph node of naive mice (68), these results are consistent with the possibility that a low baseline level of IL-12 tonically suppresses an IL-10 producing cell in lymph node and spleen. IL-12 could act directly on a cell bearing IL-12 receptor β1 and β2 chains or indirectly through the induction of IFN-γ. To clarify the role of IFN-γ in IL-12–mediated suppression of IL-10, C57BL/6 IFN-γ−/− and wild-type mice were treated with anti–IL-12 or control IgG. Splenocytes from IFN-γ−/− mice treated with anti–IL-12, but not control IgG, produced large quantities of IL-10, at levels comparable to that produced by wild-type splenocytes from identically treated donors (Fig. 4 B). Thus, IL-12 suppresses IL-10 production by an IFN-γ–independent mechanism.

To define which cell type was responsible for IL-10 production after treatment with anti–IL-12 in vivo, we compared IL-10 production by splenocytes from anti–IL-12–treated nu/nu, C.B-17 SCID, and wild-type BALB/c mice. Splenocytes from nu/nu and SCID mice failed to produce detectable IL-10, whereas spleen cells from normal BALB/c mice produced amounts of IL-10 comparable to those produced by spleen cells from similarly treated SJL mice (Fig. 5,A). As these results clearly implicate the T cell as the source of IL-10 production after anti–IL-12 treatment, we depleted various subpopulations from anti–IL-12–treated SJL spleen preparations before culture and measured IL-10 levels in supernatants 96 h later. As expected, depletion of CD4⁺ cells abrogated IL-10 production, whereas depletion of CD8⁺ and Mac1⁺ cells had no effect. Surprisingly, the B220⁺-depleted splenocytes produced significantly less IL-10 than the whole splenocyte population (Fig. 5,B). Furthermore, neither CD4⁺ T cells nor B220⁺ cells purified from spleens of anti–IL-12–treated donors produced IL-10 when cultured separately, whereas IL-10 production was restored when the two subpopulations were recombined (Fig. 5,C). CD4⁺ T cells from anti–IL-12–treated donors produced similar levels of IL-10 whether combined with purified B cells from control IgG or anti–IL-12–treated wild-type donors (Fig. 5,C), or with B cells from control IgG–treated IL-10−/− mice (Fig. 5,D). Conversely, CD4⁺ cells from control IgG-treated wild-type donors or anti–IL-12–treated IL-10−/− donors failed to produce detectable IL-10 when combined with B cells from any of the sources mentioned (Fig. 5, C and D).

The abrogation of the protective effects of anti–IL-12 by coinjection of anti–IL-10 (Fig. 3,B) and the increased incidence of EAE in IL-10−/− mice (Fig. 1) strongly suggested that IL-10 plays a downregulatory role in the generation of EAE effector cells. To directly demonstrate that the IL-10–producing CD4⁺ T cell, generated in the absence of antigenic priming, can downregulate the generation of EAE effectors, we performed an adoptive transfer study. SJL mice were injected with splenocytes from anti–IL-12– or control IgG–treated syngeneic donors and then immunized with bovine MBP according to the schedule shown in Fig. 1. Splenocytes from anti–IL-12–treated, but not control IgG–treated, SJL mice significantly inhibited the induction of EAE (Fig. 6) thereby directly demonstrating the biologic importance of the IL-10–producing CD4⁺ T cells whose presence is revealed when animals are treated with anti–IL-12.

These studies demonstrate a number of novel mechanisms by which cytokine production by the innate immune system controls an autoimmune response mediated by cells within the adaptive immune system. In the model systems we have used, IL-12 produced by macrophages and/or dendritic cells plays a critical role in the generation of autoimmune effectors, while IL-10 production by antigen nonspecific regulatory CD4⁺ T cells subserves a counterregulatory or suppressive function. During the induction of EAE in susceptible strains of mice, the homeostatic balance maintained between these antagonistic cytokines is upset to favor IL-12, the production of which is stimulated by mycobacterial components contained in the adjuvant. On the other hand, the enhanced susceptibility to EAE of IL-10−/− mice may be secondary to the loss of the dominant suppressive functions of this cytokine.

Although one might have predicted that IFN-γ would be indispensable for the generation and function of autoimmune effector cells, studies using anti–IFN-γ mAbs and IFN-γ−/− and IFN-γ–receptor−/− strains have uniformly demonstrated that IFN-γ is not required and in many cases exerts a protective effect in EAE (43, 45, 49, 51). It was therefore somewhat surprising that IL-12−/− mice were resistant to EAE and that treatment of IFN-γ−/− mice with anti–IL-12 prevented EAE. Furthermore, the generation of EAE effectors could be markedly inhibited by treatment of highly susceptible SJL mice with anti–IL-12 during the course of priming alone. Taken together, these results demonstrate for the first time that a cytokine implicated in the pathogenesis of EAE, namely IL-12, is indispensable for its manifestation, whereas the effector molecules IFN-γ, TNF-α, and LTα appear to be redundant (50, 52). These results should be contrasted with the requirements for both IL-12 and IFN-γ in mediating resistance to a number of intracellular pathogens including Toxoplasma gondii, Mycobacterium tuberculosis, and Listeria monocytogenes (69–71). Although IFN-γ–independent actions have been attributed to IL-12 in the past (72, 73, 57), in only one other case has such an action been reported to affect the pathogenesis of disease. Taylor and Murray (74) found that treatment of IFN-γ−/− mice infected with Leishmania donovani with exogenous IL-12 induced leishmanicidal activity and also partially restored the near-absent tissue granulomatous response. The action of IL-12 against L. donovani was TNF-α–dependent and involved upregulation of inducible nitric oxide synthase (iNOS). It was unclear from this study whether the protective effects of IL-12 were mediated by antigen-specific CD4⁺ T cells or by NK cells. Nevertheless, induction of the TNF-α/iNOS pathway may also be partially responsible for the IFN-γ–independent actions of IL-12 in EAE. We have detected high levels of mRNA encoding both TNF-α and iNOS in the spinal cords of actively immunized C57BL/6 IFN-γ−/− mice and markedly reduced levels after treatment with anti–IL-12 (unpublished data). Although treatment of mice with anti–TNF-α has been shown to prevent EAE, the contribution of TNF-α to the pathogenesis of demyelination has been difficult to define because the therapeutic effects of anti–TNF-α appeared to have been mediated by prevention of entry of pathogenic T cells into the CNS (32). It would therefore be of interest to assess whether treatment with anti–TNF-α and/or anti–LTα are able to protect IFN-γ−/− mice from disease at a time point when pathogenic effector cells have already entered the CNS. It would also be of interest to determine whether administration of recombinant TNF-α reverses the protection afforded by the anti–IL-12 treatment of actively immunized IFN-γ−/− mice.

Because it has previously been reported that IL-12−/− mice mount a Th2 response when infected with L. major, whereas wild-type mice mount a Th1 response (75), we intensively investigated whether anti–IL-12–treated SJL mice develop a Th2 response when immunized with MBP in CFA. Although no evidence for antigen-specific IL-4 production was observed, we consistently observed antigen-independent IL-10 production. The IL-10–producing population implicated has several unique properties that distinguish it from conventional Th2 memory cells. After anti–IL-12 treatment, it was as readily obtained from naive mice as from mice that had been previously immunized. The tonic inhibition of IL-10 production by the constitutive presence of IL-12 in vivo is an IFN-γ–independent activity as upregulation of IL-10 production was also seen when IFN-γ−/− mice were treated with anti–IL-12. Cell purification studies demonstrated that the IL-10 producing cell was a CD4⁺ T cell, but required coculture with B cells for induction of IL-10 production in vitro; we have not yet evaluated other antigen-presenting cell types for their potential to activate IL-10 production by these CD4⁺ T cells. The CD4⁺ IL-10–producing cell resembles the IL-4 producing CD4⁺, NK1.1⁺ cell in that it produces cytokines in the absence of priming and therefore appears to be a member of the newly recognized category of unconventional “natural” T cells in the innate immune system that guide adaptive responses through the production of Th1/2-modulating cytokines (76, 77). The induction of IL-4 production by CD4⁺NK1.1⁺ T cells in response to CD1d stimulation (78, 61) raises the possibility that a nonclassical MHC molecule may also serve as the ligand for the IL-10 producer and that the IL-10 producing T cell is responding to an autoantigen presented by the B cell. Currently we are conducting experiments to clarify the nature of the interaction between the B and T cells responsible for IL-10 production. It should be noted that we did not observe enhanced IL-10 production by spleen cells from IL-12−/− mice, but this may be secondary to adaptive processes that have taken place in the animal in the absence of IL-12.

The production of IL-10 by the CD4⁺ T cells is highly significant biologically as coinjection of anti–IL-10 largely reversed the autoimmune suppressive effects of anti–IL-12. Most importantly, IL-10 producing cells from anti–IL-12– treated naive mice markedly suppressed the induction of EAE after transfer into sensitized recipients. It is also likely that the increased incidence of EAE that we observed in IL-10−/− mice is due to the functional absence of this immunoregulatory cell. The concept that exposure of developing autoimmune effector cells to IL-10 hinders their development is supported by an earlier study in which EAE was suppressed by the administration of recombinant IL-10 to rats during priming (40). It remains to be seen if this CD4⁺ IL-10–producing T cell plays a broader role in immune surveillance by preventing organ-specific autoimmunity. The high incidence of spontaneous autoimmune phenomena in IL-10−/− mice suggest that this might be the case (79). The innate immune system appears to be biased towards the development of inflammatory immune responses as IL-10 production is constitutively suppressed by IL-12. Therefore, CD4⁺ IL-10–producing T cells may be needed to curtail overzealous responses to acute infections that would otherwise lead to immunopathology. The slightly delayed expression of IL-10, as compared to IL-12 and other proinflammatory cytokines, in several infectious disease models make it a particularly effective downregulator of the IL-12 response (58). The importance of this negative regulatory role of IL-10 is highlighted by the emergence of a lethal immune response, characterized by overproduction of IL-12, IFN-γ, and TNF-α, in IL-10−/− mice infected with T. gondii (80).

Thus far efforts to manipulate cytokine production to alter the course of autoimmune disease have been focused on modification of the adaptive immune response. Although positive results have been reported, these interventions were mainly effective when administered early in the evolution of the disease before the expression of clinical signs (39, 42, 64). Furthermore, recent reports of Th2-mediated autoimmunity suggest that approaches to induce immune deviation of Th1 effector cells may also ultimately result in immunopathology (54, 55). Collectively, our studies strongly suggest that manipulation of the cytokine milieu, the IL-12/ IL-10 balance, maintained by cells of the innate immune system, can have profound effects on the incidence of autoimmune disease. In chronic autoimmune diseases that progress as a result of the continuous reactivation of autoantigen-specific memory T cells, it would be unlikely that attempts to block IL-12 production or enhance IL-10 production would be therapeutically useful. On the other hand, relapsing remitting autoimmune disease may provide a unique opportunity for effective intervention employed during periods when new effectors, capable of initiating relapses, are recruited from a naive precursor pool. Studies documenting the phenomenon of determinant spreading in EAE and multiple sclerosis suggest that relapses are attributable in large part to the activation of naive T cells specific for cryptic epitopes during or after the peak of the initial episode (81–83). We have previously suggested that IL-12 production in response to an infectious insult during disease remission may lead to recruitment of Th1 cells (63); consideration should be given to the prophylactic use of IL-12 antagonists and/or IL-10 in patients with autoimmune diseases during epidemic outbreaks that have been associated with autoimmune sequelae.