IL-6 Regulates In Vivo Dendritic Cell Differentiation through STAT3 Activation¹

Before acute infection, i.e., during the steady state, the majority of dendritic cells (DCs)³ are in a resting/immature state and are not fully differentiated to perform their important role as inducers of immunity (1, 2, 3, 4, 5). DCs move through tissues, including lymphoid organs, capturing self Ags and environmental proteins. It has been demonstrated that Ag-loaded resting/immature DCs suppress T cell activation in vivo, either via promotion of T cell deletion or by causing expansion of the regulatory T cell pool (6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16). Therefore, it is thought that the immune system overcomes some of the risk of developing autoimmunity and chronic inflammation by using resting/immature DCs carrying self and environmental Ags.

When microbial invasion and tissue destruction occur, resting/immature DCs that have diverse recognition systems for monitoring danger signals become activated and undergo maturation (17, 18, 19). Activated/Mature DCs lose their Ag-capturing ability, but they acquire an Ag-presenting ability, and these DCs then migrate to the lymph nodes, where they stimulate T cells (20). Activated/Mature DCs also express high levels of MHC class II; costimulatory molecules such as CD86, CD80, and CD40; and cytokines such as IL-12 and TNF-α (4, 21, 22). Understanding what controls the in vivo differentiation from resting/immature DCs to activated/mature DCs is an important goal for vaccine development and for the prevention of unwanted immune responses, such as organ-specific autoimmune diseases.

Some evidence has suggested IL-10 as a candidate factor for controlling DC differentiation in vivo, because it inhibits the activation/maturation of DCs in vitro and its overexpression suppresses DC function in vivo (23, 24, 25, 26, 27, 28, 29, 30). However, no report describes the IL-10-mediated suppression of DC maturation; such information is needed to substantiate this idea, especially for the regulation of DCs in the steady state in wild-type (WT) animals.

Originally identified as a B cell differentiation factor (31, 32), IL-6 is now known to have pleiotropic effects on cell growth, differentiation, survival, and migration during immune responses, hemopoiesis, and inflammation (33). In addition, IL-6 affects the differentiation of myeloid lineages, including the macrophage and DC lineages, in vitro (34, 35). Moreover, a functional defect in DCs from a myeloma patient was traced to an overproduction of IL-6 (36). However, the effect of IL-6 on the transition between resting/immature and activated/mature DCs in vivo is currently unknown. The IL-6R consists of an α-chain and gp130, which is shared among the receptors for the other IL-6 family cytokines: LIF, ciliary neurotrophic factor, oncostatin M, IL-11, and cardiotrophin-1 (33). Binding of the IL-6 family cytokines to their receptors activates Janus kinases (JAK1, JAK2, and TYK2), leading to the recruitment of signal-transducing molecules such as Src homology region 2 domain-containing phosphatase 2 (SHP2) (protein tyrosine phosphatase 2) and STAT3. gp130 contains six tyrosines in its cytoplasmic region. Tyr⁷⁵⁹ is required for the tyrosine phosphorylation of SHP2, and any one of four tyrosines in the C terminus (Tyr⁷⁶⁷, Tyr⁸¹⁴, Tyr⁹⁰⁵, and Tyr⁹¹⁵), which have a glutamine at position +3 of the tyrosine motif (YXXQ), is required for the tyrosine phosphorylation of STAT3 (33). Although Tyr⁷⁵⁹ is required for the tyrosine phosphorylation of SHP-2 and the activation of ERK MAPK (33), it is also suggested to have a negative role in STAT3-mediated biological actions. We previously established a series of knockin mice that lack SHP-2 MAPK, the STAT3-dependent pathway, or both. We observed that a knockin mouse lacking the SHP-2 MAPK pathway (named the gp130^F759/F759 mouse) has splenomegaly and lymphadenopathy from an early age (37).

In this study, we analyzed the role of IL-6 on DC activation/maturation in vivo, using several mutant mice, including the IL-6 knockout (IL-6 KO) and gp130 knockin mice. We demonstrate that the DC differentiation process in vivo was affected by the activity of the IL-6 signaling pathway. IL-6 signaling increased the number of resting/immature DCs and decreased the number of activated/mature DCs in the lymph nodes; it also decreased the number of activated/mature DCs in the spleen, following LPS stimulation. In addition, we found that STAT3 was required for the IL-6-mediated suppression of LPS-induced DC activation/maturation. DC-dependent T cell activation was enhanced in IL-6 KO, but suppressed in gp130^F759/F759 mice, in which gp130-mediated STAT3 activation is enhanced. These data clearly reveal a novel in vivo function for IL-6 as a cytokine involved in the control of DC differentiation.

C57BL/6 and BALB/c mice were purchased from Japan SLC (Shizuoka, Japan); C57BL/6/Thy-1.1, C57BL/6 SJL, and BALB/c/DO11.10 mice from The Jackson Laboratory (Bar Harbor, ME); and C57BL/6 IL-10 KO, and C57BL/6 D^bK^b KO mice from Taconic Farms (Tokyo, Japan). The gp130^F759/F759 and gp130^FxxQ/FxxQ knockin mice that have a human version of gp130 (S710L) were established previously (37, 38) and backcrossed with C57BL/6 or BALB/c mice more than eight times. gp130^F759/F759/IL-6 KO mice were generated by crossing gp130^F759/F759 mice with C57BL/6 IL-6 KO mice that were established previously (39). C57BL/6/OT1 and C57BL/6/OT2 mice were kindly provided by W. Heath (Immunology Division, Walter and Eliza Hall Institute of Medical Research, Parkville, Australia) and crossed with C57BL/6/Thy-1.1. All of these mice were maintained at Institute of Experimental Animal Sciences at Osaka University Medical School.

To analyze the DC in gp130^FxxQ/FxxQ mice, which die before birth, we transplanted fetal liver cells from gp130^FxxQ/FxxQ mice at 14.5 days into lethally irradiated (950 rad) C57BL/6 SJL mice, as described previously (37).

Anti-CD8 (53-6.7), anti-CD3 (2C11), anti-CD11b (M1/70), anti-CD11c (HL3), anti-I-A^b (25-9-17), anti-CD40 (3/23), anti-CD80 (16-10A1), anti-CD86 (GL1), anti-IL-6Rα (D7715A7), anti-CD45.1 (A20), anti-CD45.2 (104), anti-Thy-1.1 (OX7), anti-phospho-STAT3 (p-STAT3), anti-KJ1.26, anti-CD40 (3/23), anti-IL-12 (p40/p70) (C15.6), anti-IL-10 (JE55-16E3), anti-TNF-α (MP6-XT22), PE streptavidin, PerCP streptavidin, and isotype controls were purchased from BD Pharmingen (San Diego, CA). The purified rat anti-mouse gp130 mAb (RX187) was provided by M. Saito (Cyugai, Tokyo, Japan) and labeled with FITC by a standard procedure. RPMI 1640 (Sigma-Aldrich, St. Louis, MO) supplemented with 10% FBS (EQUITECH-BIO, Kerrville, TX), 2-ME (50 μM), l-glutamine (2 mM), penicillin (100 U/ml), streptomycin (100 U/ml), and sodium pyruvate (1 mM) was used as the complete culture medium. Recombinant mouse GM-CSF and IL-6 were purchased from PeproTech (London, U.K.). Human rIL-6 was kindly provided by Ajinomoto (Tokyo, Japan). LPS (Escherichia coli serotype: 055:B5, 0111:B4) was purchased from Sigma-Aldrich. For intracellular cytokine staining, we used brefeldin A (Sigma-Aldrich; 2 μg/ml) and a BD Cytofix/Cytoperm Kit (BD Pharmingen) following the manufacturer’s instructions.

For the analysis of surface molecules, the spleen or superficial lymph nodes were harvested from mice, treated with collagenase D (1 mg/ml) in complete culture medium for 30 min at 37°C or room temperature, strained through 100-μm mesh (BD Falcon, Tokyo, Japan), and washed twice with cold plain RPMI 1640 medium. The washed cells were resuspended in cold MACS buffer (PBS containing 0.5% FBS and 2 mM EDTA) and treated with anti-CD11c Ab-coated microbeads (Miltenyi Biotec, Tokyo, Japan). CD11c⁺ cell-enriched samples were prepared with an AutoMacs device (Miltenyi Biotec). For the analysis of STAT3 phosphorylation, splenic and lymph node DCs were prepared, as described above, but were kept on ice throughout the experiment and were not treated with collagenase. Only anti-CD11c⁺ and anti-CD3⁻ cells were analyzed.

BMDCs were generated, as described (40). Briefly, bone marrow cells were cultured in complete culture medium containing 10 ng/ml GM-CSF or 0.5% GM-CSF supernatant (from murine GM-CSF Chinese hamster ovary cells; a gift from T. Sudo, Toray Silicon, Tokyo, Japan). The DC culture medium was changed every 2 days to remove nonadherent granulocytes, etc. Loosely adherent clustering cells were used on day 6 as resting/immature DCs. The purity of the CD11c⁺ DCs was routinely >80%. For some experiments, 6-day cultured BMDCs were stimulated with IL-6 (50 ng/ml), followed by LPS (100 ng/ml), TNF-α (40 ng/ml), or anti-CD40 Ab (40 μg/ml).

Cultured BMDCs were stimulated with IL-6 and/or LPS. Their total RNA was then isolated, and RT-PCR was performed, as described previously (37). Primers used for the analysis were: hypoxanthine phosphoribosyltransferase, 5′-CCTCCCATCTCCTTCATGACA-3′, 5′-GATTAGCGATGATGAACCAGGTT-3′; GM-CSFR α, 5′-GAGGTCACAAGGTCAAGGTG-3′, 5′-GATTGACAGTGGCAGGCTTC-3′; GM-CSFR β, 5′-TCTGAAGACTCTGCAGTGCT-3′, 5′-AGCACTGCAGAGTCTTCAGA-3′; suppressor of cytokine signaling (SOCS1), 5′-CGCCTGCGGCTTCTATTG-3′, 5′-CACGCTGAGCGCGAAGA-3′; SOCS3, 5′-GCGAGAAGATTCCGCTGGTA-3′, 5′-CGTTGACAGTCTTCCGACAAAG-3′.

Retroviruses carrying dominant-negative STAT3 (DN-STAT3) or the control vector were kindly provided by D. Link (Washington University, St. Louis, MO). For retroviral infections, 10⁶ Phoenix producer cells (provided by G. Nolan, Stanford University, Stanford, CA) were transfected with the vector DNA by the Lipofectamine 2000 reagent (Invitrogen Life Technologies, Grand Island, NY), according to G. Nolan’s instructions. The culture medium was replaced the next day and collected 24 h later. Culture supernatants containing packaged retrovirus were stored at −80°C until use. On days 3 and 4 after starting the bone marrow cell culture, the culture medium was removed and replaced with 1 ml of the virus supernatant containing 4 μg/ml polybrene (Sigma-Aldrich) and 1 ml of complete culture medium containing GM-CSF. This transduction procedure was repeated one more time.

The analysis of dying cell endocytosis was performed, as described previously (41), with the following small modification. For the preparation of splenocytes, CD11c⁻CD11b⁻ cells were negatively sorted using anti-CD11c and anti-CD11b microbeads. In addition, we used C57BL/6 D^bK^b KO instead of C57BL/6 β₂-microglobulin KO.

BMDCs were established from gp130^F759/F759 and WT mice. The cultured BMDCs were stimulated with IL-6 alone, LPS alone, or IL-6 plus LPS and seeded in triplicate with FITC-OVA (final concentration 10 μg/ml) in 2 ml of RPMI 1640 supplemented with 10% FCS in 24-well flat-bottom plates, and then cultured for 15 min in a CO₂ incubator. The percentage of phagocytic DCs was measured by a FACS device.

CD44^low T cells were purified from OT1 or OT2 mice using a nylon wool column and a MoFlo cell sorter. The resulting naive phenotype OT1 or OT2 T cells (2.5 × 10⁵) were cultured for 48 h with BMDCs (2.5 × 10⁴) that had been treated with or without IL-6, then cultured for 12 h with various concentrations of OVA protein (see Fig. 5, C and D). [³H]Thymidine incorporation was monitored for the last 8 h. For the allogenic T cell response, CD44^low T cells were harvested from BALB/c mice, and the resulting naive phenotype T cells (3 × 10⁵) were cultured with various numbers of BMDCs from C57BL/6 mice (see Fig. 5 E) that had been treated with or without IL-6 in the presence or absence of LPS, for 72 h. [³H]Thymidine incorporation was measured for the last 8 h.

T cells were negatively sorted by nylon wool (42) and anti-CD4⁺ or anti-CD8⁺ magnetic beads (Miltenyi Biotec). A total of 2 × 10⁶ Thy-1.1⁺OT1 T cells or 2.5 × 10⁶ DO11.10 T cells in PBS was i.v. injected into mice at a volume of 0.1 ml. One day later, for OT1 stimulation, the mice received 2.5 × 10⁷/0.2 ml OVA-loaded dying cells from MHC class I (D^bK^b) KO mice with 500 ng of LPS (Sigma-Aldrich), and for DO11.10 stimulation, CFA (Sigma-Aldrich) plus OVA peptide was injected at the base of the tail. The T cell number was then monitored by a flow cytometer using an anti-Thy-1.1 Ab for the OT1 or anti-KJ1.26 Ab for the DO11.10 mice.

BMDCs were starved of GM-CSF in culture medium containing 0.5% FBS for 12 h, and then stimulated with IL-6 (50 ng/ml) and/or LPS at 37°C for 30, 60, 180, and 720 min. The resulting cells were lysed, and Western blot analysis was performed, as described previously (37). For the detection, anti-p-STAT3 (New England Biolabs, Beverly, MA) and anti-STAT-3 Abs (New England Biolabs) were used.

It has been reported that IL-10 induces an inhibitory signal for DC maturation in vitro and that both IL-6 and IL-10 trigger the activation of STAT3 molecules (30, 37). Our hypothesis is that IL-6 regulates DC differentiation in vivo. To investigate the effect of IL-6 on DC differentiation in vivo, we analyzed the number of resting/immature and activated/mature DCs in superficial lymph nodes by observing the expression of MHC class II (class II) and CD86 on CD11c⁺ cells in different mouse strains. We prepared cell samples enriched in DCs from collagenase-treated lymph nodes using magnetic beads coated with an anti-CD11c Ab (N418). We then stained the resulting population with another anti-CD11c Ab (HL3), to investigate the differentiation of the CD11c⁺ DCs. We analyzed the surface markers on only the anti-CD11c Ab (HL3)-positive cells. The average percentages of CD11c⁺ cells were 50–60% at this point (Fig. 1 A).

It was recently demonstrated that CD11c⁺ DCs in the lymph nodes of control animals can be divided into three populations based on their CD205 and CD8 expressions (43). The majority are CD205⁺CD8⁻ cells, called epithelium-derived DCs, which express a high level of class II molecules (39). We observed that the cell numbers of the three populations in lymph nodes were almost the same in all mice, including the various mutant mice analyzed in this study, and that the majority were CD205⁺CD8⁻ cells (data not shown). We analyzed the CD11c⁺ DCs in lymph nodes from mice deficient in IL-6 (IL-6 KO) and others that had defective gp130-mediated signaling via SHP2 (gp130^F759/F759) or STAT3 (gp130^FxxQ/FxxQ), and mice that had both defective gp130-mediated SHP2 signaling and a lack of IL-6 (gp130^F759/F759/IL-6 KO) (see Materials and Methods). Although we observed comparable expression of gp130 and the other IL-6R subunit, IL-6Rα, on the DCs of all the mice (data not shown), we observed an enhanced expression of class II and CD86 in the DCs from IL-6 KO mice compared with the control C57BL/6 mice, indicating that IL-6 normally suppressed the activation/maturation of DCs in vivo (Fig. 1,A). Interestingly, the numbers of activated/mature DCs that expressed high levels of class II and CD86 decreased in the gp130^F759/F759 mice, which have enhanced gp130-mediated STAT3 activation, but not gp130-mediated SHP2 activation. In addition, DCs from reconstituted gp130^FxxQ/FxxQ mice, in which gp130-mediated STAT3 activation is defective, had more activated/mature DCs than reconstituted control mice did, and they had a similar phenotype to the DCs from IL-6 KO mice (Fig. 1,A). The suppression of class II and CD86 expression on DCs was recovered in the progeny of gp130^F759/F759 mice crossed with IL-6 KO mice (Fig. 1,A). These results further strengthen the idea that IL-6 plays a suppressive role in vivo. Moreover, they suggest that IL-6 suppresses DC activation/maturation through STAT3 activation in vivo. In contrast to IL-6 KO mice, the DCs from IL-10 KO mice unexpectedly had almost the same phenotype as the control mice (Fig. 1 A).

We next wished to examine the effect of IL-6 signaling on the DC transition step between the resting/immature and activated/mature states, but because we did not know which molecule or molecules normally trigger this step in the steady state in vivo, we used bacterial LPS to promote DC activation/maturation (44, 45, 46). We prepared CD11c-enriched DCs using magnetic beads coated with anti-CD11c Ab, and carefully analyzed the CD11c⁺ DCs by staining with another anti-CD11c Ab, as described in Fig. 1,A. The purity of the CD11c⁺dump⁻ population in each animal was ∼60% at this point (Fig. 1,B). Although there were many activated/mature DCs (class II⁺CD86⁺) in the lymph nodes (see Fig. 1,A), we observed that the majority of the splenic DCs were in the resting/immature state, as described previously (43) (Fig. 1,B). It has been demonstrated that epithelium-derived DCs that express a high level of class II molecules exist in lymph nodes, but not the spleen (43) (Fig. 1,B). In addition, there were comparable percentages of resting/immature and activated/mature DCs in the spleens of all of the mutant mice before LPS treatment (Fig. 1,B, upper column). Within 6 h of LPS treatment, DC activation/maturation was enhanced in the spleens of the IL-6 KO mice in vivo, when compared with the controls (Fig. 1,B). In addition, LPS-induced DC activation/maturation was suppressed in the gp130^F759/F759 mice, but enhanced in the gp130^F759/F759/IL-6 KO mice, as in the IL-6 KO mice (Fig. 1,B). Importantly, the DCs from the IL-10 KO mice had almost the same phenotype as the ones from the control mice (Fig. 1 B). These data demonstrated that IL-6 is one of the critical cytokines regulating the differentiation of DCs from a resting/immature state to an activated/mature phenotype in the spleen, at least during the initial stage following LPS injection.

To investigate the relationship between the IL-6-mediated regulation of DC differentiation and STAT3 activation, we analyzed the phosphorylation of STAT3 molecules in the DCs of superficial lymph nodes, where we observed both resting/immature and activated/mature DCs (Fig. 1,A). CD11c⁺ cells were harvested in the cold (4°C) without collagenase treatment, because we wanted to eliminate the possibility that the incubation of DCs in vitro modified the state of STAT3 phosphorylation. The phosphorylation of STAT3 molecules was evaluated by intracellular staining with an anti-pSTAT3 Ab and flow cytometry. We noticed there were p-STAT3⁺ and p-STAT3⁻ cells in the CD11c⁺ DCs of lymph nodes (Fig. 2,A). The majority of the p-STAT3⁻ cells were of the activated/mature phenotype, and the p-STAT3⁺ cells were resting/immature DCs (Fig. 2,A). We observed that most of the p-STAT3^low population consisted of CD205⁺CD8⁻ cells that might have been epithelium-derived DCs (data not shown). This observation suggested that IL-6-mediated signaling affects the phosphorylation of STAT3 molecules in DCs in vivo. To investigate this hypothesis, we analyzed the STAT3 phosphorylation in splenic DCs, because almost all of the DCs in the spleen were of the resting/immature phenotype in the steady state. The presence of p-STAT3⁺CD11c⁺ DCs was evaluated by flow cytometry, using the mean fluorescence intensity (Fig. 2,B). The STAT3 phosphorylation in splenic CD11c⁺ DCs isolated from IL-6 KO mice was lower and in those from gp130^F759/F759 mice was higher, compared with control mice (p < 0.025), suggesting that IL-6-mediated signaling was involved in the phosphorylation of STAT3 molecules in the splenic CD11c⁺ DC population (Fig. 2 B).

To investigate whether IL-6 acts directly on DCs rather than indirectly via another cell population in vivo, a pure population of DCs was prepared from bone marrow (BMDCs). We stimulated the BMDCs with IL-6 alone, LPS alone, or IL-6 plus LPS, and analyzed the expression of gp130, IL-6Rα, class II, CD86, and CD80. The expression of gp130 and IL-6Rα was altered very little by the stimuli (data not shown). The class II, CD86, and CD80 expression levels were strongly enhanced by LPS treatment, but suppressed by pretreatment with IL-6 (Fig. 3,A). In contrast, IL-6 did not inhibit the LPS-induced up-regulation of MHC class I molecules (Fig. 3,A), suggesting that IL-6 does not simply interfere with the LPS signaling pathway, which normally enhances the expression of these molecules via the TLR4. This hypothesis was confirmed by analyzing the cytokines produced by the BMDCs. We analyzed the IL-12, TNF-α, and IL-10 produced by the BMDCs after LPS, TNF-α, or anti-CD40 Ab stimulation in the presence or absence of IL-6 pretreatment. IL-6 treatment suppressed both the LPS- and TNF-α-mediated IL-12 production, but did not inhibit the LPS-induced TNF-α production (Fig. 3,B). In addition, IL-6 did not suppress the CD40-mediated production of IL-12 (Fig. 3,B). All of these results support the hypothesis that IL-6 does not simply interfere with the TLR signaling pathway. Because we did not observe IL-10 expression in the BMDCs after IL-6 stimulation (Fig. 3 B), we concluded that the IL-6 effect is not dependent on IL-10 molecules.

To investigate the importance of STAT3 activation for the IL-6-mediated inhibition of DC maturation/activation, BMDCs were prepared from gp130^F759/F759 and gp130^FxxQ/FxxQ mutants and control animals. As shown in Fig. 3, A and B, all of the BMDCs tested showed increased expression of class II and CD86, after LPS stimulation. However, pretreatment with IL-6 suppressed the LPS-mediated up-regulation of class II molecules in the BMDCs from both gp130^F759/F759 and control mice (Fig. 3,B). The inhibitory effect of IL-6 was greater for the BMDCs from gp130^F759/F759 mice than for those from control mice. In contrast, IL-6 had no inhibitory effect on the BMDCs from gp130^FxxQ/FxxQ mice (Fig. 3,B). These results suggested that the STAT3-, but not the SHP-2/Gab/MAPK-dependent pathway is critical for the IL-6-mediated suppression. Consistent with the result described above, we observed the prolonged phosphorylation of STAT3 molecules in the IL-6-treated gp130^F759/F759 BMDCs, compared with controls (Fig. 4,A). We also confirmed that IL-6 did not activate the STAT3 molecules in gp130^FxxQ/FxxQ BMDCs (Fig. 4,A). To confirm the direct involvement of STAT3 in the IL-6-mediated signaling for BMDC maturation/activation, we infected gp130^F759/F759 BMDCs with a retrovirus vector carrying a DN-STAT3 (46, 47). DN-STAT3 suppressed the IL-6-mediated inhibition of DC activation/maturation compared with control vector-infected BMDCs (Fig. 4 B). Together, these results clearly demonstrate that STAT3-mediated signaling through gp130 is crucial for the IL-6-dependent suppression of DC activation/maturation.

We conducted a series of experiments to characterize extensively the DCs from gp130^F759/F759 mice. We generated DCs from the bone marrow of gp130^F759/F759 and WT mice using GM-CSF. The yield of BMDCs on day 6 after starting the culture was almost the same for these mutant and WT mice (2.3–2.7 × 10⁶ CD11c⁺ cells per mouse without stimulation). We also confirmed the expression of SOCS1, SOCS3, and GM-CSFR (α and β) mRNA after treatment with IL-6 alone, LPS alone, and IL-6 plus LPS, because these molecules regulate the survival of BMDCs. The expression of GM-CSFR mRNA was not affected by the stimuli in the mutant or WT mice (Fig. 5,A). SOCS3 mRNA was induced by IL-6 and LPS stimulation, but SOCS1 mRNA remained at a very low level in both the WT and gp130^F759/F759 BMDCs (Fig. 5 A).

One of the most important functions of DCs in vivo is the ability to present Ag to T cells. Therefore, we analyzed the OVA protein uptake and the T cell stimulatory activity of BMDCs. We analyzed the endocytosis of BMDCs using FITC-OVA. The uptake of FITC-OVA increased in BMDCs derived from both WT and gp130^F759/F759 mice after IL-6 stimulation in either the presence or absence of LPS stimulation compared with BMDCs without IL-6 stimulation (Fig. 5,B). In addition, we observed that BMDCs from gp130^F759/F759 mice engulfed more FITC-OVA than did WTBMDCs after IL-6 treatment (Fig. 5,B). This result suggested that gp130-mediated STAT3 signaling enhances the uptake of OVA protein in BMDCs. We next analyzed the T cell stimulatory activity of BMDCs using T cell transgenic systems. We used CD4 and CD8 transgenic T cells OT1 and OT2, both of which recognize OVA protein. We observed that the BMDC-mediated CD4⁺ and CD8⁺ T cell responses were suppressed after IL-6 treatment in both WT (see Fig. 5, C and D; OVA concentration 1 mg/ml) and gp130^F759/F759 (see Fig. 5, C and D; OVA concentration 1–3 mg/ml) BMDCs. In addition, we analyzed the allogenic MLR using BMDCs derived from gp130^F759/F759 and WT mice of the C57BL/6 background. T cells from BALB/c mice were cultured with BMDCs from the C57BL/6 background (Fig. 5,E). The allogenic CD4⁺ T cell response was significantly inhibited by gp130^F759/F759 BMDCs that had been treated with IL-6 plus LPS compared with LPS alone (Fig. 5 E). All of these data demonstrated that the capability of DCs to present Ag was reduced after treatment with IL-6 in vitro.

To examine the functional significance of the IL-6 effect on DC differentiation, we tested the ability of DCs to present Ag to T cells in vivo. Because it has been reported that the presentation of Ag from dying cells is dependent on endocytosis of the cells into CD8⁺ DCs in vivo (41), we first investigated whether IL-6 affected the endocytosis of dying cells in the mutant mice. We prepared dying CD11c⁻CD11b⁻ splenocytes that were labeled with CFSE and i.v. injected them into IL-6 KO, gp130^F759/F759, and control mice. We observed that the CFSE-labeled dying cells were endocytosed equally well by the CD8⁺ DCs of IL-6 KO, gp130^F759/F759, and control animals, suggesting that IL-6-mediated signaling did not affect the endocytosis of dying cells into the CD8⁺CD11c⁺ DCs (Fig. 6 A). There was no significant difference between the samples by Student’s t test.

We then analyzed the Ag presentation ability of DCs in vivo. We prepared CD8⁺ T cells from C57BL/6/OT1/Thy-1.1 transgenic mice (see Materials and Methods), whose TCR recognizes the OVA peptide on D^b class I molecules. To exclude direct presentation of the OVA epitope on the D^b class I molecules of dying cells, but to see the indirect presentation of the epitope on the CD8⁺CD11c⁺ DCs of the mutant mice after endocytosing the dying cells, we prepared OVA-treated dying cells from the D^bK^b KO and i.v. injected them into IL-6 KO, gp130^F759/F759, and control mice 1 day after transferring the Thy-1.1⁺OT1 T cells (day 0). The actual number of OT1 T cells was calculated for days 2, 4, and 7. As shown in Fig. 6,B, the number of OT1 T cells was higher in the IL-6 KO, but lower in the gp130^F759/F759 mice, compared with control animals, especially on day 4. The differences for each animal group were significant by Student’s t test on day 4 (p < 0.05). Next, we investigated whether IL-6 affects the ability of DCs to present Ag to both CD8⁺ T cells and CD4⁺ T cells. We obtained BALB/c/DO11.10 mice (see Materials and Methods) and prepared CD4⁺ T cells, whose TCR recognizes the OVA peptide on I-A^d class II molecules. A similar result was obtained when we used the CD4⁺ T cells after stimulation with CFA plus OVA peptide, in BALB/c/gp130^F759/F759 (see Materials and Methods) and control mice (Fig. 6,C). The number of DO11.10 T cells decreased in the gp130^F759/F759 mice compared with control animals (Fig. 6 C). The differences for each animal group were significant by Student’s t test on day 4 (p < 0.025). The above observations are consistent with the hypothesis that the IL-6 regulation of DC activation/maturation represents a mechanism for controlling the efficiency of T cell stimulation for both CD8⁺ and CD4⁺ T cells in vivo.

In this study, we demonstrated that IL-6 regulates DC differentiation in vivo. We observed this phenomenon for both natural(unknown) triggers of activation/maturation in vivo, and for LPS-mediated stimulation. Even though we do not know what the natural activation/maturation signals are in vivo, we found that the IL-6-mediated modification of DC differentiation occurs in the lymph nodes at the steady state and in the spleen, particularly after experimentally boosting the activation/maturation signal with LPS (Fig. 1,B). We also observed that the IL-6 treatment needed to be applied before LPS stimulation to suppress BMDC maturation/activation (Fig. 3). Therefore, it is reasonable to assume that IL-6 continuously stimulates the resting/immature DCs in both the lymph nodes and spleen during the steady state in vivo.

Because it was reported that plasmacytoid DCs (pDCs) show an immature phenotype in vivo, we hypothesized that pDCs might be increased in the gp130^F759/F759 mice, which have excess numbers of immature DCs (Fig. 1,A). We analyzed the actual number of pDCs (CD11c⁺B220⁺Ly-6⁺) as well as conventional CD11c⁺B220⁻Ly-6⁻ DCs, T cells, B cells, and macrophages in the lymph nodes and spleen of WT and gp130^F759/F759 mice. We observed an increased number of both pDCs and conventional DCs in the gp130^F759/F759 mice (2- to 2.5-fold compared with control mice). However, we also noticed an increased number of cells in all of the populations we tested in the gp130^F759/F759 mice, at a similar ratio (2- to 3-fold compared with control mice). This was not surprising, because even before they are 6 wk old, the gp130^F759/F759 mice we used in this study have splenomegaly and lymphadenopathy, as we reported previously (36). In contrast, the IL-6 KO mice had almost the same number of cells in these populations compared with the WT control. Therefore, we suggest that the increased number of pDCs is not responsible for the high level of immature DCs in the gp130^F759/F759 mice, as shown in Fig. 1 A.

We mainly used MHC class II and CD86 molecules as the indicators of DC activation/maturation in vivo and in vitro. IL-6 stimulation clearly decreased the LPS-induced expression of not only MHC class II and CD86 molecules, but also of CD80, CD40, and IL-12 (p40/p70); however, IL-6 did not influence the LPS-mediated up-regulation of MHC class I and production of TNF-α molecules in vitro (Fig. 3). This phenomenon suggests that IL-6 does not directly suppress the main TLR-induced signaling pathway, but up-regulates the transcription of key molecules that affect the maturation of DCs. This idea is further supported by the following observations: 1) IL-6 did not inhibit TLR-induced IκB degradation, JNK activation, or ERK activation (data not shown); 2) both transcription and translation were needed for the IL-6-mediated suppression of BMDC activation/maturation (data not shown); 3) IL-6 barely induced SOCS1, which inhibits the TLR4 signal (48, 49) in both control and gp130^F759/F759 BMDCs (Fig. 5,A); 4) IL-6 suppressed not only the LPS-, but also the TNF-α- and peptidoglycan-mediated activation/maturation of BMDCs (Fig. 3,B, and data not shown); and 5) the dominant-negative form of STAT3 (50), a transcription factor, inhibited the IL-6-mediated suppression of LPS-induced maturation (Fig. 4). Therefore, we hypothesize that the IL-6-gp130-STAT3 signaling pathway induces one or more key molecules and modifies LPS-mediated DC differentiation.

Because IL-10 has been shown to inhibit DC maturation, and both IL-6 and IL-10 trigger the activation of STAT3 molecules (30, 38), our first hypothesis was that IL-6 also modifies DC activation/maturation in vivo. Although we observed that IL-6 played an important role in modifying DC differentiation in vivo, surprisingly, after a 6-h treatment of IL-10 KO mice with LPS, we did not find: 1) alteration of the DC phenotype in the lymph nodes, or 2) an enhancement of DC activation/maturation in the spleen (both monitored by class II and CD86 expression; Fig. 1, A and B). Furthermore, IL-6 stimulation did not induce IL-10 molecules in BMDCs (Fig. 3 B). These findings suggested that the IL-6-mediated inhibition of DC maturation/activation was not dependent on IL-10 secreted by the DCs themselves. These results were unexpected, given that many reports describe the IL-10-mediated suppression of DC maturation; however, these studies used in vitro induced DC systems (27, 28, 29, 30) or overexpression of the cytokine in vivo (24, 25, 26), or observed the IL-10-mediated regulation of inflammatory and immune reactions in vivo (23, 51, 52).

In this study, we demonstrated IL-6-mediated DC regulation in the steady state, but not in the state of inflammation and immune responses, when IL-10 must play a great role. The apparent role of IL-6 rather than IL-10 in regulating the DC state during the steady state may be due to local differences in the availability of each of these cytokines, because we observed that immature/resting DCs expressed both IL-10Rα and IL-6Rα molecules in the lymph nodes and spleen (data not shown). It is reported that IL-10 is highly expressed in privileged organs, where it may serve to prevent severe immune responses (53). Resting/Immature DCs in such organs in the steady state may be affected more by IL-10 than by IL-6. Therefore, we conclude in this study that IL-6, but not IL-10, is a dominant cytokine for regulating the DC and T cell state in nonprivileged organs, at least in the lymph nodes and spleen, in the steady state. This conclusion is supported by a previous report that STAT3 regulates T cell tolerance in vivo (54), and our observation that IL-6 signaling affected the STAT3 phosphorylation of DCs in the steady state (Fig. 2). We hypothesize that in vivo STAT3 activation, which is primarily regulated by IL-6 signaling in the steady state, modifies T cell responses, including activation (Fig. 5), through the alteration of DC differentiation. We currently do not know what cell population generates IL-6 in the steady state. The conventional DC population itself does not seem to be a good candidate, because a comparable amount of LPS-mediated maturation was induced in BMDCs from both IL-6 KO and control mice (data not shown).

The alteration of DC differentiation by IL-6 changed the T cell response in vivo (Fig. 5). We analyzed both CD4⁺ and CD8⁺ T cells and obtained similar results after DC-mediated stimulation. Excess amounts of IL-6-STAT3 signaling in the gp130^F759/F759 mice, in which we observed more resting/immature DCs (Fig. 1), weakened the Ag-specific CD4⁺ and CD8⁺ T cell responses. Conversely, the lack of IL-6-STAT3 signaling in the IL-6 KO mice strengthened the Ag-specific CD8⁺ T cell responses. Because we observed that the expression of MHC class I molecules was not affected by IL-6-induced signaling (Fig. 3), we believe that the alteration of both MHC class II expression and costimulatory molecules and/or cytokines such as CD86, CD80, CD40, and IL-12 in DCs influenced the T cell responses in the gp130^F759/F759 and IL-6 KO mice.

Taken together, our results show IL-6 to be a major cytokine involved in the control of DC differentiation in vivo, at least in affecting the transition from the resting/immature phenotype to the activated/mature phenotype, in the steady state. It may be more important that this modification of DC differentiation altered the magnitude of the T cell responses, as shown in Fig. 5. Increases in IL-6-induced signaling seem to induce decreases in the activation of T cells, suggesting that IL-6 acts as an immunosuppressive cytokine for DC differentiation. This is surprising, because IL-6 has been considered to be an inflammatory cytokine and is implicated in autoimmune diseases and inflammation (33, 55, 56). Our new finding that IL-6 suppresses DC activation/maturation supports the idea that cytokines act as a two-edged sword; although proinflammatory cytokines promote autoimmune disease and regulatory cytokines suppress it, this scheme is too simple to explain the mechanisms underlying autoimmunity (57). However, whatever role IL-6 plays, the results demonstrated in this study indicate that the IL-6-STAT3 pathway may represent a novel target for vaccine development and the prevention of unwanted immune responses, through the control of DC differentiation.

We are thankful to Dr. Brian C. Schaefer (Department of Microbiology and Immunology, Uniformed Services University of the Health Sciences) for his critical reading of the manuscript. We thank Dr. Daniel C. Link for providing the retrovirus carrying DN-STAT3. We also thank Ryoko Masuda and Ayako Kubota for their excellent secretarial assistance.