Th17 Cells Induce Th1-Polarizing Monocyte-Derived Dendritic Cells

This article is featured in In This Issue, p.985

Monocytes comprise >10% of circulating leukocytes in humans and ∼4% in mice (1). These cells rapidly infiltrate inflamed tissues and are renowned for their plasticity (2–4). Upon entering inflamed tissues, monocytes readily differentiate into inflammatory dendritic cells (DC) that share multiple characteristics with conventional DC and tissue-resident macrophages including F4/80 positivity and reduced CD11c and Ly6C expression (5, 6). Our previous studies indicate that inflammatory DC arise following direct interaction with CD4⁺ Th cells in humans (7). However, the role these cells play in inflammation has not yet been elucidated. Experimental mouse models of disease are ideally suited to study this question, because the key disease-causing Th cells can be generated in vitro or obtained from inflamed tissue and tested for their ability to induce the differentiation of monocytes into inflammatory DC. In the current study, we sought to investigate the biology and impact of Th cell–driven DC formation in experimental autoimmune encephalomyelitis (EAE), a mouse model of multiple sclerosis (MS) in which Th cells and circulating monocytes are known to play pathogenic roles.

In EAE, transfer of myelin-reactive Th cells robustly and reliably induces a cascade of events that results in autoimmune demyelination (8). Much is known about the roles of specific Th subsets in both the induction and progression of EAE. Autoreactive IFN-γ–producing Th1 cells or IL-17–producing Th17 cells can induce disease, whereas Th2 and regulatory T (Treg) cells are believed to be protective (9). Furthermore, mice lacking the Th1 transcription factor, Tbet, or the Th17 transcription factor, retinoic acid–related orphan receptor γt, have reduced EAE symptoms compared with control mice, demonstrating that both the Th1 and Th17 pathways are involved in EAE (10, 11). Interestingly, Hirota et al. (12) recently showed that the majority of IFN-γ–expressing Th1 cells within the inflamed spinal cord previously produced IL-17A, demonstrating the plasticity of the Th17 phenotype in EAE. However, the cell types responsible for mediating conversion of Th17 cells into Th1 cells in EAE and the significance of this process in disease resolution remain unclear.

In addition to Th cells, monocytes are thought to play an essential role in the development of EAE (13). Indeed, monocytes are often found juxtaposed with Th cells in the CNS of affected mice, and the extent of monocyte infiltration correlates strongly with the severity of disease (14). Within the inflamed CNS, activated Th cells express multiple molecules, including CD40L, IFN-γ, M-CSF, and GM-CSF, that enhance the Ag presentation and differentiation potential of monocytes. In fact, bone marrow (BM) chimera mice with myeloid cells deficient in the GM-CSFR are completely protected from EAE induction with adjuvants, often referred to as active EAE induction (15). Taken together, these findings suggest that factors expressed by CD4⁺ Th cells act on monocytes and facilitate EAE induction and progression.

Based on our previous observation that activated human Th subsets induce the formation of phenotypically distinct monocyte-derived DC (Mo-DC) (7), we hypothesized that activated murine Th subsets would behave in a similar manner. We report in this study that murine Th1, Th2, and Th17 cells are capable of inducing the formation of distinct Mo-DC, termed DC_Th1, DC_Th2, or DC_Th17, respectively. These DC_Th subsets have similar cell-surface phenotypes, but differ in their cytokine production and T cell–activating ability. Most strikingly, Th17 cells induce the formation of IL-12–producing DC_Th17 with potent Th1-polarizing capacity and the ability to convert Th17 cells into IFN-γ–expressing Th1 cells. Further, CD4⁺ T cells taken from the inflamed spinal cords of mice with Th17-mediated EAE also induce the formation of Mo-DC that secrete IL-12p70 and drive Th1 polarization. These findings suggest that such DC serve as a critical link in determining the balance between Th17 and Th1 polarization during immunity and EAE progression.

Wild-type (WT) C57BL/6, 2D2, CD40L knockout (KO), GM-CSFR KO, and IL-12p35 KO female mice were purchased from The Jackson Laboratory and used between 6 and 14 wk of age. OT-II TCR-transgenic Rag2^−/− mice were purchased from Taconic Farms. All mice were housed in an American Association for the Accreditation of Laboratory Animal Care–accredited animal facility and maintained in specific pathogen-free conditions. Animal experiments were approved and conducted in accordance with Stanford University Asia Pacific Laboratory Accreditation Cooperation #13605.

Six-week-old C57BL/6 B6 mice were placed on antibiotic-supplemented chow. One week later, they received 5 Gy total body irradiation and were injected via tail vein with 10 × 10⁶ syngeneic WT BM or BM from IL-12p35 KO mice. They were allowed to recover for 1 mo on antibiotic food before returning to a normal diet. Nine weeks later (13 wk after irradiation), reconstitution with IL-12p35 KO BM was confirmed by PCR on circulating blood cells.

Splenocytes were depleted with biotinylated anti-CD8α Ab (clone 53.6.7; BioLegend) followed by anti-biotin microbeads (Miltenyi Biotec). The remaining cells were cultured in complete media consisting of RPMI 1640 media (Life Technologies) with 100 U/ml penicillin, 100 μg/ml streptomycin, 2 mM l-glutamine, 50 μM 2-ME, and 10% FCS with 1 μg/ml anti-CD3 (BD clone 145-2C11) to generate bulk CD4⁺ T cells or in the addition of the following polarizing mixture: Th1, 10 ng/ml IL-12 (R&D Systems); Th2, 20 ng/ml IL-4 (R&D Systems), 20 μg/ml anti–IL-12 p40 (clone C17.8; eBioscience), and 20 μg/ml anti–IFN-γ (clone R4-6A2; eBioscience); and Th17, 1 ng/ml recombinant human TGF-β1 (R&D Systems), 20 ng/ml IL-6 (R&D Systems), and 10 μg/ml anti–IFN-γ. T cells were positively selected from these cultures after 96 h with Thy1.2 magnetic beads (Miltenyi Biotec). In some experiments, IL-17A–secreting Th17 cells were sorted to high purity using an FACSAria II (BD Biosciences) and IL-17A cytokine capture reagents (Miltenyi Biotec).

BM from femurs, tibias, and hips was prepared after cleaning the bones and grinding in a mortar and pestle. Monocytes were negatively selected to ∼90% purity via the murine monocyte enrichment kit (Stem Cells Inc). Monocytes were cultured for 5 d in the presence of 50 ng/ml GM-CSF (PeproTech) plus 20 ng/ml IL-4 (PeproTech) to generate DC or in 50 ng/ml M-CSF (Sigma-Aldrich) to generate macrophages.

Polarized T cells were purified after 96 h of culture using Thy1.2 positive selection (Miltenyi Biotec) and cultured at a ratio of 1 T cell/10 monocytes in the presence of 10 ng/ml rmIL-2 (R&D Systems) for 5 d in tissue culture–treated plates (Corning). At this point, some cocultures were imaged using a Leica DM IRB inverted microscope and Metamorph software (Molecular Devices). In some experiments, cells were activated via the addition of 1 μg/ml standard LPS (TLR2 and -4 agonist), 1 μg/ml ultra-pure LPS (TLR4 agonist), 200 ng/ml PAM3CSK4 (TLR1/2 agonist), 10⁸/ml heat-killed Listeria monocytogenes (HKLM; mainly TLR2 agonist), 1 μg/ml imiquimod (TLR7 agonist), 1 μg/ml polyinosinic-polycytidylic acid (TLR3 agonist), or 10 μg/ml CpG ODN 2336 (TLR9 agonist) (Invivogen). DC were then purified using CD11c-positive selection. Macrophages and monocytes were selected using CD11b-positive selection (Miltenyi Biotec).

Purified DC_Th were pulsed with 2.5 μg/ml MHC class II (MHC II)–restricted ISQ peptide (OVA 323–339; New England Peptides) or 2.5 μg/ml MHC II–restricted MOG peptide (33–55) in complete media for 90 min at 37°C. In some experiments, 1 μg/ml LPS was added at the time of pulsing. Cells were then washed twice in PBS and cultured with CD4⁺CD62L⁺ T cells from OT-II or 2D2 mice purified by CD4-negative selection and CD62L-positive selection (Miltenyi Biotec). After 3 to 4 d, cells were pulsed with 1 μCi/well [³H]thymidine and cultured for an additional 18 h before being harvested by Harvester 400 (Tomtec). Radioactivity was measured by a 1450 MicroBeta counter (LKB Wallac).

Cells were blocked with FC block (BioLegend) and then stained with fluorochrome labeled Abs to: CD4, CD11b, CD11c, Ly6c, Ly6g, CD40, CD86, CD209, MHC II, CD45, Thy1.2, IL-12p40, IFN-γ, IL-4, and IL-17A (from BioLegend, BD Biosciences, and eBioscience) and analyzed using an LSRII (BD Biosciences) and FlowJo analysis software (Tree Star). For intracellular cytokine staining, T cells were stimulated with PMA and ionomycin (Life Technologies) for 4 h in the presence of brefeldin A (BFA) (BD Biosciences) and then fixed and permeabilized with appropriate buffers (eBioscience).

Ab pairs for IL-12(p70/p40), IL-23(p19/p40), IFN-γ, IL-4, IL-17A, IL-17A/F, and GM-CSF (eBioscience) were used to detect cytokine in cell-free supernatants.

Differential interference contrast (DIC) images were acquired on cells in coverslip-bottom dishes using a Zeiss LSM 700 using Zen software (Zeiss). Bright-field images were acquired on cells in tissue culture–treated plates using a Leica DM IRB and Metamorph software (Molecular Devices).

Mice were perfused with PBS, and the spinal cords were fixed and then snap frozen in O.C.T. compound (Tissue Tek). Slides were cut to 10 μM, fixed in acetone, and stained overnight with FITC rat anti-mouse CD11b (M1/70; BD Biosciences), biotin hamster anti-CD11c (HL-3; BD Biosciences), and rabbit polyclonal anti-CD3 (Vector Laboratories). After washing, the slides were stained with donkey anti-rat A488, goat anti-hamster A647 and donkey anti-rabbit A594 (Life Technologies). After incubating in DAPI (Life Technologies)/PBS, the cells were mounted in permafluor (Thermo Scientific), imaged with a Leica DM2500 laser scanning confocal microscope, and analyzed using Leica LAS AF software (Leica Microsystems).

C57BL/6 female mice were injected with 200 μg MOG peptide aa 35–55 (Stanford Core Facility) emulsified in CFA (Sigma-Aldrich). Mice were injected on days 0 and 2 with 400 ng pertussis toxin (Sigma-Aldrich). On day 12 after injection, splenocytes and lymph node cells were frozen. Cells were thawed when needed and cultured in 10 μg/ml MOG plus 10 ng/ml IL-23 for 3 d to skew toward a Th17 phenotype prior to i.p. transfer into naive mice to induce passive EAE (16). Disease activity was scored on a 0–5 scale using standard methodology (17).

At various times after passive EAE induction, mice were euthanized and then perfused with PBS to remove circulating leukocytes from the spinal cord. The spinal cord was then removed and incubated in 2 mg/ml collagenase 4 (Worthington) and 10 U/ml DNAse I (Sigma-Aldrich) in PBS for 45 min at 37°C. The cells were then layered over a 70% Percoll (GE Healthcare) gradient to enrich for immune cells. To further enrich for Th cells, a CD4 MACS negative enrichment kit was used (Miltenyi Biotec).

An unpaired Student t test (two-tailed) with a 95% confidence interval was performed in Prism (GraphPad) to analyze all experimental data unless otherwise stated. Error bars represent ± SEM; *p < 0.05, **p < 0.01, and ***p < 0.001.

To determine whether activated murine Th cells can induce monocytes to differentiate into DC, we established in vitro monocyte and Th cell cocultures. Bulk CD4⁺ T cells were activated with an anti-CD3 Ab for 96 h (Th_Bulk), then isolated with anti-Thy magnetic beads, and cultured at a 1:10 ratio with freshly isolated autologous BM monocytes. Monocytes cultured alone, or with exogenous M-CSF or exogenous GM-CSF and IL-4 (GM IL-4 DC), served as controls. After 5 d of coculture, we imaged the cells with DIC microscopy (Fig. 1A). Monocytes cultured in complete media alone displayed poor viability. However, monocytes cultured with activated bulk CD4⁺ T cells were viable and showed similarities to both GM IL-4 DC and M-CSF macrophages in their size, shape, and granularity. They could easily be distinguished from T cells based on their larger size and dendritic morphology (Supplemental Video 1). Monocytes cultured with activated T cells displayed more surface MHC II, CD11c, CD86, and CD40 and less Ly6C than freshly isolated monocytes (Fig. 1B). Based on their morphology and surface phenotype, these Th_Bulk-induced, monocyte-derived cells resemble DC, and on this basis, we refer to them as DC_Th.

We next assessed whether DC_Th can present Ag and activate T cells, which is a defining property of DC. DC_Th were purified from our cultures using CD11c MACS positive selection, pulsed with an MHC II–restricted OVA peptide in the presence of LPS, and cultured with TCR-transgenic responder CD4⁺ T cells (OT-II). Consistent with their surface phenotype, DC_Th were ∼10-fold more potent at T cell activation than fresh monocytes. However, DC_Th were far less potent at stimulating T cell proliferation than GM IL-4 DC (Fig. 1C). To verify that DC_Th-stimulatory ability was not restricted to OVA peptide presentation, we pulsed DC_Th and fresh monocytes with an Ag relevant to EAE, MOG peptide (aa 35–55), and tested their T cell–activating capacity across much broader APC/T cell ratios in an Ag-specific T cell proliferation assay with naive MOG-reactive T cells from 2D2 mice (Fig. 1D). Under these conditions, DC_Th induced T cell proliferation at low APC/T ratios and were superior to monocytes at all ratios tested. These experiments confirm that activated CD4⁺ T cells are capable of inducing the differentiation of monocytes into functional DC.

To determine the ability of particular Th subsets to induce DC_Th formation, we polarized T cells into Th1, Th2 or Th17 cells in vitro, using standard combinations of cytokines, blocking Abs, and anti-CD3 activating Ab. T cell polarization was confirmed by intracellular staining (Fig. 2A) and ELISA of culture supernatants (Supplemental Fig. 1A). As expected, Th1 cells expressed high levels of IFN-γ, whereas Th2 cells expressed IL-4, and Th17 cells expressed IL-17A/F. In contrast to IL-17A⁺ cells in vivo (9) but in accord with IL-17A⁺ cells generated in vitro (17), our in vitro–polarized Th17 cells produced less GM-CSF upon initial polarization and restimulation than the other Th subsets (Supplemental Fig. 1B). Similar to Th_Bulk cells, activated Th1, Th2, and Th17 cells cocultured with monocytes promoted the formation of cells with dendritic morphology (Fig. 2B, Supplemental Video 2), which we refer to as DC_Th1, DC_Th2, and DC_Th17, respectively. All DC_Th subsets upregulated MHC II, CD86, and CD40 compared with fresh monocytes (Fig. 2C). Interestingly DC_Th2 had lower expression of the costimulatory molecules, CD40 and CD86, as compared with DC_Th1 and DC_Th17.

DC_Th1, DC_Th2, and DC_Th17 were purified, pulsed with OVA peptide and LPS, extensively washed, and then cultured with OT-II T cells to evaluate their functional properties. After 3 d in culture, considerable T cell activation and proliferation were observed in the DC_Th1 and DC_Th17 wells, but not in the DC_Th2 wells. DC_Th17 induced significantly more T cell proliferation than DC_Th1 (Fig. 2D), and both DC_Th1 and DC_Th17 induced significantly more T cell proliferation than macrophages or fresh monocytes. Although DC_Th2 and fresh monocytes induced detectable levels of T cell proliferation, it was minimal compared with DC_Th1 and DC_Th17. These observations show that activated Th1, Th2, and Th17 T cells promote the development of Mo-DC capable of presenting Ag and activating T cells, but only DC_Th1 and DC_Th17 are potent APC, as gauged by their ability to induce T cell proliferation.

DC production of IL-12p70 promotes the development of IFN-γ–producing Th1 cells (18). Although a Th1 response can be beneficial in the clearance of tumors and intracellular pathogens, it can be harmful in the setting of autoimmunity (19, 20). In EAE, IL-12p70 and Th1 cells are not required to initiate disease, as IL-12–specific p35 KO mice that cannot produce biologically active IL-12p70 are still susceptible to active EAE induction (21, 22). Nonetheless, IFN-γ–producing Th1 cells can contribute to inflammation (23) and are sufficient to induce EAE on their own in transgenic T cell transfer models. Surprisingly, even though Th1 cells can drive inflammation, there is evidence to suggest that a Th1 response is also important for disease resolution in EAE (24).

To determine whether the DC_Th subsets differ in their capacity to produce IL-12p70, we activated each DC_Th subset with standard LPS for 24 h and then analyzed the supernatants for the presence of this cytokine (Fig. 3A). All DC_Th subsets produced minimal IL-12p70 in the absence of stimulation. However, after stimulation with LPS, DC_Th17 produced significantly more IL-12p70 than either DC_Th1 or DC_Th2. Remarkably, DC_Th17 produced ∼10 times more IL-12p70 than commonly studied GM IL-4 DC. No intracellular IL-12p40 was detected within the Th17 cells present in the cocultures, indicating that DC_Th17 were the sole source of IL-12p70 (Supplemental Fig. 1C).

To address the question of whether stimuli other than LPS can induce DC_Th17 to produce IL-12p70, we stimulated DC_Th1, DC_Th2, or DC_Th17 cells with a panel of TLR agonists including Pam3CSK4 (TLR1/2), HKLM (primarily TLR2), polyinosinic-polycytidylic acid (TLR3), ultra-pure LPS (TLR4), imiquimod (TLR7), or CpG ODN2336 (TLR7/9) (Fig. 3B). DC_Th17 responded robustly to most of these stimuli by producing IL-12p70. Notably, Pam3CSK4 and HKLM, both TLR2 agonists, induced high levels of IL-12p70 secretion. Ultra-pure LPS, CpG, and, to a lesser extent, imiquimod also induced detectable amounts of IL-12p70. Remarkably, the ability to produce IL-12p70 was limited to DC_Th17, as DC_Th1 and DC_Th2 cultured with the same TLR agonists failed to respond by secreting IL-12p70, even though they clearly became activated in response to TLR stimulation (Supplemental Fig. 1D). Thus, although all DC_Th subsets can respond to TLR stimulation, DC_Th17 are uniquely programmed to produce high levels of IL-12p70 and do so in response to a variety of danger signals.

Many recent studies have focused on the role of IL-23 in maintaining the polarization of Th17 cells. In fact, IL-23 is critical for EAE induction in the traditional MOG/CFA model, but is not necessary to maintain disease once EAE has been induced via passive transfer of T cells (25). DC_Th subsets were assessed for IL-23 production by ELISA (Fig. 3C). As with IL-12p70, very little IL-23 was detected without stimulation. However, after LPS stimulation, both DC_Th17 and GM IL-4 DC produced large amounts of IL-23. Taken together, these results demonstrate that DC_Th, particularly DC_Th17, can respond to many different inflammatory mediators by producing both IL-12 and IL-23.

We next wanted to explore why Th17 cells, but not Th1 or Th2 cells, induced DC with the ability to produce large amounts of IL-12p70. Although Th17 cells express a unique set of cytokines and effector molecules such as IL-17A, IL-17F, and IL-22, these molecules are not known to enhance IL-12p70 production. In contrast, the TNF-α family of proteins contains multiple molecules that influence cytokine production by myeloid cells (26). Two such molecules, CD40L and RANKL, are also highly expressed by activated Th cells (27, 28). We stimulated DC_Th17 with LPS in the presence of blocking Abs to CD40L or RANKL or isotype control Ab (Fig. 4A). CD40L blockade resulted in a nearly complete lack of IL-12p70 expression as assayed via ELISA. RANKL blockade resulted in a highly significant partial reduction of IL-12p70 expression. We confirmed these findings with intracellular staining of the same cells using an Ab to IL-12p40 (Fig. 4B).

To determine if Th17 cells from CD40L KO mice are also impaired in their ability to generate IL-12p70 competent DC_Th17, we generated Th17 cells from WT or CD40L KO mice and cultured them with monocytes for 5 d and then stimulated the cocultures with LPS overnight (Fig. 4C). In line with data from the Ab-blocking experiments, IL-12p70 levels were significantly reduced in the supernatant from monocyte/CD40L KO Th17 cocultures. Similar experiments using Th cells from RANKL KO were considered, but dysregulated lymphoid cell development in these animals would have confounded such experiments (29).

Finally, we wanted to determine whether Th17-skewed cells express more CD40L and/or RANKL compared with the Th1 and Th2 skewed cells. To test this idea, we restimulated polarized Th cultures with CD3/CD28 microbeads after a 2-d rest in IL-2, which was necessary to maintain Th1 and Th2 cell viability (data not shown). Th17 cell cultures contained ∼50% more CD40L⁺ T cells than Th1 or Th2 cultures (Fig. 4D). In addition, 52% of the cells in the Th17 cultures were RANKL⁺ compared with 36% in Th1 cultures and 25% in Th2 cultures. These data demonstrate that Th17 cultures contain large numbers of CD40L- and RANKL-positive cells and that these molecules are necessary for maximal IL-12p70 expression in T cell–induced monocyte-derived DC.

Because stimulated DC_Th17 produce large amounts of IL-12p70, we hypothesized that they would be superior at inducing Th1 polarization. To test this, we pulsed the DC_Th subsets with OVA peptide and LPS and, after extensive washing, cultured purified DC_Th subsets at a 1:15 ratio with naive OT-II T cells. After 96 h of culture, intracellular staining for IFN-γ, IL-4, and IL-17A was performed (Fig. 5A). No IL-4 or IL-17A staining was detected above isotype control levels (data not shown). However, in accordance with the IL-12p70 production data, large numbers of T cells expressing high levels of IFN-γ were seen in the cultures containing DC_Th17. IFN-γ was detected in a small percentage of the T cells cultured with DC_Th1, but these cells expressed ∼2.5-fold less IFN-γ on a per-cell basis than T cells cultured with DC_Th17.

Under our experimental conditions, DC_Th subsets did not induce Th17 polarization from naive CD4⁺ T cells. However, we hypothesized that DC_Th may influence the stability of the Th17 phenotype in polarized cells. To investigate this possibility, we generated Th17 cells from OT-II RAG KO mice under polarizing conditions, rested the cells for 2 d in media in the absence of IL-2, and then cultured them with peptide-pulsed or unpulsed DC_Th1 or DC_Th17. After 4 d, the T cells were stimulated with PMA and ionomycin in the presence of BFA and then stained intracellularly for IL-17A (Fig. 5B and quantified in Supplemental Fig. 1E). Compared to Th17 cells cultured alone, IL-17A production capacity by Th17 cells was better maintained when the Th17 cells were cocultured with DC_Th1 or DC_Th17, and this effect was further enhanced in cultures treated with LPS. An identical trend was observed when the Th17 cells were rested in the presence of TGF-β, which has been reported to be necessary for Th17 maintenance in vitro (data not shown) (30). Interestingly, the addition of OVA peptide to the DC_Th resulted in the appearance of IFN-γ–producing T cells in the DC_Th17 culture, but not in the DC_Th1 culture, suggesting that DC_Th17 can promote reprogramming of Th17 cells into Th1 cells in an Ag-specific manner.

The possibility exists that the IFN-γ–producing cells in our cocultures were derived from holdover naive cells that did not become polarized to a Th17 phenotype during the initial coculture. To address this possibility, we polarized OT-II cells toward a Th17 phenotype and then FACS purified the IL-17A–secreting Th17 cells by using cytokine-capture cell sorting. These highly purified cells, which by definition produced IL-17A, were rested and then cultured with Ag and LPS-pulsed DC_Th1 or DC_Th17. In accord with the results from Fig. 5B, a significant fraction of these cells began to produce IFN-γ when cultured with DC_Th17, but not DC_Th1 (Fig. 5C, 5D). Taken together, these observations indicate that DC_Th17, and to a lesser extent DC_Th1, promote Th1 polarization from naive T cells, but in the presence of an Ag recognized by the T cells, DC_Th17 have the unique capacity to convert previously polarized Th17 cells into Th1 cells.

We induced EAE using a Th17-skewed passive-transfer model that enables investigation of the biology of EAE in the absence of confounding factors such as CFA and pertussis toxin. In this model, lymphoid cells from actively immunized mice are cultured in MOG and IL-23 to skew the cells toward a Th17 phenotype (Supplemental Fig. 1F), and then the polarized cells are transferred to naive recipients to induce EAE. Spinal cords were harvested from diseased mice at various time points, and the surface phenotype of the infiltrating cells was analyzed by flow cytometry. Other than resident microglia, immune cells were largely absent from the spinal cords of presymptomatic mice, but rapidly accumulated at disease onset and remained abundant throughout the course of disease (Fig. 6A). As expected, numerous CD4⁺ T cells infiltrated the spinal cord at the onset of disease, and these cells remained numerous throughout disease progression (Fig. 6B). CD45^hi, Ly6g⁻, CD11b⁺, CD11c⁻ monocytes were present at onset, increased at peak disease, and then decreased at recovery. CD45^hi, CD11b⁺, CD11c⁺ DC increased after onset and remained at high frequency during EAE remission. Interestingly, Ly6g⁺CD11b⁺ neutrophils accumulated with increased disease severity and then rapidly dissipated with recovery.

To determine if monocytes, DC, and T cells are associated with one another in the EAE lesions, we performed immunofluorescence staining on spinal cord tissue from Th17-skewed passive-transfer EAE at peak disease or healthy control mice (Fig. 6C). No immune cells were visualized in the healthy WT control spinal cord tissue. In contrast, multiple perivascular lesions were seen in the white matter of the EAE spinal cords containing CD11b⁺ cells (granulocytes and monocytes), CD11b⁺/CD11c⁺ cells (DC), and CD3⁺ cells (T cells). Upon further magnification (Fig. 6D), all three cell types could be seen in direct contact with one another. These data show that monocytes, T cells, and DC colocalize in EAE lesions and are consistent with the hypothesis that activated T cells drive the formation of DC_Th in vivo.

To determine if Th cells from EAE spinal cords can promote DC_Th formation, we purified CD4⁺ T cells from Th17 passive-transfer EAE mice and cultured them with BM monocytes from healthy mice for 4 d. As a control, we used splenic CD4⁺ T cells from healthy mice, as almost no T cells are present in healthy spinal cords and because splenocytes from EAE mice may contain recirculating encephalitogenic T cells. We performed this coculture in the absence of specific Ag or other stimuli, as the CNS-infiltrating T cells were already activated. However, as with the in vitro–activated Th cells, the addition of exogenous IL-2 was necessary to maintain T cell survival (data not shown). We observed by microscopy that monocytes cultured alone or with splenic T cells retained a rounded morphology and displayed poor viability (Fig. 7A). In contrast, monocytes cultured with EAE CD4⁺ T cells not only remained viable but also became considerably larger and more granular and developed numerous active dendrites, reminiscent of GM IL-4 DC and DC_Th, which could be visualized with time-lapse microscopy (Supplemental Video 3). The monocytes cultured with EAE CD4⁺ T cells also upregulated their surface expression of CD11b, CD11c, MHC II, CD40, and CD86 compared with monocytes cultured with healthy splenic CD4⁺ T cells (Fig. 7B). The cells also expressed CD209, a marker of Mo-DC found in inflammatory settings (6). Interestingly, the cells retained high Ly6C expression, reminiscent of Mo-DC recovered from the CNS in transfer experiments (13). This phenotype closely resembled that of DC_Th, and on this basis, we refer to these EAE Th cell–induced cells as DC_ThEAE.

Th cell production of GM-CSF has been implicated as a driver of pathogenesis in EAE (15, 31). In line with these studies, CNS-infiltrating Th cells produced GM-CSF ex vivo when cultured with monocytes and IL-2 (Supplemental Fig. 1G). To determine whether GM-CSF was involved in the formation of DC_ThEAE, we cultured CNS-infiltrating Th cells with monocytes from WT or GM-CSFR KO mice. Unexpectedly, signaling through the GM-CSFR on monocytes was not necessary for the induction of a DC_Th phenotype (Supplemental Fig. 1H).

To determine if DC_ThEAE could take up and present Ag to naive T cells, we used these cells in an Ag-specific T cell proliferation assay (Fig. 7C). As their phenotype would suggest, Mo-DC generated by EAE Th cells, but not monocytes cultured with splenic Th cells, were capable of robustly stimulating naive OT-II T cells when pulsed with OVA peptide.

As this EAE model was induced by Th17-skewed cells, we wanted to determine if DC_ThEAE behave like DC_Th17 in their ability to promote a Th1 response. To address this possibility, we purified DC from monocyte/EAE CD4⁺ T cell cocultures and examined IL-12p70 production following stimulation with ultra-pure LPS. Similar to DC_Th17, DC_ThEAE produced substantial amounts of IL-12p70 upon stimulation (Fig. 7D). IL-12p40⁺ DC were also detected in the spinal cords of EAE mice by IC staining in the absence of exogenous stimulation (data not shown). Moreover, when pulsed with LPS and OVA peptide and cultured with naive OT-II T cells, DC_ThEAE-polarized naive T cells to an IFN-γ⁺ Th1 phenotype as indicated by ELISA (Fig. 7E). Neither IL-17A nor IL-17A/F was detected in these same cultures (data not shown). Thus, encephalitogenic CD4⁺ T cells from the inflamed spinal cord of Th17-skewed passive transfer EAE mice induce the formation of Mo-DC with Th1-polarizing capacity.

To investigate the role of myeloid cell–derived IL-12 in vivo during EAE, we used BM chimeric mice. EAE was induced in IL-12p35 KO and WT BM chimeras via passive transfer of Th17-skewed encephalitogenic cells (Fig. 7F). Both groups of mice began to show signs of tail paralysis by day 10. However, recipients of IL-12p35 KO BM displayed milder symptoms early (days 13–15), yet were impaired in their ability to recover from peak disease (days 17–20). Although these trends did not reach statistical significance due to the low numbers of mice used (n = 4 to 5), the data support the hypothesis that myeloid-derived IL-12 contributes to disease pathology but is also important for suppressing a pathogenic Th17-mediated response in favor of a milder Th1-mediated response.

It is well documented that monocytes infiltrate inflamed tissues and differentiate into inflammatory DC (5, 6, 32). However, the physiologic mechanisms responsible for their formation and subsequent function during immunity and inflammatory disease are poorly understood. The present study demonstrates that activated Th cells can govern inflammatory DC formation and function.

The present study also demonstrates that activated CD4⁺ T cells found within EAE spinal cords elicit the differentiation of monocytes into DC, which suggests that this system is active in autoimmune disease. This finding is consistent with published studies of EAE indicating that T cell–derived molecules including CD40L, GM-CSF, M-CSF, FLT3 ligand, and CCL2 acting on infiltrating myeloid cells are key drivers of disease (13, 15, 33–35). Further, the phenotype and functions of DC_Th are influenced by the type of Th cells in the inflammatory milieu. Our data show that DC_Th1 and DC_Th17 are proficient APCs capable of presenting Ag and stimulating naive T cells. Interestingly, although Th2 cells caused the differentiation of monocytes into cells that resembled DC_Th, the resulting DC_Th2 actually had very little T cell–stimulating capacity, which might be explained by their lower CD86 expression. These findings support previously published work showing that only Th1 and Th17, but not Th2 cells, can induce EAE (9).

The cytokines produced by the various DC_Th subsets differed substantially. In contrast to DC_ThBulk, DC_Th17 produced large amounts of IL-12 and IL-23 after LPS stimulation. Importantly IL-12 was secreted from DC_Th17 after stimulation with a variety of different TLR agonists, suggesting that these cells produce IL-12 as a generalized response to danger. The induction of IL-12 in DC_Th17 was highest when the cells were stimulated with TLR2 and -4 agonists. This may be important in the context of autoimmunity because multiple endogenous TLR ligands, including products released from necrotic cells, have been shown to signal through these receptors (36, 37).

The capacity of Th17 cells to recruit and activate neutrophils is well established and can lead to the destruction of pathogens as well as injury to peripheral tissue (38). Our time-course analysis of the CNS in Th17-skewed passive transfer EAE shows that recovery from disease correlates strongly with a reduction of infiltrating neutrophils. A possible explanation for this reduction is the rise of DC _Th17, which through their production of IL-12 would be expected to inhibit the generation of Th17 cells. The rapid shutdown of IL-17 production in Th17 cells after exposure to IL-12 suggests a possible mechanism for limiting this type of pathology (30). Moreover, our data show that in the presence of specific Ag, DC_Th17, but not DC_Th1, can induce already committed Th17 cells to convert into Th1 cells, providing a potential mechanistic explanation for the high frequency of IFN-γ–producing Th1 cells that previously produced IL-17 cytokines within the inflamed spinal cord of EAE mice (12). Similar IFN-γ–producing T cells have also been described in other Th17-mediated disease such as colitis (39) and arthritis (40), so it will be important to determine if DC_Th17 play a role in these disorders as well.

Our data demonstrate that CD40L and RANKL are necessary for maximal IL-12p70 production by DC_Th17. After restimulation, Th17 cells more frequently express these molecules than either Th1 or Th2 cells, which may contribute to their ability to enhance IL-12p70 production in DC_Th. These molecules are expressed rapidly after T cell activation, which might be expected to result in the up regulation of IL-12 shortly after a Th17-mediated response. Other molecules differentially expressed by Th17 cells following activation may play a role in endowing Th17 cells with this unique ability to induce IL-12p70 in DC_Th.

One clinically important question in MS is the role of Th subsets in established disease. Whereas Th1 cells can clearly cause EAE under experimental conditions, Th17 cells have been implicated as the initiating cells in MS and EAE, based on their unique ability to traffic through the choroid plexus in a CCR6-dependent manner prior to CNS inflammation (41, 42). Th1 cells are believed to arise later, potentially as a safeguard against excessive Th17-mediated immunity. Th1 cells, which infiltrate the CNS later than Th17 cells, are considered to be more responsive to Treg-mediated immunosuppression than Th17 cells and thus drive less autoimmune pathology (43). This idea is supported by data showing that Th1, but not Th17, cells require IL-2 for their survival (44, 45). Because Treg cells regulate IL-2 levels via expression of the high-affinity IL-2R (CD25), they would be poised to control Th1 inflammation, but would be less effective at controlling Th17 inflammation (46).

The findings presented in this study are consistent with the view that although Th1 cells may contribute to EAE pathology, these cells ultimately have a moderating effect on autoimmune disease. This is supported by the fact that Th1 cells are the major T cell subset present in the spinal cord just prior to the recovery phase in Th17-initiated EAE (43). Specifically, our findings suggest a model in which Th17 cells initiate disease and then induce the formation of DC_Th17 with potent Th1-polarizing capacity. In the presence of Ag and danger signals, these DC_Th17 promote the conversion of Th17 cells already present in the inflamed CNS, as well as naive T cells, into IFN-γ⁺ Th1 cells. These Th1 cells may contribute to some CNS pathology, before ultimately being suppressed by Treg cells.

Our experiment with IL-12p35 BM chimera mice, along with work by others demonstrating that mice lacking IFN-γ, IL-12p35, or IL-12Rβ2 develop more severe EAE than control mice, reinforces this hypothesis (22, 47, 48). Because infiltrating DC produce the majority of the IL-12 in the inflamed spinal cord (49), their influence on recovery from EAE may be critical. Additionally, EAE mice lacking the IFN-γ receptor in BM-derived cells have persistent neutrophil-rich foci in the CNS, suggesting that sensitivity to IFN-γ is necessary for modulating a neutrophil response (50). Most importantly, in an EAE model similar to our own in which disease was induced by passive transfer of IL-23–treated cells, the addition of an IFN-γ–neutralizing Ab resulted in significantly worse disease (51). Intriguingly, IFN-β, which is widely used in the treatment of MS, requires intact IFN-γ signaling for efficacy in EAE (16).

Monocytes and T cells respond to many of the same chemotactic gradients including CCL2 (52), CXCL9–11, and CCL3–5 (2), so it is not surprising that inflamed tissues in multiple autoimmune diseases contain monocytes and activated T cells in direct contact (7). The induction of Mo-DC formation by activated Th cells likely has implications that extend beyond CNS inflammation. In this regard, previous work from our group investigating the skin of patients with psoriasis and atopic dermatitis showed that Th cells induce the formation of DC_Th from monocytes at sites of inflammation in humans (7). Taken as a whole, our findings raise the exciting possibility that activated T cells drive the formation of DC_Th in a diverse set of autoimmune diseases and that DC_Th17 specifically act to bridge Th17 and Th1 responses.

We thank Fionna Sun for technical assistance.

The authors have no financial conflicts of interest.