We reported previously that c-kit ligation by membrane-bound stem cell factor (mSCF) boosts IL-6 production in dendritic cells (DCs) and a Th17-immune response. However, Th17 establishment also requires heterodimeric IL-23, but the mechanisms that regulate IL-23 gene expression in DCs are not fully understood. We show that IL-23p19 gene expression in lung DCs is dependent on mSCF, which is regulated by the metalloproteinase MMP-9. Th1-inducing conditions enhanced MMP-9 activity, causing cleavage of mSCF, whereas the opposite was true for Th17-promoting conditions. In MMP-9−/− mice, a Th1-inducing condition could maintain mSCF and enhance IL-23p19 in DCs, promoting IL-17–producing CD4+ T cells in the lung. Conversely, mSCF cleavage from bone marrow DCs in vitro by rMMP-9 led to reduced IL-23p19 expression under Th17-inducing conditions, with dampening of intracellular AKT phosphorylation. Collectively, these results show that the c-kit/mSCF/MMP-9 axis regulates IL-23 gene expression in DCs to control IL-17 production in the lung.
T helper 17 cells are important in host defense against pathogens, although unbridled IL-17 production can be pathogenic. IL-23, produced by dendritic cells (DCs) and tissue macrophages, plays a quintessential role in the complete development of Th17 (1). IL-23 is composed of two subunits, p19 and p40, with the latter having the ability to also partner with p35 to form IL-12. However, the mechanisms that regulate IL-23 gene expression are not fully understood.
Mucosal adjuvants cholera toxin (CT) and CpG oligodeoxynucleotide (CpG) induce differential Th17 responses. Many functions of c-kit and its ligand stem cell factor (SCF) are well known and are generally associated with cell maturation from hematopoietic progenitors (2). In the periphery, only mature mast cells and NK cells were known to retain c-kit expression (3) prior to our description of its expression on DCs. We previously defined a functional role for c-kit ligation on DCs induced by membrane-bound SCF (mSCF) involving phosphorylation of AKT via activation of PI3K with promotion of IL-6 production and increased Th2/Th17 cytokines (4). Given the central role of SCF in allergic inflammation and our findings that the c-kit/SCF axis promotes a Th17 response, we hypothesized that modulation of SCF was critical in regulating IL-17 production.
In this study, we used two immunization regimens incorporating OVA with CT, which promotes Th17 development, or with CpG, which hinders it, to determine whether differential immune responses were controlled at the level of IL-23p19 gene expression. We show that IL-23p19 gene expression in lung DCs is negatively regulated by MMP-9 enzymatic activity acting upon the c-kit ligand, mSCF. Importantly, the c-kit–expressing DCs in which this occurs are of a proinflammatory, monocyte-derived phenotype recently implicated in chronic lung inflammation (5).
Materials and Methods
C57BL/6 and MMP-9−/− mice were purchased from The Jackson Laboratory and maintained under pathogen-free conditions in the animal facilities of the University of Pittsburgh. OT-II TCR-transgenic mice (provided by L. Cohn, Yale University, New Haven, CT) were bred and similarly maintained. Mice were aged 6–12 wk and matched for both age and sex. All experimentation was carried out according to protocols approved by the University of Pittsburgh Animal Care and Use Committee.
In vivo treatments
OVA (100 μg/25 μl) was administered with CT (1 μg, LPS undetectable) or CpG (1 μg, LPS < 0.1 ng/mg DNA) intranasally (4, 6). Unless otherwise noted, treatments were performed daily for three consecutive days. Lungs were harvested from groups consisting of a minimum of three animals 24 h following the last treatment.
Flow cytometry and cell sorting
Staining for flow cytometry was by standard methods using the following mAbs: anti-CD11c, anti-CD11b, anti-CD117 (c-kit), anti-CD64 (all from BD Biosciences), anti–MHC class II (Southern Biotec), and anti–MAR-1 (eBioscience). Appropriate isotype controls were purchased from the same company and used at concentrations identical to the test Abs. Intracellular cytokine staining was performed using Perm/Wash solution (BD Biosciences) and anti–IL-23p19 mAb (eBioscience), following a 6-h incubation with monensin (GolgiStop; BD Biosciences). All data acquisition and sorting were carried out on a FACSAria flow cytometer (BD Immunocytometry Systems) running FACSDiva software. Analysis was performed using FlowJo software (TreeStar). The position of cursors on plots was always established using isotype controls, regardless of whether these controls are presented in the figures.
Generation and treatment of bone marrow–derived DCs
Bone marrow–derived DCs (BMDCs) were generated by standard techniques (4). BMDCs were further cultured in 12-well plates (1 × 106 cells/well) in medium and 1 μg/ml CT or 5 μg/ml CpG, with or without rMMP-9 (catalytic domain; AnaSpec). After 18 h, supernatants were assayed for soluble SCF (sSCF) by ELISA (PeproTech) or immunoblotting. Total cellular extracts were prepared for immunoblotting, and RNA was isolated for analysis by quantitative real-time PCR.
Immunoblotting was performed on cell extracts prepared as described (4). Primary Abs—anti-SCF (Chemicon), anti-AKT, or anti–p-AKT (both from Cell Signaling)—were used for detection. HRP-conjugated secondary Abs were used in conjunction with ECL reagents (Thermo Scientific) for detection.
RNA isolation, cDNA preparation, and quantitative real-time PCR
Sorted lung DCs were lysed using QIAzol (QIAGEN), and total RNA was isolated using an miRNeasy Kit (QIAGEN). RNA was converted to cDNA with the High Capacity cDNA Reverse Transcription Kit (Life Technologies). Quantitative real-time PCR was carried out using TaqMan Gene Expression Assays (Applied Biosystems). All primer and probe sets were from Life Technologies (IL-23p19, Mm01160011_g1; HPRT1, Mm01545399). Target gene expression was calculated using the 2−ΔCt method, with HPRT1 as the reference gene.
Enzymatic activity of MMP-9 in total lung extracts was estimated by digestion of gelatin using Novex zymogram gels and the XCell SureLock mini-cell system (Invitrogen). Precast gelatin (10%) gels were run for 90 min at 125 V using Tris-glycine SDS running buffer. Gels were incubated with renaturing buffer and then equilibrated with developing buffer (both from Invitrogen; 30 min each at room temperature). Incubation in fresh developing buffer was continued overnight at 37°C. The gel was stained for 1 h with Coomassie blue (0.2% dye in 45% MeOH, 10% acetic acid) and then destained with repeated washes in MeOH/acetic acid until clear bands could be visualized.
CD4 T cell–DC coculture
DCs were isolated by sorting, as described above. Naive splenic OT-II CD4 T cells were prepared by magnetic bead separation and cell sorting for CD62Lhigh/CD44low cells to high purity (>95%). T cells and DCs were cultured (25:1 ratio) in the presence of specific peptide Ag (OVA323–339) for 5 d, with or without rIL-23 (20 ng/ml; gift from Dr. Mandy McGeachy, University of Pittsburgh). Supernatants were assayed for IL-17 using a commercially available ELISA kit (R&D Systems).
Cytokine-producing cells were enumerated using Ready-SET-Go! ELISPOT assay kits (eBioscience).
Comparisons of means ± SD were performed using a two-tailed Student t test with GraphPad Prism software (GraphPad, La Jolla, CA). Differences between the groups were considered significant at p < 0.05.
Results and Discussion
Selective expression of c-kit on inflammatory monocyte-derived CD11b+ DCs in the lung tissue
The pulmonary mucosa has distinct DC subsets, suggesting specific roles for each in the induction of adaptive immune responses to various pathogens and allergens (5, 7–10). Based upon our previous work (4), we asked whether c-kit is selectively expressed by a specific lung DC subset following treatment with CT or CpG and whether the Th17-promoting function of c-kit–expressing DCs can be regulated in vivo.
Following the staining strategy of Kim and Braciale (11), four general populations of lung DCs were identified (Fig. 1A): MHC class IIhighCD11b− (CD103+ DCs; green), MHC class IIhighCD11b+ (blue), MHC class IIint/lowCD11b+ monocytic DCs (moDCs; red), and a minor population of plasmacytoid DCs (not shown) (Fig. 1A). A key element of this strategy is the use of autofluorescence and MHC class II staining in the same (FITC) channel, along with Siglec F staining to exclude CD11c+ alveolar macrophages (data not shown) (5, 7, 11). No c-kit expression was detected on any population of lung tissue DCs from naive animals; upon treatment (OVA/CT or OVA/CpG), only cells among the MHC class IIhighCD11b+ population expressed c-kit (Fig. 1A).
These analyses were further refined based upon recent data that lung MHC class IIhighCD11b+ cells contain moDCs under certain conditions (5). A high Ag dose upregulated expression of MHC class II on moDCs, rendering them indistinguishable from other CD11b+ DCs, unless costained for MAR-1 and CD64 (5). These moDCs are highly inflammatory, being particularly adept at producing chemokines and expressing IL-23p19. We examined the c-kit–expressing cells among the MHC class IIhighCD11b+ DCs in treated mice and found them to be MAR-1+CD64+ (Fig. 1A, right panels), clearly identifying them as proinflammatory moDCs. Very few MHC class IIhighCD11b+ DCs without coexpression of MAR-1 and CD64 were observed with either treatment (data not shown), consistent with recent findings of low abundance of this population in the context of inflammation (5). Between the two treatments, a higher frequency of MHC class IIhighCD11b+ DCs was detected with OVA/CT (70.6% of CD11b+ DCs) compared with OVA/CpG (42.3% of CD11b+ DCs) (Fig. 1C). However, within this population, similar percentages of c-kit+MHC class IIhighCD11b+ DCs were detected with the two treatments (19.2% versus 21.1%). This effectively means that almost 2-fold more c-kit+MHC class IIhighCD11b+ DCs were induced by CT compared with CpG.
When MHC class IIhighCD11b+c-kit+ DCs were sorted and examined for cytokine expression by quantitative real-time PCR, ample IL-23p19 was observed in cells isolated from OVA/CT-treated mice (Fig. 1B), in agreement with results previously reported using the complex allergen house dust mite (HDM) (5). However, IL-23p19 RNA was not induced by OVA/CpG, suggesting the absence of ligand because c-kit expression at the cellular level was detected with CpG (Fig. 1A). Similar results were obtained when cells were analyzed for IL-23p19 protein expression by intracellular cytokine staining techniques (Fig. 1C). Expression of additional cytokine RNAs, such as IL-12p40, was similar between the treatments (data not shown).
MMP-9 activity limits expression of mSCF on DCs in vivo
Numerous cell types, including macrophages and DCs, express the c-kit ligand SCF on the cell membrane. OVA/CT treatment slightly enhanced the expression of mSCF on lung CD11c+ cells (DCs and alveolar macrophages) (Fig. 1D). An interesting and somewhat unexpected result was that, compared with untreated animals, mSCF expression on CD11c+ cells in OVA/CpG-treated mice was dramatically reduced (Fig. 1D). This suggested that mSCF was shed from the cell membrane, an effect that was ascribed to the enzymatic activity of MMP-9 (12). Therefore, we tested whether MMP-9 enzymatic activity was enhanced under Th1 conditions (OVA/CpG), leading to reduced mSCF and c-kit ligation. We found markedly increased MMP-9 enzymatic activity, as assayed by gelatin zymography, in total lung extracts from OVA/CpG-treated mice relative to those given OVA/CT (Fig. 1E). Similarly, direct treatment of BMDCs yielded higher MMP-9 enzymatic activity with CpG compared with CT or no treatment (data not shown). These results explained the apparent reduction in mSCF on CD11c+ lung cells in OVA/CpG-treated mice (Fig. 1D) (12). Importantly, OVA/CpG-treated MMP-9−/− mice did not exhibit a reduction in mSCF expression on lung DCs, suggesting that MMP-9 enzymatic activity is responsible for the phenomenon observed in wild-type (WT) animals (Fig. 1D). Given that numerous lung cell types, such as neutrophils, express MMP-9 during inflammation (13), we feel that assessment of overall MMP-9 activity in the lung is appropriate in considering effects on mSCF.
Increased IL-17 production in the lungs of MMP-9−/− mice is associated with increased IL-23 production
If mSCF expression leading to prolonged c-kit ligation and IL-23 production was maintained in the absence of MMP-9, then IL-17 production should be enhanced in MMP-9−/− mice, even in OVA/CpG-treated mice. Indeed, CD4 T cell ELISPOT assays revealed increased numbers of IL-17–producing cells in the absence of MMP-9 (Fig. 2A). This observation also extended to treatment with the common allergen HDM, which also was shown to induce IL-23 in inflammatory DCs (5). In contrast, the number of IFN-γ–producing cells was unaffected by OVA/CT or HDM treatment, and it was lower in MMP-9−/− mice treated with OVA/CpG (Fig. 2A), possibly as a result of the decrease in the relative expression of IL-12p40 (Fig. 2B).
With the significant increase in IL-17–producing cells even under Th1 conditions in MMP-9−/− mice, we found that expression of IL-23p19 was significantly higher in lung CD11c+ cells from these mice compared with cells from WT mice upon CpG treatment (Fig. 2B). IL-23 is not required for the differentiation of Th17 cells per se, but rather for stabilization of the phenotype (1, 14), suggesting that the mSCF/c-kit axis, regulated by MMP-9 activity, may be important for sustained IL-17 production. Importantly, although CpG does not favor IL-6 gene expression, expression of this cytokine was also relatively higher in MMP-9−/− mice that maintain mSCF, in agreement with our earlier findings on the positive effect of c-kit–mSCF ligation on IL-6 production (4). Thus, simultaneous increases in both IL-6 and IL-23 gene expression under MMP-9–deficient conditions would promote IL-17 production from CD4+ T cells, even when a Th1-inducing adjuvant, such as CpG, is used. In contrast, the expression of IL-12p40, which is not subject to regulation by the c-kit/mSCF axis (4), was not augmented by MMP-9 deficiency. To confirm these findings, experiments of the type shown in Fig. 2 were repeated in WT mice treated with a specific inhibitor of MMP-9 (15), with similar results on IL-23p19 expression (Supplemental Fig. 1).
We next asked whether deficient IL-23 production from DCs is indeed a major contributing factor to the inability of CpG to promote the development of IL-17+ CD4 T cells, with CpG having been shown to induce differentiation of IFN-γ+ T cells in multiple studies (4, 16, 17). We isolated total CD11b+ DCs, the majority of which had been established to be MHC class IIhighCD11b+c-kit+ DCs (Fig. 1A) expressing MAR-1 (data not shown), as would be expected under highly inflammatory conditions (5). These DCs were used to stimulate naive (CD62LhighCD44low) OVA-specific transgenic CD4 T cells. Cytokine production was monitored after a culture period of 5 d. As expected, DCs from OVA/CT-treated mice induced more IL-17 production from CD4 T cells than did those derived from OVA/CpG-treated animals (Fig. 2C). Addition of IL-23 to OVA/CpG-treated DCs increased IL-17 production to the level seen for OVA/CT-treated DCs, with the latter inducing even more IL-17 with additional exogenous IL-23 (Fig. 2C). These results suggested that the minimal expression of IL-23p19 in OVA/CpG-treated DCs, a consequence of deficient c-kit engagement, was responsible for the lack of IL-17 production.
MMP-9 enzymatic activity selectively inhibits IL-23p19 expression via cleavage of mSCF from the cell surface
Thus far, the data suggested that, in OVA/CpG-treated mice, higher MMP-9 enzymatic activity leads to decreased mSCF expression, reducing c-kit ligation and causing lower IL-23 expression. To make a direct connection between MMP-9 enzymatic activity and IL-23 gene expression, we turned to the BMDC system that we used previously (4). A key finding in our earlier work was that mSCF and c-kit are expressed simultaneously on BMDCs, thus forming a feedback-signaling loop (4). We hypothesized that exogenous MMP-9 would reduce mSCF, resulting in decreased c-kit ligation with subsequent reduction in IL-23p19. Indeed, the recombinant catalytic domain of MMP-9 cleaved mSCF from the surface of BMDCs with coincidental appearance of sSCF in the culture supernatant (Fig. 3A). The effect was dose responsive, with the highest concentration of rMMP-9 resulting in a level of sSCF comparable to that obtained with CpG treatment (Fig. 3B). Using bone marrow–derived derived mast cells, which express c-kit (18), the released sSCF in the culture supernatants (Fig. 3A, 3B) was found to be biologically active based on the profile of AKT phosphorylation (Supplemental Fig. 2). AKT phosphorylation is induced as a result of activation of PI3K by ligand-activated receptor tyrosine kinases, including c-kit (19). Similarly, release of mSCF from BMDCs upon treatment with rMMP-9 led to a reduction in AKT phosphorylation in the DCs (Fig. 3C). It is important to note that, as previously described by us (4) and other investigators (19), AKT phosphorylation was only transiently induced in response to sSCF (Supplemental Fig. 2) but was sustained after CT treatment of BMDCs (Fig. 3C), which promotes expression of mSCF in the DCs. The net effect was reduction in IL-23p19 mRNA in CT-treated cells to the level obtained using CpG (Fig. 3D).
The current study demonstrated that high-level IL-17 production requiring IL-23 is promoted by mSCF expression, which is regulated by the enzyme activity of MMP-9. A proinflammatory role for CD64+MAR-1+ moDCs via production of various mediators, including chemokines and IL-23, was proposed as being important in the pathogenesis of chronic inflammatory conditions, such as asthma (5). We now demonstrate that c-kit+ cells also express CD64 and MAR-1 and that a specific mechanism, which is regulation of mSCF expression by MMP-9, either commits them to promotion of a Th17 response or handicaps them from doing so via cleavage of mSCF. Matrix metalloproteinases (e.g., MMP-9) have been implicated in diseases that are associated with tissue destruction, such as rheumatoid arthritis, as well as in tumor growth (20). Tissue inhibitor of metalloproteinases (TIMPs) have been presumed to have a protective role in these diseases, despite limited evidence to support this concept. Thus, although inhibition of MMPs by TIMP1 should theoretically reduce disease severity in rheumatoid arthritis and cancer, the opposite is true: TIMP1 levels correlate positively with the disease state in humans and animal models (21–23). Our current finding provides an explanation for why high TIMP1 levels, which would inhibit MMP-9 activity and promote a Th17 response, would be detrimental in Th17-driven diseases. Taken together, our study identifies MMP-9 as an attractive target for therapeutic intervention whose expression can be enhanced or inhibited to appropriately regulate the Th17-immune response. Inhibition of MMP-9 activity to boost a Th17 response would be desirable during vaccination against Mycobacterium tuberculosis, when a strong Th17-immune response enhances protective Th1 immunity against the pathogen (24, 25).
We thank Drs. Mandy McGeachy and Shabaana Khader for helpful discussions and Dr. McGeachy for the gift of rIL-23.
This work was supported by National Institutes of Health Grants HL 077430 and AI 048927 (to A.R.), AI 093116 and AI 100012 (to P.R.), and HL 113956 (to A.R. and P.R.).
The online version of this article contains supplemental material.
Abbreviations used in this article:
bone marrow–derived dendritic cell
house dust mite
membrane-bound stem cell factor
stem cell factor
tissue inhibitor of metalloproteinase
The authors have no financial conflicts of interest.