TPL-2 negatively regulates interferon-β production in macrophages and myeloid dendritic cells

To determine the role of the TPL-2 signaling pathway in TLR regulation of IL-10 and IFN-β production, BM-derived macrophages (BMDMs) and BM-derived DCs (BMDCs/myeloid DCs) were generated from Tpl-2^−/− or control WT mice. Induction of IFN-β mRNA and IFN-β protein in response to LPS stimulation were significantly increased in Tpl-2^−/− BMDMs compared with control cells (Fig. 1 a). In contrast, IL-10 mRNA and IL-10 protein were significantly decreased in the absence of TPL-2. Increased production of IL-12p40 and p35 mRNA, and in IL-12p70 and p40 protein, was also observed in Tpl-2^−/− BMDMs (Fig. 1 a).

Similarly, BMDCs from Tpl-2^−/− mice produced elevated levels (mRNA and protein) of IFN-β, but reduced amounts of IL-10 (Fig. 1 b). However, the increase in IFN-β production in Tpl-2^−/− relative to WT cells was less in BMDCs than in BMDMs. The elevated amounts of IFN-β produced by Tpl-2^−/− BMDMs and BMDCs may result in part from decreased production of IL-10 because IL-10 inhibits IFN-β production (Moore et al., 2001). This may explain why IFN-β production is less affected by TPL-2 deficiency in BMDCs compared with BMDMs because the BMDCs produce lower concentrations of IL-10. Consistent with the results using Tpl-2^−/− BMDMs, Tpl-2^−/− BMDCs produced elevated levels of IL-12 p40 and p70 protein and IL-12 p35 and p40 mRNA (Fig. 1 b) compared with control BMDCs.

To determine whether impaired LPS-induced ERK activation accounted for the altered cytokine production by Tpl-2^−/− BMDMs, activation of ERK was monitored by specific immunoblotting. In WT cells, LPS-stimulated rapid ERK phosphorylation, which peaked at 15 min (Fig. 1 a). In contrast, LPS-induced ERK phosphorylation was not detected in Tpl-2^−/− BMDMs (Fig. 1 a), whereas JNK and p38 phosphorylation were unaffected (Fig. 1 a), as reported previously (Dumitru et al., 2000). To confirm that the effects of TPL-2 deficiency on TLR-induced cytokine production resulted from abrogation of ERK activation, WT cells were treated with the MEK inhibitor U0126 to block ERK activation while leaving p38 or JNK activation unaffected (Favata et al., 1998; Davies et al., 2000). U0126 increased IFN-β and IL-12 production and reduced production of IL-10 by LPS-stimulated BMDMs (Fig. S1), similar to the effects of TPL-2 deficiency in this cell type (Fig. 1 a). Comparable results were obtained with a structurally distinct inhibitor of MEK-1/2, PD184352 (a gift from P. Cohen, University of Dundee, Scotland, UK; unpublished data). These data support the hypothesis that TPL-2 promotes IL-10 production while decreasing IL-12 and IFN-β production via its activation of the ERK MAPK pathway.

We show that activation of ERK by TLR4 in macrophages required TPL-2, in agreement with previous findings (Dumitru et al., 2000; Banerjee et al., 2006; Papoutsopoulou et al., 2006). Our data also suggest that activation of the TPL-2 signaling pathway was required for optimal IL-10 production, but negatively regulated IL-12 production. These findings are in accordance with published data using pharmacological inhibitors to block ERK activation in macrophages (Feng et al., 1999; Häcker et al., 1999; Yi et al., 2002; Banerjee et al., 2006). Moreover, our data suggest that TLR4 activation of the TPL-2 pathway negatively regulates the production of IFN-β, in contrast to its positive effects on IL-10 production.

Although the consequences of TPL-2 deficiency on cytokine production after TLR stimulation were very similar in BMDMs and BMDCs (Fig. 1), it has been suggested previously that LPS does not stimulate ERK activation in BMDCs (Häcker et al., 1999). Hence, we reinvestigated whether TLR4 was actually able to induce ERK phosphorylation in BMDCs. In our hands, LPS clearly stimulated ERK phosphorylation in WT BMDCs (Fig. 1 b), although at a reduced intensity compared with WT BMDMs (Fig.1 a). This may explain the previous failure to detect ERK phosphorylation in BMDCs after LPS stimulation (Häcker et al., 1999). We show that LPS-induction of ERK phosphorylation was blocked in BMDCs from Tpl-2^−/− mice (Fig. 1 b), similarly to BMDMs. p38 MAPK phosphorylation was marginally reduced in Tpl-2^−/− BMDCs compared with WT cells after LPS stimulation, but JNK phosphorylation was unaffected. Thus, altered cytokine production in Tpl-2^−/− BMDCs after TLR stimulation was associated predominantly with a block in ERK activation.

Our data that TPL-2 negatively regulates IL-12 in BMDCs are consistent with reports from the Pulendran group that pharmacological inhibition of ERK activation increases IL-12 production by human and mouse DCs (Agrawal et al., 2003; Dillon et al., 2004). The negative regulatory effect of TPL-2 on IFN-β production that we observe in BMDCs stimulated with LPS contrasts to the requirement of TPL-2 for optimal IL-10 induction (Fig. 1 b). Thus, TLR-induction of IL-10 and IFN-β is oppositely regulated by TPL-2 in both macrophages and myeloid DCs.

Because pDCs are a major source of type I IFNs during immune responses (Stetson and Medzhitov, 2006), we determined the role of the TPL-2 signaling pathway in TLR-induction of IFN-α and IFN-β in this cell type. As pDCs express TLR9, but not TLR4 (Boonstra et al., 2003), we stimulated pDCs with CpG rather than LPS in comparison to macrophages and myeloid DCs. BMDMs and BMDCs obtained from Tpl-2^−/− mice produced elevated levels of IFN-β protein, IL-12p40 and IL-12p70 upon CpG stimulation, as compared with the same cell types from WT mice (Fig. 2, a and b). Ex vivo Tpl-2^−/− splenic CD11b⁺ macrophages purified directly from Tpl-2^−/− mice also showed highly elevated levels of IFN-β and IL-12p40 and p70 as compared with equivalent cells from control mice after CpG stimulation (Fig. S2). Furthermore, CpG-induced IL-10 production was significantly reduced in Tpl-2^−/− BMDMs (Fig. 2 a) and BMDCs (Fig. 2 b), similar to results observed after LPS stimulation. IL-10 production by purified splenic macrophages was also reduced compared with control WT cells after CpG stimulation (Fig. S2).

Opposite effects of TPL-2 deficiency were observed after stimulation of pDCs with CpG. Production of both IFN-α and IFN-β in Tpl-2^−/− BM-pDCs were significantly reduced, as compared with pDCs from WT mice (Fig. 2 c). Furthermore, again in contrast to Tpl-2^−/− BMDMs and BMDCs, a small but significant decrease in IL-12p40 and IL-12p70 was observed in Tpl-2^−/− pDCs stimulated with CpG, as compared with controls (Fig. 2 c). Similar effects were obtained with directly purified ex vivo splenic pDCs (unpublished data). Consistent with the hypothesis that pDCs are wired differently to macrophages and myeloid DCs, pDCs did not produce IL-10 in response to CpG stimulation (unpublished data). Conversely, whereas pDCs produced high levels of IFN-α after stimulation with CpG, IFN-α was not detectable in supernatants of BMDMs and BMDCs stimulated similarly, even in the absence of TPL-2 (unpublished data).

Our findings reveal cell-intrinsic differences in the requirements for TPL-2 in TLR-induced signaling in pDCs versus macrophages and myeloid DCs. Whereas TPL-2 negatively regulated the induction of optimal IFN-β and IL-12 in macrophages and myeloid DCs after TLR-9 stimulation, TPL-2 was required for TLR9 induction of both IFN-α and IFN-β by pDCs. Furthermore, these data extend previous studies showing that the signaling requirements for induction of type I IFNs in pDCs are distinct to macrophages and myeloid DCs (Colonna, 2007; Schmitz et al., 2007).

Because CpG stimulation of Tpl-2^−/− pDCs resulted in an opposite effect on type I IFN induction than that observed in macrophages and myeloid DCs, it was important to determine whether ERK was phosphorylated in pDCs in response to CpG in a TPL-2–dependent fashion. It was also important to determine whether TLR9 induction of ERK phosphorylation in BMDMs and BMDCs required TPL-2, as it has been recently proposed that TLR9 stimulation does not induce ERK phosphorylation in this cell type (Häcker et al., 1999) and that CpG activation of ERK is not affected by TPL-2 deficiency in macrophages (Sugimoto et al., 2004).

We show here that CpG induced a more gradual increase in ERK phosphorylation in WT BMDMs (Fig. 3 a) than LPS did (Fig. 1 a), and that this was abrogated in BMDMs from Tpl-2^−/− mice (Fig. 3 a). CpG clearly stimulated ERK phosphorylation in WT BMDCs (Fig. 3 b), although at a reduced intensity compared with WT BMDMs, possibly explaining data from Sugimoto et al. (2004), who failed to detect any reduction in ERK phosphorylation after CpG stimulation of Tpl-2^−/− macrophages. The reduced signal may also explain the previous failure to detect CpG stimulation of ERK phosphorylation in BMDCs (Häcker et al., 1999). CpG-induced ERK phosphorylation in BMDCs was blocked in the absence of TPL-2 (Fig. 3 b). Using a high sensitivity substrate for chemiluminescent detection of phospho-ERK antibody, CpG was also found to induce phosphorylation of ERK in pDCs, which was dependent on TPL-2. CpG-induced p38 and JNK phosphorylation in pDCs were minimally affected by the absence of TPL-2 (Fig. 3 c).

Our data show that ERK activation after TLR9 ligation required TPL-2 in macrophages, myeloid DCs, and pDCs, although the effects of TPL-2 deficiency on cytokine induction were distinct in pDCs compared with the other two cell types. The very low levels of ERK phosphorylation induced by CpG stimulation of pDCs compared with BMDMs and BMDCs may explain this differential requirement for TPL-2 in cytokine induction in these different cell types; it has been established in PC12 cells and fibroblasts that the amplitude and duration of ERK signaling can dictate the biological effect of this signaling pathway (Marshall, 1995; Murphy et al., 2002). Alternatively, it is possible that ERK signaling is interpreted differently in pDCs compared with macrophages and myeloid DCs as a consequence of integration of ERK activation with other signaling pathways triggered by TLR9. In accordance with the latter possibility, pDCs have been shown to use unique signaling pathways for the induction of type I IFNs (Honda et al., 2006; Stetson and Medzhitov, 2006; Colonna, 2007; Schmitz et al., 2007; Cao et al., 2008; Guiducci et al., 2008). Consistent with cell type–specific differences for ERK in regulating IFN-β production, a previous study has shown that ERK positively regulates IFN-β production by mouse embryonic fibroblasts in response to myxoma virus infection (Wang et al., 2004), similar to CpG induction of IFN-α and IFN-β in pDCs, but opposite to its negative effects on IFN-β production in macrophages and myeloid DCs (Fig. 2).

To determine the requirement for protein synthesis in the negative regulation of IFN-β mRNA induction by TPL-2, Tpl-2^−/− and WT control BMDMs were stimulated with CpG in the presence or absence of cycloheximide (CHX). Induction of IFN-β mRNA was increased when WT BMDMs were stimulated in the presence of CHX compared with untreated control, indicating that IFN-β expression was negatively regulated by a protein synthesis–dependent mechanism (Fig. 4 a). The elevated expression of IFN-β mRNA observed in the Tpl-2^−/− BMDMs, however, was not affected by CHX, demonstrating that the negative regulation of IFN-β production by protein synthesis is TPL-2 dependent. Similar results were obtained for IL-12p40 expression (Fig. 4 a); however, CHX increased IL-12p40 mRNA not only in WT BMDMs but also, to a lesser extent, in Tpl-2^−/− BMDMs, suggesting TPL-2–dependent and –independent mechanisms for IL-12p40 mRNA regulation. Similar data were obtained with BMDMs stimulated with LPS (unpublished data). This protein synthesis requirement for negative regulation of IFN-β mRNA expression could indicate the action of a TPL-2–dependent cytokine or transcription factor that negatively regulates IFN-β transcription.

Because IL-10 has profound inhibitory effects on cytokine production by TLR-stimulated macrophages (Moore et al., 2001), it was possible that the increased levels of IFN-β and IL-12 in Tpl-2^−/− BMDMs resulted from diminished IL-10 production. However, we show that stimulation of IL-10^−/− BMDMs with LPS (not depicted) or CpG (Fig. 4 b) in the presence of U0126 inhibitor resulted in elevated levels of IFN-β and IL-12, as compared with controls. In addition, in the absence of U0126, the production of IFN-β and IL-12 was increased in IL-10^−/− BMDMs compared with WT cells. Thus, ERK negatively regulates the production of IFN-β and IL-12 independent of IL-10, but TLR-induced production of these cytokines is additionally regulated by the autocrine effects of IL-10, which is itself positively regulated by ERK.

It has been suggested that TLR stimulation coordinately induces the TRAF3-dependent production of IFN-β and IL-10 (Häcker et al., 2006). However, we have shown previously that IFN-β and IL-10 are not always co-regulated and that LPS activates signaling pathways that have different effects on the expression of these two cytokine genes (Boonstra et al., 2006). In keeping with this, we show that TPL-2 activation of ERK is required for TLR4 and TLR9 induction of IL-10, but negatively regulates IFN-β production via both IL-10–dependent and –independent mechanisms.

Our results indicate that negative regulation of IFN-β expression by TPL-2–mediated activation of ERK was dependent on protein synthesis. To identify which transcription factors might regulate IFN-β transcription in an ERK-dependent fashion, LPS and CpG-induced gene expression in BMDMs stimulated in the presence or absence of the MEK inhibitor UO126 was analyzed by Affymetrix gene array. Transcription of the IFN-β gene is known to be regulated by NF-κB, AP-1, and IRFs (for review see Colonna, 2007). Analysis of the gene array data revealed that expression of mRNAs encoding NF-κB and IRF family members was not affected by U0126 (unpublished data). However, U0126 substantially reduced LPS and CpG induction of mRNA encoding the AP-1 transcription factor c-fos (unpublished data), which was shown to regulate IL-10 and IL-12 expression (Dillon et al., 2004). Previous studies with MEK inhibitors have shown that ERK signaling is required for IL-10 production by myeloid cells triggered via TLR (Yi et al., 2002; Dillon et al., 2004). This was suggested to be mediated by transcriptional induction of IL-10 by c-Fos, whose expression is positively regulated by ERK signaling. It has also been suggested that negative regulation of IL-12p40 by ERK was mediated indirectly as a consequence of reduced IL-10 production (Yi et al., 2002). Although ERK down-regulation of IL-12p40 production has also been proposed to be mediated via c-Fos (Dillon et al., 2004), the regulation of IFN-β production was not addressed. Our data obtained with IL-10^−/− macrophages demonstrate that ERK negatively regulates IL-12 and IFN-β production independent of IL-10 (Fig. 4 b).

We show that c-fos mRNA expression in macrophages is positively regulated by TPL-2 because CpG induced reduced levels of c-fos mRNA in Tpl-2^−/− compared with WT control BMDMs (Fig. 4 c). Similar results were obtained with LPS (unpublished data). At the peak time of c-fos mRNA induction, however, no differences in mRNA expression of AP-1 transcription factors, Jund1 or ets-1, were observed between WT and Tpl-2^−/− BMDMs (Fig. S3). Reduced c-fos mRNA in Tpl-2^−/− BMDMs stimulated with CpG translated into reduced c-Fos binding activity, since c-Fos binding to a specific AP-1 DNA oligonucleotide was found to be significantly reduced in the absence of TPL-2 (Fig. 4 d).

To investigate the potential role of c-Fos in negative regulation of IFN-β and IL-12, BMDCs from Tpl-2^−/−, IL-10^−/−, and WT mice were transduced with a retroviral construct containing c-fos-IRES-GFP or Mock-IRES-GFP and flow cytometry purified on the basis of GFP expression. Expression of c-fos decreased the levels of IFN-β and IL-12 in Tpl-2^−/−, IL-10^−/−, and WT cells (Fig. 5) induced after CpG stimulation. Whereas negative regulation of IL-12 by c-Fos had been proposed previously, it remained unclear whether this was independent of effects on IL-10 expression (Agrawal et al., 2003; Dillon et al., 2004). We show that c-Fos regulates IL-12 production in the presence and absence of IL-10. Furthermore, the negative regulation of IFN-β by c-Fos suggests a potential mechanism by which the TPL-2 signaling pathway negatively regulates IFN-β production.

The TPL-2 MAPK signaling pathway has attracted considerable interest as a possible target for antiinflammatory drugs (Hall et al., 2007). Because of its critical role in the induction of TNF (Dumitru et al., 2000; Rousseau et al., 2008), several large pharmaceutical companies are developing small molecule inhibitors of TPL-2 as potential drugs to treat inflammatory diseases (Hall et al., 2007). However, our work suggests that TPL-2 inhibition may have undesirable proinflammatory effects caused by its complex regulation of cytokine production. Thus, although TPL-2 inhibition might decrease TNF production, our data suggest that the production of the proinflammatory cytokines IL-12 and IFN-β would be increased in macrophages and myeloid DCs, whereas production of the antiinflammatory cytokine IL-10 would be decreased. We also show that TPL-2 regulates cytokine production in a cell-specific fashion, positively regulating TLR-induced IL-12 and IFN-β production in pDCs. Our data, therefore, raise questions about the validity of TPL-2 as an antiinflammatory drug target.

C57BL/6, Tpl-2^−/− (Dumitru et al., 2000), and IL-10^−/− (Kühn et al., 1993) mice were bred at the National Institute for Medical Research under specific pathogen–free conditions. All protocols for breeding and experiments with animals were performed and approved by the Home Office, UK, Animals (Scientific Procedures) Act 1986 (Project License Number: PPL 80/2236). Macrophages, myeloid DCs, and pDCs were generated from BM as previously described (Boonstra et al., 2006). BMDM (F4/80⁺) and BMDC (CD11c⁺) purities were ≥95%, and pDC (CD11b⁻ CD11c⁺ B220⁺ or 120G8⁺) purities were ≥98% (MoFlo cytometer; Cytomation). Similarly, spleen cell suspensions were depleted of T cells, and DCs and macrophages were purified as CD11b⁺ (≥95%; Fiorentino et al., 1991).

The antibodies used for immunoblotting have been previously described (Papoutsopoulou et al., 2006). mAbs used for cell purification were as follows: anti-F4/80-PE (Invitrogen), anti-CD11c-PE (BD), anti-CD11b-APC (eBioscience), and 120G8-Alexa Fluor 488 (Boonstra et al., 2006). Cells were stimulated with Salmonella Minnesota LPS (Enzo Biochem, Inc.) or phosphorothioate CpG DNA (CpG1668: TCCATGACGTTCCTGATGCT; TriLink Biotech). GM-CSF was obtained from Schering-Plough, and Flt3 ligand was purchased from Shanghai Genomics. The MEK-1 inhibitor U0126 (Favata et al., 1998) was obtained from BIOMOL International; DMSO, cycloheximide, and polybrene were purchased from Sigma-Aldrich; Liberase Cl and FuGENE-6 were obtained from Roche. The cDNA encoding c-fos was donated by A. Behrens (Cancer Research Institute, London, UK).

Supernatants from BMDMs and BMDCs (2 × 10⁵/200 µl) stimulated with LPS (100 ng/ml) or CpG1668 (500 nM) for 24 h were analyzed by commercial ELISA kits for IFN-β (PBL), IL-12p70, and IL-10 (eBioscience). IL-12p40 was detected as previously described (Boonstra et al., 2006).

BMDMs and BMDCs were cultured in medium containing 1% FCS for 5 h before stimulation with LPS or CpG and washed in PBS before lysis (1% NP-40, 0.1% SDS, 0.5% deoxycholate acid, 50 mM Tris HCl, pH 8.0, 50 mM NaCl, 2 mM EDTA, 2 mM sodium-pyrophosphate, 50 mM sodium fluoride, 100 mM vanadate [all from Sigma-Aldrich], and complete EDTA-free protease inhibitor cocktail [Roche]). Immunoblotting of proteins was performed as previously described (Saraiva et al., 2005; Papoutsopoulou et al., 2006) and visualized with ECL (GE Healthcare) or SuperSignal West Femto Substrate (Thermo Fisher Scientific). pDCs showed low background levels of MAPK phosphorylation without resting and were used directly, but stimulated in 1% FCS.

Total RNA was isolated from BMDMs, BMDCs, and pDCs, using the RNeasy kit (QIAGEN) and reverse transcribed with the High Capacity cDNA Reverse Transcription kit (Applied Biosystems). The expression of IFN-β, IL-12p35, IL-12p40, IL-10, c-fos, JunD1, and ets-1 was quantified by Real-Time PCR (7900HT; Applied Biosystems) and normalized against ubiquitin or HPRT mRNA levels, as previously described (Rajsbaum et al., 2008). The following primers and probes were obtained from Applied Biosystems: c-fos (Mm00487425); Jund1 (Mm00495088); ets-1 (Mm00468970).

Nuclear extracts of BMDMs stimulated with CpG (500 nM) were prepared with the Nuclear Extract kit and assayed with the TransAM ELISA kit (both from Active Motif).

Amphoteric c-fos recombinant retrovirus was produced by transfection of c-fos cDNA (subcloned into pMXI-IRES-eGFP vector and sequenced to verify correct insertion; Shoemaker et al., 2006) into the Plat-E packaging cell line (Morita et al., 2000) using FuGene-6 transfection reagent (Roche). Transfected cells were cultured at 37°C for 48 h (DMEM plus 10% FCS and antibiotics). Culture supernatants were centrifuged at 48,000 g for 4 h at 4°C, sediments resuspended and filtered (0.4 µm) as concentrated virus. BMDCs from WT, Tpl-2^−/−, or IL-10^−/− mice were prepared as previously described (Boonstra et al., 2006), with minor modifications. BM cells were cultured in 6-well plates (Corning) in complete RPMI containing GM-CSF as previously described (Boonstra et al., 2006) at 2 ml/well. After 48 h of incubation, 0.5 µg/ml polybrene and 200 µl of concentrated virus was added per well and plates were centrifuged at 2,000 g for 1 h at 25°C. On days 3, 4, and 5, each well was further supplemented with 1 ml complete BMDCs medium containing 100 µl of virus. On day 7, nonadherent cells were harvested and sorted for GFP⁺ (MoFlo; Cytomation) and 96% of cells were CD11c⁺, and 99% GFP⁺.

Analyses were performed using Prism software (GraphPad).

Fig. S1 shows the contribution of ERK activation to LPS-mediated cytokine production. Fig. S2 shows that TPL-2 negatively regulates TLR-induced IFN-β and IL-12 production, but is required for optimal IL-10 production in splenic macrophages stimulated with CpG. Fig. S3 shows that transcription of Jund1 and Ets1 is not regulated by TPL-2.

We would like to thank Aaron Rae, Graham Preece, and Nadine Biboum for help with flow cytometric cell sorting.

A. O'Garra, D. Cook, S. Papoutsopulou, H-T. Yang, and S. Ley are supported by the Medical Research Council (MRC) UK. A. O'Garra has additional funding from the sixth Framework EU Program, DC-VACC, and DC-THERA. P. Ricciardi-Castagnoli had funding from DC-VACC and DC-THERA for this project. R. Rajsbaum was supported by DC-THERA. F. Kaiser was supported by the Arthritis Research Campaign UK (ARC 16053), and S. Grant (AICR 03-297) by the Association for International Cancer Research. F. Kaiser was also subsequently by MRC. P.N. Tsichlis is supported by the National Institutes of Health grant R01CA095431.

The authors have no conflicting financial interests.