Tristetraprolin Is Required for Full Anti-Inflammatory Response of Murine Macrophages to IL-10¹

One of the key players in immune homeostasis is IL-10, a cytokine that was discovered 18 years ago as a T cell-secreted factor that inhibited cytokine production by Th1 cells (1). Over the years, it became clear that IL-10 is produced by many cell types including T and B cells, macrophages, dendritic cells, mast cells, keratinocytes, or epithelial cells (2). It is generally considered that the main biological function of IL-10 is to limit or shut down inflammatory responses. This notion is supported by the phenotype of IL-10-deficient mice that develop severe inflammatory bowel disease due to spontaneous chronic inflammation and are susceptible to endotoxin treatment because of acute overproduction of proinflammatory cytokines (3, 4). Many of these phenotypes can be recapitulated by T cell-specific deletion of the IL-10 gene, indicating that T cell-derived IL-10 is primarily responsible for chronic inflammation (5). Under conditions of acute inflammation, the main source of IL-10 are macrophages and dendritic cells (6). On the cellular level, IL-10 inhibits production of proinflammatory cytokines and regulates differentiation and proliferation of various immune cells. These effects depend entirely on the activation of the transcription factor Stat3 by IL-10 (7, 8, 9, 10, 11). The events downstream of Stat3 activation that mediate the anti-inflammatory functions of IL-10 remain an area of active research. It is becoming increasingly clear that multiple mechanisms mediate the IL-10 function. On the one hand, IL-10 inhibits transcription of a subset of proinflammatory genes (12, 13). In a mouse model lacking 3′ untranslated regions (UTR)⁴ in the Tnf gene, the transcriptional mechanism was found to play a major role in IL-10 responses (13). On the other hand, IL-10 was also reported to act posttranscriptionally by increasing the rate of mRNA decay of inflammatory cytokines such as TNF-α or by inhibiting translation (12, 14, 15). The posttranscriptional regulation depends on 3′ UTR containing AU-rich elements (AREs) (16). However, far less is known about the IL-10-regulated effector genes that control the anti-inflammatory response. Several candidates have been described, but so far none have been shown to account for the majority of the anti-inflammatory effects (17). IL-10 was also demonstrated to up-regulate the dual-specificity phosphatase 1 (DUSP1) in LPS-stimulated macrophages. causing a more rapid inactivation of p38 MAPK (18). The reduction of p38 MAPK activity may lead to a decreased stability of ARE-containing mRNAs and/or reduced transcription by transcription factors that depend on p38 MAPK. Recently, the transcriptional repressor ETV3 and the corepressor SBNO2 were characterized as IL-10/Stat3-induced genes that may contribute to the anti-inflammatory IL-10 effects (19). An important aspect of the above-mentioned studies that remains to be resolved is how IL-10 inhibits specifically only a subset of inflammatory genes. These studies indicate that yet unknown IL-10/Stat3 target genes need to be discovered to complement the current knowledge about the anti-inflammatory effects of IL-10.

In this study, we demonstrate that IL-10-mediated reduction of TNF-α and IL-1α production by LPS-treated mouse bone marrow-derived macrophages (BMDMs) is less efficient in cells lacking the RNA-destabilizing factor tristetraprolin (TTP). TTP is known to bind and destabilize mRNAs of several proinflammatory cytokines (e.g., TNF-α) containing class II AREs in their 3′ UTRs (20). TTP facilitates degradation of the bound mRNA by initiating the assembly of RNA decay machinery (21, 22). The RNA degradation activity of TTP is negatively regulated by p38 MAPK-dependent signaling (23, 24, 25). Mice lacking TTP (encoded by the Zfp36 gene) develop multiple chronic inflammatory syndromes ranging from arthritis to cachexia to dermatitis that can all be relieved by reduction of TNF-α levels (26). By comparing LPS-treated wild-type and TTP-deficient BMDMs, we show that IL-10 accelerates, in a TTP-dependent way, the decay of TNF-α mRNA, resulting in a reduction of secreted TNF-α. Furthermore, IL-10 increases TTP expression in LPS-treated wild-type (WT) but not Stat3-deficient BMDMs. However, we show that the increased TTP levels are not sufficient to mediate the IL-10 effects. Instead, the IL-10- mediated reduction of p38 MAPK activity in LPS-treated BMDMs that is known to be caused by up-regulation of the p38 MAPK phosphatase DUSP1 (18) is required to act in concert with TTP to reduce mRNA levels of proinflammatory cytokines such as TNF-α and IL-1α. We propose that the sustained activation of Stat3 by IL-10 causes both an increased TTP expression and a reduction in p38 MAPK activity. The combination of these effects results in a reduction of mRNA stability and attenuation of cytokine production.

Stat3, p38 MAPK, and DUSP1 (sc-1102) Abs were from Santa Cruz Biotechnology, the ERK Ab was from BD Transduction Laboratories, and the phosphotyrosine-Stat3 Ab (pY-Stat3) was from Cell Signaling and New England Biolabs. Rabbit Ab to TTP was obtained by immunizing rabbits with 44 C-terminal amino acids of TTP fused to GST. IL-10 and IL-6 (Sigma-Aldrich) were used at a concentration of 10 ng/ml, LPS from Salmonella minnesota (Alexis) was used at a concentration of 5–10 ng/ml, and anisomycin and actinomycin D (both Sigma-Aldrich) were used at a concentration of 100 ng/ml and 5 μg/ml, respectively. SB203580 (SB; Sigma-Aldrich) was dissolved in DMSO and used at final concentration 4 μM. ATP (1 μM; Sigma-Aldrich) was added 1 h before collecting the samples.

Primary macrophages were grown in L cell-derived CSF-1 as previously described (27). Zfp36^−/− and Dusp1^−/− mice (26) were on the C57BL/6 background. The Zfp36 gene encodes TTP, which is the better known name and therefore is used throughout. For experiments with cells derived from TTP^−/− and TTP^+/+ mice, littermates originating from the TTP^+/− colony were used. The LysMcre-Stat3^fl/fl mice were obtained by crossing Stat3^fl/fl mice (28) with LysMcre, strain B6.129P2-Lyzs^tml(cre)Ifo (The Jackson Laboratory), all on the C57BL/6 background. Mice, 8–12 wk old at the time of bone marrow collection, were housed under specific pathogen-free conditions. Mouse experiments were conducted in compliance with national laws. Mouse macrophage line RAW 264.7 was grown in DMEM supplemented with 10% FCS.

The open reading frame of the mouse TTP was PCR-amplified from cDNA. The PCR was used to add Flag tag DYKDDDDK at the C terminus. The TTP-Flag fragment was cloned into the pGL2-basic (Promega) containing a tetracycline-responsive element (TRE) in the promoter. The resulting plasmid pTRE-TTPfl was transfected into HeLa-Tet-off cells (BD Clontech) using nucleofection (Amaxa). After transfection, the cells were incubated overnight in medium with (no TTP expression) or without (TTP expression) tetracycline (1 μg/ml).

For ELISA, BMDMs were seeded the day before use at 2 × 10⁵ cells/well in a 24-well plate. Supernatants were diluted 1/8 in DMEM (for TNF-α) or 1/2 (for IL-1α) and cytokines were assayed using ELISA kits (R&D Systems) according to the manufacturer’s instructions.

After treatment, whole cell extracts were prepared and assayed by Western blotting as described elsewhere (29). Detection and quantitation of signals were performed using the infrared imaging system Odyssey (LI-COR Biosciences).

Total RNA was isolated using TRIzol reagent (Invitrogen). Reverse transcription was performed with Moloney murine leukemia virus reverse transcriptase (Fermentas). The following primers were used: for HPRT, the housekeeping gene used for normalization, HPRT forward (fwd) 5′-GGATTTGAATCACGTTTGTGTCAT-3′ and HPRT reverse (rev) 5′-ACACCTGCTAATTTTACTGGCAA-3′; for TTP, TTP fwd 5′-CTCTGCCATCTACGAGAGCC-3′ and TTP rev 5′-GATGGAGTCCGAGTTTATGTTCC-3′; for TNF-α, TNF-α fwd 5′-CAAAATTCGAGTGACAAGCCTG-3′ and TNF-α rev 5′-GAGATCCATGCCGTTGGC-3′; for suppressor of cytokine signaling (SOCS) 3, SOCS3 fwd 5′-GCTCCAAAAGCGAGTACCAGC-3′ and SOCS3 rev 5′-AGTAGAATCCGCTCTCCTGCAG-3′; and for TTP primary transcript, TTPpt fwd 5′-GACTGGCAAGCTCGTGAAGT-3′, pt-TTP-rev 5′-CAGTCAGGCGAGAGGTGA-3′. For determination of mRNA decay by qRT-PCR, the primers TNF-α fwd 5′-TTCTGTCTACTGAACTTCGGGGTGATCGGTCC-3′ and TNF-α rev 5′-GTATGAGATAGCAAATCGGCTGACGGTGTGGG-3′ and for IL-1α, the primer set QT00113505 from Qiagen was used. Amplification of DNA was monitored by SYBR Green (Molecular Probes (30).

To prepare extracts, cells were washed with cold PBS and lysed in buffer containing 10 mM Tris-HCl (pH 7.5), 50 mM NaCl, 30 mM NaPP_i, 50 mM NaF, 2 mM EDTA, 1% Triton X-100, and a protease inhibitor mixture (Roche). Extracts were cleared by centrifugation at 15,000 rpm. Twelve-microliter cell extracts (30 μg of protein) from RAW 264.7 cells or 5-μl cell extracts (15 μg of protein) from pTRE-TTPfl-transfected HeLa-Tet-off cells were incubated with 0.5 μl of poly(U) RNA (100 ng/μl), 0.5 μl of Cy5.5-labeled TNF-α ARE (1 pmol/μl), 1 μl of RiboLock RNase Inhibitor (Fermentas), and 2.5 μl of 5× Gelshift buffer (200 mM KCl, 5 mM MgCl₂, 0.5 mM EGTA, 2.5 mM DTT, 100 mM HEPES-KOH (pH 7.9), and glycerin 50% (v/v)) for 20 min at room temperature. For supershift assays, 0.5 μl of TTP antiserum was added. Samples were then separated on a 6% polyacrylamide gel. The Cy5.5 signal was detected and quantified using the infrared imaging system Odyssey (LI-COR Biosciences). Poly(U) RNA and Cy5.5′-labeled TNF-α ARE RNA were purchased from Microsynth. The sequence for Cy5.5′-labeled TNF-α ARE was as follows: 5′-AUUAUUUAUUAUUUAUUUAUUAUUUA-3′.

Data from independent experiments were analyzed using univariate linear regression models and the SPSS program. For qRT-PCR normalized copy numbers and for ELISA pg/ml were log-transformed. Residuals were plotted, visually inspected, and tested for normality. Design matrices were specified such that the coefficients for the relevant comparisons could be calculated, e.g., between the baseline and induced states and between genotypes. Only the significance levels are reported.

TTP expression has been reported to be controlled by the transcription factors Stat1 (in response to IFNs) and Stat6 (in response to IL-4) (27, 31). Affymetrix analysis revealed that LPS-induced TTP expression is further enhanced by IL-10 (data not shown) at 30 min after treatment of Il10^−/− BMDMs with IL-10 and LPS, suggesting that TTP may represent a Stat3 target gene involved in the anti-inflammatory responses to IL-10. To examine the effect of IL-10 on TTP expression in more detail, we investigated TTP gene expression in BMDMs conditionally deleted for Stat3 (LysMcre-Stat3^fl/fl) and control LysMcre-Stat3 cells. LysMcre is known to delete loxP-flanked alleles in macrophage and neutrophil lineages with >90% efficiency (7). Treatment of BMDMs with IL-10 and LPS caused a 2- to 3-fold increase in TTP mRNA (Fig. 1,A) and protein levels (Fig. 1,B) compared with LPS treatment alone. The IL-10-mediated increase in TTP expression required Stat3 since in LysMcre-Stat3^fl/fl cells the TTP levels remained unchanged upon IL-10 treatment. The expression of Stat3 was reduced by >90% in LysMcre-Stat3^fl/fl cells (Fig. 1,B), thereby confirming the deletion efficiency by LysMcre. Analysis of TTP primary transcripts by PCR (Fig. 1 C) and nuclear run-on assays (data not shown) revealed that the IL-10-mediated induction of TTP was caused by increased transcription.

These findings document that IL-10-activated Stat3 increases expression of TTP in LPS-stimulated macrophages. IL-10 alone only weakly increased TTP transcription, and this induction was not sufficient to generate detectable levels of TTP protein. This way of regulation resembles that of IFN-induced TTP expression, which was shown to be activated by IFNs and the IFN-activated Stat1 only if p38 MAPK signaling (by e.g., LPS) is stimulated in parallel (27). p38 MAPK is known to increase by yet unclear mechanisms the transcriptional activity of several Stat members (32, 33, 34).

To test whether TTP was an effector of the IL-10 anti-inflammatory responses, we measured the amount of secreted TNF-α by ELISA in BMDMs from TTP^−/− and control WT littermates that were treated with LPS with or without pretreatment with IL-10. We measured TNF-α production at 6, 8, and 10 h after LPS stimulation. After 10 h, TNF-α was diminished by IL-10 to ∼20% in control cells, whereas in TTP^−/− cells only a reduction to 60% was achieved (Fig. 2,A). Similar results were obtained if cells were treated simultaneously with LPS and IL-10 (Fig. 2,B), although the contribution of TTP to the IL-10 response was higher in the pretreatment protocol (Fig. 2Α). These data show that TTP contributes to IL-10-mediated inhibition of TNF-α cytokine production.

To investigate whether the incomplete IL-10-mediated inhibition of TNF-α production in TTP^−/− BMDMs resulted from differences in mRNA amounts or in mRNA decay, we analyzed the amounts of TNF-α mRNA in LPS-stimulated TTP^−/− and control BMDMs with or without pretreatment with IL-10. In WT cells, IL-10 caused a reduction of TNF-α mRNA to 40% of the amount present in cells treated with LPS alone, whereas in TTP^−/− cells IL-10 caused a reduction to only 80% of the level in cells treated with LPS alone (Fig. 2^,C). In these experiments, TNF-α mRNA was measured after 2 h of LPS treatment since the amount of TNF-α mRNA peaks at this time point (Ref. 13 and data not shown). To further illustrate the role of TTP in IL-10-mediated decrease of TNF-α mRNA, we analyzed the rate of TNF-α mRNA decay in LPS-stimulated TTP^−/− and WT BMDMs with or without IL-10 pretreatment. Transcription was stopped after 3 h of LPS stimulation by addition of actinomycin D, and the degradation of TNF-α mRNA was followed in 15- min intervals for a total of 45 min after imposing the transcriptional stop (Fig. 3^,A). In LPS-treated control BMDMs, the decay rate of TNF-α mRNA was increased 2.5-fold by IL-10 treatment (half-life t_½ without IL-10 = 32 min; t_½ with IL-10 = 13 min). In LPS-treated TTP^−/− BMDMs, the residual TNF-α mRNA decayed with a 2.5-fold longer half-life (t_[1/2] = 82 min) compared with WT. This difference is similar to the one reported previously (35). Importantly, IL-10 treatment of LPS-stimulated TTP^−/− cells did not increase the decay rate and the remnant mRNA levels were comparable to those in cells treated with LPS alone. LPS-stimulated macrophages produced endogenous IL-10 that is known to mask to some extent the effect of added IL-10 (see Ref. 18). Since TTP was recently shown to target IL-10 mRNA for degradation (36), we asked whether TTP^−/− BMDMs produce more IL-10 after stimulation with LPS. Cytokine measurement revealed that TTP-deficient cells secrete 2- to 3-fold more IL-10 than the WT cells (supplemental Fig. S1⁵). To assess the contribution of endogenous IL-10 to the IL-10-mediated TNF-α mRNA decay, we compared the TNF-α mRNA levels and decay rates in IL-10^−/− and WT cells. IL-10^−/− BMDMs expressed 2-fold more TNF-α than WT cells (Fig. 3,B). Importantly, the reduction of TNF-α mRNA by exogenous IL-10 was more pronounced in IL-10^−/− cells compared with WT cells. The decay rate of TNF-α mRNA was not affected by IL-10 treatment of WT cells stimulated for 1 or 2 h with LPS (Fig. 3,C). However, IL-10 accelerated the decay rate in IL-10^−/− cells treated for 2 h with LPS, but not for 1 h (Fig. 3,D). Therefore, endogenous IL-10 can mask to a certain extent the effect of exogenous IL-10 on the mRNA decay. In addition, the IL-10-mediated increase in mRNA decay rate becomes more apparent at later time points of LPS treatment, consistent with the need for new protein synthesis (e.g., TTP). Note that the duration of LPS treatment in Fig. 3,A was 3 h and that the differences in IL-10-imposed inhibition of TNF-α production in TTP^−/− vs WT cells increased with time of LPS treatment (Fig. 2, A and B).

To further substantiate the role of TTP in the anti-inflammatory effects of IL-10, we measured the IL-10-mediated reduction of IL-1α protein and mRNA in TTP^−/− and WT cells (Fig. 3, E and F). Efficient LPS-stimulated production of IL-1α, a known IL-10 target (13), depends on a second stimulus, such as ATP, that activates the IL-1-processing function of the inflammasome (37). Although IL-1α is not a direct substrate of caspase 1, production of mature IL-1α has been shown to be inflammasome/caspase 1 dependent (37). BMDMs from WT and TTP^−/− animals were stimulated with LPS (10 h) and ATP (1 h before the collection of supernatants) in the presence or absence of IL-10. IL-10 caused a decrease of IL-1α protein to 20% of LPS plus ATP-treated WT cells, whereas in TTP^−/− cells the IL-1α production was reduced only to 60% of the LPS plus ATP-treated samples (Fig. 3,E). Similar differences between WT and TTP^−/− cells were determined also for the IL-10-mediated decrease in IL-1α mRNA (Fig. 3 F).

To rule out that the absence of TTP affected activation of Stat3 by IL-10 that might result in reduced IL-10 responsiveness of TTP^−/− cells, the IL-10-induced tyrosine phosphorylation of Stat3 was determined in LPS-stimulated TTP^−/− and WT BMDMs in the presence or absence of IL-10. The level of tyrosine-phosphorylated Stat3 was under all conditions comparable in both genotypes (supplemental Fig. S2A). In addition, the stimulatory effect of LPS was similar in TTP^−/− and WT cells as judged by the activation of p38 MAPK (supplemental Fig. S2B). Thus, a different activation of the critical proinflammatory (p38 MAPK) and anti-inflammatory (Stat3) components in the WT and TTP^−/− cells could be excluded as a reason for the observed differences in IL-10 responses.

These data establish that TTP plays an important role in IL-10-mediated down-regulation of two critical inflammatory cytokines (TNF-α and IL-1α).

IL-10 and IL-6 are both known to activate Stat3 yet only IL-10 exhibits anti-inflammatory properties (38, 39, 40). To examine whether and to what extent IL-6 was able to stimulate TTP expression, BMDMs were treated with LPS with or without IL-6. Interestingly, IL-6 was able to increase TTP expression in LPS-treated macrophages almost to the same extent as IL-10 (Fig. 4,A). Yet, consistent with the known properties of IL-6, IL-6 was not able to inhibit TNF-α (Fig. 4,B) and IL-1α (Fig. 4,C) in either WT or TTP^−/− BMDMs. These data also implicate that the up-regulation of TTP expression by IL-10 cannot solely explain the role of TTP in the anti-inflammatory effects of this cytokine. We speculated that the role of TTP in IL-10 responses might be explained by an IL-10-dependent increase in TTP activity. TTP function is known to be negatively regulated by p38 MAPK and its downstream kinase MK2 (41). IL-10 has been reported to modestly inhibit p38 MAPK in the later phases of LPS treatment (18). The reduction of p38 MAPK activity is caused by the IL-10-mediated up-regulation of the dual-specificity phosphatase DUSP1 (Ref. 18 and Fig. 4,F). We reasoned that such a reduction in p38 MAPK could relieve the p38 MAPK-dependent inhibition of TTP, thereby increasing the ability of TTP to down-regulate its target mRNAs. To investigate whether IL-6 was also able to reduce p38 MAPK activity, BMDMs were stimulated with LPS alone or along with either IL-10 or IL-6 and p38 MAPK activity was monitored after 2, 4, and 6 h. Treatment of LPS-stimulated cells with IL-10 caused a modest (3-fold) but consistent inhibition of p38 MAPK at 2, 4, and 6 h compared with the samples treated with LPS alone (Fig. 4,D). To better document the IL-10 effect on p38 MAPK phosphorylation, the complete time course of IL-6- and IL-10-treated samples was run on a single gel (supplemental Fig. S3A). The LPS-induced activation of p38 MAPK at an early time point (30 min) was not affected by IL-10 cotreatment (Fig. 4,E). Importantly, p38 MAPK activity was not reduced by treatment with IL-6 at any time point examined (Fig. 4, D and E, and supplemental Fig. S3A). These findings are in agreement with the induction of mRNA for the MAPK phosphatase DUSP1 by IL-10 but not IL-6 in LPS-treated macrophages (18). To further support the role of DUSP1 in the IL-10-mediated effect on p38 MAPK activity, we examined p38 MAPK phosphorylation in DUSP1^−/− BMDMs treated with LPS alone or along with either IL-10 or IL-6 for 4 h. In DUSP1^−/− cells, IL-10 was no longer able to reduce p38 MAPK phosphorylation after treatment (Fig. 4 F). Consistent with the IL-10-mediated decrease in p38 MAPK phosphorylation, at the 4 h-time point, the induction of DUSP1 by IL-10 plus LPS compared with LPS alone was more apparent than after a shorter treatment (1 h; supplemental Fig. S3B). The different effect of IL-10 and IL-6 on p38 MAPK is in agreement with the anti-inflammatory properties that are exhibited by IL-10 but not IL-6. The difference between IL-10 and IL-6 with regard to their anti-inflammatory properties has been attributed to SOCS3 that, as part of the negative feedback loop, binds to the gp130 subunit of the IL-6 receptor but not to the IL-10 receptor (38, 39, 40). Thus, SOCS3, a Stat3 target gene, is able to inhibit signaling elicited by IL-6 but not by IL-10. Consequently, IL-10 induces a prolonged Stat3 activation, whereas IL-6-mediated Stat3 activation is rapidly shut down. Consistently, in LPS-treated BMDMs IL-10 and IL-6 caused a comparable Stat3 activation after 30 min of cytokine treatment, whereas after 2 h Stat3 remained active only in cells stimulated with IL-10 but not IL-6 (supplemental Fig. 3C). These data suggest that a sustained Stat3 activation is needed for inhibition of p38 MAPK.

Although p38 MAPK is needed for TTP expression and protein stability (24, 25, 42), the kinase negatively regulates TTP activity at least at two levels. First, phosphorylation of TTP at Ser⁵² and Ser¹⁷⁸ by MAPK-activated protein kinase kinase 2 (MK2), a kinase downstream of p38 MAPK, provides binding sites for 14-3-3 proteins that reduce the destabilizing activity of TTP (23, 43). Second, phosphorylation of TTP by MK2 was reported to negatively regulate its binding to AREs (25). To show that the IL-10-mediated decrease in p38 MAPK activity was able to change TTP properties in terms of its phosphorylation and binding to AREs, we analyzed the mobility of TTP in SDS-PAGE and binding of TTP to AREs in EMSA experiments. TTP appears in SDS-PAGE in the form of multiple bands that reflect various degree of predominantly p38 MAPK-dependent phosphorylation (24). To demonstrate p38 MAPK effects on TTP phosphorylation and binding to AREs, we first used an inducible expression of TTP in HeLa-Tet-off cells. This system allows manipulation of p38 MAPK activity without affecting the transcription of the TTP gene that is known to require p38 MAPK activity (27, 42). Stimulation of HeLa-Tet-off cells expressing TTP (i.e., without tetracycline) with anisomycin (a p38 MAPK agonist (44)) resulted in a more pronounced appearance of a slower migrating hyperphosphorylated TTP band, whereas the p38 MAPK inhibitor SB reduced the amount of the hyperphosphorylated TTP and also the total TTP level due to TTP protein destabilization (Fig. 5,A). The analysis of the same extracts in a RNA EMSA experiment revealed that anisomycin reduced binding of TTP to the TNF-α ARE by 50% (Fig. 5,B), whereas inhibition of p38 MAPK resulted in a similar amount of ARE-bound TTP as in untreated cells (Fig. 5,B) despite a reduced TTP protein level present in that sample (compare SB-labeled lanes in Fig. 5,A and 5B). Consistent with the IL-10-mediated p38 MAPK inhibition, the treatment of BMDMs with LPS plus IL-10 resulted in a shift to less phosphorylated TTP bands in SDS-PAGE if compared with cells treated with LPS alone or with LPS plus IL-6 (Fig. 5,C). This shift to faster migrating TTP bands was more pronounced after 6 h of treatment. To address the IL-10 effect on TTP binding to AREs, we used the murine macrophage cell line RAW 264.7 since we were not able to detect TTP-ARE complexes in RNA EMSA experiments if BMDMs were used. LPS-stimulated RAW 264.7 cells express at least 3- to 5-fold more TTP than BMDMs (data not shown). Similar to BMDMs, LPS plus IL-10 treatment resulted in a more pronounced appearance of the hypophosphorylated TTP if compared with treatment with LPS alone or LPS plus IL-6 (Fig. 5,D). Consistently, LPS plus IL-10 treatment caused an ∼50–80% increase in the formation of TTP-ARE complexes in RNA EMSA experiments compared with LPS alone or LPS plus IL-6 treatments (Fig. 5,E). These data show that IL-10 decreases the phosphorylation of TTP and enhances the in vitro binding of TTP to TNF-α ARE in a manner similar to the pharmacological inhibition of p38 MAPK. These findings are in agreement with an increased TTP activity under conditions of reduced p38 MAPK activation as is the case in IL-10-treated cells. Interestingly, despite the reduced p38 MAPK activity and hence increased proteasome-mediated TTP degradation in IL-10-treated cells, the TTP protein levels were not diminished compared with LPS plus IL-6-treated cells (Fig. 5, C and D). We explain this observation by the sustained Stat3 activation and, hence, a strong Stat3-driven transcription of the TTP gene in IL-10-treated cells (supplemental Fig. 3C).

To test whether p38 MAPK activity protects mRNA from TTP-mediated decay, we examined the stability of TNF-α and IL-1α mRNAs in the presence of the p38 MAPK inhibitor SB in TTP^−/− and control WT BMDMs. Two hours after LPS stimulation, SB was added in combination with actinomycin D. p38 MAPK inhibition caused a 4-fold decrease of mRNA stability of TNF-α (TNF-α without SB: t_½ = 31 min; TNF-α with SB: t_½ = 8 min) and an ∼8- to 10-fold decrease in IL-1α mRNA stability (IL-1α without SB: t_½ >2 h; IL-1α with SB: t_½ = 14 min; Fig. 6). In TTP^−/− cells, the inhibition of p38 MAPK caused only a modest (2-fold) decrease in mRNA stability of TNF-α mRNA (TNF-α without SB: t_½ = 80 min; TNF-α with SB: t_½ = 37 min with SB) and no detectable decrease of IL-1α mRNA stability (IL-1α without SB: t_½ = >2 h; IL-1α with SB: t_½ = >2 h). These data indicate that the augmentation of mRNA decay by inhibition of p38 MAPK is to a large part TTP dependent. The contribution of TTP to the effects of p38 MAPK inhibition depends on the nature of the target mRNA.

We conclude that the IL-10-mediated reduction of p38 MAPK activity increases TTP-dependent mRNA decay. The combined effect of IL-10 on p38 MAPK activity and TTP expression contributes to the anti-inflammatory properties of IL-10.

This study provides evidence that the mRNA- destabilizing factor TTP plays an important role in anti-inflammatory effects of IL-10. BMDMs deficient in TTP show a strongly reduced anti-inflammatory response to IL-10. Two cytokines, TNF-α and IL-1α, were here demonstrated to be inhibited by IL-10 in a TTP-dependent manner. IL-10 appears to regulate TTP function in two ways (Fig. 7). First, IL-10 increases TTP expression in LPS-treated BMDMs. This augmented expression depends on Stat3, the key immediate effector of IL-10 effects. Second, IL-10 reduces at later time points the activity of p38 MAPK in LPS-treated BMDMs, hereby releasing TTP from the p38 MAPK-mediated inhibition. Since the increased expression of TTP is not sufficient to initiate an anti-inflammatory response (as demonstrated by IL-6-mediated up-regulation of TTP), we conclude that the TTP-dependent IL-10 effects can be best explained by the combination of higher TTP expression and reduction of p38 MAPK activity.

TTP is known for its mRNA-destabilizing activity most notably, but not exclusively, toward TNF-α mRNA via AREs in 3′ UTR. Thus, the IL-10-mediated rise in TNF-α decay rate is consistent with the elevated TTP levels and activity. In addition to TNF-α, our data suggest that IL-1α is also a TTP target. The 3′ UTR of the murine IL-1α (GenBank accession no. NM_010554; http://www. ncbi.nlm.nih.gov/GenBank/) contains three elements (UAUUUAUA, AUAUUUAU, and UAUUAUUUAU) with similar sequences as the nonameric canonical sequence (UUAUUUAUU) recognized by TTP (45, 46). IL-1α has so far not been described as TTP target despite of several recently published global screens for novel TTP targets (36, 47). However, these searches revealed that in such global screens important targets might be missed (e.g., TNF-α in the study of Stoecklin et al. (36)). In addition, the finding that TTP regulates also mRNAs with no obvious AREs in 3′ UTR adds another degree of complexity to searches for TTP targets (47). The role of TTP in IL-10 responses is also in agreement with a study reporting IL-10-mediated ARE-dependent destabilization of CXCL1 mRNA (16). Importantly, in that study, the IL-10-induced mRNA decay was not detectable before 2 h of LPS/IL-10 treatment. This time point correlates well with the appearance of TTP expression (42) and with our data showing that IL-10-mediated decay of TNF-α mRNA is not apparent before 2 h of LPS treatment.

TTP-deficient cells are known to produce 2- to 3-fold more TNF-α (both mRNA and cytokine) than WT cells (26, 48). At the same time, IL-10 mRNA was shown to be targeted by TTP for degradation (36) and we found that the TTP-deficient BMDMs produce more IL-10 cytokine (supplemental Fig. S1). Thus, TTP cells produce more of both the proinflammatory TNF-α and the anti-inflammatory IL-10. Yet, the TTP cells still respond to IL-10 treatment (by e.g., Stat3 activation) and LPS treatment (by e.g., p38 MAPK activation) similarly as the WT cells (Fig. 4). We conclude that despite the 2- to 3-fold higher production of pro- and anti-inflammatory cytokines the TTP-deficient cells display comparable immediate-early response to pro- (LPS) or anti-inflammatory (IL-10) stimuli as WT cells.

We were not able to detect any differences in the TTP expression or p38 MAPK activity, the two key factors influencing the TTP function, when we compared the two protocols for IL-10 treatment. Thus, the higher absolute contribution of TTP to IL-10 responses in the pretreatment protocol suggests that IL-10 induces a cofactor of TTP during the pretreatment period. TTP serves as an adaptor protein linking mRNA to the mRNA processing and degradation machinery located in stress granules and processing bodies (21, 22). Some of the components of these complexes may be up-regulated by IL-10 during the pretreatment phase to enhance TTP function. Alternatively, IL-10-activated Stat3 may increase transcription of micro-RNAs such as micro-R16 that was shown to assist TTP in degradation of ARE-containing mRNAs (49). The influence of the experimental protocol on the IL-10 effects may also partially explain the discrepancy between our study and the study of Kontoyiannis et al. (14) who found no evidence for a role of TTP in IL-10 responses in the simultaneous treatment protocol. Instead, they proposed a rapid (within 15 min of IL-10 treatment) inhibition of p38 MAPK as the main mechanism of IL-10 action. We did not detect any inhibition of p38 MAPK activity by IL-10 at early time points regardless of applied protocol for IL-10 treatment (supplemental Fig. S2, A and B). These data are in agreement with several other studies showing no effect of IL-10 on p38 MAPK activity at early time points (2, 12, 50, 51). In a kinetics analysis of p38 MAPK activity, IL-10 was found to exhibit a modest inhibitory effect on p38 MAPK activity at later time points (after 3 h of LPS plus IL-10 treatment) (18). The IL-10-mediated inhibition of p38 MAPK correlated with the induction of the dual-specificity phosphatase DUSP1, a MAPK phosphatase (18). We observed a similar reduction in p38 MAPK phosphorylation in terms of both, the magnitude and kinetics, and found that DUSP1 was required for this effect. The inhibition of p38 MAPK was caused specifically by IL-10 but not by IL-6 and correlated with the ability of IL-10 (but not IL-6) to induce sustained Stat3 activation. Consistently, IL-10 but not IL-6 was reported to increase expression of DUSP1 in LPS-treated macrophages (18). DUSP1 function as a critical p38 MAPK-inactivating enzyme provides the most likely explanation for the susceptibility of DUSP1^−/− mice to LPS-mediated toxic shock (52, 53). We speculate that the inhibition of the proinflammatory p38 MAPK by IL-10 at the later stages of macrophage stimulation is a key factor in immune homeostasis since p38 MAPK influences inflammation-related transcription, RNA stability, translation, as well as secretion. To mimic the inhibition of p38 MAPK by IL-10 at the late phase of stimulation, we used the p38 MAPK inhibitor SB and examined the effect of p38 MAPK on the TTP-dependent decrease of target mRNA molecules. Interestingly, whereas in the case of TNF-α a different decay rate (i.e., different decay rate in TTP^+/+ and TTP^−/− cells) was observed also without p38 MAPK inhibition, for IL-1α mRNA the TTP-dependent decay was detectable only if p38 MAPK was inhibited (Fig. 5). A similar requirement for p38 MAPK inhibition has been recently described also for the TTP target CXCL1 (KC) mRNA (54). These findings indicate that the p38 MAPK-mediated control of TTP activity has a differential impact on target mRNAs. Although the TTP-dependent decay of some mRNAs (e.g., TNF-α) proceeds in the presence of p38 MAPK activity, the degradation of other mRNAs (e.g., IL-1α and CXCL1) is strongly dependent on the kinase inhibition. This mechanism may play an important role in the specificity of TTP-mediated RNA decay: depending on the activation status of p38 MAPK or other kinases implicated in regulation of TTP activity (e.g., ERK of MK2), TTP would discriminate between various targets. The critical role that p38 MAPK plays in TTP-mediated RNA decay also suggests that the time point that is taken to determine RNA stability has a decisive effect on the outcome of the assay. The activation/inactivation profile of p38 MAPK is likely to vary between different experimental settings (e.g., amount and quality of LPS, the use of primary or immortalized cells, the origin of primary cells such as bone marrow or peritoneum) so that this aspect may also contribute to the variability of published data. For example, we did not observe an IL-10-mediated increase in TNF-α mRNA decay in peritoneal-derived macrophages.

Suppression of inflammatory responses by IL-10 is one of the key features in immune homeostasis. Despite many years of research, the question of how a single cytokine can specifically inhibit various inflammatory reactions with such a high efficiency remains unresolved. Recent studies suggest that known as well as yet unknown IL-10 effectors interfere on different levels and by different mechanisms with the intracellular inflammatory networks. This multitasking system employed by IL-10 is likely to be essential for the efficiency and specificity of the anti-inflammatory properties of IL-10. For example, the efficient and dominant inhibition of Tnf gene transcription by IL-10 still requires the remaining TNF-α mRNA to be removed from the system, i.e., by TTP. Although the primary IL-10-elicited signaling events that involve the IL-10 receptor, Jak1 and Tyk2 kinases, as well as the transcription factor Stat3, are to a large part common to all cell types, the more complex effects downstream of Stat3 may be cell type dependent. Thus, the function and activity of the Stat3-induced IL-10 effectors may be regulated by the environment within the particular cell type, hereby helping to explain the still incoherent and sometimes contradictory studies of the anti-inflammatory effects of IL-10.

We thank S. Badelt for excellent technical help and Bristol-Myers Squibb for the permission to use Dusp1^−/− mice.

The authors have no financial conflict of interest.