Antibodies against total ERK1/2 (#9102), phospho-ERK1/2 (Thr202/Tyr204; #9101), total IκBα (#9242), total JNK1/2 (#9252), phospho-p38MAPK (Thr180-Tyr182; #9211), total MK2 (#3042), phospho-MK2 (Thr334; #3042), total MSK1 (#3489), phospho-MSK1 (Thr581; #9595), phospho-c-Jun (Ser73; #9164) and phospho-IKKα/β (Ser176/180; #2697) were purchased from Cell Signaling Technology. Antibodies to Phospho-CREB (Ser133, #06-519) was from Millipore, anti-p38α (#sc-535) and anti-DUSP1 (#sc-373841) were from Santa Cruz, anti-active phospho-JNK1/2 (Thr183-Tyr185; #MAB1205) from R&D System, and anti-α-Tubulin (#T9026) from Sigma. Secondary antibodies from Invitrogen (Waltham, Massachusetts, United States) included Alexa Fluor 680 donkey α-sheep IgG (H + L) (#A21102), Alexa Fluor 680 goat α-rabbit IgG (H + L) (#A21109), Alexa Fluor 700 goat α-mouse IgG (H + L) (#A21036).

Kinase inhibitors SB203580 (inhibits p38α/p38β (Kuma et al., 2005)) was purchased from Selleckchem, and SB747651A (inhibits MSK1, (Naqvi et al., 2012)) from Axon Medchem. PD184352 (MKK1 inhibitor, (Kuma et al., 2005)) and BIRB0796 (p38α/p38β/p38γ and p38δ inhibitor, (Kuma et al., 2005)) were from the Division of Signal Transduction Therapy (DSTT); University of Dundee (Dundee, UK). JNK-IN-8 (JNK inhibitor (Zhang et al., 2012)) was from Calbiochem.

Activated GST-p38α, -p38β, -p38γ and p38δ were obtained from the Division of Signal Transduction Therapy (DSTT); University of Dundee (Dundee, UK) (https://mrcppureagents.dundee.ac.uk). pGEX4T-p38γD171A (GST-p38γKD), pGEX4T-p38δD168A (GST-p38δKD), pGEX4T-p38δ (GST-p38δ), pGEX6P-MSK1(GST-MSK1) and pGEX6P-CREB (GST-CREB) were from the DSTT, expressed in E. coli strain BL21 and purified as described in (Knebel et al., 2001).

All mice were housed in specific pathogen‐free conditions in the CNB‐CSIC animal house. Animal procedures were performed in accordance with national and EU guidelines, with the approval of the Centro Nacional de Biotecnología Animal Ethics Committee, CSIC and Comunidad de Madrid (Reference: PROEX 316/15 and PROEX 071/19). Adult mice 12-week-old C57BL/6J-WT, p38γ/δ−/−, -p38γ−/− and -p38δ−/− were used in this work.

Bone marrow derived macrophages (BMDM) lacking p38γ, p38δ or p38γ/δ were obtained from adult mouse femur and tibia as described elsewhere (Risco et al., 2012; Alsina-Beauchamp et al., 2018). Briefly, bone marrow cells were differentiated for 6 days on bacteria-grade plastic dishes in DMEM with 20% FBS and 30% L929 cell-conditioned media. Adherent cells were collected and plated (0.5 × 106 cells/plate) in DMEM with 0.05% FBS. After 12 h, BMDMs were stimulated in 0.1%–1% serum with 100 ng/mL LPS (Sigma-Aldrich) or with 250 ng/mL unmethylated CpG oligonucleotide (CpG-ODN, ODN-1668) (InvivoGen). Murine macrophage Raw 264.7 cells were cultured in DMEM with 10% FBS, Penicillin (100 U/mL), Streptomycin (100 μg/mL) and L-glutamine (2 mM), and stimulated with 100 ng/mL LPS. When indicated, cells were pre-treated for 1 h with DMSO, SB203580, BIRB0796, PD184352 or SB747651A. Cells were lysed in lysis buffer ([50 mM Tris-HCl (pH 7.5), 1 mM EGTA, 1 mM EDTA, 50 mM sodium fluoride, 10 mM sodium β-glycerophosphate, 5 mM pyrophosphate, 0.27 M sucrose, 1% (vol/vol) Triton X-100] plus 0.1% (vol/vol) 2-mercaptoethanol, 0.1 mM phenylmethylsulfonyl fluoride, 1 mM benzamidine and 1 mM sodium orthovanadate). Lysates were centrifuged at 20,800 g for 15 min at 4°C, the supernatants removed, quick frozen in liquid nitrogen and stored at −80°C until used.

For mRNA expression analysis, BMDM were lysed with NZYol (NZYtech) and the RNA extracted using a standard protocol with chloroform-isopropanol-ethanol.

p38MAPKs were assayed using Myelin Basic Protein (MBP) as substrate (Cuenda et al., 1997). Briefly, kinase assay were set in a 30 µL final phosphorylation reaction mixture containing MBP (0.33 mg/mL), active p38MAPK (0.5 U/mL) and 50 mM Tris-HCl pH 7.5, 0.1 mM EGTA, 10 mM MgCl2 and 0.1 mM [γ32P]ATP (Amersham; specific activity: ∼3 × 106 cpms). The reactions were carried out at 30°C for 60 min and stopped by spotting the phosphorylation reaction mixture onto P81 filtermats, washed four times in 75 mM phosphoric acid to remove ATP, washed once in acetone, and then dried and counted for radioactivity incorporated into MBP. To study the MSK1 phosphorylation by active p38MAPKs, these kinases were matched for activity against MBP. Each p38MAPK (0.5 U/mL) was incubated for 60 min or the indicated times at 30°C with Mg [γ32P]ATP (specific activity: ∼3 × 106 cpms) or Mg-ATP plus 1 µM GST-MSK1. The samples were denatured by adding 4 x SDS-PAGE sample buffer containing 1% (v/v) 2-mercaptoethanol, electrophoresed and autoradiographed. Phosphorylated MSK1 was quantified using the Fiji program.

MSK1 (2 μg, 0.9 µM) was first activated with p38α or p38δ (0.5 U/mL) in kinase assay buffer (50 mM Tris-HCl pH 7.5, 0.1 mM EGTA and 10 mM MgCl2) and 0.1 mM ATP. After the indicated times at 30°C, fractions containing active MSK1 were diluted 1:100 in kinase assay buffer plus 0.1 mM [γ32P]ATP (specific activity: ∼3 × 106 cpms) or 0.1 mM ATP, and GST-CREB (1 μg, 0.74 µM). The reactions were carried out at 30°C for 15 min and terminated by adding 4 x SDS-PAGE sample buffer containing 1% (v/v) 2-mercaptoethanol. Reaction samples were electrophoresed and autoradiographed. Phosphorylated MSK1 and CREB were quantified using the Fiji program.

Protein samples were resolved in sodium dodecyl sulphate-polyacrylamide gel electrophoresis (SDS-PAGE) and transferred to nitrocellulose membranes, blocked (30 min at 25°C) in 50 mM Tris/HCl pH 7.5, 0.15 M NaCl and 0.05% (v/v) Tween (TBST buffer) with 10% (w/v) dry milk. Then membranes were incubated in TBST buffer with 5% (w/v) dry milk and 0.5–1 μg/mL antibody (2 h at 25°C or overnight at 4°C). Proteins were detected using fluorescently labelled secondary antibodies and the Odyssey infrared Imaging System (LI-COR Biosciences, Lincoln, Nebraska, United States).

cDNA for real-time quantitative PCR (qPCR) was generated from total RNA using the High Capacity cDNA Reverse Transcription Kit (Applied Biosystems). Real-time qPCR reactions were performed in triplicate as described (Risco et al., 2012; Alsina-Beauchamp et al., 2018) in MicroAmp Optical 384-well plates (Applied Biosystems). PCR reactions were carried out in an ABI PRISM 7900HT (Applied Biosystems) and SDS v2.2 software was used to analyse results by the Comparative Ct Method (ΔΔCt). X-fold change in mRNA expression was quantified relative to non-stimulated wild-type cells, and β-actin mRNA was used as control. Primers used were:

IL-1Ra:

forward 5′-GGC​AGT​GGA​AGA​CCT​TGT​GT and

revers 5′-CAT​CTT​GCA​GGG​TCT​TTT​CC;

β-actin:

forward 5′-AAG​GAG​ATT​ACT​TGC​TCT​GGC​TCC​T and

revers 5′-ACT​CAT​CGT​ACT​CCT​GCT​TGC​TGA​T;

DUSP1:

forward 5′-TGG​GAG​CTG​GTC​CTT​ATT​TAT​T and

revers 5′-GAC​TGC​TTA​GGA​ACT​CAG​TGG​AA.

Data were processed using Student’s t-test. In all cases, p values < 0.05 were considered significant. Data are shown as mean ± SEM.

Signalling pathways activated in response to the Toll-like receptor 4 (TLR4) ligand lipopolysaccharide (LPS) (Figure 1A) were analysed in bone marrow derived macrophages (BMDM) from WT, p38γ−/−, p38δ−/− and p38γ/δ−/− mice. As reported before, the activation of ERK1/2 was impaired in p38γ/δ−/− cells, whereas p38α and JNK1/2 activation was not affected, as determined by immunoblotting with phosphospecific antibodies (Risco et al., 2012; Alsina-Beauchamp et al., 2018) (Figure 1B). IKKα/β phosphorylation and TLR4-induced NF-κB inhibitor IκBα proteolysis were also unaffected in all genotypes (Figure 1B). In contrast, the phosphorylation of the Mitogen- and Stress-activated Kinase (MSK) 1 was notably diminished in LPS stimulated p38δ−/− and p38γ/δ−/− cells (Figure 1C). MSK1 total levels were similar in all genotypes (Figure 1C). MSK1phosphorylation was also decreased in p38δ−/− BMDM stimulated with the TLR9-ligand the unmethylated CpG oligonucleotide (ODN) (Figure 1D), which shows that this effect is not restricted just to TLR4 signalling. MSK1 is activated downstream of ERK1/2 and p38α (Deak et al., 1998; Reyskens and Arthur, 2016). Consistent with this, MSK1 phosphorylation was decreased in WT LPS-stimulated Raw 264.7 macrophages and BMDM treated with the MKK1 inhibitor PD184352, to block ERK1/2 activation, or with the p38α/p38β inhibitor SB203580, to block p38α, or with both inhibitors together (Figures 2A–D). Treatment with high concentration (10 µM) of the pan-p38MAPK inhibitor BIRB0796, which at 0.1 µM inhibits p38α/p38β, at 1 µM p38γ, and at 10 µM inhibits p38δ (Kuma et al., 2005), caused a decreased in MSK1 phosphorylation larger than incubation with lower concentrations (0.1 or 1 µM) of BIRB0796 or with SB203580 alone (Figures 2A–D), supporting the idea that p38δ regulates MSK1 activation in macrophages in response to LPS. As expected, the compound PD184352 impaired ERK1/2 phosphorylation, and both SB203580 and BIRB0796 the phosphorylation of MK2, which is a p38α substrate, in LPS-stimulated macrophages (Figures 2A, C). BIRB0796 (10 µM) also inhibits JNK1/2 (Kuma et al., 2005), we then treated BMDM with the specific JNK inhibitor, JNK-IN-8 (Zhang et al., 2012), to examine if the decrease on MSK1 phosphorylation was mediated by JNK1/2 inhibition, and found that MSK1 phosphorylation was not blocked by JNK-IN-8 (Figures 2A–D), which inhibited c-Jun phosphorylation (Figure 2A).

MSK1 phosphorylation was reduced in p38δ−/− macrophages. Since MSK1 is an ERK1/2 and p38α substrate, but the activation of these two kinases was not affected in p38δ−/− macrophages in response to LPS, we hypothesised that p38δ could directly phosphorylate MSK1. Thus, we next examined if recombinant MSK1 was phosphorylated by active recombinant p38δ in in vitro kinase assay using Mg [γ32P]-ATP and p38α, p38β and p38γ as comparative controls. All p38MAPKs were used at the same specific activity towards myelin basic protein (MBP), which is a pan-p38MAPK substrate. GST-MSK1 was phosphorylated by p38α, p38β and p38δ, but not by p38γ (Figures 3A, B). The rate of phosphorylation of MSK1 by p38α and p38δ was similar under our experimental conditions (Figure 3C).

MSK1 autophosphorylates in multiple sites (McCoy et al., 2005), thus, there is a possibility that the presence of p38δ might be helping MSK1 autophosphorylation. To examine this, we incubated MSK1 and Mg[γ32P]-ATP, in the presence or absence of recombinant non-activated p38δ (p38δ(na)), that was not previously activated by MKK6 in vitro, p38δ inactive mutant (p38δD168A, a p38δ kinase dead (p38δKD)), or p38γ inactive mutant (p38γD171A; p38γKD) as control (Figure 3D). All recombinant p38δ preparations contained the GST-tagged (GST-p38s) and the non-GST-tagged (p38s) protein, probably due to the cleavage of the GST part after purification (Figure 3D). We found similar MSK1 autophosphorylation in the absence of p38 and in the presence of p38δKD and p38γKD; however, MSK1 autophosphorylation/phosphorylation was increased in the presence of p38δ (Figure 3D). In addition, we observed autophosphorylation of the wild type p38δ(na), but not p38δKD or p38γKD (Figure 3D). These results indicate that basal kinase activity of recombinant p38δ could account for the increase in MSK1 phosphorylation (Figure 3D). Additionally, both p38δ and p38δKD phosphorylation was significantly increased in the presence of MSK1 (Figure 3D), suggesting that p38δ is directly phosphorylated by MSK1.

To check if MSK1 and p38δ phosphorylate each other we studied their phosphorylation in the presence or absence of 10 µM BIRB0796 to inhibit p38δ, or 100 µM SB747651A to inhibit MSK1 (Kuma et al., 2005; Naqvi et al., 2012). As expected, incubation with the BIRB0796 inhibitor blocked p38δ autophosphorylation and decreased MSK1 phosphorylation by p38δ to similar levels to that of MSK1 autophosphorylation (Figure 3E). BIRB0796 did not impaired either p38δKD or p38δ phosphorylation in the presence of MSK1 (Figure 3E). These data confirmed that p38δ directly phosphorylates MSK1 and strongly suggest that MSK1 directly phosphorylates p38δ (Figure 3F). Incubation with SB747651A blocked MSK1 autophosphorylation, but did not inhibit the phosphorylation of MSK1 by p38δ (Figure 3G). However, SB747651A impaired p38δKD and p38δ phosphorylation by MSK1, but not p38δ autophosphorylation (Figure 3G), showing that MSK1 directly phosphorylates p38δ (Figure 3H).

In vitro MSK1 phosphorylation by p38δ was confirmed by western blot using the antibody against Phospho-T581 (Figure 4A), which is the proline-direct phosphorylation site essential for MSK1 activation directly phosphorylated by ERK1/2 and p38α (McCoy et al., 2005). We then study whether or not phosphorylation by p38δ activates MSK1 in vitro. For this, we use the transcription factor CREB as MSK1 substrate. CREB is a MSK1 physiological substrate (Reyskens and Arthur, 2016). We found that the incubation with Mg[γ32P]-ATP and activated p38δ or p38α enhanced CREB phosphorylation. (Figure 4B). As a positive control, we confirmed that MSK1 was activated with active p38α. The ability of p38δ or p38α to activate GST-MSK1 correlated with the extent of phosphorylation of this kinase (Figure 3C). MSK1 activation by p38δ was confirmed analysing CREB phosphorylation by immunoblot, using the anti-Phospho-CREB (S133) antibody (Figure 4C). This antibody recognized the CREB residue (S133) specifically phosphorylated by MSK1 (Deak et al., 1998). These data show that MSK1 is phosphorylated and activated by p38δ in vitro.

We then evaluated the role of p38δ in mediating the phosphorylation of CREB at S133, and also of its close relative transcription factor ATF1 at the equivalent residue, S63, in cells. We found that CREB and ATF1 are phosphorylated after treatment of WT and p38γ−/− BMDM with LPS (Figure 5A). In contrast, the lack of p38δ significantly decreased CREB and ATF1 phosphorylation in LPS-stimulated p38δ−/− BMDM, and also in p38γ/δ−/− BMDM, although to a less extent (Figure 5A).

One role of MSK in macrophages is to regulate the expression of immediate early genes, such as dual specificity protein phosphatase 1 (DUSP1), through CREB phosphorylation (Arthur et al., 2004; Ananieva et al., 2008). We found that LPS-induced DUSP1 mRNA expression was significantly decreased in p38δ−/− and p38γ/δ−/− BMDM compared to WT cells (Figure 5B). In all WT, p38δ−/− and p38γ/δ−/− macrophages, DUSP1 mRNA expression, in response to LPS, was affected in the presence of p38 inhibitors, SB203580 or BIRB0796, and this effect was more pronounced in p38γ/δ−/− cells and at high concentration of BIRB0796 (Figure 5B). LPS stimulation also induced DUSP1 protein expression in WT macrophages, which was significantly impaired in p38γ/δ−/− and p38δ−/− BMDM (Figures 5C–E). Consistent with this finding, in WT macrophages, the expression of DUSP1 protein induced by LPS was blocked by preincubation with high concentrations of BIRB0796 to levels comparative to those observed after preincubation with a combination of both PD184352 and SB203580 (Figure 5F). Preincubation with PD184352 or SB203580 alone, or with JNK-IN-8 did not affect DUSP1 protein expression (Figure 5F). All these results indicate that p38δ is a key player in the regulation of DUSP1 expression in response to TLR4 activation in macrophages.

MSKs also regulate the transcription of the anti-inflammatory molecule, the IL-1 receptor antagonist (IL-1Ra) in macrophages (Darragh et al., 2010). Consistent with the involvement of p38δ in controlling MSK activation, IL-1Ra mRNA expression was significantly blocked in LPS-stimulated p38δ−/− and p38γ/δ−/− BMDM, compared to WT cells (Figure 5G). All these data indicate that p38δ positively regulates anti-inflammatory signalling.