TNFα was purchased from PeproTech Inc. (Rocky Hill, NJ, United States). 1-Oleoyl-sn-glycerol 3-phosphate sodium salt (LPA, 18:1) was purchased from Sigma-Aldrich Canada (Oakville, ON, Canada). Antibodies against human phospho-MSK1 (Ser-376)/MSK2 (Ser-360), phospho-MSK1 (Ser-212), MSK2 and GAPDH were from R&D Systems Inc. (Minneapolis, MN, United States). Antibodies against human phospho-CREB (Ser-133), phospho-MSK1 (Ser-360), phospho-MSK1 (Thr-581) and MSK1 were purchased from Cell Signaling Technology (Beverly, MA, United States). Antibody to actin was from Sigma–Aldrich Canada (Oakville, ON, Canada). Inhibitors of p38MAPK (SB203580), ERK (PD98059), Rho kinase (Y27632), and NF-κB (Bay11-7082) were purchased from Calbiochem (San Diego, CA, United States). The PI3K inhibitor wortmannin was from Millipore Corporation (St. Charles, MO, United States). MSK inhibitor SB-747651A was obtained from Axon Medchem (Groningen, The Netherlands). Human IL-8 ELISA kit was purchased from BioSource International Inc. (Camarillo, CA, United States). The human cytokine/chemokine Luminex multiplex immunoassay kit (Milliplex^® MAP Kit, detecting MCP-1, IL-6, IL-8, IP-10, and RANTES) was from Millipore Corporation (St. Charles, MO, United States). FBS and DMEM were from Wisent Inc. (St-Bruno, QC, Canada). Propidium iodide and Annexin V-eFluor450 were from BD Pharmingen (Oakville, ON, Canada).

Human primary FLS were obtained from RA patients who were diagnosed according to the criteria developed by the American College of Rheumatology and were undergoing joint surgery (Arnett et al., 1988). Human primary FLS at passage 0 or 1 were purchased from Asterand (Detroit, MI, United States) or isolated from RA synovial membrane specimens collected from RA patients, after informed and written consent was obtained and with the approval of the CHU de Québec-Laval University ethics committee (B14-04-1946). FLS were isolated from synovial membrane specimens as described previously (Faour et al., 2003; Zhao et al., 2008). Cells were cultured under standard conditions (37°C and 5% CO₂) and grown in DMEM supplemented with 10% FBS, penicillin (100 IU), and streptomycin (100 M) as described previously (Zhao et al., 2008, 2014). Cells from three different donors were miscellaneously used in all experiments at passages 2–7.

For the experiments, semi-confluent RAFLS were starved in FBS-free DMEM for 24 h. To evaluate the effect of TNFα, starved cells were pre-incubated with TNFα (80 ng/ml) for 8 h prior to LPA treatment (Supplementary Figure 1). Where indicated, cells were pre-treated with the inhibitors of p38MAPK (SB203580), ERK (PD98059), Rho kinase (Y27632), PI3K (wortmannin), NF-κB (Bay11-7082), and MSK (SB-747651A) at indicated concentrations for 30 min as described previously (Zhao et al., 2008, 2014). In some experiments combinations of signaling inhibitors were also used. Cells were then washed with serum-free DMEM and stimulated with 5 μM LPA in fresh serum-free DMEM containing testing compounds (Supplementary Figure 1). For measurement of cytokine/chemokine synthesis cells were stimulated for 24 h as described previously (Zhao et al., 2008, 2014). Cell culture supernatants were collected and stored at -80°C until the ELISA and the multiplex immunoassay for human cytokines/chemokines were performed. Data points in each of the independent experiments were performed in duplicate. To monitor the levels of MSK phosphorylation RALFS were stimulated with 5 μM LPA for 5 min, a time at which phosphorylation of MSK1 at Ser-376 or of MSK2 at Ser-360 is maximal (Zhao et al., 2014).

IL-8 was monitored using ELISA according to the manufacturer’s protocol. Optical densities were determined using a SoftMaxPro5 plate reader at 450 nm. The detection range of the IL-8 ELISA kit is 15.6-1000 pg/ml.

Human cytokines/chemokines IL-6, IL-8, IP-10, MCP-1, and RANTES were monitored using a Luminex multiplex immunoassay (Milliplex^® MAP Kit) according to the manufacturer’s instructions. The dynamic range of the assay is 3.2-10000 pg/ml. The assay sensitivities (minimum detectable concentrations, mean ± SD, pg/ml) were 0.9 ± 1.3, 0.4 ± 0.7, 8.6 ± 14, 1.9 ± 3.4, and 1.2 ± 1.9 for IL-6, IL-8, IP-10, MCP-1, and RANTES, respectively.

RAFLS were lysed in boiling sample buffer [50 mM Tris-HCl (pH6.8), 10% (v/v) glycerol, 50 mM DTT, 4% (v/v) SDS] for 7–10 min. Equal amounts of protein were loaded on gels and separated by 10% SDS-polyacrylamide gel electrophoresis. Proteins were later transferred from polyacrylamide gel to methanol-soaked PVDF membranes (Pall Canada Ltd., Ville St-Laurent, QC, Canada). Primary antibody incubation was performed overnight at 4°C, 1 h at 37°C, or 2 h at room temperature according to optimization tests. The membranes were washed three times and incubated with appropriate horseradish peroxidase-conjugated secondary antibodies at room temperature for 1 h. Membranes were when washed three times and antibody-antigen complexes were revealed using Western Lightning chemiluminescence reagent according to the manufacturer’s instructions (Perkin Elmer Life Sciences, Woodbridge, ON, Canada).

Unless otherwise stated, the data are from at least three independent experiments presented as mean values ± SEM. Prism 7.0 software was used for all statistical analyses. Statistical significance of the difference between samples of two different treatments was determined by t-test (two-tailed p-value). For studies using samples from more than one treatment, statistical significance between control and treated cells was determined by one-way ANOVA or two-way ANOVA multiple comparison test according to the context. P values less than 0.05 were considered statistically significant.

Our previous study has shown the key role of the MSK-CREB signaling axis in LPA-mediated IL-8 secretion in RAFLS (Zhao et al., 2014). LPA was reported to induce a transient phosphorylation of MSK1 at Ser-376 and/or MSK2 at Ser-360 (p-MSK1/2), as well as phosphorylation of CREB at Ser-133 (Zhao et al., 2014). Since cell preconditioning with TNFα synergistically enhances LPA-mediated cytokine/chemokine synthesis (Zhao et al., 2008), we first examined the effect of TNFα on the kinetics of MSK1/2 and CREB phosphorylation. TNFα also induced a transient increase in p-MSK1/2 that was maximal at 15 min (∼6-fold increase) and returned to near basal levels by 4 h (Figures 1A, 2A,B). After 8 h stimulation with TNFα phosphorylation of MSK1 at Ser-212 (p-MSK1) remained significantly elevated (80 ± 32%, p < 0.001) when compared to the non-treated cells (Figures 2C,D). Further efforts to detect MSK1 phosphorylation at Ser-360 or Thr-581 were fruitless due to the detection threshold of the antibodies and of the chemiluminescence reagent used.

Stimulation of RAFLS with LPA for 5 min led to a six-fold increase in the levels of p-MSK1/2 (Figures 2A,B). LPA stimulation for 5 min after treatment with TNFα for 8 h did not increase the levels of p-MSK1/2 (Figures 2A,B). The levels of p-MSK1 were increased by ∼2-fold in response to LPA stimulation for 5 min. (Figures 2C,D). TNFα treatment for 8 h significantly enhanced the levels of p-MSK1 induced by LPA (Figures 2C,D). The levels of p-MSK1 were increased by 56 ± 16% (p < 0.0001) when compared to LPA-treated samples.

CREB phosphorylation in response to TNFα was maximal at 30 min (∼4.5-fold increase) and slowly declined to baseline levels by 8 h (Figure 1A). In all subsequent experiments with TNFα the time point 8 h, when p-CREB returned to baseline levels, was selected to monitor LPA-induced signaling and functional responses in primed cells. RAFLS stimulation with LPA for 5 min led to a ∼5-fold increase in the levels of p-CREB (Figures 2E,F). TNFα pretreatment for 8 h significantly enhanced LPA-mediated CREB phosphorylation at 5 min but not at the later time points tested (Figures 2E–J).

Overall the data suggest that TNFα priming for 8 h tends to increase LPA-mediated phosphorylation of MSK1/2 and CREB. Since TNFα has no effect on MSK1, MSK2, and CBEB protein expression (Figure 1B), the data would suggest that increased signaling through the MSK1/2-CREB axis may in turn enhance LPA-induced responses in TNFα-primed RAFLS.

Previous studies showed that signaling through ERK, p38MAPK, Rho kinase, and MSK-CREB regulates LPA-induced cytokine/chemokine secretion (Zhao et al., 2008, 2014). In addition to Rho kinase, ERK and p38MAPK, PI3K was also reported to be involved in LPA-mediated IL-6 secretion (Chou et al., 2005) and cell migration (Sengupta et al., 2003; Du et al., 2010; Riaz et al., 2016). MSKs are phosphorylated by ERK and p38MAPK (Arthur, 2008; Vermeulen et al., 2009). To test whether ERK and p38MAPK signal via MSKs in our model, we used kinase-selective inhibitors and first monitored the levels of p-MSK1/2 and p-MSK1 as a read out.

Incubation of RAFLS with inhibitors of p38MAPK, Rho kinase, or PI3K did not affect the level of p-MSK1/2 or p-MSK1 induced by LPA (Figures 3A,B,D, 4A,B,D). In contrast, the ERK inhibitor significantly decreased the levels of p-MSK1/2 (50 ± 20%, p < 0.01) (Figure 3C), and reduced that of p-MSK1 (42 ± 7%, p = 0.0694) induced by LPA (Figure 4C).

Inhibition of ERK activation after priming with TNFα for 8 h significantly reduced the levels of p-MSK1/2 and p-MSK1 induced by LPA by 54 ± 18% (Figure 3C) and 66 ± 8% (Figure 4C), respectively. Inhibition of PI3K in TNFα-primed RAFLS stimulated with LPA significantly decreased the levels of p-MSK1/2 by 48 ± 10% (Figure 3D) but did not affect those of p-MSK1 (Figure 4D).

The next series of experiments evaluated the effects of inhibiting ERK in combination with one or more signaling pathways. Inhibition of both ERK and p38MAPK did not further reduce MSK phosphorylation compared to cells treated with ERK inhibitor alone (Figures 5A,C). Simultaneous inhibition of ERK signaling together with Rho kinase, or p38MAPK and Rho kinase, or p38MAPK and PI3K, or all of the signaling pathways together did not further reduce p-MSK1/2 (Figure 5A). In contrast, the same combinations of inhibitors slightly attenuated p-MSK1 phosphorylated at Ser-212 when compared to cells treated with the ERK inhibitor alone (Figure 5C). The most potent combinations were inhibition of ERK, p38MAPK, and Rho kinase, and that of ERK, Rho kinase, and PI3K signaling (Figure 5C).

Since inhibition of p38MAPK, Rho kinase, or PI3K alone partially inhibited LPA-induced p-MSK1/2 in TNFα primed RAFLS (Figures 3A,B,D, 5B), we performed experiments to determine whether simultaneous inhibition of at least two pathways could further diminish MSK phosphorylation. As shown in Figure 5B, all combinations of inhibitors moderately decreased the levels of p-MSK1/2. However, inhibition of Rho kinase and p38MAPK or Rho kinase and PI3K did not decrease the levels of p-MSK1/2 compared to cells treated with the inhibitor of Rho kinase alone. Combination of p38MAPK and PI3K, or of p38MAPK, PI3K, and Rho kinase inhibitors reduced p-MSK1/2 to an extent greater than inhibition of either signaling pathway alone (Figure 5B). In contrast, these combinations of inhibitors did not reduce p-MSK1 at Ser-212 (Figure 5D). Taken together the data indicate that in both control and TNFα-primed RAFLS stimulation with LPA results in the phosphorylation of MSK1 at Ser-376 and Ser-212 and/or of MSK2 at Ser-360 via ERK. Although combination of p38MAPK and Rho kinase, p38MAPK and PI3K, or of p38MAPK, PI3K, and Rho kinase inhibitors reduced LPA-mediated phosphorylation event at Ser-376 of MSK1 (or Ser-360 of MSK2) in TNFα-primed cells, phosphorylation at Ser-212 of MSK1 was not inhibited.

LPA-mediated CREB phosphorylation at Ser-133 is MSK-dependent in RAFLS (Zhao et al., 2014). In the next series of experiments we assessed the impact of blocking MSK activity or the upstream MAPKs phosphorylation of MSK1/2 on LPA-mediated CREB phosphorylation in RAFLS. As anticipated the ERK inhibitor decreased LPA-mediated CREB phosphorylation by 41 ± 15% and 41 ± 11% (p < 0.01) in control and TNFα-primed cells, respectively (Figures 6A,B). Whereas the inhibitor of ERK or p38MAPK alone reduced LPA-mediated CREB phosphorylation by ∼50%, their combination completely blocked CREB phosphorylation in TNFα-primed cells (Figures 6E,F). The MSK inhibitor SB-747651A fully inhibited LPA-mediated CREB phosphorylation in unprimed cells and reduced the levels of phosphorylated CREB by 54 ± 4% (p < 0.01) in cells pre-treated with TNFα for 8 h (Figures 6C,D). Taken together, the data suggest that in both control and TNFα-primed cells LPA-induced CREB phosphorylation relies on p38MAPK and ERK signaling.

As show in Figure 7 and as previously reported, priming of RAFLS with TNFα for 8 h strongly enhances the subsequent release of IL-8 as well as of other cytokines/chemokines in response to stimulation with LPA (Zhao et al., 2008, 2014). Early reports also highlighted that this super-production of cytokines/chemokines by TNFα-conditioned cells was dependent on LPA₁ and LPA₃ receptor activation (Zhao et al., 2008). In TNFα-primed cells, selective LPA_1/3 antagonist inhibited the super-production of IL-8 (and IL-6 as well) induced by LPA (Zhao et al., 2008). TNFα-primed air pouch tissues in LPA₃ knockout mice release less cytokines/chemokines in response to LPA when compared to control mice, and treatment with a selective LPA_1/3 antagonist was necessary to totally block cytokine/chemokine production in this mouse model of inflammation (Zhao et al., 2011). Though LPA was reported to induce cytokine/chemokine secretion through activation of multiple signaling pathways including MSK1, MSK2, ERK, and p38MAPK (Zhao et al., 2008, 2014), the contribution of MSKs and other kinase cascades to LPA-mediated cytokine/chemokine production after priming with TNFα has not been investigated. In the next series of experiments we investigated in primed and unprimed cells the impact of inhibiting p38MAPK, ERK, Rho kinase, PI3K, MSKs, or the transcription factor NF-κB on LPA-mediated IL-8 secretion. Consistent with previous studies, inhibition of p38MAPK, ERK, Rho kinase, or MSKs significantly reduced LPA-mediated IL-8 synthesis (Figures 8A–D). Inhibition of PI3K and NF-κB also decreased LPA-induced IL-8 secretion (Figures 8E,F).

IL-8 secretion due to TNFα priming was sensitive to inhibition by SB-747651A (Figure 8D), wortmannin (Figure 8E), and the highest dose of Bay11-7082 (Figure 8F), whereas the other signaling inhibitors were without effect. However, control experiments revealed that the effect of the NF-κB inhibitor (Bay11-7082) at 10 μM was cytotoxic for RAFLS whereas all other compounds had no significant impact on cell viability (Supplementary Figures 2, 3). Surprisingly, LPA-induced IL-8 secretion became insensitive to the inhibitors of p38MAPK, ERK, Rho Kinase, MSKs, and NF-κB in TNFα-primed cells (Figures 8A–D,F). Only LPA-induced IL-8 secretion remained sensitive to inhibition by wortmannin after priming with TNFα (Figure 8E).

As neither the p38MAPK, ERK, Rho kinase, nor the NF-κB inhibitors (1 μM) acting independently could inhibit IL-8 in response to LPA in TNFα-primed cells, we next tested various combinations of signaling inhibitors at concentrations that had no significant impact on cell viability (Supplementary Figure 3). The combined inhibition of p38MAPK and ERK that completely blocked the MSK-CREB signaling axis (Figures 5A,C, 6E,F) significantly decreased the production of IL-8 by 61 ± 3% (Figure 9A) when compared to TNFα-primed RAFLS stimulated with LPA. Though not statistically significant, combinations of the p38MAPK inhibitor with those of Rho kinase, PI3K, or NF-κB showed a tendency to more strongly decrease IL-8 secretion compared to other combinations of inhibitors (Figure 9A), thereby suggesting a role for p38MAPK in exacerbating LPA-induced IL-8 release in TNFα-primed RAFLS. In contrast, combined inhibition of Rho kinase and PI3K or of Rho kinase and NF-κB increased the release of IL-8 by 72.5% (p < 0.05) and 40 ± 30%, respectively (Figure 9A).

The pattern of IL-6 secretion by cells treated with the same combinations of signaling inhibitors was similar to that of IL-8. LPA-induced IL-6 secretion after TNFα priming, which was not inhibited by a pre-incubation with either the inhibitor of p38MAPK (not shown) or of ERK (Figure 9B), was reduced by 41 ± 3% when these two pathways were inhibited (Figure 9B). Combination of Rho kinase and PI3K inhibitors significantly enhanced (260%, p < 0.01) the release of IL-6 compared to the control TNFα-primed cells stimulated with LPA. MCP-1, IP-10, and RANTES exhibited different inhibition patterns with combinations of inhibitors. The release of IP-10 and of RANTES was not affected by any of the combinations of inhibitors (Figures 9C,E) whereas combinations of Rho kinase and PI3K, Rho kinase and NF-κB, or PI3K and NF-κB inhibitors decreased MCP-1 secretion to 48 ± 23%, 34 ± 21%, and 53 ± 17% of control LPA-stimulated cells, respectively (Figure 9D).