MSC protects against ALF via PGE2
Paralyzed with our previous study13, MSC protected against ALF through PGE2, as revealed by plasma levels of ALT and AST (Figure 1A) and liver damage in HE staining (Figure 1B). TUNEL and PI staining confirmed the hepatocyte necrosis and apoptosis was decreased when MSC infusion, especially in MSC-COX2(+) group, however, MSC-COX2(-) failed to decrease hepatocyte necrosis and apoptosis (Figure 1C and 1D). Taken together, MSC protected hepatocyte death against ALF via PGE2.
MSC ameliorates inflammation in ALF
Liver inflammation is associated with hepatocyte damage and death, therefore we examined the inflammatory response in liver after MSC infusion. Notably, mRNA expression levels of inflammatory cytokines and chemokines, including CCL2, IL-1β, iNOS and TNF-α were significantly inhibited in MSC group, and in MSC-COX2(+) group, the levels of these cytokines decreased more obviously. By contrast, MSC-COX2(-) group, which inhibited secretion of PGE2 from MSC, failed to decrease these inflammatory cytokines (Figure 2A). NF-κB signaling plays a great role in inflammation15. We demonstrated that LPS/D-Gal induced ALF significantly activated NF-κB signaling in liver, as the increased expression of the phosphorylation of IKK-β and P65 (p-IKK-β and p-P65) (Figure 2B). MSC or MSC-COX2(+) group exhibited an ameliorated phenotype, while MSC-COX2(-) did not show this phenotype (Figure 2B). Next, we aimed to explore the mechanism of MSC on NF-κB signaling inhibition. TGF-β-activated kinase 1 (TAK1) is an intracellular hub molecule that regulates NF-κB signaling pathways which plays a key role in cell survival and death16. Our results demonstrated that TAK1 signaling was inhibited after MSC infusion, as the expression of the phosphorylation of TAK1 (p-TAK1), JNK (p-JNK), P38 (p-P38) and c-Jun (p-c-Jun) were significantly reduced in MSC or MSC-COX2(+) group, while these TAK1 substrates still expressed high in MSC-COX2(-) group (Figure 2C).
To confirm the role of TAK1 in ALF, we further used 5Z-7-ox, a specific TAK1 inhibitor, to inhibit TAK1 before LPS/D-Gal administration. Notably, TAK1 inhibition protected against liver damage, shown by normalized serum ALT and AST and reduced necrosis on histological analysis (Figure 2D and 2E). Treatment of 5Z-7-ox significantly suppressed TAK1 activation, reduced activation of its downstream signaling pathways, including p-JNK, p-P38, p-c-Jun and NF-κB signaling (Figure 2E). Collectively, these observations suggested that MSC-derived PGE2 protected ALF through suppressing liver inflammation in a TAK1-NF-κB pathway.
MSC inhibits inflammasome activation of macrophages in ALF
Previous studies have confirmed the role of inflammasome activation in LPS/D-Gal induced liver injury, which accelerated liver inflammation17. Next we explored whether MSC infusion could inhibit NLRP3 inflammasome activation in liver. The protein levels of NLRP3, caspase1 p20 and mature-IL-1β were decreased in MSC or MSC-COX2(+) group (Figure 3A). In parallel, serum concentrations of IL-1β and the activity of caspase 1 enzyme in liver tissues confirmed the inhibition of NLRP3 inflammasome by MSC-derived PGE2 (Figure 3B and 3C). However, MSC-COX2(-) failed to inhibit NLRP3 inflammasome activation in liver tissues. In order to identify the cell population responsible for increased production of NLRP3 inflammasome, the double immunohistochemistry staining was performed. The results demonstrated that NLRP3 was mostly activated in macrophages in liver, MSC or MSC-COX2(+) could inhibited NLRP3 activation in liver macrophages (Figure 3D). To confirm this result, we treated mouse hepatocyte cell line AML12 or BMDM with 1μg/ml LPS for 24h in vitro. As showed in Figure 3E and 3F, the activation of NLRP3 and levels of IL-1β secretion was much higher in BMDM, which indicated that macrophages were more sensitive to DAMPs than hepatocytes. Collectively, our results indicated that MSC-derived PGE2 inhibited liver macrophages inflammasome activation to protect against liver injury.
To confirm the role of MSC on NLRP3 inhibition, we treated BMDM with LPS and nigericin, a typical inducer of NLPR3. MSC or MSC-COX2(+) conditioned medium significantly inhibited NLPR3 activation and IL-1β secretion (Figure 4A and 4B), while MSC-COX2(-) conditioned medium did not show this phenomenon. In parallel, exogenous PGE2 exhibited same effects on NLRP3 (Figure 4C and 4D). Previous studies have confirmed the role of NF-κB in NLRP3 inflammasome activation18, thus we explored whether MSC inhibited NLRP3 through TAK1-NF-κB. Our results showed that MSC conditioned medium or PGE2 could inhibit TAK1 and NF-κB activation in BMDM induced by LPS and nigericin (Figure 4A and 4C). Meanwhile, TAK1 inhibitor could also inhibit NLRP3 activation in BMDM (Figure 4E-F). In vivo study confirmed the effects of TAK1 inhibitor on NLRP3 in ALF model (Figure 4G-H). Overall, our results indicated that MSC-derived PGE2 could inhibited NLPR3 inflammasome activation of liver macrophages through TAK1- NF-κB pathway.
MSC induces M2 macrophage polarization in ALF
Mouse models of acute liver injury have revealed the role of macrophages in accelerating damage through cytokines and chemokines releasing. However, liver macrophages are plastic and adapt their phenotype to promote tissue repair according to signals derived from the hepatic microenvironment6. Several studies confirmed the role of MSC on macrophage polarization in different models19, 20. Thus we tended to explore whether MSC could induce M2 macrophages in ALF to promote inflammation resolution. We found that genes expression associated with M2-like (Arg1, Mgl1, Mgl2 and Ym1) macrophages were increased in MSC and MSC-COX2(+) group (Figure 5A), meanwhile, the serum levels of IL-10, which produced by M2 macrophages, were also increased in MSC and MSC-COX2(+) group (Figure 5B). However, MSC-COX2(-) failed to induce M2 macrophages. In vitro studies confirmed the MSC-derived PGE2 on macrophage polarization, as the expression of M2 markers and numbers of CD206 positive cells of BMDM treated with MSC conditioned medium or PGE2 were increased (Figure 5C and 5D). Taken together, our results demonstrated that MSC induced M2 macrophages to resolute inflammation in ALF through PGE2.
Next we investigated the mechanism of MSC-derived PGE2 on macrophage polarization. Western blotting results showed that not only the classical M2 pathway, STAT6, was activated when MSC or PGE2 treatment, but also resulted in mTOR activation, as increased phosphorylation of the mTOR substrates mTOR, S6K, AKT and GSK-3β (Figure 6A), which correlated with previous studies21. In contrast, MSC-COX2(-) conditioned medium failed to activated mTOR and STAT6 pathway (Figure 6A). Treatment of BMDM with mTOR inhibitor, rapamycin (Rap) or AKT inhibitor, MK-2206, confirmed the role of mTOR signaling in MSC-derived PGE2 on M2 macrophage polarization, as the mRNA levels of M2 markers were significantly increased after administration of Rap or MK-2206 (Figure 6B-6D). Taken together, our results indicated that MSC-derived PGE2 induced M2 macrophages through mTOR and classical STAT6 pathway.
MSC protects against liver inflammation via PGE2 receptor (EP) 4
Our previous studies showed that MSC-derived PGE2 promoted hepatocyte proliferation through EP413, we explored whether EP4 also played a role in liver inflammation resolution. We treated mice with EP4 inhibitor (EP4i), GW627368X. The mRNA levels of inflammatory cytokines increased when EP4i treatment (Figure 7A). Western blot results confirmed that inhibition of TAK1 and NF-κB signaling by MSC were also mediated by EP4, as the expression of TAK1 and NF-κB signaling substrates was higher when EP4i administration (Figure 7B and 7C). Meanwhile, the expression of NLRP3 and the activity of caspase 1 enzyme were also higher when EP4i administration, even in MSC-COX2(+) group (Figure 7D and 7E). In vitro studies using EP4i or siRNA to inhibition or deplete EP4 in BMDM, the western blot and IL-1βsecretion confirmed the role of EP4 on NLRP3 inflammasome inhibition (Figure 7F,7G). Taken together, these results showed that MSC-derived PGE2 ameliorated liver inflammation through EP4.
EP4 mediates MSC-derived PGE2 on macrophage polarization
Next we explored that whether M2 macrophage polarization was also mediated by EP4. The mRNA levels of M2 macrophages markers and serum levels of IL-10 were also inhibited when EP4i treatment (Figure 8A and 8B). And mRNA levels of PGE2 receptors (EP1-EP4) in BMDM confirmed the role of EP4 on M2 macrophage polarization (Figure 8C). EP4i or siRNA treatment demonstrated the role of EP4 in macrophage polarization, as the mRNA levels of M2 markers were decreased when EP4i or siRNA treatment (Figure 8E and 8G). Meanwhile, the protein levels of STAT6 and mTOR signaling substrates confirmed the role of EP4 on STAT6 and mTOR signaling (Figure 8D and 8F). Overall, these results demonstrated that EP4 played a major role in MSC-derived PGE2 on macrophage polarization.