PGAM5 enhanced M1 polarization and inhibited M2 polarization of BMDMs
To further illustrate the role of PGAM5 in macrophage polarization, we isolated BMDMs from Pgam5 cKO and Pgam5 fl/fl mice and induced them to an M1-polarized state by LPS plus IFN-γ for 24 hr in vitro. PGAM5 deletion in BMDMs significantly suppressed the mRNA levels of proinflammatory genes, including iNOS, IL1α, IL1β, IL6, and IL12, compared with that in Pgam5 fl/fl macrophages (Fig. 4a). The number of CD86+ cells was significantly lower in Pgam5 cKO BMDMs than in Pgam5 fl/fl BMDMs after M1 induction, as determined by flow cytometry (Fig. 4b). Enzyme-linked immunosorbent assay (ELISA) of the supernatant of BMDMs treated with LPS plus IFN-γ also verified that PGAM5 deletion led to decreased secretion of proinflammatory cytokines in macrophages (Fig. 4c). In addition, the protein level of iNOS was significantly lower in Pgam5 cKO BMDMs than in Pgam5 fl/fl macrophages after M1 induction (Fig. 4d). To clarify the influence of PGAM5-deficient macrophages on cartilage, BMDMs from Pgam5 cKO and Pgam5 fl/fl mice were cocultured with chondrocytes after M1 induction in vitro. The protein levels of ACAN and MMP3 and the mRNA levels of COL2A1 and SOX9 in chondrocytes indicated that PGAM5 deletion in macrophages limited the proinflammatory phenotypes of chondrocytes (Fig. 4e, f).
Next, we aimed to investigate whether PGAM5 plays an essential role in M2 macrophage polarization. Pgam5 cKO and Pgam5 fl/fl BMDMs were isolated and stimulated with IL4 for 24 hr for M2 induction. After IL-4 stimulation, Pgam5 cKO BMDMs expressed enhanced mRNA levels of Arg1, CD206, PPARγ, and IL10, compared to Pgam5 fl/fl BMDMs (Fig. 4g). The number of CD206 + cells was significantly higher in Pgam5 cKO BMDMs than in Pgam5 fl/fl BMDMs after M2 induction, as determined by flow cytometry (Fig. 4h). The protein level of CD206 significantly increased in Pgam5−cKO BMDMs compared to Pgam5 fl/fl macrophages after M2 induction (Fig. 4i). In conclusion, Pgam5 led to an increased M1 response and a decreased M2 response in macrophages in vitro.
PGAM5 induced M1 polarization via the AKT-mTOR/P38/ERK signaling pathway, whereas inhibited M2 polarization through STAT6-PPARγ signaling pathway
In previous studies, a series of signaling pathways were activated by induction of M1 polarization, such as the AKT-mTOR and MAPK signaling pathways.[12] We further examined whether PGAM5 promotes M1 polarization through these specific signals. Treated with LPS and IFNγ, BMDMs from Pgam5 cKO mice significantly showed lower protein levels of p-p38 and p-ERK than Pgam5 fl/fl macrophages and nearly identical expression of total p38 and total ERK, indicating that PGAM5 activates the p38 and ERK signaling pathways to enhance M1 polarization since MAPK signaling has been proven to increase M1 polarization in previous studies[15] (Fig. 5a). However, we did not find different expression pattern of p-JNK in Pgam5 cKO and Pgam5 fl/fl BMDMs (Fig. 5a). p-AKT and p-mTOR significantly decreased in Pgam5 cKO macrophages compared with Pgam5 fl/fl macrophages stimulated by LPS and IFNγ (Fig. 5a). To explore the possible role of AKT and mTOR in mediating the intensive role of PGAM5 in macrophage inflammation, we employed the specific mTOR activator MHY1485[16] to activate the mTOR pathway. Activating mTOR rescued the decreased levels of iNOS, IL1α, IL1β and IL-6 in PGAM5-deficient macrophages after LPS and IFNγ stimulation, as examined by qPCR (Fig. 5b), indicating that mTOR activity is involved in the regulation of M1 macrophage polarization by PGAM5. The protein level of iNOS was also enhanced in the presence of MHY1485 in Pgam5−cKO macrophages (Fig. 5c). The activation of mTOR, also presented by phosphorylation of mTOR, increased in BMDMs when treated with doses of 0, 1, 5, and 10 µM of MHY1485 (Fig. 5d), and the mRNA levels of iNOS, IL1α, IL1β, and IL6 increased in Pgam5 cKO macrophages in a dose-dependent manner (Fig. 5e), indicating that PGAM5 promoted the M1 phenotype by activating the mTOR signaling pathway.
To explore the mechanisms by which PGAM5 modulates M2 polarization, we focused on the STAT6-PPARγ signaling pathway, which regulates M2 polarization.[17] After stimulation with IL4 for 24 hr, the protein levels of PPARγ and p-STAT6 enhanced in Pgam5−cKO BMDMs compared to Pgam5 fl/fl BMDMs (Fig. 5f), indicating that the STAT6-PPARγ signaling pathway might be involved in PGAM5-mediated M2 polarization. Next, inhibition of PPARγ by a specific inhibitor, T0070907,[18] markedly reduced the protein level of CD206 in a dose-dependent manner (Fig. 5g). Furthermore, T0070907 decreased the M2 polarization response in Pgam5 cKO macrophages, as indicated by significantly decreased mRNA levels of Arg1, chitinase-like 3 (Ym1), and CD206 (Fig. 5h). Besides, the protein level of CD206 was also markedly reduced by treatment with T0070907 in Pgam5−cKO macrophages stimulated by IL4 (Fig. 5i), suggesting that M2 polarization was regulated by PGAM5 through the STAT6-PPARγ signaling pathway. To identify the role of STAT6 in the PGAM5 modulation of M2 polarization, specific inhibition of STAT6 by AS1517499[19] was added to Pgam5 cKO and Pgam5 fl/fl macrophages at a dose of 10 µM. As a result, the protein expression of PPARγ and CD206 significantly reduced, indicating that STAT6 functions as a regulator of PPARγ (Fig. 5j). In conclusion, PGAM5 regulates M2 polarization via the STAT6-PPARγ signaling pathway.
PGAM5 regulated macrophage polarization by targeting the β-catenin pathway via dephosphorylation of DVL2.
Although we have identified the related signaling pathways that regulate PGAM5-mediated macrophage polarization, the direct target of PGAM5 in regulating macrophage polarization has not been verified. PGAM5 was reported to inhibit the Wnt/β-catenin signaling pathway on the mitochondrial membrane in human cells and Xenopus embryogenesis.[20] Moreover, the β-catenin signaling pathway is closely correlated with macrophage activation and polarization.[21–23] Nevertheless, whether PGAM5 modulates β-catenin in synovial macrophage polarization and the related targets have not been reported.
Thus, we first examined the potential regulation of PGAM5 on β-catenin in macrophage polarization. Pgam5 cKO and Pgam5fl/fl BMDMs were isolated and stimulated to induce M1 or M2 polarization. After 24 hr of stimulation, Pgam5 cKO BMDMs showed increased phosphorylation of β-catenin at Ser675 and decreased phosphorylation at Thr41/Ser45 in both the M1- and M2-polarized states, and increased protein levels of β-catenin were detected in the M1-polarized state compared to Pgam5fl/fl BMDMs, indicating that PGAM5 triggered inhibition of β-catenin in both M1 and M2 polarization (Fig. 6a). To better confirm the target of PGAM5 on β-catenin in macrophage polarization, we focused on whether PGAM5 could dephosphorylate Dishevelled Segment Polarity Protein 2 (DVL2), which is a inhibitory regulator in the upstream of β-catenin signaling pathway. Interestingly, the phosphorylation sites of DVL2, S143 and T224, were both increased in Pgam5 cKO macrophages when induced to M1 and M2-polarized states compared to Pgam5fl/fl BMDMs (Fig. 6a), indicating potential interaction of PGAM5 and DVL2. Thus, we detected whether PGAM5 could directly bind to DVL2 and found that PGAM5 could be coimmunoprecipitated with DVL2 in both M1 and M2 macrophages (Fig. 6b), indicating PGAM5 directly dephosphorylated DVL2 via binding to it. We then focused on whether β-catenin was involved in PGAM5-modulated macrophage polarization. Specific inhibition of β-catenin by ICG-001[24] increased the protein levels of p-p38, p-ERK and iNOS. ICG-001 also reduced the expression of p-STAT6, PPARγ, and CD206, indicating that β-catenin functions as a regulator of macrophage polarization (Fig. 6c, d). In addition, as predicted, the expression of β-catenin increased in the OA synovium of Pgam5 cKO mice compared to Pgam5fl/fl mice (Fig. 6e). Thus, these results indicated that PGAM5 modulated macrophage polarization by inhibiting the β-catenin pathway via directly targeting DVL2.
To verify the role of β-catenin in PGAM5-mediated macrophage polarization in synovium, we generated mice in which both PGAM5 and β-catenin were ablated in macrophages, herein referred to as DKO mice, by crossing β-cateninfl/fl mice with Pgam5 cKO mice. Knee joints of DKO and Pgam5 cKO male mice were collected 28 days after DMM surgery for further investigations. DKO mice significantly exacerbated OA symptoms compared to Pgam5 cKO mice, as determined by safranin O staining (Fig. 6f). Expression of ACAN decreased while expression of MMP13 increased in the knee joint of DKO mice compared to Pgam5 cKO mice, indicating that the relieved OA symptom by Pgam5 deficiency in macrophages was partly mediated by the enhanced activity of β-catenin (Fig. 6f). In addition, iNOS-positive cells were significantly increased and CD206 + cells were significantly decreased in the synovium of DKO mice compared to Pgam5 cKO mice (Fig. 6f, g), suggesting that PGAM5 regulates synovial macrophage polarization by inhibiting the β-catenin signaling pathway, further aggravating the progression of OA. Peritoneal macrophages in DKO mice showed increased mRNA levels of iNOS and CD80 and decreased mRNA levels of CD206 (Fig. 6h). In addition, flow cytometry of peritoneal macrophages also indicated that DKO mice showed more proinflammatory phenotypes than Pgam5 cKO mice (Fig. 6i). In conclusion, PGAM5 regulates macrophage polarization by inhibiting the β-catenin signaling pathway, and inhibition of β-catenin extensively reversed the alleviation of OA symptoms in Pgam5 cKO mice.
Targeted knockdown of PGAM5 in synovial macrophages by MFP9-2/siPGAM5 relieved OA symptoms.Based on the mechanisms of PGAM5 in regulating macrophage polarization, we aimed to establish targeted deletion of PGAM5 in synovial macrophages to treat OA in early stage. RNA interference (RNAi) is a powerful technique to treat various diseases via specific gene silence.[25] However, targeted delivery of siRNA into synovial macrophages is challenging owing to the complicated synovial fluid composition and extracellular matrix in the joint microenvironment.[26] Here, we developed a series of mannose modified fluoropolymers for macrophage-targeted siRNA delivery to treat OA via inhibition of PGAM5. ε-PLL was conjugated with fluoroalkanes (F7-F17, Fig. 7a) via amine-epoxide reactions, and different feeding ratios were chosen to get PLL modified with average numbers of about 10 and 15 fluoroalkanes, respectively. The obtained fluoropolymers were further grafted with mannose via an amine-isocyanate reaction, and an average number of 1 mannose was conjugated on each polymer calculated by 1H NMR. Take F7 for example, the mannose grafted polymer conjugated with 10 F7 ligands and 15 F7 ligands were termed MFP7-1 and MFP7-2, respectively. The siRNA delivery efficacy of the synthesized nanoparticles (NPs) was first screened on Raw264.7 cells. MFP9-2/siPGAM5 complexes exhibited the highest gene knockdown efficiency, which was higher than that of the Lipofectamine 2000 (Lipo)/siPGAM5 complexes (Fig. 7b). The MFP9-2 transported the FAM-labelled siRNA into the CD68 + synovial macrophages rather than chondrocytes efficiently after intraarticular injection (Fig. 7c), which is beneficial for achieving macrophage-targeted RNAi. To validate the therapeutic efficacy of the MFP9-2 based siRNA delivery system, MFP9-2/siPGAM5 complex was injected into the joint of WT mice twice weekly in early stage of OA established by DMM surgery and the joints were collected 28 days after DMM surgery. Sham group, DMM group with no injection and DMM group with injection of MFP9-2 combined with siNC (MFP9-2/siNC) were established as controls (Fig. 7d). Intraarticular injection of MFP9-2/siPGAM5 greatly relieved the OA symptoms compared to DMM group and DMM with injection of MFP9-2/siNC group, indicated by increased safranine O staining area, decreased expression of MMP13 and enhanced level of ACAN in cartilage (Fig. 7d). Besides, the amount of iNOS positive cells in synovium treated with MFP9-2/siPGAM5 greatly decreased, while CD206 positive cells increased compared to DMM group and DMM with injection of MFP9-2/siNC group (Fig. 7e, g), indicating the availability of macrophage modulation in OA synovium by MFP9-2/siPGAM5. To further confirm the modulatory function of MFP9-2/siPGAM5, we detected whether the relief of OA symptoms was the result of PGAM5 inhibition in synovial macrophage by MFP9-2/siPGAM5. We found that injection of MFP9-2/siPGAM5 greatly decreased the level of PGAM5 in CD68 positive macrophages in synovium compared to injection of MFP9-2/siNC with FAM (Fig. 7f, g), indicating successful targeting and inhibiting of PGAM5 in synovial macrophages, which could further achieve better OA outcomes.
To conclude, PGAM5 serves as a novel factor of regulating macrophage polarization in osteoarthritis via dephosphorylating DVL2, resulting in increased activity of GSK3β and degradation of β-catenin,[27] which disables the translocation of β-catenin into nucleus to bind to promotors for downstream signaling pathways, further contributes to increased M1 and decreased M2 phenotypes via specific signals. To better treat OA via early intervention of macrophage PGAM5, specifically inhibition of PGAM5 in macrophages was achieved by intraarticular injection of MFP9-2/siPGAM5, which could significantly target synovial macrophages and reduce the expression of PGAM5 in macrophages, resulting in the relief of OA symptoms. Together, we have clarified the modulatory role of PGAM5 in OA macrophage and designed a functional macrophage-targeted therapy, which might contribute to early and precise immunological interventions in OA in clinic.