Alteration of macrophage phenotype during SCI
To investigate the macrophage phenotypes during SCI, we established a mouse model of SCI using Balb/c mice following the methods described previously . To identify the SCI model, we measured the BMS scores the content and distribution of macrophage marker F4/80, M1 macrophage marker CD16/32, and M2 macrophage marker Arg1 in spinal cords from the sham group and SCI mouse on day 1, 3, 7, 14, 28 after the operation. As revealed by IF staining, compared to the sham group, the fluorescence intensity representing F4/80 (red) and CD16/32 (green on the upper panel) gradually increased in a time-dependent manner in SCI group (Fig.1A); however, the fluorescence intensity representing Arg1 (green on the lower panel) increased moderately on day 1 and 3, reached a sharp peak value on day 7, and then gradually decreased on day 14 and day 28 (Fig.1A). Consistently, the mRNA expression and the protein levels of M1 macrophage markers CD16, iNOS, and IFN-γ and M2 macrophage markers Arg1, CD206, and IL-4 emerged similar trend. M1 macrophage markers increased after SCI in a time-dependent manner from day 1 to day 28, while M2 macrophage markers increased moderately on day 1 and 3, reached sharp peak values on day 7 or 14, and then decreased gradually (Fig.1B-C). The BMS score results confirmed that the SCI model was successfully established (Fig.1D). These data indicate that the macrophages in the spinal cord are skewed toward M1 after SCI thatmight participate in the dysfunction of SCI repair.
Selection of lncRNAs and genes related to macrophage polarization
As we have mentioned, the deregulation and dysfunction of lncRNAs in the process of a pro-inflammatory (M1) to an anti-inflammatory phenotype (M2) have been observed [38, 53]. In the present study, we attempted to identify lncRNAs related to the skewing macrophage toward the M1 phenotype after SCI. The study downloaded and analyzed microarray profiles (GSE117040 and GSE5099), which reported upregulated lncRNAs in M1 macrophages. A total of 4 lncRNAs were reported to be upregulated in M1 macrophages by both profiles; there are homologous genes for lncRNA GBP1P1 (GBP9 in mice) and LINC00869 (Fam91a1 in mice) in mice (Fig. S1A).
Next, to further validate the involvement of these lncRNAs, we isolated BMDMs and identified them by examining macrophage markers F4/80 and CD11b using IF staining (Fig.2A). BMDMs (M0 macrophages) were then induced for differentiating towards M1 or M2 macrophages; the mRNA expression and protein levels of M1 macrophage markers iNOS and CD16 and M2 macrophage markers Arg1 and CD206 were examined to identify different subtypes. As shown in Fig.2B-C, iNOS and CD16 were significantly upregulated in M1 subtype while downregulated in M2 subtype; Arg1 and CD206 were remarkably upregulated in M2 subtype while downregulated in M1 subtype, indicating the successful induction. As revealed by real-time PCR, the expression of GBP9 and Fam91a1 were both significantly upregulated in M1 macrophages, GBP9 more upregulated (Fig.2D), indicating that GBP9 might be involved in macrophage M1/2 polarization. Reportedly, GBP1P1 is a pseudogene of the guanylate-binding protein of guanylate-binding protein (GBP); this family is also involved in macrophage functions, such as IFN-γ-mediated macrophage activation and immune defense . More importantly, based on microarray profile or RNA-seq analyses GSE5099 (Fig.S1B), GSE117040 (Fig.S1C), E-GEOD-57494 (Fig.S1D), E-MTAB-2399 (Fig.S1E), and GSE40885 (Fig.S1F), lncRNA GBP1P1 is specifically highly-expressed in human M1 macrophages and could be rapidly up-regulated after treatment with M1-inducing factors LPS and IFN-γ. Thus, GBP1P1 (lncGBP9 in mice) was selected for further experiments.
We then performed search tool for the retrieval of interacting genes/proteins (STRING) analyses on differentially-expressed genes in M1 macrophages reported previously [38, 55] to identify key regulators of the switch from a pro-inflammatory (M1) to an anti-inflammatory phenotype (M2). As revealed by STRING analyses, Tnf, SOCS3, and STAT1 are key factors in macrophage polarization (Fig.2E). Next, the mRNA expression and protein levels of M1-related STAT1 and p-STAT1, M2-related STAT6 and p-STAT6, and SOCS3 were examined in M0, M1, and M2 macrophages. As shown in Fig.2F and G, p-STAT1 protein levels and SOCS3 mRNA and protein levels were dramatically upregulated in M1 macrophages while p-STAT6 was upregulated in M2 macrophages. In the meantime, the production of STAT1 downstream cytokines, including IL-6 and IL-12, was increased in M1 macrophages while that of STAT6 downstream IL-10 and TGF-β1 was increased in M2 macrophages (Fig.2H). These data indicate that lncGBP9 and SOCS3 expression are upregulated in M1 macrophages and might be related to M1/2 polarization.
Effects of lncGBP9 on macrophage polarization in vitro
After selecting lncGBP9 for further experiments, we next evaluated its effects on macrophage polarization in vitro and in vivo. BMDMs were induced toward M0 macrophages for 7 days as described. Then, the silencing and overexpression of lncGBP9 were conducted in M0 macrophages for 48 h, and the transduction was effective and confirmed by real-time PCR (Fig.3A). After transduction for 24 h, LncGBP9-silenced and lncGBP9-overexpressing M0 macrophages were stimulated by LPS + IFN-γ for M1 polarization. In M1 macrophages, lncGBP9 silencing or overexpression caused no significant changes in STAT1 mRNA expression; lncGBP9 silencing significantly downregulated, while lncGBP9 overexpression upregulated SOCS3 mRNA expression (Fig.3B). In M1 macrophages, the protein levels of p-STAT1 and SOCS3 were reduced considerably by lncGBP9 silencing while increased by lncGBP9 overexpression (Fig.3C); consistently, the production of IL-6 and IL-12 was also inhibited by lncGBP9 silencing while promoted by lncGBP9 overexpression in M1 macrophages (Fig.3D). On the contrary, lncGBP9-overexpressing or lncGBP9-silenced macrophages were stimulated with IL-4 for M2 polarization. In M2 macrophages, lncGBP9 overexpression or silencing caused no significant changes in STAT6 mRNA expression; lncGBP9 overexpression significantly upregulated, while lncGBP9 silencing downregulated SOCS3 mRNA expression (Fig.3E). In M2 macrophages, lncGBP9 overexpression decreased p-STAT6 and increased SOCS3 protein, while lncGBP9 silencing increased p-STAT6 and decreased SOCS3 (Fig.3F); consistently, the production of IL-10 and TGF-β1 was suppressed by lncGBP9 overexpression while promoted by lncGBP9 silencing in M2 macrophages (Fig.3G). These data indicate that lncGBP9 might modulate macrophage M1/2 polarization through affecting SOCS3 and the phosphorylation of STAT1/STAT6.
To further investigate the speculation, the study then co-transfected M0 macrophages with Adv-lncGBP9 and SOCS3-overexpressing vector (SOCS3 OE) or with Adv-lncGBP9 and si-SOCS3, induced transfected M0 macrophages toward M1 polarization, and examined for macrophage M1/2 polarization. In Adv-sh-lncGBP9 and SOCS3 OE co-transfected M1 macrophages, lncGBP9 silencing significantly inhibited SOCS3 mRNA expression (Fig.3H), decreased SOCS3 protein level and STAT1 phosphorylation (Fig.3I), and reduced the concentrations of IL-6 and IL-12 (Fig.3J). SOCS3 overexpression in M1 macrophages exerted opposite effects. The effects of lncGBP9 silencing on M1 macrophages were significantly reversed by SOCS3 overexpression. Next, M0 macrophages were co-transfected with Adv-lncGBP9 and si-SOCS3, induced toward M2 polarization, and examined accordingly. In Adv-lncGBP9 and si-SOCS3 co-transfected M2 macrophages, lncGBP9 overexpression significantly upregulated SOCS3 mRNA expression (Fig.3K), increased SOCS3 protein level, inhibited STAT6 phosphorylation (Fig.3L), and reduced the concentrations of IL-10 and TGFβ (Fig.3M). SOCS3 silencing in M2 macrophages exerted opposite effects. The effects of lncGBP9 silencing on M2 macrophages were significantly reversed by SOCS3 silencing.
Effects of lncGBP9 on macrophage polarization in vivo
Next, the effects of lncGBP9 on macrophage polarization were evaluated in vivo. LncGBP9 silencing was conducted in the SCI mice model via injecting to the epicenter of the injured spinal cord with Adv-sh-lncGBP9. On day 28 of SCI, Adv-sh-lncGBP9 effectively reduced the level of lncGBP9 in the injured spinal cords (Fig.4A). Next, we evaluated the BMS scores on day 1, 3, 7, 14, and 28 after the operation to access the effects of lncGBP9 silencing on SCI severity. As shown in Fig.4B, lncGBP9 silencing significantly increased the BMS scores in SCI mice on day 14 and 28 after the operation, indicating lncGBP9 silencing in SCI mice promoted the repair after SCI.
At the same time, the content and distribution of M1 macrophage marker CD16/32 and M2 macrophage marker Arg1 determined in lncGBP9-silenced SCI mice by IF staining and Immunoblotting on day 28 after the operation to investigate macrophage polarization. As shown in Fig.4C-D, the fluorescence intensity representing M1 marker CD16/32 was significantly inhibited in Adv-sh-lncGBP9-infected mice on day 28, while M2 marker Arg1 was increased in Adv-sh-lncGBP9-infected mice on day 28, compared to those in Adv-sh-NC group. Consistently, Fig.4E showed that the protein levels of p-STAT1 and SOCS3 were significantly decreased, while the protein levels of p-STAT6 were increased by lncGBP9 silencing in SCI mice on day 28 after the operation. These data indicate that lncGBP9 silencing might promote M2 and inhibit M1 polarization via STAT1/6 and SOCS3, therefore modulating the repair after SCI.
LncGBP9 modulates SOCS3 through miR-34a in macrophages
LncRNAs could serve as ceRNAs for miRNAs to counteract miRNA-mediated suppression on miRNA downstream transcripts, therefore exerting their biological functions [33-35]. Essandoh reported a number of miRNAs that might promote M2 polarization ; among them, miR-124 and miR-34a were predicted to target SOCS3 and only miR-34a was predicted to target lncGBP9. More importantly, miR-34a could promote M2 macrophage polarization . Thus, we hypothesize that miR-34a might participate in lncGBP9 function on macrophage polarization.
To validate the hypothesis, we examined the expression of miR-34a in vivo and in vitro, including in M0, M1, and M2 macrophages. In SCI mice, miR-34a expression was significantly downregulated on day 28 after the operation (Fig.S2A); in SCI mice infected with Adv-sh-lncGBP9, miR-34a expression was significantly upregulated on day 28 after the operation, compared to Adv-sh-NC group (Fig.S2B). As shown in Fig.5A, miR-34a expression was dramatically upregulated in M2 macrophages. In M1 macrophages, miR-34a expression was significantly increased by lncGBP9 silencing (Fig.5B). To investigate the cellular effects of miR-34a, we conducted miR-34a overexpression in M1 macrophages by transfection of miR-34a mimics into M0 macrophages before polarization, as confirmed by real-time PCR (Fig.5C). LncGBP9 expression was significantly downregulated by miR-34a overexpression in M1 macrophages (Fig.5D). Consistently, the SOCS3 protein level was also decreased by miR-34a overexpression in M1 macrophages (Fig.5E).
In M2 macrophages, miR-34a expression was significantly downregulated by lncGBP9 overexpression (Fig.5F). Here, we conducted miR-34a inhibition M2 macrophages by transfection of miR-34a inhibitor, as confirmed by real-time PCR (Fig.5G). In M2 macrophages, lncGBP9 expression was significantly upregulated by miR-34a inhibition (Fig.5H). Consistently, the SOCS3 protein level was increased by miR-34a inhibition in M2 macrophages (Fig.5I). These data indicate that lncGBP9 might regulate SOCS3 through miR-34a to participate in M1/2 macrophage polarization.
LncGBP9 serves as a ceRNA for miR-34a to counteract miR-34a-mediated SOCS3 suppression
To validate the predicted targeting of miR-34a to lncGBP9 and SOCS3, we performed luciferase reporter assays by constructing wild- and mutant-type GBP9 and SOCS3 3'-UTR luciferase reporter vectors (wt-GBP9/SOCS3 3'-UTR or mut-GBP9/SOCS3 3'-UTR) as described in M&M section (Fig.6A-B). Next, 293T cells were co-transfected with the above-described vectors and miR-34a mimics/inhibitor and examined for the luciferase activity. As shown in Fig.6A-B, the luciferase activity of wt-GBP9 and wt-SOCS3 3'-UTR vectors could be significantly inhibited by miR-34a overexpression and enhanced by miR-34a inhibition; in responding to the mutation at the putative miR-34a binding sites, the changes in the luciferase activity were abolished. Moreover, in the RNA derived from precipitated AGO2 protein, lncGBP9 and miR-34a levels were significantly higher than those in IgG in M1 macrophages (Fig.6C). We also performed RIP assay in M1 macrophages transfected with NC mimics or miR-34a mimics and then detected lncGBP9 and miR-34a levels associated with AGO2; the results shown in Fig.6D confirmed the interaction between lncGBP9 and miR-34a. Furthermore, in lncGBP9-overexpressing M1 macrophages, the level of lncGBP9 detected was dramatically higher than that of NC group. While the levels of SOCS3 was lower than that of NC group (Fig.6E), indicating that lncGBP9 and SOCS3 could bind miR-34a, respectively; lncGBP9 competes with SOCS3 for miR-34a binding.
LncGBP9/miR-34a axis modulates macrophage polarization via affecting the balance of STAT1/STAT6
After confirming the binding of miR-34a to lncGBP9 and SOCS3, next, we evaluated the dynamic effects of lncGBP9 and miR-34a on STAT1/STAT6 and macrophage polarization. M0 macrophages were co-transfected with Ad-sh-lncGBP9 and miR-34a inhibitor and then polarized to M1 macrophages, the mRNA expression and protein levels of STAT1, p-STAT1, SOCS3, iNOS, and CD16, and the production of IL-6 and IL-12 were examined. As shown in Fig.7A, C, and E, lncGBP9 silencing significantly reduced, while miR-34a inhibition significantly increased the mRNA expression and protein levels of p-STAT1, SOCS3, iNOS, and CD16/32, as well as the production of IL-6 and IL-12 in M1 macrophages; the effects of lncGBP9 silencing could be significantly reversed by miR-34a inhibition.
M0 macrophages were co-transfected with Ad-lncGBP9 and miR-34a mimics and then polarized to M2 macrophages, the mRNA expression and protein levels of STAT6, p-STAT6, SOCS3, Arg1, and CD206, and the production of IL-10 and TGF-β1 were examined. As shown in Fig.7B, D, and E, lncGBP9 overexpression significantly increased SOCS3 mRNA expression and protein level, decreased p-STAT6, Arg1, and CD206 mRNA and protein levels, and suppressed the production of IL-10 and TGF-β1; miR-34a overexpression exerted opposing effects on these indicators; the effects of lncGBP9 overexpression could be significantly reversed by miR-34a overexpression. These data indicate that the lncGBP9/miR-34a axis modulates M1/2 macrophage polarization through SOCS3 and STAT1/STAT6.
STAT6 binds miR-34a promoter to activate its transcription
As predicted by the online tool, STAT6 might bind the promoter region of miR-34a to activate its transcription. STAT6 overexpression or silencing was conducted in M0 macrophages by transfection of STAT6-overexpressing or si-STAT6 vector following M2 polarization, as confirmed by immunoblotting (Fig.8A). In M2 macrophages, the expression of miR-34a was significantly upregulated by STAT6 overexpression while downregulated by STAT6 silencing (Fig.8B). Next, wild- and mutant-type miR-34a luciferase reporter vectors are constructed; the mut-miR-34a vector contained a 9-bp mutation in any of the predicted STAT6 binding sites (Fig.8C). STAT6 and wt- or mut-miR-34a promoter were then co-transfected in M0 macrophages followed by M2 polarization; the luciferase activity was determined. As shown in Fig.8D, the promoter activity of wt-miR-34a was dramatically increased by STAT6 overexpression; however, after mutating any of the predicted binding sites, STAT6 overexpression-induced increase in promoter activity was abolished (Fig.8D). Moreover, the ChIP assay showed that the level of STAT6 antibody binding to miR-34a binding element in the miR-34a promoter was much greater than that of IgG in M2 macrophages (Fig.8E), suggesting that STAT6 might bind to the promoter of miR-34a to activate its expression in M2 macrophages.