Expression and characterization of circN4bp1 in macrophages
Based on our previous microarray analysis that circN4bp1 was increased in the pulmonary macrophages isolated from the lung homogenates of sepsis-induced ARDS model , we verified the upregulation of circN4bp1 in macrophages of the BALF and lung tissue by qRT-PCR (Figure 1A). Then, the sequence of circN4bp1 was confirmed with Sanger sequencing assays (Figure 1B). Next, we tested the property of circN4bp1 by RNase R treatment, as shown in Figure 1C, linear N4bp1 was mostly digested, while circN4bp1 remained almost unchanged, thus suggesting a circular configuration. Additionally, the half-life of circN4bp1, its expression level, together with the N4bp1 level was examined following actinomycin D transcriptional inhibition. We found that circN4bp1was more stable than N4bp1 (Figure 1D). Lastly, we determined the subcellular localization of circN4bp1 by conducting FISH assay and found that circN4bp1 was mainly localized in the cytoplasm of macrophages (Figure 1E).
Upregulation of CircN4bp1 correlates with poor prognosis of sepsis-induced ARDS patients
To further explore the role of circN4bp1 in ARDS, the expression levels of circN4bp1 in the peripheral blood of a cohort comprising of 40 sepsis-induced ARDS patients and 40 healthy controls were quantified using qRT-PCR. The detailed clinical characteristics of the ARDS patients and healthy controls are listed in supplementary Table S1. We found that circN4bp1expression levels were significantly upregulated in the PBMCs of ARDS patients relative to the controls (Figure 2A). Furthermore, circN4bp1was found to be significantly higher in non-survivors than survivors (Figure 2B). To assess the role of circN4bp1 in the regulation of subtypes of PBMCs, we isolated PBMCs into T cells, B cells, and monocytes to detect the expression levels of circN4bp1. We observed that circN4bp1was mainly expressed in monocytes rather than T and B cells (Figure 2C). Furthermore, higher expression of circN4bp1 in monocytes was found in ARDS patients than healthy controls, with much higher levels in non-survivors than those survived (Figure 2D, 2E).The Spearman correlation analysis was conducted to further evaluate the prognostic value of circN4bp1in monocytes. Interestingly, the circN4bp1 expression levels in monocytes of the ARDS group was positively correlated with mSOFA score (r = 0.646, P<0.001) (Figure 2F). Therefore, these findings suggested that circN4bp1 upregulation in the PBMCs and monocytes was correlated with a poor prognosis in ARDS patients, and that monitoring over-expression of circN4bp1 might predict poor outcomes of these patients.
CircN4bp1 is involved in promoting M1 macrophage polarization while inhibiting M2 activation in two ex-vivo macrophage cell lines
To determine the functional influence of circN4bp1on macrophage differentiation, we first specifically knocked down the expression of circN4bp1in RAW264.7 and MH-S cells with small interfering RNAs (siRNAs) targeting the circN4bp1 junction site and over-expressed of circN4bp1with a circN4bp1 lentivirus plasmids transfected into the two macrophage cell lines in contrast to the NC group (Figure 3A). Next, we cultured RAW264.7 and MH-S cells in the presence or absence of LPS (M1 polarization) or IL-4 (M2 phenotype) as previously reported  and treated them with Si-circN4bp1 or circN4bp1 lentivirus plasmids for 24 h. We found that genetic knock down of circN4bp1 significantly inhibited M1 polarization as evidenced by a down-regulation of M1 marker INOS (Figure 3C, 3D, supplementary Figure S1A,1B) and related cytokines as IL-6 and TNF-α (Figure 3D, supplementary Figure S1C), while augmented the M2 polarization, as indicated by up-regulation of M2 marker Arg-1 (Figure 3C, 3D, supplementary Figure S1A,1B) and associated cytokine IL-10 (Figure 3D, supplementary Figure S1C). To the contrary, overexpression of circN4bp1 was found to promote the M1 polarization while inhibit the M2 differentiation exhibited by the increase of INOS, IL-6, TNF-α, and a reduction of IL-10 and Arg-1 (Figure 3B, 3C, 3D, supplementary Figure S1). Taken together, these findings suggested that circN4bp1 can promote M1 macrophage activation while inhibit M2 polarization in ex-vivo macrophage cell lines.
As we had previously documented the involvement of STAT1 pathway in M1 polarization and PPAR-γ in M2 differentiation , we further investigated the effects of circN4bp1on these pathways. Intriguingly, western blotting showed that circN4bp1 knockdown did indeed result in markedly decreased levels of p-STAT1 in M1 polarized macrophages while augmented levels of PPAR-γ in M2 polarized macrophages. Conversely, over-expression of circN4bp1 exhibited a totally opposite performance (Figure 3C, 3D, supplementary Figure S1). Overall, these findings implied that circN4bp1can promote M1 macrophage activation by enhancing STAT1 signaling, while inhibit the M2 macrophage polarization by suppressing PPAR-γ signaling pathways.
CircN4bp1 functions as a molecular sponge for miR-138-5p
It is well known that circRNAs act as miRNA sponges , thus the ability of circN4bp1 to bind miRNAs was explored. Based on bioinformatics prediction by the miRanda &Targetscan data base, miR-138-5p was considered to potentially bind with circN4bp1 and circN4bp1–miR-138-5p–mRNA network was shown in Figure 4A. In addition, the miR-138-5p level was significantly inhibited in PBMCs and monocytes of patients with sepsis induced ARDS patients in comparison with healthy subjects (Figure 4B). Based on these findings, we hypothesized that circN4bp1 could regulate macrophage function by sponging miR-138-5p. Firstly, we observed a negative association between circN4bp1 and miR-138-5p in circN4bp1-overexpressed two macrophage cell lines (Figure 4C). Then a dual-luciferase assay was performed showing high binding affinity between circN4bp1 and miR-138-3p. Besides, miR-138-5p significantly reduced luciferase reporter activity when compared to the control (Figure 4D). A RIP assay showed that both circN4bp1 and miR-138-5p were elevated in the immunoprecipitates of the anti-Ago2 group (Figure 4E,4F). Both circN4bp1 and miR-138-5p inhibitors were reduced in miR-138-5p inhibitor-treated group compared with the IgG group (Figure 4E,4F). Thus, our results suggested that circN4bp1 could function as molecular sponge for miR-138-5p in macrophages.
Next, we investigated the influence of miR-138-5p on macrophage differentiation in RAW264.7 and MH-S cells with miR-138-5p mimic and inhibitor. We found that
miR-138-5p inhibitor could significantly promoted M1 macrophage polarization with an upregulation of INOS (Figure 4G,4H), IL-6 and TNF-α (supplementary material, Figure S2A,2B,2D,2E), but a downregulation of M2 associated proteins as IL-10(supplementary material, Figure S2C, 2F) and Arg-1 (Figure 4G,4H) in contrast to miR-138-5p-mimic and miR-NC group. As we had observed that circN4bp1 could promote M1 polarization while inhibit M2 activation, we further adopted the rescue tests using miRNA mimics, which showed that miR-138-5p mimics could suppress the effect of circN4bp1on macrophage polarization (Figure 5, supplementary Figure S3). Putting together, circN4bp1 regulated macrophage differentiation via miR-138-5p sponge in vitro through a ceRNA mechanism.
CircN4bp1-miR-138-5p ceRNA modulates macrophage differentiation via targeted-regulating EZH2
To further investigate the downstream mRNA targets of circN4bp1-miR-138-5p
ceRNA network, we performed bioinformatics analysis in TargetScan database and found miR-138-5p could target the 3′-untranslated region (UTR) of EZH2 (Figure 6A). Besides, a negative association between EZH2 and miR-138-5p was found in macrophages (Figure 6B). The luciferase reporter assay demonstrated that EZH2 was a target of miR-138-5p, but the rescue test by use of miR-138-5p inhibitors reversed its effect on EZH2(Figure 6C). Therefore, EZH2 might be the targeted gene of miR-138-5p. As we had reported that EZH2 was involved in the activation of M1 macrophage and inhibition of M2 differentiation , we hypothesized that circN4bp1 regulates macrophage differentiation through preventing EZH2 downregulation by miR-138-5p. To test this hypothesis, we overexpressed miR-138-5p mimics in the aforementioned two macrophage cell lines and found that the expression level of EZH2 was increased in the presence of circN4bp1 overexpression (Figure 6C), whereas miR-138-5p mimics significantly inhibited the upregulation of EZH2 after circN4bp1 overexpression (Figure 6D, 6E). CircN4bp1 could promote the expression of p-STAT1 in M1 polarization macrophages and inhibit the expression of PPAR-γ in M2 polarized macrophages (Figure 6D, 6E). However, miR-138-5p mimics could partially rescue this effect via cirN4bp1/ miR-138-5p sponge (Figure 6D, 6E). These data demonstrated that circN4bp1 might regulate macrophage polarization as a miR-138-5p sponge to modulate the circN4bp1/miR-138-5p/EZH2 axis.
Knock down of circN4bp1 in macrophages alleviated lung injury induced by sepsis after CLP surgery through inhibition of M1 macrophage activation
Next, we evaluated the effect of circN4bp1 in murine sepsis induced ALI models.
As we had documented that CLP induced a dramatic increase in the expression of circN4bp1 in the macrophages isolated from BALF and lung tissues of ALI mice (Figure 1A). Hence, we hypothesized that circN4bp1 plays a role in the modulation of macrophage actions in inflammation and injury of ARDS. To address this hypothesis, mice were intravenously injected with Si-circN4bp1 lentivirus plasmids to knock down circN4bp1in macrophages (circN4bp1-KD) prior to CLP, followed by measuring the indices of lung injury in the lung 24 h post CLP. We observed that the circN4bp1-KD-macrophage treated animals displayed significantly higher long-term survival compared to the vector group (Figure 7A). As shown in Figure 7B, the expression level of circN4bp1 in the alveolar macrophages was down-regulated in the circN4bp1-KD mice comparing with vector-treated CLP mice. The circN4bp1-KD CLP mice also exhibited decreased ALI parameters as evidenced by a reduction of morphological disruption of lung tissue architecture (Figure 7C), a reduced wet/dry ratio (Figure 7D) and BAL protein leakage (Figure 7E) in comparison with the vector group.
In order to further evaluate effects and potential molecular mechanism of circN4bp1 on macrophage polarization, we next set out to assess the phenotype changes in sepsis-induced ARDS mice of by isolating macrophages from BALF. We noted a remarkably up-regulated protein expression of iNOS while down-regulated expressions of Arg-1. Furthermore, circN4bp1-KD group mice exhibited significantly lower levels of IL-6 and TNF-α, accompanied by higher levels of IL-10 in the macrophages isolated from BALF, relative to sham and vector groups (Figure 7F). As we had documented that circN4bp1 promote M1 macrophage polarization through miR-138-5P/EZH2 signaling, we further verify these results in vivo. In consistent with the findings in vitro, we observed a lower expression of miR-138-5p and a higher level of EZH2 in macrophages isolated from the vector ARDS mice in comparison with sham group, circN4bp1-KD treatment can restore the expression of miR-138-5p but decrease the levels of EZH2 (Figure 7G,7H). Besides, we identified that p-STAT1 were significantly inhibited while PPAR-γ were activated in the macrophages from BALF of circN4bp1-KD ARDS mice comparing with sham and vector-treated ARDS mice (Figure 7H).
Upregulation of circN4bp1 in macrophages of sepsis-induced ARDS is partially attributed to m6A modification
Recent evidence shows that circRNA is modified by m6A, which affects circRNA levels, and m6A modification could promote macrophage polarization in vitro . To explore the potential mechanism involved in the upregulation of circN4bp1in macrophages of CLP-induced ARDS mice, we first determined the m6A level of circN4bp1 in isolated macrophages by MeRIP-qRT-PCR. As shown in Figure 8A, the relative m6A level of circN4bp1was remarkably elevated in the macrophages of CLP mice in comparison with sham controls. Next, we examined the relative expression of m6A related genes, including methyltransferase(writer), demethylase (eraser) and reader protein (reader) in macrophages. As shown in Figure 8B, the mRNA expression levels of METTL3, FTO and YTHDF2 in CLP mice were obviously elevated compared with control groups, and METTL3 upregulation was the most significant among all these genes (Figure 8B).
To identify the role of METTL3 in modulating the m6A modification of circN4bp1, qRT-PCR analysis of circN4bp1 levels was conducted in ex-vivo LPS-stimulated pulmonary macrophages (MH-S) after METTL3 silencing or treatment with 3-deazaadenosine (DAA), a global methylation inhibitor. Interestingly, the increased circN4bp1was almost reduced to the normal level with knock down or pharmacological inhibition of METTL3 (Figure 8C). Moreover, we found two highly conservative m6A sites on circN4bp1 based on the online m6A SRAMP database (http://www.cuilab.cn/sramp) (Figure 8D). We mutated them and conducted the luciferase reporter assay which showed an augment of the luciferase activity of wild-type vector, but not the mutated vector (Figure 8E). These data imply that m6A modification might be involved in the upregulation of circN4bp1 in macrophages of sepsis-induced ARDS.