Airn expression were downregulated in diabetic hearts, while overexpression of Airn ameliorates diabetes-induced cardiac dysfunction
STZ were used to induce diabetes in mice. As shown in Figure S1A&B, intraperitoneal injections of STZ successfully induced diabetic mellitus (DM) in the mice, which was confirmed by significantly elevated blood glucose and decreased body weight. Cardiac function were determined by echocardiographic analysis. As shown in FigureS1C-G, mice with DM exhibited a significant impairment in cardiac function, as evidenced by decreased LVEF and LVFS. To explore whether Airn was involved in diabetic cardiac fibrosis, cardiac fibrosis indicators and Airn expression were determined in control and diabetic hearts. Compared with the control heart, diabetic hearts exhibited significant cardiac fibrosis, as evidenced by elevated collagen I content and increased Masson staining positive area percentage (Fig. S1 H). In contrast, Airn expression in diabetic hearts decreased gradually as DM progressed (Fig. S1 K). Moreover, high-glucose (HG)-treatment also leads to a significant reduction of Airn expression both in neonatal rat cardiac fibroblasts and cardiomyocytes (Fig. S1 L-N). These data suggested a strong linkage between Airn inhibition and cardiac fibrosis in diabetic mice.
Next, we aimed to determine whether Airn could protect against cardiac fibrosis in the diabetic cardiomyopathy (DCM) model. Gain-of-function studies were performed by employing recombinant adeno-associated virus (AAV)9-Airn to up regulate Airn expression in diabetic heart through intramyocardial injection (Fig. 1A). After eight weeks of AAV injection, Airn expression was detected in heart tissues by polymerase chain reaction (PCR), and results showed successful virus injection with high efficiency (Fig. 1B). Overexpression of Airn alleviated cardiac dysfunction in diabetic mice, as evidenced by significant increases in left ventricular ejection fraction (LVEF) and left ventricular fractional shorting (LVFS), as well as decreases in left ventricular systolic internal dimension and left ventricular diastolic internal dimension (LVIDd) (Fig. 1C-G). Furthermore, Doppler echocardiography showed that Airn reconstitution also alleviated diastolic dysfunction in DM mice with an enhanced E/A ratio (Fig. 1H and I). Cardiac hypertrophy is an important structural feature of pathology in DCM(39). Compared with the control group, DM mice showed significant cardiac hypertrophy as evidenced by hematoxylin, eosin, and wheat germ agglutinin staining (Fig. 1J and K). DM-induced cardiac hypertrophy was efficiently ameliorated in the hearts of DM mice receiving AAV-Airn. Taken together, these data indicate that overexpression of Airn could ameliorate diabetes-induced cardiac dysfunction. Additionally, we investigated whether cardiac function was perturbed by the loss of Airn function in wild-type (WT) mice receiving AAV-shAirn. Notably, WT mice with Airn knockdown displayed impaired systolic and diastolic function (Fig. 1C-I) and cardiac hypertrophy (Fig. 1J and K) as compared with control mice. Collectively, we observed the downregulation of Airn both in diabetic mice heart tissues and HG-treated CFs and found that Airn is an essential LncRNA for DM-induced cardiac dysfunction.
Overexpression of Airn protects against diabetes-induced cardiac fibrosis, while Airn knockdown leads to cardiac fibrosis in the normal heart
Efficient Airn overexpression abrogated DM-induced cardiac fibrosis, as demonstrated by Masson’s staining (Fig. 2C and D). Western blotting and immunohistochemical staining showed that expression of the fibrotic marker, collagen I, was significantly decreased in the heart tissues of diabetic mice after AAV-Airn treatment, indicating that Airn reduced the secretion of collagen I in diabetic heart tissues (Fig. 2A and B). Because the aberrant proliferation and fibroblast-to-myofibroblast transition (FMT) of CFs are key events during cardiac fibrosis(39), we performed immunofluorescence assays for Ki67 and alpha smooth muscle actin (α-SMA) to detect these two cell phenotypes. Immunofluorescence showed excessive proliferation and FMT of CFs in the hearts of the diabetic mice. Moreover, these phenotypes were significantly attenuated by Airn overexpression (Fig. 2E-H). In addition, downregulation of Airn augmented the production of collagen I, proliferation, and FMT, and further exacerbated myocardial fibrosis in non-diabetic mice compared to that in CTRL mice. Taken together, these results indicate that Airn is an important negative regulator of fibrogenesis induced by HG and is a potential therapeutic target for the treatment of DCM fibrosis.
Airn attenuates HG-induced CF proliferation and FMT.
We evaluated the function of Airn in cardiac fibrosis in vitro. CFs were treated with HG medium to mimic diabetic conditions. As expected, HG treatment alone upregulated α-SMA and collagen I expression (Fig. 3A). However, Airn overexpression blocked the HG-induced expression of α-SMA and collagen I (Fig. 3A). Using the CCK8 assay, we found that elevated Airn significantly reduced the absorbance of HG-treated CFs compared to CFs treated with HG only, which was consistent with the in vivo results (Fig. 3B). EDU immunofluorescence staining also confirmed that the percentage of EDU-positive cells was significantly lower in Airn-overexpressing CFs (Fig. 3C). Meanwhile, immunostaining of α-SMA indicated that upregulated Airn levels blunted HG-induced morphological changes in FMT (Fig. 3D).
To investigate whether the loss of Airn is essential for CF activation independent of glucose level changes, we used siRNA-Airn to suppress the expression of Airn in CFs treated with normal glucose (NG) medium. Western blotting revealed that Airn knockdown enhanced α-SMA and collagen I expression (Fig. 3E). CCK8 and EDU assays indicated that si-Airn treatment accelerated the proliferation of CFs (Fig. 3F and G). As shown by α-SMA immunofluorescence, downregulation of Airn activated the FMT of CFs under NG conditions (Fig. 3H). Thus, we concluded that Airn plays an essential role in maintaining the normal form and function of CFs in vivo.
Airn exerts its anti-fibrotic effects by directly binding to insulin-like growth factor 2 mRNA-binding protein 2 (IMP2)
To identify the molecular mechanisms underlying the regulation of DCM fibrosis by Airn, its downstream molecules were investigated according to previously reported methods. IMP2 is a potential binding partner of Airn(35). RNA pull-down assays showed that IMP2 was directly bound to sense Airn and not to the antisense strand in CFs (Fig. 4A). Furthermore, RIP assay confirmed binding between Airn and IMP2 (Fig. 4B). Confocal fluorescence imaging analysis revealed colocalization of IMP2 and Airn in the cytoplasm, but not in the nucleus (Fig. 4C). IMP2, an RNA-binding protein, binds to a series of mRNAs and regulates several biological processes by influencing their fate(40). Thus, we speculated that IMP2 may be involved in Airn-mediated phenotypic changes in CFs. To test our hypothesis, we transfected CFs with si-IMP2 and Airn-overexpressing plasmids. We found that IMP2 knockdown significantly restored the expression of α-SMA and collagen I in CFs with Airn overexpression following HG treatment (Fig. 4D). Furthermore, suppression of IMP2 significantly enhanced the proliferative capacity of Airn-overexpressing CFs, as indicated by EDU assay (Fig. 4E). α-SMA immunofluorescence also indicated a reversed role of si-IMP2 in the Airn overexpression-mediated inhibition of FMT in CFs (Fig. 4F). Taken together, these results show that IMP2 knockdown reversed the inhibitory effect of Airn on the proliferation and FMT of CFs.
Airn protects IMP2 from ubiquitin-proteasome-dependent degradation, thus maintaining its stability
Decreased expression of the IMP2 protein was observed in CFs treated with HG, while overexpression of Airn significantly increased IMP2 protein levels in HG treatment (Fig. 5A). However, the mRNA levels of IMP2 remained unchanged (Fig. 5B). Next, we assessed the effect of si-Airn on IMP2 expression. Depletion of Airn resulted in the decline of IMP2 at the protein level, but not at the mRNA level (Fig. S2A and B). Based on these results, we speculated that IMP2 may be regulated by HG treatment and Airn at the post-transcriptional level. To test our hypothesis, we treated CFs with the protein synthesis inhibitor cycloheximide (CHX, 20 µg/mL) and harvested cells at the indicated time points (0, 3, 6, and 9 h). Compared to NG treatment, HG treatment significantly shortened the half-life of the IMP2 protein (Fig. 5C and D). However, Airn overexpression extended the protein half-life of IMP2 in CFs cultured in HG, suggesting that Airn may be a protector of IMP2 protein stabilization (Fig. 5C and D). To investigate the degradation patterns of IMP2 under HG conditions, CFs were treated with MG132 (0.1 mol/L), a potent proteasome and calpain inhibitor, to inhibit the proteasomal system and thus increase ubiquitin accumulation. MG132 rescued the HG-mediated decrease in IMP2 expression (Fig. 5E), suggesting that degradation of IMP2 may be linked to the accumulation of ubiquitin mediated by HG. Moreover, Airn overexpression significantly weakened the enhanced ubiquitination of IMP2 by MG132 (Fig. 5G). Additionally, the ubiquitination of IMP2 was increased significantly in HG-treated CFs as compared to NG-treated CFs, while Airn overexpression restrained the effects of HG-induced ubiquitination of IMP2 in CFs (Fig. 5F and H).
To locate the binding sites of IMP2 targeted by Airn, plasmids expressing FLAG-tagged WT or IMP2 ubiquitin mutants (K77R and K139R) were constructed (18) (Fig. 5I), which were conserved among species (Additional file). After transfection with the relevant plasmids, RIP assay showed that the K139 mutation weakened the binding between Airn and IMP2 (Fig. 5K). Compared to the WT and K77R, the K139 mutation significantly decreased IMP2 ubiquitination (Fig. 5J). Taken together, the above observations indicate that Airn could promote the protein stability of IMP2 by suppressing its ubiquitination through masking of the K139 ubiquitin site and thus slowing IMP2 degradation under HG stimulation.
Airn /IMP2 inhibits HG-induced CF proliferation and activation by maintaining p53-mediated cell cycle arrest.
As an m6A-binding protein, IMP2 is essential for mRNA stability. To check whether Airn could regulate IMP2-mRNA-binding, RIP-chip assay data with anti-IMP2 antibodies in Airn knockdown were obtained from the GSE87221 dataset in the GEO database. Six representative mRNAs associated with fibrosis (Gadd45b, Smad5, Mapk7, Mmmp14, Tcf7l1, and p53) were found to bind to IMP2 (Fig. 6A). We proposed that Airn may assist in the binding of IMP2 to mRNAs and may even affect the expression levels of these candidates. We then measured the mRNA expression levels of the six candidates after knocking down Airn and IMP2. The results showed that only p53 mRNA exhibited the same decreasing trend in response to Airn and IMP2 knockdown (Fig. 6B). Meanwhile, RIP and qRT-PCR assays identified that IMP2 did bind to p53 mRNA (Fig. 6C); thus, it was selected for further analysis. Immunofluorescence images revealed the colocalization of IMP2 protein with p53 mRNA in the cytoplasm (Fig. 6D). p53 is a negative regulator of fibrogenesis. However, the biological function of p53 in CFs from diabetic patients remains unclear. Western blotting results showed increased p53 expression in the heart tissues of diabetic mice, which is consistent with previous findings (41) (Fig. S3A). Interestingly, the patterns of p53 expression in cardiomyocytes and CFs in response to HG were different, as p53 was upregulated in cardiomyocytes but downregulated in CFs after HG treatment (Fig. S3B and C). As shown in Fig. S3E, inhibiting p53 expression promoted the levels of α-SMA and collagen I in CFs cultured in NG, whereas p53 overexpression downregulated α-SMA and collagen I levels induced by HG treatment. Next, we examined whether p53 was a downstream target of Airn. Results of EDU staining and immunofluorescence for α-SMA showed that p53 knockdown significantly restored the proliferation and FMT of CFs with Airn overexpression and HG treatment (Fig. 6E and F). Meanwhile, Western blotting showed that p53 knockdown blunted the inhibitory effects of Airn on fibrosis, as evidenced by the enhanced fibrosis markers (α-SMA and collagen I) (Fig. 6G). As cell cycle progression and proliferation are regulated by p53, we propose that Airn may inhibit the activation of CFs by promoting p53-mediated cell cycle arrest. Moreover, HG treatment triggered downregulation of p21 and p53 and upregulation of cyclin-dependent kinases (cyclin D1, CDK2, and CDK4), which were abolished by Airn overexpression (Fig. 6G). However, the suppression of p53 inhibited the inhibitory effects of Airn on HG treatment. Next, we detected the effects of Airn on cell cycle distribution by Western blotting and flow cytometry. Western blotting results showed that Airn rescued the HG-mediated decrease in p53 levels. Consistent with the increased p53 expression, Airn triggered the upregulation of p21 and downregulation of cyclin D1, CDK2, and CDK4 (Fig. 6H). Airn silencing showed opposite trends in the expression of p53 and its downstream molecules. Flow cytometry demonstrated a decrease in the number of cells in the G1 and G2/M phases in the HG group, whereas the proportion of cells in the S phase increased markedly, suggesting increased proliferation of CFs. However, Airn overexpression significantly arrested the activated proliferation induced by HG. When Airn was silenced in the NG group, the number of cells in the S phase increased, whereas those in the G1 and G2/M phases declined significantly (Fig. 6I). These results show that Airn was sufficient to inhibit CF proliferation and FMT, thus alleviating cardiac fibrosis by regulating the p53-mediated cell cycle pathway.
Airn /IMP2 stabilizes p53 mRNA in an m6A-dependent manner.
Previous studies have shown that IMP2 plays an important role in mRNA stabilization by binding to m6A-mRNAs (38). We hypothesized that Airn/IMP2 regulates p53 by affecting p53 mRNA stabilization. Therefore, we performed double-label immunofluorescence and RNA decay assays to determine the stability of p53 mRNA. The double-label immunofluorescence assay showed that Airn knockdown inhibited the expression of IMP2 and p53 mRNA and weakened the binding of IMP2 and p53 mRNA, which is consistent with the si-IMP2 findings (Fig. 7A). RNA decay assays and qRT-PCR showed decreased expression and stability of p53 mRNA in CFs treated with si-Airn and si-IMP2, respectively (Fig. 7B and C). Next, we examined whether Airn is a cofactor of IMP2 that regulates p53 expression. Exogenous overexpression of IMP2 enhanced binding with p53 mRNA and thus elevated the mRNA level of p53 (Fig. 7D and E). However, even when IMP2 was overexpressed, binding of IMP2 to p53 mRNA was clearly blunted by treatment with si-Airn (Fig. 7D). Therefore, Airn plays an essential role in the binding between IMP2 and p53 mRNA. Consistent with the results of double-label immunofluorescence, the weakened binding between IMP2 and p53 mRNA induced by si-Airn resulted in p53 mRNA instability in CFs with upregulation of IMP2 (Fig. 7F). Taken together, the above results suggest that Airn could contribute to the binding of IMP2 with p53 mRNA and synergistically work with IMP2 to maintain the stability of p53 mRNA.
To evaluate whether m6A modification is functionally important for IMP2-mediated p53 mRNA stabilization, we transfected CFs with si-METTL3 (methyltransferase-like protein 3), a core methyltransferase of the N6-adenosine-methyltransferase complex, which is known to modulate multiple biological processes (42, 43). As reported previously, the consensus sequence ‘UGGAC’ motif is an m6A core motif enriched in the targets of the IMP family (IMP1, 2, and 3) (38) (Fig. 7G). The predicted potential binding site of p53 for IMP2 localization was in the 3’UTR, which coincides with the m6A distribution patterns (Fig. 7G). Upon METTL3 silencing, m6A modification of p53 decreased compared to that in the negative control group, as evidenced by the results of Me-RIP-PCR (Fig. 7H). The RIP assay demonstrated that IMP2 bound to enriched m6A-modified p53, and the binding between p53 and IMP2 was impaired upon METTL3-knockdown (Fig. 7I), suggesting that the recognition of p53 by IMP2 was based on m6A modifications. Meanwhile, decreased m6A modification by knocking down METTL3 triggered a decline in the expression of p53 at the protein level (Fig. 7J). More importantly, similar to the results of knocking down IMP2 and Airn, the effects of METTL3 downregulation on p53 showed a reduction in the stability of p53 mRNA and its subsequent levels in CFs (Fig. 7K-M). Overall, our data showed that m6A modifications are required for the binding of IMP2 to p53 mRNA and IMP2-mediated regulation of p53 expression.