Identification and characterization of circArhgap26 in ischemia‒reperfusion injury

To identify potential dysregulated circular RNAs (circRNAs) associated with acute myocardial infarction (AMI), we performed circRNA expression profiling in cardiac tissues from AMI and sham-operated mice via heatmaps and volcano plots. (Supplementary Fig. 1a and b). By integrating the sequence features, fold changes, and conservation profiles of these circRNAs, we selected three circRNAs, namely, circRNA_41355, circRNA_018468 (circArhgap26), and circRNA_31781, with the most significant alterations in their characteristics for further investigation. Unexpectedly, the microarray results obtained via qRT‒PCR revealed that there were no significant changes in the expression of these three circRNAs in cardiac tissues from AMI mice or hypoxia-induced cardiomyocytes (Fig. 1a and b). This finding suggests that the aforementioned circRNAs may not play a central role in AMI, as we initially hypothesized. Several studies have suggested that the gene may not undergo substantial changes during ischemia but shows significant alterations during reperfusion, emphasizing its pivotal role in reperfusion injury.25,26 Next, we constructed a model of myocardial I/R injury (Supplementary Fig. 2a–g) and examined the changes in the expression of these three circRNAs. The results revealed that the levels of circRNA_41355, circRNA_018468 (CircArhgap26) and circRNA_31781 were significantly decreased in I/R hearts and H2O2 or H/R-stimulated cardiomyocytes (Fig. 1c–e). These results indicate that these three circRNAs may be involved mainly in I/R progression. Among these, circRNA_018468, which we named circArhgap26, was selected for in-depth investigation because its significant downregulation was observed in both I/R-injured hearts and H2O2 or H/R-induced cardiomyocytes. Additionally, the circArhgap26 level was significantly decreased in the serum of mice with I/R hearts (Fig. 1f). Interestingly, according to the website, we found that circArhgap26 was highly homologous (91.65%) with human circular RNA (h-circArhgap26) (Supplementary Fig. 3a). H-circArhgap26 expression in plasma samples from PCI-treated patients and healthy volunteers was further analyzed and compared, and h-circArhgap26 was significantly lower in the plasma of patients with PCI than in that of healthy subjects (Fig. 1g). Additionally, we compiled key demographic and clinical characteristics, including age, sex and other relevant clinical parameters, for both healthy volunteers and patients with PCI, as shown in Supplementary Table 1. H-circArhgap26 expression was also significantly downregulated in H2O2-stimulated AC16 cells (Supplementary Fig. 3b and c). To characterize the distribution of circArhgap26 in cardiac tissues, we performed Sanger sequencing of its back-spliced junction, which revealed that head-to-tail splicing occurred in exons 2 to 6 of the Arhgap26 gene (Fig. 1h). According to circBase (http://www.circbas e.org), h-circArhgap26 arises from exons 2 to 11 of the h-Arhgap26 gene with a length of 953 nucleotides, and its back-splicing site was confirmed by Sanger sequencing (Supplementary Fig. 3d). Additionally, the resistance to digestion by RNase R exonuclease suggested that circArhgap26 harbored a loop structure (Fig. 1i). H-circArhgap26 was also resistant to degradation by RNase R, while h-Arhgap26 mRNA was degraded, suggesting that h-circArhgap26 has a closed loop structure (Supplementary Fig. 3e). Subcellular localization analysis via both fluorescence in situ hybridization (FISH) and cellular fractionation demonstrated that circArhgap26 was primarily distributed in the cytoplasmic compartment in vitro (Fig. 1j and k). Moreover, accumulating evidence suggests that certain circRNAs have the potential to encode biologically active peptides.27,28,29 To determine the protein encoding capacity of circArhgap26, we first analyzed the putative open reading frame (ORF) of circArhgap26 (https://indra.mullins.microbiol.washington. edu/sms2/orf _find.html). Sequence analysis revealed that circArhgap26 harbors a spanning junction ORF with the potential to encode a 131 amino acid protein (Supplementary Fig. 4a). A circArhgap26-3 flag vector was subsequently constructed and transfected into cardiomyocytes to investigate the translatability of circArhgap26. However, the flag antibody failed to detect an ~14 kDa band in cells transfected with the 3 flag-circArhgap26 vector or in cells transfected with the control vector (Supplementary Fig. 4b), which indicated that circArhgap26 does not have protein-coding potential. Taken together, our findings demonstrated that the highly conserved circArhgap26 was downregulated in cardiac tissues following I/R injury, H2O2-stimulated cardiomyocytes, and particularly in PCI serum, suggesting its potential involvement in the pathogenesis of I/R.

Fig. 1

CircArhgap26 (a novel circRNA derived from Arhgap26) is decreased in the serum of patients with I/R, in the infarction border zone of I/R mouse cardiac tissues, and in H/R cardiomyocytes. a, b qRT‒PCR analysis of the expression of circRNAs in cardiac tissue from sham and AMI mice (n = 4) and in cardiomyocytes subjected to hypoxia (n = 4). c–e qRT‒PCR and quantification of circRNA levels in cardiac tissue from sham and I/R-treated mice (n = 4) and in cardiomyocytes subjected to H2O2 and H/R (n = 4–6). f CircArhgap26 expression, as measured by qRT‒PCR, was lower in the serum of mice with I/R than in that of control subjects (n = 6). g qRT‒PCR was used to measure the plasma levels of h-circArhgap26 in PCI patients (n = 30) relative to those in healthy volunteers (n = 30). h Schematic illustration showing the back-splicing of circArhgap26 from the Arhgap26 primary transcript, which was validated by Sanger sequencing. i qRT‒PCR analysis of the expression of circArhgap26 and Arhgap26 mRNAs in cardiomyocytes treated with or without RNase R (n = 3). j The distribution of circArhgap26 in cardiomyocytes was detected via FISH experiments (n = 6). Red, 18S, circArhgap26, U6; blue, DAPI. k The levels of circArhgap26 in the nucleus and cytoplasm of cardiomyocytes were detected via qRT‒PCR, and GAPDH and U6 were used as markers in the cytoplasm and nucleus (n = 4). The data are expressed as the means ± SEMs. P < 0.05 was considered statistically significant

The cardiac-specific circArhgap26 in cardiomyocytes participates in the regulation of I/R-induced cardiac injury

To assess the functional impact of circArhgap26 upregulation in I/R injury in vivo, we delivered adeno-associated virus serotype 9 (AAV9) via tail vein injection. Experimental group: AAV9-cTnT-CircArhgap26 (cardiac-specific CircArhgap26 overexpression driven by the troponin T promoter, AAV9-CircArhgap26). Control group: AAV9-cTnT-Vector (AAV9-Vector) (Fig. 2a, Supplementary Fig. 5a). Compared with AAV9-Vector mouse hearts, circArhgap26-overexpressing hearts presented improved cardiac function following I/R injury, as evidenced by increased ejection fraction (EF) and fractional shortening (FS) (Fig. 2b and c) and decreased left ventricular end-diastolic diameter (LVID;d) and left ventricular end-systolic diameter (LVID;s) (Supplementary Fig. 5b). Consistently, overexpression of circArhgap26 in cardiomyocytes diminished the increase in LDH in the serum and the increase in the concentration of caspase-3 in cardiac tissue induced by I/R (Fig. 2d and e). We also quantified the cTnT levels in plasma samples collected from the mice following the I/R surgery and the respective treatments. The results demonstrated that circArhgap26 overexpression reduced the plasma concentration of cTnT following I/R injury (Fig. 2f). In addition, overexpression of circArhgap26 in cardiomyocytes also ameliorated cardiomyocyte apoptosis in the I/R group (Fig. 2g). Moreover, the results of the Evans blue/TTC double-staining experiment demonstrated that the overexpression of circArhgap26 in cardiomyocytes significantly reduced the infarct size in the mice subjected to I/R injury (Fig. 2h). Moreover, cardiac-specific overexpression of circArhgap26 had antiapoptotic effects on hearts subjected to I/R injury, as evidenced by significant downregulation of the proapoptotic protein Bax and concomitant upregulation of the antiapoptotic protein Bcl2 (Fig. 2i).

Fig. 2

Overexpression of circArhgap26 ameliorates cardiac ischemic injury. a Schematic diagram of this study. b, c Representative echocardiography images and quantification of EFs and FSs (n = 6). d Serum LDH levels (n = 6). e Caspase-3 activity in the heart (n = 6). f cTnT levels in plasma samples (n = 6). g Representative TUNEL-stained images of heart sections and quantitative data (n = 6). h Representative Evans blue/TTC dual-stained heart sections and quantitative data (n = 6). Scale bar = 1 mm. i The protein levels of Bax (n = 8) and Bcl2 (n = 6) were detected via western blotting. The data are expressed as the means ± SEMs. P < 0.05 was considered statistically significant

We subsequently investigated whether circArhgap26 downregulation could induce pathological injury under physiological conditions. The expression of circArhgap26 was downregulated, but Arhgap26 expression was not different between AAV9-cTnT-Sh-circArhgap26 (AAV9-Sh-circArhgap26) and AAV9-cTnT-Sh-NC (AAV9-Sh-NC) (Supplementary Fig. 6a and b). Echocardiographic evaluation revealed a significant decrease in EF and FS upon circArhgap26 silencing (Supplementary Fig. 6c and d). Additionally, an increase was observed in LVID;d and LVID;s (Supplementary Fig. 6e). CircArhgap26 knockdown increased LDH levels in the serum and caspase-3 activity (Supplementary Fig. 6f and g). As expected, circArhgap26 silencing significantly induced cardiomyocyte apoptosis (Supplementary Fig. 6h). Consistently, downregulation of circArhgap26 led to increased expression of the Bax protein but decreased expression of the Bcl2 protein (Supplementary Fig. 6i).

To further validate the role of circArhgap26 silencing in the heart, adult male mice were intravenously injected with AAV9-Sh-circArhgap26 or AAV9-Sh-NC followed by I/R injury (45 min/24 h) or sham surgery 3 weeks after AAV9 injection (Supplementary Fig. 7a). Echocardiography revealed that cardiac function was strongly impaired by circArhgap26 downregulation under cardiac I/R injury, as indicated by further decreases in EF and FS (Supplementary Fig. 7b and c) and increases in LVID;d and LVID;s (Supplementary Fig. 7d). Compared with control mice subjected to I/R stress, AAV9-sh-circArhgap26 mice presented increased serum LDH levels and caspase-3 activity after cardiac I/R injury (Supplementary Fig. 7e and f). Additionally, cardiomyocyte death was also greater in AAV9-Sh-circArhgap26 mice than in control cardiac I/R injury model mice during cardiac I/R injury (Supplementary Fig. 7g). Furthermore, the infarct size was significantly greater in the hearts of the AAV9-Sh-circArhgap26 mice after cardiac I/R injury than in those of the mice subjected to I/R (Supplementary Fig. 7h). Consistently, silencing circArhgap26 resulted in the upregulation of Bax protein expression while concurrently decreasing the level of the Bcl2 protein (Supplementary Fig. 7i).

Because sex is crucial for the robustness and generalizability of our conclusions, we conducted an additional set of experiments using female C57BL/6 mice, following the same experimental protocols as those used for male mice in the original manuscript. Echocardiographic analysis revealed that, compared with that in AAV9-Vector control mice, circArhgap26 overexpression was associated with improved cardiac function after I/R injury, as demonstrated by increased EF and FS (Supplementary Fig. 8a and b). We also quantified cTnT levels in plasma samples, and the results demonstrated that circArhgap26 overexpression reduced the plasma concentration of cTnT following I/R injury (Supplementary Fig. 8c). In addition, overexpression of circArhgap26 in cardiomyocytes significantly attenuated cardiomyocyte apoptosis in the I/R group (Supplementary Fig. 8d). Moreover, Evans blue/TTC double-staining experiments demonstrated that circArhgap26 overexpression in cardiomyocytes significantly reduced the infarct size in mice subjected to I/R injury. Conversely, knockdown of circArhgap26 led to an increase in the infarct area (Supplementary Fig. 8e and f). In summary, the data from female mice strongly support our initial findings, indicating that the reported phenomenon is not sex specific.

We also evaluated the effects of circArhgap26 overexpression on H2O2-treated cardiomyocytes in vitro. As depicted in Supplementary Fig. 9a–e, overexpression of circArhgap26 in cardiomyocytes resulted in increased cell viability, decreased LDH release and reduced caspase-3 activity in H2O2-stimulated cardiomyocytes. Moreover, we found that circArhgap26 overexpression led to a reduction in TUNEL-positive nuclei in H2O2-induced cardiomyocytes (Supplementary Fig. 9f). As expected, Bax protein expression was decreased, whereas Bcl2 protein expression was increased in circArhgap26-overexpressing cardiomyocytes after H2O2 stimulation (Supplementary Fig. 9g).

Next, we assessed the function of silencing circArhgap26 via siRNA in cardiomyocytes in vitro. CircArhgap26 knockdown attenuated cell viability while promoting LDH release and caspase-3 activity in cardiomyocytes (Supplementary Fig. 10a–d). In addition, decreased expression of circArhgap26 significantly increased the number of TUNEL-positive nuclei in cardiomyocytes (Supplementary Fig. 10e). As expected, downregulation of circArhgap26 promoted Bax protein upregulation and suppressed Bcl2 expression (Supplementary Fig. 10f). Collectively, our results suggest that downregulation of circArhgap26 facilitates the progression of myocardial I/R.

Taken together, these gain- and loss-of-function results suggest that cardiomyocyte-specific circArhgap26 participates in the regulation of I/R-induced cardiac injury and dysfunction both in vivo and in vitro.

H-circArhgap26 is involved in regulating apoptosis in AC16 cell lines

Given its high degree of structural conservation, whether circArhgap26 plays similar functional roles in a human cardiac myocyte cell line (AC16 cell line) remains to be determined. To validate this notion, we first examined the overexpression and knockdown efficiency of h-circArhgap26 and the effect of h-circArhgap26 on parental gene expression (Supplementary Fig. 11a and b). The data revealed that the overexpression of h-circArhgap26 attenuated H2O2-induced apoptosis in AC16 cells, as demonstrated by increased cell viability and decreased LDH release, caspase-3 activity and TUNEL-positive nuclei (Supplementary Fig. 11c–f). Moreover, the expression of h-Bax was suppressed, whereas h-Bcl2 expression was increased by h-circArhgap26 overexpression in H2O2-treated AC16 cells (Supplementary Fig. 11g). Conversely, the knockdown of h-circArhgap26 inhibited cell viability and promoted LDH release, caspase-3 activity and TUNEL-positive nuclei in AC16 cells (Supplementary Fig. 11h–k). In addition, h-circArhgap26 knockdown resulted in the upregulation of Bax expression, accompanied by a decrease in the Bcl2 level in AC16 cells (Supplementary Fig. 11l). Overall, we proved that the h-circArhgap26 regulated apoptosis in AC16 cells, suggesting its possible involvement in human-related diseases.

Identification of PKP1 as a circArhgap26-associated protein

To further investigate the molecular mechanism by which circArhgap26 regulates cardiomyocyte apoptosis, RNA pull-down mass spectrometry was performed to identify the circArhgap26-binding proteins in cardiomyocytes (Supplementary Fig. 12a). We then analyzed the functional annotation of these pulled-down proteins via the Metascape database, and functional clustering analyses revealed that the proteins were predominantly associated with the apoptotic cleavage of cell adhesion proteins, including Dsg1a, PKP1, and Dsp (Fig. 3a). To further validate the results of the mass spectrometry analysis, a RIP assay was applied to confirm the interaction between circArhgap26 and PKP1, and the results revealed that circArhgap26 binds to PKP1 but not to Dsg1a or Dsp (Fig. 3b). Mass spectrometry assays suggested that PKP1 potentially interacts with circArhgap26 in cardiomyocytes (Fig. 3c). In addition, we observed an increase in PKP1 levels in H2O2-stimulated cardiomyocytes (Fig. 3d). Furthermore, PKP1 expression was significantly elevated in cardiac tissues after myocardial I/R injury compared with that in sham-operated controls (Fig. 3e). The interaction between circArhgap26 and PKP1 was subsequently validated through RNA pull-down assays and molecular docking analysis (Fig. 3f and g). Interestingly, potential colocalization of circArhgap26 and PKP1 was observed in the cytoplasm of cardiomyocytes, suggesting a possible interaction between circArhgap26 and PKP1 (Fig. 3h). To investigate the structural basis of the interactions between circArhgap26 and PKP1, the online database catRAPID was used to analyze the precise interaction between these two molecules (Supplementary Fig. 12b). We also constructed full-length PKP1 and various truncated variants on the basis of different domains identified in the UniProt and catRAPID databases (Fig. 3i). The RIP assay provided additional experimental evidence that circArhgap26 could be enriched in the full-length PKP1 and ARM1 domains of PKP1; these results suggested that circArhgap26 mainly interacted with the ARM1 region of PKP1 (Fig. 3j). These results indicate that circArhgap26 is highly likely to exert its functional effects through interactions with PKP1.

Fig. 3

Identification of PKP1 as a circArhgap26-associated protein. a Functional clustering analyses of the proteins pulled down by circArhgap26. b A RIP assay was performed to evaluate the interaction between PKP1 and circArhgap26 in cardiomyocytes (n = 3). IgG served as the control. c A representative PKP1 peptide identified by LC‒MS/MS mass spectrometry. d qRT‒PCR (n = 6) and western blotting (n = 3) analyses of PKP1 protein levels in mouse cardiomyocytes harvested from the control and H2O2 groups. PKP1 levels were normalized to those of β-actin in each group. e Heart samples were collected for qRT‒PCR (n = 4) and western blotting (n = 3) after I/R surgery. f CircArhgap26 pull-down followed by western blot analysis validated the interaction between circArhgap26 and PKP1 (n = 3). g The binding pose of circArhgap26 to PKP1 is shown. The green sticks indicate the structure of circArhgap26. The cyan sticks represent the key amino acids of the PKP1 residues that bind with circArhgap26. h FISH and IF costaining indicating the colocalization of circArhgap26 (red) and PKP1 (green) in cardiomyocytes. Scale bar = 3.5 μm (n = 3). i Schematic diagram of full-length and truncated PKP1. j RIP assay for full-length and truncated PKP1 followed by qRT‒PCR analysis (n = 3). The data are expressed as the means ± SEMs. P < 0.05 was considered statistically significant

Cardiac-specific PKP1 in cardiomyocytes participates in the regulation of I/R-induced cardiac injury

To determine the function of PKP1 silencing in I/R injury, C57BL/6 mice were injected with AAV9 via the tail vein via intravenous delivery. AAV9-cTnT-Sh-PKP1 (referred to as AAV9-Sh-PKP1) and AAV9-cTnT-Sh-NC (referred to as AAV9-Sh-NC) were constructed under the control of the cardiac troponin T promoter. Three weeks post-administration, the mice were subjected to myocardial I/R injury via ligation of the left anterior descending artery for 45 minutes of ischemia followed by 24 hours of reperfusion. (Fig. 4a, Supplementary Fig. 13a). Cardiac function was subsequently assessed by echocardiography in all groups, as shown in Fig. 4b and c and Supplementary Fig. 13b. Cardiac dysfunction induced by I/R injury, characterized by decreased EF and FS, as well as increased left ventricular internal dimension at LVID;d and LVID;s, was significantly ameliorated in AAV9-Sh-PKP1-treated mice. In addition, LDH activity and caspase-3 concentration induced by cardiac I/R were effectively suppressed in AAV9-Sh-PKP1 mice (Fig. 4d and e). Moreover, the hearts of the AAV9-Sh-PKP1 group exhibited significant attenuation of myocardial cell death induced by cardiac I/R, as evidenced by the reduction in TUNEL-positive nuclei (Fig. 4f). Similarly, PKP1 deficiency markedly reduced the myocardial infarct size caused by myocardial I/R (Fig. 4g). In addition, PKP1 downregulation effectively attenuated Bax levels and enhanced Bcl2 expression in cardiac I/R-induced hearts (Fig. 4h). Taken together, these results suggested that PKP1 deficiency attenuated I/R-induced apoptosis and myocardial dysfunction in vivo.

Fig. 4

Inhibition of PKP1 ameliorates ischemia/reperfusion (I/R)-induced cardiac injury. a Schematic diagram of the experimental design. b, c Echocardiographic evaluation of EF and FS (n = 6). d Serum levels of the cardiomyocyte death marker LDH in sham and AAV9-Sh-PKP1 mice after I/R surgery (n = 6). e Caspase-3 activity in sham and AAV9-Sh-PKP1 mouse hearts after I/R treatment (n = 6). f Representative images and quantification of TUNEL-positive cells in sham and AAV9-Sh-PKP1 mouse heart sections following I/R injury; scale bar = 50 μm (n = 6). g Evans blue/TTC dual-staining of cross sections from mouse hearts post I/R with WT or AAV9-Sh-PKP1 treatment (n = 6); scale bar = 1 mm. h Representative western blots and statistical data for Bax (n = 6) and Bcl2 (n = 8) in sham and AAV9-Sh-PKP1 mouse hearts after I/R stimulation. The data are expressed as the means ± SEMs. P < 0.05 was considered statistically significant

To investigate the effects of cardiac-specific PKP1 overexpression on I/R injury, we administered AAV9-cTnT-PKP1 (referred to as AAV9-PKP1) or AAV9-cTnT-Vector (referred to as AAV9-Vector) via tail vein injection into mice. Cardiac I/R injury or sham surgery was subsequently performed 3 weeks after viral vector administration. (Supplementary Fig. 14a and b). Cardiac function analyses by echocardiography revealed a significant reduction in EF and FS, an increase in LVID;d and LVID;s in cardiac I/R injury. Notably, cardiac function was further deteriorated in the mice treated with AAV9-PKP1 (Supplementary Fig. 14c–e). As shown in Supplementary Fig. 14f and g, overexpression of PKP1 significantly aggravated cardiac I/R-induced cardiomyocyte injury, as indicated by increased LDH release and caspase-3 activity. In addition, TUNEL staining revealed a significant increase in cardiomyocyte apoptosis in these mice in response to cardiac I/R injury, which was further exacerbated in the mice that received AAV9-PKP1 treatment (Supplementary Fig. 14h). Moreover, Evans blue/TTC double staining revealed an increased infarct size during I/R remodeling, which was further exacerbated in the AAV9-PKP1-treated mice (Supplementary Fig. 14i). Furthermore, compared with I/R injury alone, AAV9-PKP1 exacerbated the I/R-induced upregulation of Bax while further suppressing Bcl2 protein expression (Supplementary Fig. 14j).

We further investigated the potential functions of PKP1 in vitro by evaluating the effects of PKP1 knockdown mediated by siRNA in H2O2-treated cardiomyocytes. The H2O2 model significantly reduced cell viability, increased LDH release and increased caspase-3 activity in cardiomyocytes. However, effective silencing of PKP1 successfully prevented these alterations (Supplementary Fig. 15a–d). Additionally, PKP1 downregulation alleviated the H2O2-induced increase in TUNEL-positive nuclei (Supplementary Fig. 15e). Furthermore, Bax levels were significantly increased, whereas Bcl2 expression was markedly inhibited under H2O2 stimulation. However, these effects were effectively alleviated in the groups with PKP1 knockdown. (Supplementary Fig. 15f).

To investigate the role of PKP1 overexpression in H2O2-treated cardiomyocytes in vitro, we constructed a PKP1 overexpression plasmid, and the overexpression efficiency was validated by western blotting (Supplementary Fig. 16a). Notably, overexpressing PKP1 further inhibited the viability of cardiomyocytes stimulated with H2O2 (Supplementary Fig. 16b). PKP1 upregulation significantly increased the serum LDH level and caspase-3 activity in H2O2-stimulated cardiomyocytes (Supplementary Fig. 16c and d). H2O2-induced apoptosis was also exacerbated in PKP1-overexpressing cardiomyocytes (Supplementary Fig. 16e). Moreover, under H2O2 stimulation, PKP1-overexpressing cardiomyocytes presented upregulated Bax and suppressed Bcl2 compared with H2O2-treated controls (Supplementary Fig. 16f).

PKP1 mediates the effects of circArhgap26 on I/R-induced cardiac injury

Next, we determined the functional relationship between CircArhgap26 and PKP1. In addition, cardiac-specific overexpression of PKP1 prevented the improvements in EF, FS, LVID;d and LVID;s observed in AAV9-CircArhgap26 mice with cardiac I/R injury (Fig. 5a–c and Supplementary Fig. 17a). Moreover, CircArhgap26 significantly suppressed LDH levels, caspase-3 concentrations and apoptosis under I/R stimulation. However, PKP1 overexpression partially reversed these effects (Fig. 5d–f). Moreover, treatment with AAV9-PKP1 effectively blocked the inhibitory effects on the I/R-induced increase in infarct size in AAV9-CircArhgap26 mice (Fig. 5g). Accordingly, the changes in the levels of Bax and Bcl2 observed in AAV9-CircArhgap26 mice after I/R surgery were abolished by PKP1 upregulation (Fig. 5h). We further verified that the function of AAV9-Sh-PKP1 in AAV9-Sh-CircArhgap26 mice ameliorated I/R-induced cardiac injury (Supplementary Fig. 17b). The further decrease in cardiac function observed in AAV9-Sh-CircArhgap26 mice after I/R surgery was abolished via PKP1 silencing, as reflected by the increased EF and FS and decreased LVID;d and LVID;s (Supplementary Fig. 17c–e). PKP1 knockdown alleviated cardiac I/R injury in AAV9-Sh-circArhgap26 mice, as indicated by decreases in LDH release, caspase-3 activity, TUNEL-positive nuclei and infarct size (Supplementary Fig. 17f–i). Moreover, PKP1 downregulation reduced the level of Bax and upregulated Bcl2 expression in AAV9-Sh-CircArhgap26 mice after cardiac I/R injury (Supplementary Fig. 17j). These results suggest that PKP1 is critically involved in the regulatory mechanism by which circArhgap26 participates in the regulation of cardiac effects in vivo.

Fig. 5

Overexpression of PKP1 affects circArhgap26 upregulation in I/R-induced cardiac injury. a Schematic of the study of virus injection and the establishment of a myocardial ischemia‒reperfusion model. b, c Echocardiographic assessment of EF (n = 6) and FS (n = 6) in the indicated groups. d LDH levels in the mice in the indicated groups (n = 4). e Caspase-3 activity in the hearts of the indicated groups of mice subjected to I/R (n = 6). f Representative images of TUNEL assays of heart sections from each group after treatment and quantitative analysis (n = 6); scale bar = 50 μm. g Representative images of heart sections subjected to Evans blue/TTC dual-staining (n = 6); scale bar = 1 mm. h Representative western blotting results of Bax (n = 5) and Bcl2 (n = 6) in the hearts of the indicated groups of mice subjected to I/R. The data are expressed as the means ± SEMs. P < 0.05 was considered statistically significant

We also investigated the potential role of PKP1 in mediating the biological function of circArhgap26 and conducted a functional rescue assay using cardiomyocytes in vitro. In addition, the antiapoptotic effects of circArhgap26 overexpression were blocked by the upregulation of PKP1, as evaluated by analyses of CCK8, LDH, and caspase-3 activity; TUNEL staining; and western blot detection of apoptosis-associated genes (Bax and Bcl2) (Supplementary Fig. 18a–e). In brief, our data revealed that PKP1 mediated the effects of circArhgap26 on H2O2-induced cardiomyocyte apoptosis.

CircArhgap26 inhibits PKP1 protein stability by suppressing its palmitoylation

To elucidate the mechanism by which circArhgap26 affects PKP1, the expression of PKP1 at both the mRNA and protein levels was assessed in cardiomyocytes or tissues with either overexpressed or knocked down circArhgap26. qRT‒PCR analysis revealed that there was no significant difference in PKP1 mRNA levels between the overexpression group and the control group or between the knockdown group and the control group. Notably, despite the lack of effect on PKP1 mRNA levels, circArhgap26 overexpression robustly decreased PKP1 protein levels in cardiomyocytes; however, circArhgap26 silencing increased PKP1 protein levels in cardiomyocytes (Fig. 6a and b). These findings suggest that circArhgap26 regulates PKP1 expression posttranscriptionally. To evaluate the impact of circArhgap26 on PKP1 protein stability, cardiomyocytes were treated with cycloheximide (CHX), a protein translation inhibitor. In the presence of CHX, PKP1 protein levels decreased more rapidly in the circArhgap26-overexpressing group than in the vector-only control group, suggesting that circArhgap26 promotes PKP1 protein degradation and thereby reduces its stability (Fig. 6c). To elucidate the mechanism by which circArhgap26 inhibits the protein expression of PKP1, we performed functional clustering analyses on the basis of the microarray results to identify the potential biological function of circArhgap26. The results revealed that circArhgap26 was associated with lipid modification (Fig. 6d). The process of protein palmitoylation is one of the most important posttranslational lipid modifications that regulate protein localization, accumulation, secretion and stability.19,30 We subsequently investigated whether palmitoylation was responsible for PKP1 protein stability. We investigated whether palmitoylation affects PKP1 degradation by applying 2-bromopalmitate (2-BP), a general palmitoylation inhibitor, to reduce palmitoylation in cardiomyocytes. The protein expression of PKP1 was dose-dependently decreased by 2-BP treatment. (Fig. 6e). Using an acyl-biotin exchange (ABE) assay, we verified that 2-BP treatment reduced the PKP1 palmitoylation level (Fig. 6f). Given that palmitoylation regulates protein stability, we next determined whether the palmitoylation of PKP1 affects its stability. As expected, the half-life of PKP1 further decreased rapidly in 2-BP-treated cells (Fig. 6g). Intriguingly, the PKP1 palmitoylation level was also reduced in circArhgap26-overexpressing cardiomyocytes (Fig. 6h).

Fig. 6

CircArhgap26 inhibits PKP1 protein stability by suppressing its palmitoylation. a, b Representative qRT‒PCR (n = 6) and western blot (n = 6) data and statistical data of PKP1 in cardiomyocytes infected with circArhgap26 or si-circArhgap26 (n = 3-4). c A protein stability assay in which CHX (100 μg/mL) was used to treat cells for different durations was performed to evaluate the effect of circArhgap26 overexpression, followed by western blot analysis (n = 5). d Functional clustering analysis based on the microarray results. e The protein expression of PKP1 in cardiomyocytes treated with 0 μM, 25 μM, or 50 μM 2-BP for 24 h was measured by western blot analysis (n = 5). f Palmitoylated PKP1 levels were detected via a palmitoylation assay in cardiomyocytes treated with 50 μM 2-BP for 24 h (n = 3). g A protein stability assay in which CHX (100 μg/mL) was used to treat cells for different durations was performed to evaluate the effect of 2-BP, followed by western blot analysis (n = 4). h Palmitoylated PKP1 levels were detected via a palmitoylation assay in cardiomyocytes treated with the circArhgap26 overexpression plasmid for 48 h (n = 3). The data are expressed as the means ± SEMs. P < 0.05 was considered statistically significant

CircArhgap26 competitively binds to PKP1 and inhibits ZDHHC1-mediated palmitoylation

Palmitoyltransferases serve as critical enzymatic mediators in the palmitoylation process. To identify the predominant palmitoyltransferases for PKP1, we established a screening approach that combines protein expression and a palmitoylation level test. First, we comprehensively explored the potential functions of ZDHHCs, which contain the conserved DHHC domain and play important roles in various physiological and pathological processes associated with cardiac I/R. We therefore investigated the expression of all ZDHHCs in the cardiac I/R data (NCBI/GEO/GSE61592 and GSE58486). The analysis results suggested that only two ZDHHCs, ZDHHC1 and ZDHHC17, were upregulated, which was further validated experimentally. Both ZDHHC1 and ZDHHC17 levels were elevated in H2O2-induced cardiomyocytes and in the cardiac tissues of mice subjected to I/R (Supplementary Fig. 19a and b). Interestingly, silencing ZDHHC1 but not ZDHHC17 suppressed PKP1 palmitoylation (Fig. 7a, Supplementary Fig. 19c). Furthermore, inhibition of ZDHHC1 reduced the stability of the PKP1 protein, suggesting that ZDHHC1 is involved in the palmitoylation of PKP1 mediated by circArhgap26 (Fig. 7b). Additionally, a co-IP assay demonstrated an interaction between PKP1 and ZDHHC1 (Fig. 7c). Interestingly, circArhgap26 overexpression led to reduced interactions between PKP1 and ZDHHC1 in cardiomyocytes (Fig. 7d). To further identify the in-depth molecular mechanism by which circArhgap26 affects the interactions between PKP1 and ZDHHC1, we first performed molecular docking, and the results revealed that ZDHHC1 could bind mainly directly to the ARM1 domain of PKP1 (Fig. 7e). Furthermore, we transfected HEK293T cells with truncated variants of PKP1 constructed on the basis of its molecular domain organization and performed co-IP analysis. Our results demonstrated that ZDHHC1 directly interacts with the ARM1 domain of PKP1, suggesting that the ARM1 domain functions as a binding site for both circRNA and PKP1 (Fig. 7f). To further explore the role of palmitoylation in PKP1, we utilized the CSS-Palm database to predict potential palmitoylation sites on PKP1. On the basis of palmitoylation site prediction analysis, we identified cysteine residues 14, 136, 457, and 672 of PKP1 as potential palmitoylation sites (Supplementary Fig. 19d). Among the four predicted palmitoylation sites in PKP1, mutation of C14S and C136S significantly reduced the palmitoylation of PKP1, indicating that these two cysteine residues at positions 14 and 136 within the ARM1 domain serve as the principal sites for palmitoylation in PKP1 (Fig. 7g). Moreover, simultaneous mutation of both C14S and C136S of PKP1 further abrogated its palmitoylation (Fig. 7h). To further confirm these results, the stabilities of WT PKP1 and nonpalmitoylated PKP1 C14S/C136S mutants were analyzed. Compared with that of WT PKP1, the degradation rate of the PKP1 C14S/C136S protein was significantly accelerated (Fig.