Long non-coding RNA CCRR controls cardiac conduction via regulating intercellular coupling

Abstract

Long non-coding RNAs (lncRNAs) have emerged as a new class of gene expression regulators playing key roles in many biological and pathophysiological processes. Here, we identify cardiac conduction regulatory RNA (CCRR) as an antiarrhythmic lncRNA. CCRR is downregulated in a mouse model of heart failure (HF) and in patients with HF, and this downregulation slows cardiac conduction and enhances arrhythmogenicity. Moreover, CCRR silencing induces arrhythmias in healthy mice. CCRR overexpression eliminates these detrimental alterations. HF or CCRR knockdown causes destruction of intercalated discs and gap junctions to slow longitudinal cardiac conduction. CCRR overexpression improves cardiac conduction by blocking endocytic trafficking of connexin43 (Cx43) to prevent its degradation via binding to Cx43-interacting protein CIP85, whereas CCRR silence does the opposite. We identified the functional domain of CCRR, which can reproduce the functional roles and pertinent molecular events of full-length CCRR. Our study suggests CCRR replacement a potential therapeutic approach for pathological arrhythmias.

Fig. 1

Expression deregulation of lncRNA-CCRR in heart failure (HF). a Real-time RT-PCR (qPCR) results showing the decrease of CCRR (cardiac conduction regulating RNA) expression in a mouse model of HF mice. ^**P < 0.01 HF vs. Sham-control; n = 18 mice for each group. b Downregulation of CCRR expression in patients with HF as compared to non-HF human subjects, determined by qPCR. ^**P < 0.01 HF (n = 5) vs. non-HF; n = 3 patient samples for each group. c Fluorescence in situ hybridization (FISH) images (200× magnification) showing the decrease of CCRR expression in HF mice. Left panel: representative images showing the reduced signal intensity (red) indicative of reduced CCRR level in the cytoplasm. Right panel: averaged values of fluorescence intensity for quantification of CCRR expression. NC: negative control probe. ^**P < 0.01 HF vs. Sham-control; n = 3 hearts for each group. d Northern blot analysis showing the decrease of CCRR expression in HF mice. ^*P < 0.05 HF vs. Sham; n = 4 mice for each group. (Mean ± SEM; analysis of variance—ANOVA followed by Dunnett’s test for comparisons among multiple groups, and Student t test for comparisons between two groups)

Fig. 2

Regulation of cardiac conduction and arrhythmias by CCRR in HF mice. a Slowing of conduction velocity (CV) in HF and restoration by CCRR overexpression. CV was determined by optical mapping techniques with a voltage-sensitive dye to define the cardiac activation. CV values were calculated from the gradient of the scalar field of 12-ms isochronal activation maps along the septal apex–base axis: CV = distance/12 ms. Note that CV was substantially decreased in HF and this conduction slowing was restored in the hearts pretreated with the lentivirus carrying the CCRR gene for overexpression (Lv-CCRR), but not with the negative control construct (Lv-NC). Viral vectors were administered by intracavity injection (directly injected into the left ventricular chamber). The Sham group underwent the same surgical procedures without TAC. ^*P < 0.05 HF or Lv-NC vs. Sham-control; ^#P  < 0.05 Lv-CCRR vs. HF; n = 4 mice for each group. b Antiarrhythmic effects of CCRR overexpression in a mouse model of HF. The incidence and duration of ventricular tachycardia (VT) induced by programmed stimuli were determined from ECG recordings. Note that the arrhythmogenicity was significantly enhanced in HF hearts, which was considerably suppressed by Lv-CCRR, but not by Lv-NC. The red lines in the ECG traces indicate VT duration, and the values within the parentheses in the bar charts indicate VT incidence. ^*P < 0.05 HF or Lv-NC vs. Sham-control; ^#P < 0.05 Lv-CCRR vs. HF. c Slowing of cardiac conduction induced by CCRR knockdown in healthy mice. The lentivirus vector engineered to contain a CCRR siRNA (Lv-siCCRR) was injected into the left ventricular chamber to silence myocardial CCRR. Lv-siCCRR caused a remarkable decrease in cardiac CV, whereas the negative control (Lv-siNC) failed to elicit any significant changes. The Sham group received the same surgical procedures without injecting vectors. ^*P < 0.05 Lv-siCCRR vs. Sham-Control; n = 4 mice for each group. d Pro-arrhythmic effects of CCRR knockdown in healthy mice. A prominent finding here is that knockdown of endogenous CCRR by Lv-siCCRR was sufficient to induce VT in otherwise healthy hearts. ^*P < 0.05 Lv-siCCRR vs. Sham-Control. (Mean ± SEM; ANOVA followed by Dunnett’s test for multiple group comparisons, Student t test for two group comparisons, and χ²-test for nonparametric data set comparisons)

Fig. 3

CCRR alleviates destruction of intercalated discs and gap junction in HF mice. a Typical examples of electron microscopic images (30,000× magnification) showing the deranged intercalated discs in HF mice (left panels) and patients (right panels), and the effects of CCRR overexpression by Lv-CCRR on the microstructure of left ventricular cardiac muscles in HF mice. Marked disruption of intercalated discs and gap junction was consistently observed in HF myocardium (indicated by arrows), and such derangement was ameliorated and restored to normal conditions after pretreatment with Lv-CCRR, but not with Lv-NC. Similar observations were repeated in another three HF mice. b Effects of CCRR overexpression by Lv-CCRR on the protein levels of connexin43 (Cx43) and CIP85 (the Cx43-interacting protein that regulates the endocytic trafficking of Cx43 for degradation) in HF mice. Upper panel: examples of Western blot bands; lower panel: averaged data on Cx43 and CIP85 protein levels. Comparisons of Cx43 and CIP85 at the protein levels were made between the membrane and cytoplasm fractions. Cx43 level was markedly decreased in the membrane, but increased in the cytoplasm, which resulted in an overall reduction of total protein level, indicating the enhanced endocytic trafficking of Cx43 in HF mice compared to the Sham-operated control animals. The membrane bands were normalized to N-cadherin (N-Cad), and the cytosolic and total protein bands were normalized to GAPDH (GAP). ^*P < 0.05, ^**P < 0.01, ^***P < 0.001; ^#P < 0.05, ^##P < 0.01, ^###P < 0.001 Lv-CCRR vs. HF; n = 5 heart samples of membrane fraction for each group, n = 6 for cytoplasmic fraction, and n = 6 for total protein. Similar changes were observed with CIP85, but its total protein level remained unaffected. ^*P < 0.05, ^**P < 0.01; ^##P < 0.01, Lv-CCRR vs. HF; n = 5 heart samples of membrane fraction for each group, n = 6 for cytoplasmic fraction, and n = 6 for total protein

Fig. 4

CCRR knockdown causes destruction of intercalated discs and gap junction in healthy mice. a Representative images of electron microscopy (30,000× magnification) showing the effects of knockdown of endogenous CCRR by Lv-siCCRR on the microstructure of left ventricular cardiac muscles in healthy mice. Note the profound derangement and loss of intercalated discs in the hearts pretreated with Lv-siCCRR (dark arrows), resembling those noted in HF heart, which was not observed with Lv-siNC and Sham-Control (subjected to the surgical procedures but without lentivirus injection). Similar results were observed in another three independent experiments. b Effects of CCRR knockdown by Lv-siCCRR on the protein levels of Cx43 (left panel) and CIP85 (right panel) in healthy mice, with comparisons between the membrane and cytoplasm fractions. Lv-siCCRR elicited similar alterations of Cx43 and CIP85 levels in normal hearts as those seen above with HF hearts. Cx43 was decreased in the membrane fraction, but increased in the cytoplasm, with a net reduction of total protein levels. ^*P < 0.05, ^**P < 0.01, ^***P < 0.001 Lv-siCCRR vs. Sham; n = 6 mice for each group. CIP85 showed similar changes but with the total protein level unaffected. ^*P < 0.05, ^**P < 0.01 Lv-siCCRR vs. Sham; n = 5 heart samples of membrane fraction for each group, n = 6 for cytoplasmic fraction, and n = 6 for total protein. (Mean ± SEM; ANOVA followed by Dunnett’s test for multiple group comparisons, and Student t test for two group comparisons)

Fig. 5

Effects of CCRR on the gap junction distribution of Cx43 and CIP85. a Immunostaining of cardiac slices showing the co-localization (yellow) of CIP85 (red) and Cx43 (green) proteins at the intercalated discs in healthy hearts (upper panel), and the effects of CCRR on the gap junction distribution of these proteins (600× magnification, lower panel). Knockdown of endogenous CCRR by Lv-siCCRR drastically diminished the presence of Cx43 (stained in green) in the intercalated discs in healthy mouse hearts, resembling the changes caused by HF. In contrast, CCRR overexpression by Lv-CCRR normalized the HF-induced expression repression of Cx43. DAPI (blue) was used to stain nuclei and α-actinin (red) to identify the cell contour. Similar results were observed in another three separate experiments. b Immunostaining of cardiac slices showing the co-localization of CIP85 (stained in red) and Cx43 (green) proteins at the intercalated discs in healthy control subjects (600× magnification). Cx43 in the intercalated discs in HF patient significantly diminished compared with healthy control subjects. DAPI (blue) was used to stain nuclei and α-actinin (red) to identify the cell contour. Similar results were observed in another two separate experiments

Fig. 6

CCRR improves gap junction communication in AC16 human heart cells. AC16 human adult ventricular cardiomyocytes were loaded with a diffusible fluorescent dye (Lucifer yellow), and the diffusion of the dye was monitored under a laser confocal microscope (200× magnification). Note that CCRR overexpression after infection of Lv-CCRR significantly promoted, whereas CCRR knockdown by Lv-siCCRR nearly completely blocked, the diffusion of Lucifer yellow among AC16 cells (green). The high molecular weight marker dye conjugate Rhodamine B was used as a negative control (red). Similar data were acquired from another three independent experiments

Fig. 7

Interactions between CCRR and CIP85 and between CIP85 and CX43. a RNA immunoprecipitation (RIP) showing the physical interaction between CCRR and CIP85 proteins in cultured neonatal mouse ventricular myocytes (NMVMs). Most left panel: the electrophoresis gel image showing the anticipated fragment representing CCRR that was pulled down by CIP85 antibody with IgG antibody as a negative control. Middle left panel: the quantified results of the electrophoretic bands. ^**P < 0.01 anti-CIP85 vs. anti-IgG; n = 4 batches of cells for each group. Middle right panel: an example of immunoblotting bands showing the pulldown of CIP85 protein by CCRR or other RNAs as labeled on the left. ZFAS1: an lncRNA used as a negative control; AS: antisense to CCRR used as another negative control; AR: androgen receptor 3′ UTR: a commercially obtained positive control; Input: purified CIP85 as a positive control; NC: negative control probe; HuR: Hu-antigen R. Right panel: mean relative band densities representing CIP85 protein with varying treatments. ^***P < 0.001 AS, ZFAS1, NC vs. CCRR; n = 4. b Protein co-immunoprecipitation (Co-IP) confirming the physical interaction between CIP85 and Cx43 in cultured NMVMs. Upper panel shows the presence of CIP85 in an anti-Cx43 pulldown sample, and lower panel shows the presence of Cx43 in an anti-CIP85 pulldown sample. c RIP results showing the physical interaction between FD and CIP85 proteins in heathy mice. Right panel: the quantified results of the electrophoretic bands. ^**P < 0.01 anti-CIP85 vs. anti-IgG; n = 3 mice for each group. d Immunoblotting image showing the pulldown of CIP85 protein by FD. Sham: non-treated sample; Lv-FD: lentiviral vector carrying FD. Lv-FD-NC: lentiviral vector carrying negative sequence. Lower panel: mean relative band densities representing CIP85 protein with varying treatments. ^**P < 0.01 Lv-FD vs. Lv-FD-NC; n = 4. (Mean ± SEM; ANOVA followed by Dunnett’s test for multiple group comparisons, and Student t test for two group comparisons)

Fig. 8

Effect of FD on gap junctional integrity. a Typical examples of electron microscopic images (30,000× magnification) showing the deranged intercalated discs in MI mice, and the effects of FD overexpression by Lv-FD on the microstructure of left ventricular cardiac muscles. Marked disruption of intercalated discs and gap junction was consistently observed in infarcted myocardium (dark arrows), and such derangement was ameliorated and restored to normal conditions after pretreatment with Lv-FD, but not with Lv-FD-NC. Similar observations were repeated in another three MI mice. b Effect of FD overexpression on the protein level of connexin43 (Cx43) in MI mice. Left panel: an example of Western blot bands; right panel: averaged data on Cx43 protein level. Cx43 expression was significantly repressed by MI and rescued by Lv-FD, relative to control and Lv-FD-NC. ^*P < 0.05 MI or Lv-FC-NC vs. Sham; ^#P < 0.05 Lv-FD vs. MI; n = 6 mice for each group. (Mean ± SEM; ANOVA followed by Dunnett’s test for multiple group comparisons, Student t test for two group comparisons)

Fig. 9

Functional domain of CCRR (FD) improves gap junction communication. a Effects of conserved FD on gap junction communication assessed by dye transfer techniques in AC16 human adult ventricular cells (200× magnification). AC16 cells were loaded with a diffusible fluorescent dye (Lucifer yellow), and the diffusion of the dye was monitored under a laser confocal microscope. Note that transfection of pFD for FD overexpression significantly promoted, whereas pFD-Mut did not alter, the diffusion of Lucifer yellow among AC16 cells (green). The high molecular weight marker dye conjugate Rhodamine B was used as a negative control (red). Similar data were acquired from another three experiments. b Effect of FD on the interaction between Cx43 and CIP85, as reported by co-immunoprecipitation in cultured NMVMs transfected with the plasmid carrying the FD (pFD) for overexpression. Note that the protein level of Cx43 co-immunoprecipitated by CIP85 antibody was significantly reduced. pFD-Mut: the mutant construct of pFD with nucleotide substitution to the putative binding site of CCRR or FD to CIP85. ^**P < 0.01 pFD vs. pFD-Mut; n = 3 batches of cells for each group. (Mean ± SEM; ANOVA followed by Dunnett’s test for multiple group comparisons, and Student t test for two group comparisons). c Effect of CCRR or FD on the recruitment of Cx43 to lysosome and the presence of Cx43 in the plasma membrane determined by co-immunostaining of Cx43 and lysosome marker Lamp1 in cultured NMVMs (600× magnification). Note that either pCCRR or pFD weakened the overlapping staining of Cx43 and Lamp1, and enhanced the Cx43 staining to the cytoplasmic membrane. n = 3 batches of cells for each group

Fig. 10

Schematic illustration of the proposed signaling pathway for CCRR actions. CCRR binds to CIP85 to prevent the binding of CIP85 to Cx43 to reduce the internalization of Cx43 and block CIP85-induced endocytic trafficking of Cx43 thereby rescuing the functional presence of Cx43 in the cytoplasmic membrane