The G4 resolvase Dhx36 modulates cardiomyocyte differentiation and ventricular conduction system development

Abstract

Extensive genetic studies have elucidated cardiomyocyte differentiation and associated gene networks using single-cell RNA-seq, yet the intricate transcriptional mechanisms governing cardiac conduction system (CCS) development and working cardiomyocyte differentiation remain largely unexplored. Here we show that mice deleted for Dhx36 (encoding the Dhx36 helicase) in the embryonic or neonatal heart develop overt dilated cardiomyopathy, surface ECG alterations related to cardiac impulse propagation, and (in the embryonic heart) a lack of a ventricular conduction system (VCS). Heart snRNA-seq and snATAC-seq reveal the role of Dhx36 in CCS development and in the differentiation of working cardiomyocytes. Dhx36 deficiency directly influences cardiomyocyte gene networks by disrupting the resolution of promoter G-quadruplexes in key cardiac genes, impacting cardiomyocyte differentiation and CCS morphogenesis, and ultimately leading to dilated cardiomyopathy and atrioventricular block. These findings further identify crucial genes and pathways that regulate the development and function of the VCS/Purkinje fiber (PF) network.

Fig. 1. Dhx36 protein expression in embryonic and adult hearts.

a Immunohistochemical (IHC) analysis of ED10.5, ED11.5, and ED12.5 embryonic hearts stained with IgG (upper panels) or α-Dhx36 (lower panels). Panel i presents a magnified view of a specific region (boxed area) within the left ventricle of an ED12.5 heart. b IHC staining of adult hearts with IgG (left upper panel) or α-Dhx36 (left lower panel). Panels i and ii show magnified views of the corresponding boxed areas. c Western blotting showing Dhx36 expression in cytoplasmic (Cyt.) and nuclear (Nuc.) subcellular compartments of total adult hearts. Specific expression was validated by analyzing total extracts of C2C12 myocytes and mouse brain. Anti-tubulin (Tub.) was blotted in the same membrane after α-Dhx36 and served as a cytoplasmic control. Samples derived from the same experiment. d IHC of Dhx36 expression in WT and Dhx36 ^Nkx2-5 ED12.5 embryonic hearts (top left panels). The right panels depict magnifications of regions of the right ventricles (RV, top) or left ventricles (LV, bottom). As a control for Dhx36 staining, the bottom left panels show IHC in the neural tube of the same WT and Dhx36 ^Nkx2-5 embryos. All micrographs and western blots shown are representative of three independent experiments rendering similar results. Uncropped blot for Fig. 1c is provided in the Source data file. Scale bars are expressed in μm and apply to corresponding images, as they present the same magnifications.

Fig. 2. Dhx36 deletion induces dilated cardiomyopathy, reduced ejection fraction, and surface ECG alterations.

a Kaplan-Meier survival curve comparing WT mice (Dhx36 ^f/f) with Dhx36^Tnnt2 mutant mice. b Gross morphology of hearts from 21-day-old WT (top left) and Dhx36^Tnnt2 conditional KO (cKO.1) mice (bottom left). Right panels show hematoxylin & eosin (H&E)-stained sections of the corresponding hearts. c Left panels: a 90-day old Dhx36^Tnnt2 mouse (cKO.2) exhibiting overt signs of dilated cardiomyopathy and non-compaction of the left ventricular (LV) myocardium. The left atrium displays a prominent thrombus in the H&E-stained section. Right panels: picrosirius red stained sections highlight larger LV fibrotic areas (lower) compared to representative WT (upper) sections. The pictures in b and c are representative of 5 and 3 independent experiments, respectively, analyzing WT and mutant hearts of around similar ages, all showing similar phenotypes. d Box plots representing the distribution of IQR values for LV ejection fractions (LVEF; left) and the LV end-diastolic (LVd) volumes (right) of Dhx36^Tnnt2 mutant mice (n = 10) and WT (n = 10). Lower and upper hinges of the boxes correspond to Q1 and Q3 (25th and 75th percentiles), with the median represented by the horizontal line inside the box, while whiskers extend from hinges to minima and maxima values. The exact values of all these data are annotated in the Source Data file. e, Comparisons of PR interval (left), QRS complex duration (middle), and amplitude (right) between Dhx36^Tnnt2 mutant mice (n = 10) and WT (n = 9). f Kaplan-Meier survival curve comparing WT mice (Dhx36 ^f/f) with Dhx36^Myh6 mutant mice. g Same analysis as in b and c but of hearts from WT and Dhx36 ^Myh6 mutant mice aged 40 days (cKO.1) or 82 days (cKO.2). The cKO.2 heart exhibited a large thrombus in the left atrium. The pictures shown are representative of 4 and 3 independent experiments, respectively, analyzing WT and mutant hearts of similar ages, all showing similar phenotypes. h. Box plots comparing (as in d) LVEFs (left) and LVd volumes (right) between Dhx36 ^Myh6 mutant mice (n = 10) and WT (n = 9). The values are represented in the Source Data file. i Comparisons of PR interval (left), QRS complex duration (middle), and amplitude (right) between Dhx36^Myh6 mutant mice (n = 7) and WT (n = 6). Asterisks indicate left atrial thrombi, and arrowheads show myocardial areas with non-compacted trabeculae. No phenotypic differences were observed between sexes (sex of the mice are annotated in the Source data file). Significance was determined using unpaired two-sided t-test, except for a and f, in which a χ² was used. p-values are shown in the corresponding figures. Source data for d, and h are provided in the source data file and in Supplementary Data 1 for e and i. Scale bars are as in Fig. 1.

Fig. 3. Dhx36 deletion increases abnormal paroxysmal arrhythmic events.

a, b Comparative analysis of RR intervals (heart rate calculated as [1000/RR interval in ms] x 60) between Dhx36^Tnnt2 mutant mice and WT (a), and Dhx36^Myh6 mutants and WT (b). c Incidence of relevant atrioventricular (AV) block events during ECG recordings of WT mice and Dhx36^Tnnt2 and Dhx36^Myh6 mutant mice. d Representative ECG tracings depicting a WT mouse and a Dhx36^Tnnt2 mutant mouse with a paroxysmal AV block event. e Incidence of sinus node dysfunction during ECG recordings in WT mice and Dhx36^Tnnt2 and Dhx36^Myh6 mutant mice. f RR variability representation of Dhx36^Tnnt2 mice (in light red) as compared to WT mice (in grey). Each dot represents a RR interval. g Sample tracing with overt signs of sinus node dysfunction (arrowhead) during the ECG recording of a Dhx36^Tnnt2 mutant mouse. h Incidence of premature ventricular complexes (PVC) during ECG recordings in WT mice and Dhx36^Tnnt2 and Dhx36^Myh6 mutant mice. i Sample tracing of a potential nonsustained bidirectional ventricular tachycardia episode in an ECG recording of a Dhx36^Tnnt2 mutant mouse. Green arrowheads show one PVC morphology, probably from Purkinje fibers at the root of the His-Purkinje system, and blue arrowheads show a second PVC morphology, in the absence of overt P waves. In panels c, e, and h, WT mice were grouped, as they did not exhibit any significant cardiac rhythm alterations. The mice analyzed were the same that in Fig. 2e, i. Statistical significance was determined using unpaired two-sided t-test (a and b) and the data are presented as mean ± SEM. The source data are provided in the Source data file and Supplementary Data 1.

Fig. 4. Impact of perinatal and embryonic Dhx36 deletion on CCS morphogenesis.

a Whole-mount (WM) confocal immunofluorescence (IF) of hearts of 41-day-old (PD41) WT and Dhx36^Myh6 littermates, revealing an underdeveloped ventricular conduction system/Purkinje fiber (VCS/PF) network in the mutant. The pictures are representative of 3 independent experiments, analyzing WT and mutant hearts of similar ages, all rendering similar results. b Left: gross morphology of hearts from PD28 WT and Dhx36^Tnnt2 littermates. Right: WM confocal IF of the same hearts, indicating an undetectable VCS/PF network in the mutant. Representative images are presented from a total of three Dhx36^Myh6 and five Dhx36^TnnT2 analyzed mutants. The pictures are representative of 5 independent experiments, analyzing WT and mutant hearts of similar ages, all rendering similar results. c–g ISH analysis of PD7 (P7) wild type and Dhx36^Tnnt2 hearts hybridized with various CCS markers. Representative images are shown, three wild type and mutant embryos were examined; rv, right ventricle; lv, left ventricle. Scale bars, 100 µm in a, b (confocal images), 1 mm in b (whole hearts images). 200 µm in (c–g, general heart views), 100 µm in (c, d, e, g, high magnification views), and 50 µm (f, high magnification, showing Hcn4 staining in the AVN). The depicted scale bars are valid for both corresponding WT and mutant pictures.

Fig. 5. Single-nucleus RNA sequencing (snRNA-seq) and snATAC-seq multiome of neonatal WT and Dhx36^Tnnt2 mutant hearts.

a Multimodal Uniform Manifold Approximation Projection (UMAP) visualization of cell clusters in WT hearts, randomly colored by identity. b Heatmap illustrating expression of Dhx36 in each cell cluster, projected on the UMAP graph of WT heart. c Violin plots presenting expression of Dhx36 in individual cell clusters. d Multimodal UMAPs displaying all WT cells in blue (n = 6499) and cKO cells (n = 5426) in red (upper panel). Below, WT and KO cells are separated in their multimodal UMAPs, randomly colored by identity as in (a). e, Fraction (%) of cell population clusters in each sample. Refer to Supplementary Fig. 6 for additional details. EC coronary = Coronary endothelial cells; EC Endoc.= Endocardial endothelial cells; SMCs= Smooth muscle cells.

Fig. 6. Single-nucleus RNA sequencing (snRNA-seq) and snATAC-seq multiome identify differences between WT and mutant hearts in cardiomyocyte populations.

a Zoomed UMAP visualization of cardiomyocyte clusters colored by identity in the WT (left) and the mutant heart (right). b Violin plots depicting the expression of specific representative genes for each cardiomyocyte subcluster. Refer to Results for definition of CM clusters.

Fig. 7. Widespread transcriptomic alterations of mutant versus WT cardiomyocytes.

a Quantitative PCR of selected cardiac conduction system genes and genes deregulated in the snRNA-seq between WT (n = 3) and mutant neonatal PD7 (KO; n = 3) hearts. b CTCF motif enriched at mutant cardiomyocyte-active open chromatin regions compared to WT (left). UMAPs of a zoom of the KO and WT cardiomyocytes, depicting their level of open chromatin at CTCF motifs (right). Violin plots of Z scores of CTCF open chromatin in specific CM1/mCM0, CM2/mCM2 and auricular CM7/mCM7 CM-specific clusters are shown below. Inset boxplots show the median, lower and upper hinges as well as whiskers as in Fig. 2d and i. Significance was determined in a using unpaired two-sided t-test; the data are presented as mean ± SEM with the p-values inserted. Source data for a and b are provided in the source data file.

Fig. 8. G4 structures and transcriptional alterations of Dhx36 gene targets.

a Examples (Ntn1 and Nr4a1) of genes downregulated in KO CMs (Supplementary Data 10) with G4s overlapping open chromatin in their promoters (TRUE). b Example (Esrrb) of a gene downregulated in KO CMs with G4 not overlapping open chromatin in its promoter (FALSE). c Example (Rai2) of a gene upregulated in KO CMs with G4s overlapping open chromatin in their promoters (TRUE). On the left, the plots show the ATAC-seq data with the peaks of open chromatin in the promoter regions of the genes in all ventricular WT CMs (CM1) or mCM0, in WT + KO atrial CMs (CM7), and in VCS/PF (CM6) cardiomyocytes. The corresponding gene (with accessible peak represented as a grey box), the TRUE G4 (blue colored box) or FALSE G4 (salmon colored box) and the links between the regulatory regions (lower) are also shown. On the right, the violin plots of RNA-seq expression of the genes are shown in each cluster.

Fig. 9. G4-resolvase-dependent transcriptional activation of Dhx36 CM target genes.

a Luciferase transcriptional activation assays in HeLa cells of promoter regions of Dhx36, Kcne1, Bcl2l11, Ntn1, Ppargc1a (Pgc1a), Fhl2, and Nr4a1 cloned in pGL4.25 luciferase expressing vector. The luciferase activity, measured in Relative light units (R.L.U.) of the different constructs, are shown in the presence or absence of 20 μM of the G4 stabilization drug TMPyP4. b same as in (a) with Dhx36, Drd2, Hcn4 and Cntn2 promoters. The pGL4.25 vector, presenting a minimal promoter, is shown as a negative control. Significance was determined in a and b using unpaired two-sided t-test; the data with the p-values included are presented as mean ± SEM (n = 3). The experiments shown in (a) and (b) are representative of 3 independent experiments (total n = 9), with similar results. Source data are provided in the source data file.

Fig. 10. Model of the impact of Dhx36 deletion in the heart.

Cardiac phenotypes observed upon Dhx36 deletion with the indicated Cre-driver lines described in this study and in Nie et al and Huang et al. For each mutant line, the dotted red line indicates the time window of Cre-driver allele action, and red shading indicates the expected anatomical recombination domain and deletion strength within the heart. Timelines depict the approximate extent of survival, the onset of lethality, and the phenotype penetrance for each driver. The panel to the right indicates the range and intensity of the cardiac phenotypes reported here and in previous studies. AVB, atrioventricular block; CCS, cardiac conduction system; LA, left atrium; LVNC, left ventricular non-compaction; RV, right ventricle; RVNC, right ventricular non-compaction; VCS, ventricular conduction system.