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

The heart depends on specialized cardiomyocytes within the CCS to ensure rapid and synchronized electrical impulse propagation. The CCS includes the SAN, AVN, His bundle, and bundle branches integrated within the Purkinje system, and its development extends from embryonic to early neonatal stages. Despite progress, the transcriptional and posttranscriptional mechanisms that control CCS development and working cardiomyocyte differentiation remain incompletely defined. This study investigates whether guanine-rich nucleic acid secondary structures (G-quadruplexes; G4) in DNA and RNA, which are resolved by the helicase Dhx36, regulate heart development and CCS formation. The central question is how Dhx36-mediated G4 resolution influences cardiomyocyte gene expression programs, differentiation, and CCS morphogenesis, and how its loss leads to structural and electrophysiological defects.

Prior single-cell transcriptomic and epigenomic studies have mapped cardiac cellular diversity and dynamics during neonatal regeneration and after injury, identifying major cardiac cell types and transcriptional networks. Early lineage studies identified progenitors expressing Nkx2-5 and Isl1 and delineated zone-specific cell types and signatures. Dhx36 (RHAU/G4R1) is known to resolve RNA G4s and modulate mRNA stability and translation in cardiomyocytes (e.g., Nkx2-5, Hexim1, Hey2, Yap1), yet its direct role in transcription, despite its capacity to resolve DNA G4s, is less explored. Single-cell approaches (scRNA-seq/scATAC-seq) overcome limitations of bulk RNA-seq in conditional knockouts by deconvolving cell type–specific effects and have been instrumental in characterizing CCS development and function in mouse and human hearts.

• Mouse genetics: Conditional deletion of Dhx36 using cardiomyocyte-specific Cre drivers with different temporal windows: Tnnt2-Cre (embryonic onset), Myh6-Cre (perinatal/postnatal onset), and additional lines (MCK-Cre; Nkx2-5CreERT2 for validation of expression). Survival and phenotypes were assessed across ages.
• Histology and protein localization: Immunohistochemistry (IHC) and in situ hybridization (ISH) were used to assess Dhx36 protein localization and CCS marker expression during embryogenesis (ED10.5–ED12.5, ED16.5) and postnatally (PD7). Western blotting of adult heart subcellular fractions determined Dhx36 nuclear and cytoplasmic localization.
• Cardiac imaging and electrophysiology: Echocardiography measured LV function (LVEF, LV end-diastolic volume). Surface ECGs in sedated mice quantified PR interval, QRS duration and amplitude, RR variability, and paroxysmal arrhythmias (AV block, PVCs, episodes suggestive of bidirectional VT).
• CCS morphogenesis assays: Whole-mount confocal immunofluorescence of the pan-VCS marker Cntn2 quantified VCS/Purkinje fiber network development in WT and mutants.
• Bulk RNA-seq: Whole hearts from 2–3-week-old mice (WT vs Dhx36 cKO) were sequenced (pools of 5–6 hearts, in triplicate). Differential expression and GO/Enrichr analyses focused on downregulated genes relevant to ion channel activity and conduction.
• Single-nucleus multiome (snRNA-seq + snATAC-seq): PD7 hearts (WT and Dhx36 cKO) were profiled using 10x Genomics Multiome. Data processing included SCTransform, PCA/UMAP for RNA, LSI/SVD for ATAC, WNN integration, and Louvain clustering. Cell types were annotated by marker genes. Differential expression (Wilcoxon) and differential accessibility (logistic regression with fragment count as latent variable) analyses were performed. Peak-to-gene links were defined by correlation with bias correction. Motif enrichment (JASPAR) and motif activity (chromVAR) were computed. Predicted G4s overlapping DA peaks were evaluated (GAIA/G4RNA screener) with hypergeometric testing for enrichment.
• Promoter cloning and luciferase assays: Promoter regions (with/without predicted G4s) from key genes (Hcn4, Cntn2, Kcne1, Drd2, Bcl2l11, Ppargc1a, Ntn1, Fhl2, Nr4a1, Dhx36) were cloned into pGL4.25. HeLa cells were transfected and treated with TMPyP4 (20 μM) to stabilize G4s. Luciferase activity was measured to test G4-dependent transcriptional regulation.
• Statistics: Unpaired two-tailed t-tests for most comparisons; χ² for survival curves; Bonferroni adjustments in multiomic DE/DA analyses; significance at p<0.05. Data deposited in EBI (E-MTAB-14440/14441/14442).

• Dhx36 expression and localization: Dhx36 is highly expressed in embryonic hearts (peak ED11.5), enriched in trabecular cardiomyocytes (ED12.5), and detectable in adult atria; present in both nuclear and cytoplasmic fractions of adult hearts.
• Cardiac phenotypes upon deletion:
  • Embryonic cardiomyocyte deletion (Tnnt2-Cre): Mice show dilated cardiomyopathy with LV non-compaction, thin ventricular walls and septum, atrial dilation, interstitial fibrosis; reduced contractility (increased LVd volume) and surface ECG abnormalities including PR and QRS prolongation; higher incidence of paroxysmal AV block (including 2:1, Mobitz II/advanced, complete) and sinus node dysfunction; increased PVCs and occasional nonsustained episodes compatible with bidirectional VT. Left atrial thrombosis observed in end-of-life animals.
  • Perinatal/postnatal deletion (Myh6-Cre): Premature death with dilated cardiomyopathy, early atrial dilation and fibrosis, reduced LVEF and increased LVd; no significant PR/QRS alterations vs WT.
• CCS morphogenesis: Whole-mount Cntn2 IF shows hypoplastic VCS/PF network in Myh6-Cre mutants and undetectable Cntn2 immunoreactivity (apparent VCS agenesis) in Tnnt2-Cre mutants. ISH at ED16.5 shows downregulation of VCS markers (Etv1, Gja5, Slit2, Irx3) with relative preservation of Nppa and upregulation of Nrg1; at PD7, reduced Hcn4 and expanded Nppa expression with stable Irx3. Findings indicate impaired CCS development and maturation.
• Bulk RNA-seq (2–3 weeks): 703 upregulated (238 ≥2-fold) and 354 downregulated (100 ≤−2-fold) transcripts in Dhx36 cKO vs WT. GO terms depleted in cardiac action potentials and ion transport; molecular functions depleted in inward rectifier and voltage-gated potassium channel activity. Key conduction genes downregulated (e.g., Hcn4, Kcne1, Kcnd2, Kcnv2). qPCR validates downregulation of Hcn4 (~2.0-fold) and Kcne1 (~2.27-fold) among others in Myh6-Cre mutants.
• Single-nucleus multiome (PD7): 11 cardiac cell types identified; ~47% cardiomyocytes. Mutant hearts show similar cellular composition but distinct cardiomyocyte transcriptomes/chromatin. WT cardiomyocytes partition into seven clusters (CM1–CM7), including a VCS/PF cluster (CM6) expressing Cntn2, Tbx3, Robo1, Cpne5, Sema3a, Ncam1; atrial CM7. Mutants lack the WT CM3 (Ddc+), CM4, CM5, and CM6 clusters, and instead show a predominant mCM0 cluster with stress/non-regenerative signature (elevated Xirp2, Enah; reduced Fhl2). Atrial (mCM7) and proliferating (mCM2) clusters persist.
• Differential accessibility and DEG integration: 679 differentially accessible (DA) regions in KO cardiomyocytes; intersecting DA-linked genes and DEGs identifies 58 downregulated genes including transcription factors (Gata6, Fosl2, Ppargc1a, Nr4a1/3) and cardiac function genes (Hcn4, Ntn1, Bcl2l11, Fhl2, Plekhh1, Irs2). qPCR validates selected targets at PD7.
• Motifs: 25 cardiomyocyte-enriched motifs; chromVAR reveals 206 differentially active motifs with strong enrichment of CTCF in KO open chromatin, while NFY (Nfya/b/c), Mef2, Tfeb/Tfe3, and Nkx family motifs are underrepresented in KO.
• G4 involvement: Of 679 DAPs, 173 contain predicted G4s; G4s are enriched in DA peaks (1.32-fold; p=4×10^−5). Downregulated genes with promoter G4s overlapping DAPs include Hcn4, Drd2, Cntn2, Kcne1, Bcl2l11, Ppargc1a, Ntn1, Fhl2, Nr4a1. In the VCS/PF program, Kcne1 has a promoter G4; Hcn4, Cntn2, Drd2 harbor potential G4s.
• Functional reporter assays: Promoters of Dhx36, Kcne1, Bcl2l11, Ppargc1a, Ntn1, Fhl2, Nr4a1, and CCS markers (Hcn4, Cntn2, Drd2) drive luciferase expression in HeLa cells; G4 stabilizer TMPyP4 significantly reduces activity for all tested promoters except Dhx36, supporting G4-resolvase-dependent transcription.
• Additional observations: Nkx2-5 mRNA increases but protein decreases in KO hearts, indicating Dhx36-dependent translational control; mutant hearts show increased proliferating fibroblasts and endothelial cells, consistent with remodeling; upregulation of Nrg1 suggests compensatory signaling.

The data demonstrate that Dhx36 is a key regulator of cardiomyocyte differentiation and CCS morphogenesis via transcriptional control that requires resolution of promoter G-quadruplexes. Loss of Dhx36 during embryogenesis disrupts formation of the VCS/PF network, downregulates CCS gene programs (e.g., Hcn4, Cntn2, Kcne1), and causes conduction delays (PR and QRS prolongation) and advanced AV block, culminating in dilated cardiomyopathy and thrombus-prone atrial dilation. Deletion initiated postnatally yields a hypoplastic but present VCS with predominant systolic dysfunction, highlighting temporal dependence of Dhx36’s roles in CCS development versus working myocardium homeostasis. Multiomic profiling shows that Dhx36-deficient cardiomyocytes lose specific WT clusters, including VCS/PF (CM6) and Ddc+ CM3, and instead adopt a stressed, non-regenerative mCM0 state with altered chromatin accessibility and transcription factor motif usage (increased CTCF, decreased NFY, MEF2, NKX activities). Integration of snATAC and snRNA data points to genes whose promoters harbor G4s overlapping DA peaks, consistent with a model where Dhx36 resolves dG4s to maintain open, transcriptionally competent chromatin at key cardiac genes. Luciferase assays with TMPyP4 further support G4-dependent promoter activity for multiple targets. Pathway analyses suggest axon guidance signals (Robo/Slit, Netrin-1/DCC) and secreted factors like Nrg1 may contribute to VCS morphogenesis and possibly compensate when Dhx36 is absent. Dhx36 also modulates Nkx2-5 post-transcriptionally, but additional Dhx36-dependent transcriptional defects likely underlie the full phenotype. Collectively, Dhx36 coordinates gene networks across CCS and working cardiomyocytes, linking G4 resolution to chromatin accessibility, transcription, and cardiac conduction.

This work establishes Dhx36 as an essential G4-resolving helicase that controls cardiomyocyte differentiation and CCS development by regulating chromatin accessibility and transcription of key cardiac genes. Conditional Dhx36 deletion reveals timing-dependent effects: embryonic loss ablates VCS/PF development and causes conduction block with dilated cardiomyopathy, while perinatal loss yields a hypoplastic VCS and predominant systolic dysfunction. Multiomic analyses identify gene networks and new candidate PF markers, highlight G4-enriched promoters among Dhx36 targets, and implicate axon guidance pathways in PF morphogenesis. Future studies should dissect cell-autonomous roles of Dhx36 in CCS sublineages, define the 3D chromatin architecture and CTCF–G4 interactions at Dhx36 target loci (e.g., Ddc), validate newly identified PF markers and pathways (Robo/Slit, Netrin-1) in vivo, and explore the relevance of DHX36 variation in human cardiac conduction and cardiomyopathy. Therapeutic modulation of G4 dynamics or Dhx36 activity may represent a strategy to correct conduction and contractile defects.

• Genetic models used broad cardiomyocyte Cre drivers, limiting the ability to define cell-autonomous effects within specific CCS subpopulations; phenotypes may reflect both intrinsic and non–cell-autonomous influences.
• Inability to generate Dhx36;Cx40-eGFP compound mutants (genomic proximity) precluded certain lineage-tracing and visualization strategies.
• Absence of the VCS/PF cluster in KO snRNA-seq data limited direct interrogation of Dhx36’s transcriptional effects within PF cells.
• Use of whole hearts for bulk RNA-seq confounds cell-type-specific interpretation; although addressed with single-nucleus multiome, spatial context is lacking.
• HeLa-based promoter assays may not fully recapitulate cardiomyocyte-specific chromatin context.
• Findings are in mice; no DHX36 mutations are currently linked to human heart disease, so translational extrapolation is uncertain.