TAD boundary deletion causes PITX2-related cardiac electrical and structural defects

Abstract

While 3D chromatin organization in topologically associating domains (TADs) and loops mediating regulatory element-promoter interactions is crucial for tissue-specific gene regulation, the extent of their involvement in human Mendelian disease is largely unknown. Here, we identify 7 families presenting a new cardiac entity associated with a heterozygous deletion of 2 CTCF binding sites on 4q25, inducing TAD fusion and chromatin conformation remodeling. The CTCF binding sites are located in a gene desert at 1 Mb from the Paired-like homeodomain transcription factor 2 gene (PITX2). By introducing the ortholog of the human deletion in the mouse genome, we recapitulate the patient phenotype and characterize an opposite dysregulation of PITX2 expression in the sinoatrial node (ectopic activation) and ventricle (reduction), respectively. Chromatin conformation assay performed in human induced pluripotent stem cell-derived cardiomyocytes harboring the minimal deletion identified in family#1 reveals a conformation remodeling and fusion of TADs. We conclude that TAD remodeling mediated by deletion of CTCF binding sites causes a new autosomal dominant Mendelian cardiac disorder.

Fig. 1. Overlapping deletion in seven families presenting a new cardiac entity associating electrical disorder and developmental defects recapitulated in a mouse model.

a Pedigree of seven families presenting a new cardiac entity. Symbols marked by a slash indicate that the subject was deceased. Males were indicated by squares and females by circles. Black symbols represent individuals who present at least one of the cardiac phenotypes listed in Table 1 according to clinical examination. “+” symbol denotes deletion carriers and “−” symbol non-carrier. The proband is indicated by the arrow. b Overview of the 4q25 human locus in UCSC Genome browser with the reference genome assembly hg19. Genes are represented in black and blue. 7 Deletions sizes are represented in different horizontal color lines: red/Family#1 (chr4:112555060-112570371), green/Family#2, brown/Family#3, yellow/Family#4, Blue/Family#5, dark blue/Family#6 and purple/Family#7. Human-mouse conservation is measured by Cons 46-way (phastCons and phyloP score) provided by UCSC Genome browser. c Graphs depict individual and average measurements for RR interval and cumulative number of patients presenting sinus node dysfunction (SND), atrial fibrillation (AF), atrial septal defect (ASD) in deletion carriers (RR interval n = 70, SND n = 69, AF n = 58, ASD n = 58) and non-carriers (RR interval n = 31, SND n = 33, AF n = 32, ASD n = 27) in the seven families. Significance for each parameter was determined with two-sided Mann–Whitney test (RR interval p < 1.0E−04) and two-sided Fisher’s exact test (SND p < 2.2E−16, AF p = 2.1E−05, ASD p = 3.7E−03). d Representation of the reverse orthologous mouse locus in UCSC Genome browser with the reference genome assembly mm10. Genes are represented in black and blue. Homozygous 15Kb deletion generated in the mouse model is represented in red delimited by the dashed black lines. e Graphs show individual and average RR interval, Standard deviation of NN intervals (SDNN; heart rate variation), the duration of atrial arrhythmia (AA) episode (WT, n = 25, DelB, n = 15) after two pacing passes, number of mice with at least one episode lasting >1 s is indicated above each bar graph of control (n = 25) and DelB (n = 15) mice. Number of mice from control (n = 10) and DelB (n = 12) presenting atrial septal defect are depicted above bar graph. Significance for each parameter was determined with two-sided Welch’s t test (RR p = 0.0002), two-sided Mann–Whitney test (SDNN p = 6.3E−05, AA duration p = 1.18E−5), or two-sided Fisher’s exact test (AA inducibility p = 0.0021, ASD Adult p = 0.0396). Individual data are available in Supplementary Data 1 and in the Source Data file.

Fig. 2. The 15Kb region contains CTCF TAD boundary.

Annotation of the 4q25 region with Hi-C map interaction profile from hiPSC-CM WT. Dashed triangles delimit the PITX2-TAD, Non-coding-TAD, ANK2-TAD. Coding genes are represented in blue and non-coding genes in green and 7 Deletions sizes are represented in different horizontal color lines: red/Family#1 (chr4:112555060-112570371), green/Family#2, brown/Family#3, yellow/Family#4, Blue/Family#5, dark blue/Family#6 and purple/Family#7. ATAC-seq tracks from adult human ventricular tissue from CARE database, ATAC-seq from hiPSC-CM WT, CUT&RUN tracks from hiPSC-CM WT using H3K27ac, H3K4me3 and H3K4me1 and CTCF antibodies are displayed. Zoom in of the 15Kb region annotated with ATAC and CUT&RUN CTCF tracks are represented in the bottom panel. The blue line highlights the open region containing the CTCF binding sites. Black arrows represent CTCF binding site orientation. Genome tracks were generated using UCSC Genome browser with the reference genome assembly hg19.

Fig. 3. 3D chromatin remodeling induces new interaction between PITX2 and SAN specific regulatory region resulting to PITX2 expression remodeling.

a Hi-C subtraction map of hiPSC-CM WT and hiPSC-CM 15Kb−/− (red indicates gain of interaction and blue indicates loss of interaction). Dashed triangles delimit the fused TAD. Coding genes are represented in blue and non-coding genes in green. b Annotation of the 4q25 region with ATAC-seq from adult atrial and ventricular tissues from CARE database (hg38) and from hiPSC-derived pacemaker-like cardiomyocytes. Genome conservation across 100 vertebrates is presenting with phyloP score. Location of AF associated region in the genome is highlight with a green horizontal line and location of PLAYRR in purple horizontal line. The Blue line highlights the open region containing the CTCF binding sites and gray lines highlight regulatory regions newly interacting with PITX2 (R1 and R2). Annotation of the 4q25 region with, ATAC-seq from hiPSC-CM WT, CUT&RUN tracks from hiPSC-CM WT using H3K27ac, H3K4me3 and H3K4me1. Annotation of R1 and R2 regulatory regions with human and mouse binding sites for cardiac transcription factors (ISL1, GATA4, NKX2-5, MEIS1, TEAD and TBX5). Genome tracks were generated using UCSC Genome browser with the reference genome assembly hg19.

Fig. 4. Pitx2 expression is deregulated in the DelB SAN and ventricles.

a Fold change expression levels of detectable TAD genes in P21 DelB (n = 9) vs. control littermates (dotted line) in the left and right atria and ventricles. Red and blue bars highlight upregulation (p = 0.0009) and downregulation (p = 0.0007) of Pitx2 respectively. Statistical significance was determined across four genotypes and within each tissue type using Kruskal–Wallis test followed by pairwise comparisons with Dunn’s multiple comparison tests in right and left atrial, and Welch’s ANOVA followed by Dunnett’s T3 multiple comparison tests in ventricles. Significant p values adjusted for multiple testing are shown in the graph. b Whole-tissue RNA sequencing was performed on micro-dissected SAN (in red) and RA regions (PCM pacemaker cardiomyocyte, LA left atrium, PV pulmonary vein, SVC superior caval vein, IVC inferior caval vein, RA right atrium, SAN sinoatrial node). Expression analysis of microdissected SANs of wild type and DelB mice show that Pitx2 and Larp7 are the only differentially expressed TAD genes between control (n = 8) and DelB (n = 7) adult SANs. Differential expression analysis was performed using the DESeq2 package. p values were corrected for multiple testing using a false discovery rate with 0.05. c Expression analysis of microdissected SANs of wild type and DelB mice show differential expression of genes between control and DelB adult mice. Blue dots highlight genes downregulated and red dots highlight genes upregulated in DelB mice. The differential expression analysis was performed using the DESeq2 package with a two-sided Wald test and multiple testing correction using a false discovery rate with 0.05. d Scatter plot showing the relative expression of genes enriched in the pacemaker cardiomyocyte cluster alongside the differential expression of genes between control (n = 8) and DelB (n = 7) SAN region. e Relative PITX2 expression in hiPSC-CM WT and hiPSC-CM-CTCF−/− ventricle like cardiomyocytes (hiPSC-VCM n = 8, n = 8) and pacemaker cell like cardiomyocytes (hiPSC-PCM n = 13, n = 19). Source data are provided as a Source Data file.

Fig. 5. The mutually exclusive expression of the lncRNA Playrr and Pitx2 is disrupted in the DelB SAN.

a RNA-seq datasets showing that Playrr is expressed in the embryonic day (E)14.5 sinoatrial node (SAN) and less so in the E14.5 right atrium (RA) and in the prenatal human SAN at 9 weeks. Genome tracks were generated using UCSC Genome browser with the reference genome assembly mm10. b Playrr expression level normalized to Myh6 in the mouse SAN (n = 3) and RA (n = 3) at E14.5, postnatal day P4 and P14 (data from ref. ). Error bars represent SD. c PLAYRR expression level normalized to MYH6 in the human SAN (n = 2) and RA (n = 2) at 9/12 weeks of gestation (data from ref. ). d Playrr and Pitx2 expression in adult WT (n = 8) and DelB (n = 7) micro-dissected SAN normalized to Myh6. Error bars represent SD. Source data are provided as a Source Data file.