Engineered heart tissues facilitate functional characterization of noncoding variants implicated in Hypertrophic and Dilated Cardiomyopathy

doi:10.21203/rs.3.rs-7696371/v1

Download PDF

Research Article

Engineered heart tissues facilitate functional characterization of noncoding variants implicated in Hypertrophic and Dilated Cardiomyopathy

https://doi.org/10.21203/rs.3.rs-7696371/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

Background

Inherited cardiomyopathies frequently arise from rare, highly penetrant coding variants with variable clinical expressivity. Recent biobank-scale genome-wide association studies (GWAS) suggest significant polygenic contributions to cardiovascular diseases, including cardiomyopathy. Most GWAS loci map to noncoding regions, which are poorly conserved across species, requiring a human genome context for experimental validation.

Methods

We created engineered heart tissues (EHTs) from human induced pluripotent stem cell (hiPSC)-derived cardiomyocytes and primary cardiac fibroblasts. We assayed single-cell gene expression and chromatin accessibility to generate comprehensive genome-wide regulatory maps. Open chromatin regions (OCRs) were integrated with chromatin contact information and used to functionally fine-map single nucleotide polymorphisms (SNPs) in cardiomyopathy GWAS. SNPs and their associated regulatory regions were assessed using reporter assays, genome editing, and expression profiling.

Results

Single-cell RNA-seq of EHTs confirmed populations recapitulating major cell types found in hearts, with advanced cardiomyocyte maturation compared to monolayer hiPSC-cardiomyocytes. More than 400,000 OCRs were resolved to cell type and assayed for canonical transcription factor footprints. Functional fine-mapping of GWAS loci prioritized 5817 variants, and reporter assays on select variants validated allele-specific enhancer activity. We identified a locus harboring significant GWAS signals from both dilated cardiomyopathy and left ventricle ejection fraction in an intergenic region at chr3p25.1. Several of these variants lie in OCRs participating in long range chromatin interactions with SLC6A6 and GRIP2. Haplotype-resolved and synthetic reporter assays confirmed enhancer activity and narrowed candidate SNPs. CRISPR-deletion of this region reduced expression of both SLC6A6 and GRIP2, indicating the enhancer regulates the expression of more than one gene.

Conclusions

EHTs derived from hiPSCs are an experimentally tractable platform for testing the function of noncoding variants as modifiers of cardiomyopathy. Variants fine-mapped from cardiomyopathies using EHT regulatory maps have functional consequences and provide a set of prioritized sites to advance the study of polygenic heart failure liability.

Epigenetics & Genomics

Medical Genetics

Cardiac & Cardiovascular Systems

Engineered Heart Tissue

Cardiomyopathy

Functional Genomics

Noncoding Variants

Advances in human biobank genomics have dramatically increased our ability to identify candidate genetic risk variants for both common and rare diseases [1]. These conditions include cardiomyopathies –both dilated [2-6] and hypertrophic [2, 7, 8], as well as arrhythmogenic conditions such as Long QT Syndrome [9] and Brugada Syndrome [10, 11]. For example, two recent large-scale genome-wide association studies (GWAS) of dilated cardiomyopathy (DCM) report 80 and 70 risk loci respectively [5, 6], and one hypertrophic cardiomyopathy study (HCM) highlights at least 70 risk loci [8, 12]. These common variants generally map to regions noncoding regions, often distinct from monogenic genes for cardiomyopathy. The noncoding GWAS signals likely impart phenotypic effects through distal gene-enhancer interactions that regulate gene expression.

The dozens of GWAS loci associated with both human HCM and DCM provide a much-expanded genomic context for variation contributing to these diseases. Nevertheless, it is challenging to identifying the regulatory variants that directly affect gene targets versus those that are simply linked to functional regions. The lack of conservation in noncoding regions across species renders assessment nontrivial in traditional model systems like mice. Human-specific myocardial disease models have largely relied on monoculture induced pluripotent stem cell derived cardiomyocytes (hiPSC-CMs) [13, 14]. These models provide a critical experimental substrate, namely the complete human genome. Nonetheless, hiPSC-CMs are limited in both physiological and transcriptional maturity [15-17]. They also lack the cellular diversity present in developing and adult heart tissue, limiting the study of cell-type interactions and genetic pleiotropy.

The development of three dimensional differentiated organoid cultures [18-20], and engineered tissues [21, 22] has progressively begun to mitigate concerns about maturation level and cell-type diversity to increase model fidelity and increase translational capacity in cardiac research. Here, we generated engineered heart tissues (EHTs) comprised of hiPSC-derived cardiomyocytes and primary human cardiac fibroblasts in a collagen-based scaffold in long term co-culture [23]. EHTs were interrogated using single-cell RNA and ATAC-sequencing, highlighting their enhanced maturity compared to hiPSC-CMs in 2D monolayer cultures. Using comprehensive regulatory maps generated from EHTs, 2D monolayer hiPSC-CMs, and ENCODE adult left ventricle ATAC-Seq [24, 25], we functionally fine-mapped 122 independent genomic loci from 12 GWAS cohorts of DCM and HCM. We identified thousands of prioritized variant candidates positioned to contribute to HCM and DCM, including their potential cell-type specificity. We validated several prioritized regions as enhancers, revealing haplotype-specific differences and links to specific effector genes, including SLC6A6 which encodes a taurine receptor and GRIP2, encoding a protein that interacts with ephrin-B ligands. These data additionally provide a resource of epigenetic and expression data, along with functional fine-mapping results, to identify and test the effects of genetic variation on myocardial function.

Differentiation of human iPSC to Cardiomyocytes

Ventricular hiPSC-CMs were generated by standard methods using small molecule modulation of WNT signaling [26, 27] adapted from previously published protocols. Briefly, when hiPSCs reached ~90% confluency, differentiation was initiated using 6mM CHIR99021(Tocris) for 24 hours in cardiomyocyte differentiation medium 3 (CDM3): RPMI 1640 (Thermo Fisher Scientific) supplemented with 2mM L-glutamine (Gibco), 213 mg/mL L-ascorbic acid 2-phosphate (Wako Chem), and 500mg/mL recombinant human albumin (Sigma). Media containing CHIR99201 was exchanged for fresh CDM3 at 24 hours. On day 3 cells were treated with 2mM Wnt-C59 (Tocris) in CDM3. On day 5, c59-containing media was changed for CDM3 and subsequently exchanged every two days until cells matured into beating monolayers. On day 10 of differentiation, beating cardiomyocytes were dissociated to single cells using 200units/mL collagenase IV (Worthington Biochemical Corporation) in Hank’s Balanced Salt Solution (Thermo) with 10mM HEPES, 2mM Thiazovivin (STEMCELL Technologies) and 30 mM N-Benzyl-p-Toluenesulfonamide (TCI) for 2 hours at 37°C. The single cell suspension of iPSC-CMs was filtered through a 100mm cell strainer and purified via PSC-Derived Cardiomyocyte Isolation Kit, human (Miltenyi Biotec), which routinely yields >90% purity, as assessed by flow cytometry from Cardiac Troponin T (BD Biosciences).

Generation of Engineered Heart Tissues

Engineered Heart Tissues were generated according to our previously published protocol [23], derived from the procedure originally outlined in by Tiburcy et al., [21, 22]. Briefly, we purified eGFP expressing hiPSC-CMs using column a column-based purification and combined them with human cardiac fibroblasts (Promocell) in a 9:1 ratio for a total of 0.55x10⁶ cells per EHT in custom 48-well plates (Myriamed). Cells were suspended in RPMI containing B-27 supplement (Thermo Fisher Scientific) and combined over ice with 44 μL collagen type I (6.5 mg/mL, Millipore Sigma), 44 μL of 2× RPMI (Thermo Fisher Scientific), and 6 μL of 0.1N NaOH for a total of 198 μL per EHT, creating a hydrogel substrate mixture. 180 μL of the mixture was then transferred evenly into each well of the 48-well plate for EHT formation, resulting in a final cell density of 500,000 cells/well. Following the transfer, EHTs were incubated the 48-well plate at 37° C for 1 hour and before supplementing with EHT media containing DMEM low glucose (Millipore Sigma), 10% horse serum (heat inactivated, New Zealand origin, Gibco), 1% Penicillin/Streptomycin (Thermo Fisher Scientific), and 0.1% human insulin (10 mg/mL, Millipore Sigma). For 72 hours following EHT generation, the cells were supplemented with 5 μg/mL recombinant human TGF-β1 (CHO derived, PeproTech). On day 3 after tissue casting, TGF-β1 was removed from culture medium and media was exchanged with 500 μL fresh media per well. EHT media was exchanged every 2 days throughout the course of tissue maturation. EHTs were harvested for downstream applications at day 20 post tissue casting.

Immunofluorescence Imaging of EHTs and 2D hIPSC-CMs

For iPSC-CM monolayer imaging (2D), 3x10⁵ cells were plated in 24‐well plate 12 mm on Matrigel‐coated micro cover glass slips (Electron Microscopy Sciences). Cells were washed with 30mM KCl in PBS and fixed with 10% formalin for 10 minutes, and then permeabilized with 0.2% Triton. They were then blocked for one hour with 10% donkey serum-PBS and then incubated overnight at 4°C with MYBPC3 Antibody (E-7, Santa Cruz) diluted 1:400 in 10% Donkey Serum-PBS, followed by a wash with 0.1% Tween-PBS. Next, they were incubated 1 hour at room temperature in 1:400 Anti-mouse Alexa 594 (ThermoFisher) in 10% donkey serum and then washed again with 0.1% Tween-PBS, then PBS. Nuclei were stained with DAPI (1:1000) in PBS, washed again with PBS, and then mounted onto glass slides with Prolong Gold (Thermo) and imaged on a Zeiss Axio Imager M2.

EHTs were removed from the plates, washed with 30mM KCl in PBS, and fixed overnight in 10% formalin at 4°C. The EHTs were then washed in PBS and permeabilized with 0.5% Triton in PBS, three times for 30 minutes each time. They were then washed in PBS and equilibrated with 15% sucrose in PBS for 5 minutes, followed by 30% sucrose-PBS for 5 minutes. These equilibrated EHTs were then placed into Tissue-Tek® O.C.T. Compound (Sakura) and flash frozen in LiN2. Frozen EHTS were sectioned at 15µm onto glass slides. These sections were then blocked and stained using the same method as described for the 2D monolayers.

Single-Cell RNA-Sequencing of EHTs

To produce inputs for single cell transcriptomic datasets of EHTs, 2 x 48-well plates of EHTs were generated, representing two independent differentiations of input cardiomyocytes. The tissues were then pooled and incubated them at 37° C for 1-2 hours in a dissociation solution containing 1mg/ml Collagenase II (200 units/mg, Worthington),10 mM HEPES pH 7.4 (Gibco), 2 uM Thiazovivin (STEMCELL Technologies), and 30 uM N-Benzyl-p-Toluenesulfonamide, a contractility inhibitor, (TCI) to obtain a single cell suspension. A single cell RNA-sequencing library targeting 10,000 cells was prepared from this suspension using the 10X Genomics Chromium Platform at the NUSeq Core Facility at Northwestern University, Feinberg School of Medicine. We then repeated this procedure for a second time with two additional EHT differentiations to prepare an independent, replicated experiment. Both libraries were sequenced on the Illumina NovaSeq platform in 50bp paired-end format. Reads were aligned to the GRCh38.p13 transcriptome using the STAR [28] aligner as integrated into the 10X Genomics Cell Ranger count pipeline [29]. Cell x gene count matrices were then filtered on mitochondrial read content and total RNA content, as well as predicted doublet status, using Scrublet [30]. The datasets were then clustered, integrated, and analyzed using the Seurat package [31, 32] for the R programming language.

Bulk RNA-Sequencing of 2D hiPSC-Cardiomyocytes

Day 10 differentiated hiPSC cardiomyocytes were released from cell culture plates using trypsin-EDTA (Thermo Fisher Scientific), pelleted and washed in Dulbecco’s phosphate-buffered saline solution (DPBS) without calcium or magnesium (Thermo Fisher Scientific) and RNA was extracted using Qiagen RNeasy mini kit (Qiagen). Libraries were prepared using the NEBNext low input RNA library prep kit for Illumina (NEB) with NEBNext Illumina index primers and adapters. Sequencing of the RNA-Seq libraries was performed at the University of Chicago on an Illumina NovaSeq 6000 platform, and reads were aligned to the GRCh38/hg38 human transcriptome reference with NCBI refseq transcript models. Library strand orientation was confirmed using kallisto [33] and count matrices were generated by HTSeq-count [34]. Differential expression analyses were performed in R using the limma [35], and edgeR [36] packages.

ATAC-Seq in EHTs and 2D hiPSC-CMs

Day 30 EHTs were digested to a single cell suspension as described above and sent for fluorescence-activated cell sorting (FACS), gating on expression of eGFP. eGFP- fibroblasts were separated from the eGFP+ iPSC-CMs, producing CM enriched (CM+) and fibroblast enriched (FB+) cell pools. The sorted cells were washed again with 10mL EHT media and resuspended in 1mL with 2%BSA. The cells were then counted and separated into aliquots of 50,000. Assay for transposase-accessible chromatin sequencing (ATAC-Seq) was performed as described by Grandi, F. C. et al. [37]. Briefly, 50k cells were collected at 500g for 5min at 4°C and gently resuspended in 50µl ice-cold lysis buffer: 10 mM Tris–HCl pH 7.5, 10 mM NaCl, 3 mM MgCl2, 0.1% NP40 (Sigma), 0.1% Tween-20 (Thermo Fisher Scientific), and 0.01% digitonin (Promega). Lysis proceeded for 3min on ice and was neutralized with 1ml of ice-cold lysis buffer excluding detergents digitonin and NP40. Following neutralization, permeabilized nuclei were pelleted at 4°C and resuspended in 50µl transposition master mix: 25µl (2X) Illumina TD buffer, 16.5µl PBS, 5µl H2O, 0.5µl 1% digitonin, 0.5µl 10% Tween-20, and 2.5µl Illumina TDE1 enzyme (Illumina), and incubated for 30min @ 37°C on a thermomixer. Tagmentation was terminated and DNA collected using Zymo Clean and Concentrator-5 (Zymo) following the manufacturers recommendations. Libraries were barcoded with 5 cycles of preamplification, and the PCR reaction was paused for quantification using NEBNext Library Quant Kit for Illumina (New England Biosciences). Diluted final libraries were mixed and sequenced on an Illumina Nova-Seq 6000 using paired end 50bp reads with a target depth of ~300M reads/sample. Detailed descriptions of read processing, peak calling, and downstream analyses of ATAC-Seq libraries can be found in the Detailed Methods section of the supplement.

Effect Size Estimates and Functionally informed Fine-mapping

Summary statistics for fine-mapped traits (Supplemental Table 1) were lifted over to GRCh38.p13/hg38 in R v.4.1.0 using the liftOver package (https://doi.org/10.18129/B9.bioc.liftOver). Effect-size enrichment estimates and SNP-level prior weights were obtained in each dataset using TORUS [38], where loci were split into 1600 distinct LD blocks across the human genome for estimation. The prior weights were then utilized in functionally informed fine-mapping in R with the susieR [39] and mapgen [40] packages to obtain estimates of posterior inclusion probabilities at each locus in both HCM and DCM. The estimates were compared with uniform prior fine-mapping in the same tools, and loci were prioritized for further study based on their apparent localization to distal regulatory elements and the exclusion of high probability coding variants in the credible sets.

Luciferase Assays

Genomic sequences used in luciferase reporter assays for fine-mapping at PRKCA, VTI1A / TCF7L2, and the GATA4 locus were designed as 1000-bp sequences centered on the variant of interest (Supplemental Table 2), with an additional 5’ homology sequence (GGCCTAACTGGCCGgtacctgagctcgctagcctcga) and 3’ homology sequence (atcaagatctggcctcggcggccaaGCTTAGACACTA) for Gibson assembly into the Promega pgL4.23[luc2/minP] backbone. These sequences were synthesized as linear IDT gBlocks. The backbone was propagated in NEB Stable Competent E. coli, linearized by XhoI and EcoRV (NEB), and assembled on a thermocycler using NEB Gibson Master Mix. Constructs were dialyzed in deionized water using Millipore MF membrane filters and transformed into NEB stable E coli for propagation, isolation, and sequencing.

Longer haplotype sequences for luciferase assays of the chr3p25.1 enhancer (Supplemental Table 2) were synthesized and delivered as complete plasmids by VectorBuilder, and were similarly propagated, isolated, and sequenced before transfection into hiPSC-CMs. For transfection and activity readings, hiPSCs were differentiated to cardiomyocytes using methods described above, isolated at day 10 using column purification, then plated at 30,000 cells per well into clear bottomed, opaque-walled 96-well plates designed for use with Promega plate-readers (Corning costar). Cells were maintained in RPMI 1640 supplemented with B27 and L-Glutamine. After 24 hours, media was exchanged for media containing 2 uM CHIR99201 and was replaced every 48 hours for the duration of the culture. When cells reached 90% confluency, they were co-transfected with experimental constructs and Renilla normalization controls (Promega) into cells. 20 hours-post transfection, cells were washed, and media was replaced with CHIR99201 containing media. At 48-hours post transfection, cells were lysed, and luciferase activity was read using Promega Glowmax Multi-detection system and Promega Dual-Luciferase Reporter Assay System kit.

CRISPR Editing of hiPSCs

To obtain hiPSC cell lines with one or more copies of the chr3p25.1 enhancer region removed, we made use of the IDT CRISPR Alt-R genome editing system. Briefly, we obtained four custom crRNAs targeting each of the flanks of the enhancer region (Supplemental Table 2) and complexed them with Alt-R tracrRNAs (IDT) and Alt-R S.p. Cas9 V3 (IDT) to form four distinct editing ribonuclear proteins (RNPs). These RNPs were pooled in equal quantity and nucleofected into approximately 1x10⁶ hiPSCs per manufacturers’ recommendations using the Lonza 4D Nucleofector System and the Amaxa P3 Primary Cell X Kit L (Lonza). After nucleofection, cells were plated in a single well of a 6-well plate with media containing ROCK inhibitor. Nucleofected cells were left to recover for 48 hours then sorted using FACS at one cell/well into Matrigel coated 96-well plates. Cell colonies were then expanded and passaged until enough material for genotyping could be collected. To genotype edited cells, cell scrapes were taken from 24-well plates and DNA was extracted using Lucigen QuickExtract Buffer (Biosearch Technologies). The genotype of the cells was assayed using PCR with custom primers spanning the target deletion breakpoints, as well as internal primers used to distinguish heterozygous cells. Selected cells were further expanded and banked. Their genotypes were confirmed via additional DNA isolations of larger cell pellets using the Promega Wizard Genomic DNA purification kit, PCR, and Sanger sequencing.

Generation of Engineered Heart Tissues as a Model of Human Myocardium

Although monolayer hiPSC-CM cultures have revolutionized cardiac disease research [13, 14], their cellular immaturity limits their potential as an experimental system. Engineered heart tissue or engineered heart myocardium [18-22] systems have been developed by mixing hiPSC-CMs with fibroblasts and other cell types, and then placing cell mixtures onto bio-scaffolds where they mature under tension. There are multiple formats for tissue generation, varying in bio-scaffold content and form, ranging from simple strips to rings or even bio-printed ventricular shapes [41]. Here, we molecularly characterized the transcriptomic and open chromatin landscapes of human EHTs and compared these data to hiPSC-CMs in 2D cultures and to left ventricle (LV) myocardium.

To generate EHTs, eGFP-expressing human iPSCs were differentiated to cardiomyocytes in monolayer cell culture for 10 days using WNT modulation followed by depletion of stem cells and enrichment for cardiomyocytes [21-23, 42]. Enriched cardiomyocytes were combined 9:1 with fibroblasts isolated from human hearts in a three-dimensional collagen-based scaffold, forming approximately 15mm ring-like EHTs around two posts. The tissues were then matured for an additional 20 days, totaling 30 days from hiPSC-CM induction to mature EHTs (Figure 1A). Immunofluorescence imaging with an antibody to cMyBP-C in EHTs revealed cardiomyocyte alignment as well as sarcomere alignment parallel to the axis of contraction (Figure 1B). By comparison, 2D monolayer hiPSC-CMs show sarcomeres radially aligned around central nuclei, with little cell-cell alignment (Figure 1C). These characteristics agree with previous morphology described in EHTs [42-44], and further suggest that this cellular alignment more closely models in vivo cardiomyocytes.

Single-cell gene expression in EHTs

To assess the cell-type composition and gene expression patterns within EHTs, we dissociated pools of day 30 EHTs, collected from 2 independent differentiations, into single cell suspensions, and generated single-cell RNA seq libraries using the 10x Genomics Chromium^TM platform. After sequencing, quality control, and filtering, we obtained 17,081 cells. Unbiased clustering of the transcriptomic data revealed 14 communities (Supplement 1A-1C) comprising six distinct cell-types (Figure 1D). By cell number, cardiomyocytes (42.4 %) and cardiac fibroblasts (37.9 %) represented most of the tissue. This cardiomyocyte percentage is similar to estimates found in human hearts from the Heart Cell Atlas (HCA) project, where cardiomyocytes constitute 49% of ventricle cells [45]. SMC/pericyte-like, mesothelial progenitors, neural-like, and endothelial cells contribute the remaining 19.7% of the cell population. EHT cardiomyocytes expressed genes for canonical sarcomere components and known cardiac transcription factors (Figure 1E, Supplement 1D). Furthermore, average expression of the adult ventricular myosin heavy chain, MYH7, exceeded that of the fetal isoform MYH6 in EHTs. In general, each major cell-type expressed canonical markers such as FN1 in cardiac fibroblasts, RGS5 in pericyte/SMC cells, PECAM1 in endothelial cells, and SOX2 in neural-like cells (Figure 1F). As a validation of these cell type annotations, we compared our labels with predictions from CellTypist [46, 47] models trained on the human Heart Cell Atlas. We found general agreement in the predictions of the major cell types, adding confidence regarding EHT composition (Supplement 1B).

To assess potential heterogeneity within the EHT-CM population, we subclustered EHT-CMs at higher resolution with particular attention to sarcomere and transcription factor gene expression. This analysis revealed populations spanning the maturation spectrum (Figure 2A), aligning well with the second principal component of variation in the expression data (Figure 2B). We noted a continuum of early, still cycling, and late cardiomyocytes, characterized by differential regulation of canonical adult and neonatal sarcomere genes as well as the specific transitions between MYH6 and MYH7 (Figure 2C). We also noted subsets of cardiomyocytes across maturation states expressing extracellular matrix components (noted as ECM+). We identified a cardiac progenitor population defined by high expression of the transcription factors GATA4, TBX20, and MYOCD (Figure 2C). To compare cardiac gene expression between input hiPSC-CMs and EHT-CMs, we performed RNA-seq on 2D hiPSC-CMs. Gene expression was moderately and significantly correlated (r: 0.476334. P-value: 0.0025) between the datasets (Figure 2D). However, several genes characteristic of mature cardiomyocytes were expressed at substantially higher levels in EHTs, including the sarcomere components TPM1, ACTC1, TTN, MYL2, MYL3, TNNC1, and MYH7. Conversely, we observe only one gene relatively upregulated in 2D hiPSC-CMs, MYH6, the MYH isoform that characterizes the developing ventricle, highlighting a difference in maturity between the systems.

Open chromatin profiling of engineered heart tissues

Given the advanced maturation of gene expression patterns in EHT-CMs relative to 2D hiPSC-CMs, we speculated that the enhancer landscape of myocardial genes may also more closely resemble that of the adult heart. To test this hypothesis, we generated deep ATAC-Seq datasets from day 30 EHTs and 2D hiPSC-CMs. To separate cell populations into cardiomyocyte-enriched (CM) and fibroblast enriched (FB) pools from which to generate libraries, we sorted the input cell populations using FACS, gated for eGFP signal (Figure 3A). For an adult myocardial comparison, we also reprocessed sequencing data from eight libraries of four high quality bio samples taken from left ventricle (LV) tissue originally published for the ENCODE project [24, 25] (Supplemental Table 1). After peak calling and filtering, this analysis produced a set of 160,270 fixed-width open chromatin regions (OCRs) in the reference ENCODE LV dataset and 153,189 OCRs in the 2D hiPSC-CM ATAC-Seq (Figure 3B). In the EHT-CM and EHT-FB datasets, we found substantially more OCRs: 221,530 and 185,303 regions respectively, for a combined dataset of 406,833 OCRs. This increased number of OCRs likely reflects the model, which includes developmental and adult gene expression patterns, further enhanced by the uniquely deep sequencing applied to the EHTs.

The pattern of genomic context of the identified OCRs was consistent with what is expected from ATAC sequencing [48], where the plurality of regions mapped to introns, followed by intergenic regions, promoters, exons, and finally transcription termination sites (Figure 3C). Comparing the EHT ATAC-Seq results to ENCODE LV data reprocessed using the same metrics, we detected 206,077 EHT intersecting at least one element in the ENCODE LV set (Figure 3D). Of the EHT-CM OCRs, we also found 130,571 annotations (58.9%) overlapped at least one region in the 2D hiPSC-CM set; demonstrating broad overlap with both the adult and early states. We illustrate one example proximal to the cardiac troponin C1 gene, TNNC1, which is absent in ENCODE LV but found in both EHT and 2D hiPSC-CMs (Figure 3E, arrow). We also observe many sites where the EHT dataset mimics a hybrid state between the early open chromatin structure of 2D hiPSC-CMs and the mature structure of ENCODE LV tissue. This hybrid state is exemplified at the MYH6-MYH7 locus where 2D hiPSC-CMs display strong open chromatin signal at a proximal promoter element of the fetal ventricular heavy chain MYH6 (Figure 3F, arrows). In contrast, the adult LV shows open chromatin approximately 1900 bp upstream, with diminished signal at the proximal promoter.The EHT CM OCR dataset captures both elements with comparable signal strength. These findings suggest EHTs can model genomic perturbations relevant to cardiac development and the more mature myocardium. As an additional validation of the cardiac regulatory elements, we analyzed transcription factor binding sites within EHT CM OCRs. We found canonical cardiac transcription factors motifs among the most enriched in the EHT OCRs (Figure 3G) including those of MEF2C, GATA4, TBX20, NKX2-5, and TBX5.

CM- and FB-enriched ATAC-Seq data also enable deconvolution of the bulk ENCODE LV signal into relative cell type contributions. For instance, at DDR2, a gene encoding a collagen-binding receptor tyrosine kinase involved in fibrosis, we detect several promoter elements as well as distal intronic OCRs in the fibroblasts rather than the cardiomyocytes (Figure 3H, arrows). Motif enrichment within fibroblast footprints suggests an active state, likely reflecting the TGFb supplementation used during the tissue consolidation process. These include motifs corresponding to TCF1, ETS1, FLI1, and NR2F2 (Figure 3I). These fibroblast-specific sites provide an additional opportunity to assess potential fibroblast contributions to cardiac disease.

Functional Fine-mapping of Cardiomyopathies from EHTs

The depth of OCR annotations in EHTs enabled fine-mapping of cardiomyopathy GWAS loci [2, 5-8]. Functionally informed fine-mapping is a two-step, empirical Bayes approach to variant prioritization that uses genomic annotations to estimate how strongly GWAS effect sizes are enriched in specific functional regions. These enrichment estimates are then used as prior probabilities and combined with linkage disequilibrium data to pinpoint variants likely to contribute to trait biology at the locus (Figure 4A). We generated effect-size enrichment estimates using TORUS [38] for each OCR dataset, and applied this analysis to EHT-CM, EHT-FB, iPSC-CM, and ENCODE LV data. We used summary statistics from a multi-trait GWAS (MTAG) conducted on cohorts with hypertrophic cardiomyopathy [8] and dilated cardiomyopathy [6], left ventricle ejection fraction (LVEF) and left ventricle end-diastolic volume (LVEDVi) [49]. We also selected two traits unrelated to myocardial tissue: body mass index (BMI) [50] and schizophrenia (SCH) [51] to serve as trait controls.

Our estimates reveal strong GWAS effect size enrichment in all cardiac OCRs across each of the cardiac traits (Figure 4B). Both EHT-CM and ENCODE LV OCRs were enriched for variants associated with contractile traits like LVEF and LVEDVi, as well as DCM and HCM, reflecting the contributions of cardiomyocytes and fibroblasts underlying those phenotypes. By comparison, unrelated traits of body mass index and schizophrenia showed no enrichment (Figure 4B). The strong enrichment in cardiomyopathy associated GWAS effects may reflect increased maturation of gene regulatory circuits in EHTs across entire pathways, relative to 2D hiPSC-CMs. Assaying pathway-level gene expression in EHT-CMs, we note higher expression in EHT-CMs in the sarcomere, as well as in cardiomyopathy and several other disease-associated gene sets, and sets related to cardiac development and contractile function (Figure 4C). Taken together, these estimates demonstrate that the open chromatin regions in EHTs are globally enriched for SNPs identified in GWAS of cardiac traits at levels comparable to adult left ventricle.

We next used the enrichment estimates to derive SNP-level prior probabilities for statistical fine-mapping of each GWAS and OCR dataset with SuSieR [39, 52]. SuSieR uses summary statistics data from GWAS studies and linkage disequilibrium data (LD) to perform fine-mapping. At each locus, we obtained a group of SNPs that together have a 90% chance of containing a functional variant, termed a “credible set.” Each SNP was then assigned a posterior inclusion probability (PIP), reflecting the probability that the variant is functional. In total, 5817 SNPs were mapped to at least one credible SNP set in HCM (2545 SNPs across 83 loci) and/or DCM (3870 SNPs across 87 loci) (Supplemental Tables 3-5).

We found that average credible set sizes were reduced using this prioritized fine-mapping compared to uniform fine-mapping of the same loci. EHT-CM fine-mapping especially outperforms the other analyses in reducing large credible sets of 30 or more SNPs (Supplement 2A). Most fine-mapped SNPs were selected into more than one credible set across each OCR input, indicating a measure of robustness in the strategy (Supplement 2B). These SNPs included in multiple credible sets tended to show higher PIPs when using OCR prior probabilities (max functional PIP) when compared to fine-mapping with uniform priors (uniform PIP) (Figure 4D). This finding suggests that SNPs within open chromatin regions are more likely to be prioritized as potentially functional, as opposed to simply being in LD. A substantial proportion of the entire set of 5817 SNPs were found to intersect with at least one OCR dataset (17.3%), with DCM SNPs in OCRs at a slightly higher rate (20.5%) than HCM SNPs (15.7%) (Supplement 2C). Most of this overlap could be reproduced in EHT datasets alone (Combined: 13.6%, HCM: 12.4%, DCM 16.2%). These fine-mapping results allowed us to identify individual variants of interest which are likely to be functionally relevant for cardiomyopathy expression, improving the efficiency of downstream variant-to-function analysis.

Interrogation of Noncoding Variants in HCM and DCM

When fine mapping the HCM and DCM GWAS data, we found that most credible set SNPs mapped within noncoding regions of the genome, including those SNPs identified as having the highest probability of being functional. The annotation indicates that these SNPs are likely to modify gene expression via cis-regulatory mechanisms, as has been suggested for HCM and DCM associated GWAS signals [53]. To further prioritize loci potentially acting via cis-regulatory mechanisms, we intersected many these credible sets with chromatin conformation capture data which we previously generated in hiPSC-CMs [27].

We identified four HCM GWAS variants at chr17q24.2 that were strongly prioritized by fine-mapping. These SNPs, rs17633401, rs67700546, rs12601850, and rs9892651, were in a locus spanning the PRKCA gene, which encodes protein kinase C alpha (Figure 5A & Supplement 3A). Both rs67700546 and rs17633401 were in OCRs shared by EHT-CMs and EHT-FBs, while rs12601850 intersects a fibroblast specific element, and rs9892651 did not intersect an OCR in any dataset (Figure 5A). In enhancer assays in hiPSC-CM, the regions displayed activity, with rs12601850 and rs9892651 conferring allele-specific expression effects (Figure 5B). These variants were also identified in GTEx [54] as eQTLs for expression of PRKCA in left ventricle in the expected direction, adding external support to the presence of one or more functional variants at the locus(Figure 5C). These results illustrate that the genetic architecture of a given GWAS-associated locus may involve more than one functionally relevant variant affecting different enhancers [55, 56], potentially in different cell types.

We fine-mapped an additional HCM signal within an intron of VTI1A, a gene implicated in vesicle transport on chr10 at 10q25.2 (Supplement 4A). The credible set for this signal spans approximately 38 kilobases (kb), with both ends forming distal chromatin interactions with theTCF7L2 promoter approximately 245 kb away. Interestingly, the EHT-CM and ENCODE LV fine-mapping prioritized two different variants; rs7096151 (EHT CM PIP = 0.878) and rs10885378 (ENCODE LV PIP = 0.542), which may be due to the structure of the underlying open chromatin data. rs7096151 lies in an OCR found in EHT-CM and 2D hiPSC-CMs. Conversely, rs10885378 is at the edge of an ENCODE LV OCR (Supplement 4A). To test whether either variant was located in cardiac enhancers, we expressed 1000-bp sequences centered on each SNP in hiPSC-CMs to assay luciferase reporter activity. rs10885378-T showed enhancer activity, while rs10885378-C was not significantly different from gene desert or rs10885378-T, indicating modest genotype-specificity (Supplement 4B). Similarly, rs70961651-A trended towards higher enhancer activity than rs70961651-G, but was not significantly different (P=0.09) (Supplement 4B). The transcriptional regulator TCF7L2 is differentially active in failed hearts [57], thus, it is possible that this SNP might demonstrate stronger effects under conditions of heart failure, or that both regions may contribute and act synergistically.

We also dissected a credible set associated with DCM locus at chr8p23.1. The credible set at this locus spans 30 kb and forms long-range chromatin interactions over 280 kb in hiPSC-CM with the developmental cardiac transcription factor gene, GATA4 (Supplement 5A). A single variant, rs13260325, strongly prioritized in EHT-CM fine-mapping (PIP = 0.46), ENCODE LV (PIP = 0.577) and 2D-hiPSC-CM (PIP = 0.436), overlapped an OCR shared in all datasets (Supplement 5A). When placed into a luciferase reporter and transfected into hiPSC-CMs, rs13260325 acts as an enhancer of luciferase gene expression (Supplement 5B). These data suggest that noncoding variants, albeit at a distance from GATA4, may impart their effects on through cis-regulation of GATA4 expression in addition to the established roles of rare coding variants in GATA4 in DCM [58].

Characterization of the 3p25.1 HCM Locus

We identified a DCM-associated region at chr3p25.1 harboring a GWAS signal concentrated in an intergenic region near the LSM3 gene previously highlighted in Garnier et al, 2021 [3]. Fine-mapping of the locus prioritized a credible set spanning 6 kb overlapping a dense set of OCRs shared in multiple cardiac data sets (Figure 6A). This cluster of OCRs forms long-range chromatin interactions with SLC6A6, and GRIP2. Each fine-mapping approach strongly prioritized SNP rs6807275 (EHT-CM PIP = 0.886). Variants in this DCM-chr3p25.1 credible set are arranged into four predominant haplotypes in European-associated ancestry populations; designated haplotypes ‘A-D’. Haplotypes ‘A’ and ‘B’ occur most frequently (Allele frequencies: A = 0.455; B = 0.240) (Figure 6B). Both haplotypes carry the prioritized SNP rs6807275, but differ in three other linked variants (rs6786141, rs7650762, and rs12634565) common to the ‘C-D’ group sequences. Luciferase reporter assays confirmed enhancer activity of each of the four haplotypes expression with levels significantly higher than our negative control non-enhancer sequence (A; P = 9.19x10^-8, B; P = 6.80x10^-10, C; P = 1.38x10^-6, D; P = 6.96x10^-6) (Figure 6C). Furthermore, we observed that the ‘A’ haplotype – which is associated increased DCM risk, showed the lowest average enhancer activity, and differed significantly from the ‘B’ haplotype (P=6.72x10^-3). We hypothesized that one or more of the SNPs differing between ‘A’ and ‘B’ conferred cis-compensatory, stabilizing enhancer function to mitigate the effect of rs6807275. To test this hypothesis, we created synthetic haplotypes on the ‘A’ background, introducing the single substitutions of each ‘B’ specific variant and similarly tested these sequences in hiPSC-CMs (Figure 6D). Two of the tested B-Haplotype SNPs, rs6786141 (P =2.29x10^-2) and rs7650762 (P =2.25x10^-2) restored enhancer activity, consistent with a compensatory effect (Figure 6E). Notably, both SNPs are also shared on the two minor ‘C-D’ group haplotypes. Reporter constructs for these haplotypes exhibited intermediate activity relative to A and B, suggesting that regulatory output at this locus reflects combinatorial interactions between at the three variants, highlighting a complex cis-regulatory architecture centered on rs6807275.

Finally, we deleted a 6 kb region containing the enhancer haplotypes in hiPSCs using CRISPR-Cas9 genome editing to test its regulatory influence on gene expression, generating heterozygous and homozygous knockout lines. We then differentiated these hiPSCs to CMs and assessed the transcriptome by RNA-seq (Figure 6F). In enhancer deleted hiPSC-CMs, the expression of both SLC6A6 (P=9.27x10^-4)and GRIP2 (P=1.77x10^-2) was significantly reduced (Figure 6G-H). Despite being highly expressed in the heart, GRIP2 has not been well studied for its cardiac role. There is evidence for role of the taurine transporter SLC6A6 in cardiomyopathy [2, 3, 59], and our data indicates this genomic region regulates both genes and the possibility that both genes contribute to trait liability at the locus. Together, these results identify the GWAS signal at chr3p25.1 as localizing to a distal enhancer, positioned as a modulator of SLC6A6 and GRIP2 expression. These data indicate multiple variants within a single region can induce changes in expression of more than one target gene and may have broader impacts outside of individual-trait biology.

The genetic understanding of both hypertrophic and dilated cardiomyopathy is rapidly expanding to include significant contributions of common, noncoding variation in the genome. In the last five years, the number of common-variant associated loci have quadrupled from approximately 30 to more than 120 across the two conditions. Consequently, the development of comprehensive functional annotations in cardiac cells coupled with functional genomic analysis carried out in human models of myocardium are increasingly important to bridge the gap between genetic inference and cell biological mechanisms.

In this work, we used human EHTs to provide a model of human myocardium that is cellularly diverse, and more morphologically and transcriptionally mature than 2D hiPSC-CMs. Because EHTs are derived from human IPSCs, they carry the human genome and are well suited to examine noncoding variation that may not be well conserved across species, yet relevant to the expression of human cardiac traits. Although not fully mature, the intermediate maturation state of EHTs yields an experimental system with broad flexibility as a platform for human cardiac genetics since it can be used to study both developmental and adult-onset conditions. This is illustrated in the functional fine-mapping of GWAS traits, where those traits can have early and late effects mediated by developmentally important genes, genes responsible for long-term maintenance of function, or environmentally responsive genes. We showed that the open chromatin regions from EHT datasets could be integrated into statistical fine-mapping procedures to help concentrate evidence on smaller credible set of genomic variants and narrow downstream priorities for further study. At the same time, EHTs can provide insight into which cell-type components may be more likely to contribute effects on a single-variant basis.

Through this work, we generated unique datasets that add to existing data derived from human hearts. Combined with statistical fine-mapping tools, this integrated information provides a roadmap for examining genetic variation in noncoding regions that may not have comparable regions in model systems like mice. We found that many GWAS variants located to active cardiac enhancers and that those variants can have allele-specific effects. This data provides experimental evidence for the role of noncoding cis-regulatory contributions to genetic cardiomyopathy liability. Moreover, this risk is not restricted to genetically mediated cardiomyopathies as these nocoding regions may influence cardiac function in a variety of heart failure etiologies.

Testing a subset of variants identified using this approach, we demonstrate the functional consequence of specific regulatory elements, supporting their contribution to modifying gene expression. Illustrating the power of a combination of refined genomic annotations in EHTs with statistical methods for fine-mapping loci associated with cardiomyopathies, we demonstrate a specific distal enhancer-gene interaction at chr3p25.1 between an upstream regulatory element localized to a DCM GWAS signal. Although this region falls closest to LSM3, the 3D chromatin mapping data link this region to a region containing the SLC6A6 and GRIP2 genes, both of which are expressed in human myocardium. SLC6A6 encodes the taurine receptor, and loss of function variants in SLC6A6 cause a potentially treatable form of cardiomyopathy associated with low taurine content [59]. Taurine supplementation has been studied as a in people with heart failure and cardiomyopathy [60], however, these studies were not conducted with consideration of underlying genotypes that influence expression of its receptor. Notably, this region was previously identified in GWAS for DCM [3], and it was suggested this region likely regulated SLC6AC. However, we demonstrated that deleting this region from chr3p25.1 markedly reduced both SLC6AC and GRIP2, indicating this region regulates more than one gene. GRIP2 is enriched in cardiac expression and encodes an ephrin B1 interacting protein [61], and ephrin B1 is implicated in heart development and function [62]. This example demonstrates how the fine-mapping and refined analysis, combined with functional genomic annotations in EHTs, can serve as a critical ancillary tool to link genetic variation associated with cardiomyopathies with genes and their function.

Additionally, the EHT OCR dataset generates a major addition to previously available data. Using these datasets, we can detect previously unannotated open chromatin signals, capture temporal or transient regulatory dynamics in cardiac biology, open avenues for study in fibroblast specific regulation of cardiac traits, and leverage powerful functional fine-mapping paradigms to increase the efficiency of discover from genome-wide association studies. Finally, this work provides a wealth of new, publicly available resources broadly applicable to cardiac disease genetics. Our comprehensive OCR datasets add substantially to known open chromatin annotations of hiPSC-differentiated cardiomyocytes and primary cardiac fibroblasts. Additionally, our fine-mapping results serve as an anchor point for a wide variety of downstream assays and avenues of future study.

ATAC-seq Assay for Transposase-Accessible Chromatin Sequencing

CM Cardiomyocyte

DCM Dilated Cardiomyopathy

ECM Extracellular Matrix

EHT Engineered Heart Tissue

FACS Fluorescence-Activated Cell Sorting

FB Fibroblast

HCM Hypertrophic Cardiomyopathy

hiPSC Human Induced Pluripotent Stem Cells

hiPSC-CM Human Induced Pluripotent Stem Cell Derived Cardiomyocyte

GWAS Genome Wide Association Study

OCR Open Chromatin Region

SNP Single Nucleotide Polymorphism

Acknowledgments:

This research was supported in part through contributions provided by the Genomics Compute Cluster which is jointly supported by the Feinberg School of Medicine, the Center for Genetic Medicine, and Feinberg's Department of Biochemistry and Molecular Genetics, the Office of the Provost, the Office for Research, and Northwestern Information Technology. The Genomics Compute Cluster is part of Quest, Northwestern University's high performance computing facility, with the purpose to advance research in genomics. We thank the Robert H. Lurie Comprehensive Cancer Center of Northwestern University in Chicago, IL, for the use of the Flow Cytometry Core Facility, which provided cell sorting services.

Sources of Funding:

This work is supported by NIH HL128075 (EMM), NIH HL168239 (TOM), HL131914 (EMM), CA060553 (Flow cytometry core); American Heart Association Strategically Focused Research Network on Arrhythmia and Sudden Cardiac Death; Leducq TransAtlantic Foundation.

Disclosures:

EMM consults for PepGen, Tenaya Therapeutics, Novartis, and is a founder of Ikaika Therapeutics and Kardigan. These activities are unrelated to the content of this manuscript.

Wang, Q., et al., Rare variant contribution to human disease in 281,104 UK Biobank exomes. Nature, 2021. 597(7877): p. 527-532.
Tadros, R., et al., Shared genetic pathways contribute to risk of hypertrophic and dilated cardiomyopathies with opposite directions of effect. Nat Genet, 2021. 53(2): p. 128-134.
Garnier, S., et al., Genome-wide association analysis in dilated cardiomyopathy reveals two new players in systolic heart failure on chromosomes 3p25.1 and 22q11.23. Eur Heart J, 2021. 42(20): p. 2000-2011.
Villard, E., et al., A genome-wide association study identifies two loci associated with heart failure due to dilated cardiomyopathy. Eur Heart J, 2011. 32(9): p. 1065-76.
Zheng, S.L., et al., Genome-wide association analysis provides insights into the molecular etiology of dilated cardiomyopathy. Nat Genet, 2024. 56(12): p. 2646-2658.
Jurgens, S.J., et al., Genome-wide association study reveals mechanisms underlying dilated cardiomyopathy and myocardial resilience. Nat Genet, 2024. 56(12): p. 2636-2645.
Harper, A.R., et al., Common genetic variants and modifiable risk factors underpin hypertrophic cardiomyopathy susceptibility and expressivity. Nat Genet, 2021. 53(2): p. 135-142.
Tadros, R., et al., Large-scale genome-wide association analyses identify novel genetic loci and mechanisms in hypertrophic cardiomyopathy. Nat Genet, 2025.
Lahrouchi, N., et al., Transethnic Genome-Wide Association Study Provides Insights in the Genetic Architecture and Heritability of Long QT Syndrome. Circulation, 2020. 142(4): p. 324-338.
Barc, J., et al., Genome-wide association analyses identify new Brugada syndrome risk loci and highlight a new mechanism of sodium channel regulation in disease susceptibility. Nat Genet, 2022. 54(3): p. 232-239.
Ishikawa, T., et al., Brugada syndrome in Japan and Europe: a genome-wide association study reveals shared genetic architecture and new risk loci. Eur Heart J, 2024. 45(26): p. 2320-2332.
Miyazawa, K., et al., Cross-ancestry genome-wide analysis of atrial fibrillation unveils disease biology and enables cardioembolic risk prediction. Nat Genet, 2023. 55(2): p. 187-197.
Gai, H., et al., Generation and characterization of functional cardiomyocytes using induced pluripotent stem cells derived from human fibroblasts. Cell Biol Int, 2009. 33(11): p. 1184-93.
Zhang, J., et al., Functional cardiomyocytes derived from human induced pluripotent stem cells. Circ Res, 2009. 104(4): p. e30-41.
Tu, C., B.S. Chao, and J.C. Wu, Strategies for Improving the Maturity of Human Induced Pluripotent Stem Cell-Derived Cardiomyocytes. Circ Res, 2018. 123(5): p. 512-514.
da Rocha, A.M., et al., hiPSC-CM Monolayer Maturation State Determines Drug Responsiveness in High Throughput Pro-Arrhythmia Screen. Sci Rep, 2017. 7(1): p. 13834.
Wu, P., et al., Maturation strategies and limitations of induced pluripotent stem cell-derived cardiomyocytes. Biosci Rep, 2021. 41(6).
Kim, H., et al., Progress in multicellular human cardiac organoids for clinical applications. Cell Stem Cell, 2022. 29(4): p. 503-514.
Li, R.A., et al., Bioengineering an electro-mechanically functional miniature ventricular heart chamber from human pluripotent stem cells. Biomaterials, 2018. 163: p. 116-127.
Mills, R.J., et al., Functional screening in human cardiac organoids reveals a metabolic mechanism for cardiomyocyte cell cycle arrest. Proc Natl Acad Sci U S A, 2017. 114(40): p. E8372-E8381.
Tiburcy, M., et al., Generation of Engineered Human Myocardium in a Multi-well Format. STAR Protoc, 2020. 1(1): p. 100032.
Tiburcy, M., et al., Defined Engineered Human Myocardium With Advanced Maturation for Applications in Heart Failure Modeling and Repair. Circulation, 2017. 135(19): p. 1832-1847.
Selgrade, D.F., et al., Susceptibility to innate immune activation in genetically mediated myocarditis. J Clin Invest, 2024. 134(13).
Consortium, E.P., et al., Identification and analysis of functional elements in 1% of the human genome by the ENCODE pilot project. Nature, 2007. 447(7146): p. 799-816.
de Souza, N., The ENCODE project. Nat Methods, 2012. 9(11): p. 1046.
Burridge, P.W., et al., Chemically defined generation of human cardiomyocytes. Nat Methods, 2014. 11(8): p. 855-60.
Montefiori, L.E., et al., A promoter interaction map for cardiovascular disease genetics. Elife, 2018. 7.
Dobin, A., et al., STAR: ultrafast universal RNA-seq aligner. Bioinformatics, 2013. 29(1): p. 15-21.
Zheng, G.X., et al., Massively parallel digital transcriptional profiling of single cells. Nat Commun, 2017. 8: p. 14049.
Wolock, S.L., R. Lopez, and A.M. Klein, Scrublet: Computational Identification of Cell Doublets in Single-Cell Transcriptomic Data. Cell Syst, 2019. 8(4): p. 281-291 e9.
Hao, Y., et al., Integrated analysis of multimodal single-cell data. Cell, 2021. 184(13): p. 3573-3587 e29.
Stuart, T., et al., Comprehensive Integration of Single-Cell Data. Cell, 2019. 177(7): p. 1888-1902 e21.
Bray, N.L., et al., Near-optimal probabilistic RNA-seq quantification. Nat Biotechnol, 2016. 34(5): p. 525-7.
Anders, S., P.T. Pyl, and W. Huber, HTSeq--a Python framework to work with high-throughput sequencing data. Bioinformatics, 2015. 31(2): p. 166-9.
Ritchie, M.E., et al., limma powers differential expression analyses for RNA-sequencing and microarray studies. Nucleic Acids Res, 2015. 43(7): p. e47.
Chen, Y., et al., edgeR v4: powerful differential analysis of sequencing data with expanded functionality and improved support for small counts and larger datasets. Nucleic Acids Res, 2025. 53(2).
Grandi, F.C., et al., Chromatin accessibility profiling by ATAC-seq. Nat Protoc, 2022. 17(6): p. 1518-1552.
Wen, X., Molecular QTL discovery incorporating genomic annotations using Bayesian false discovery rate control. The Annals of Applied Statistics, 2016. 10(3).
Zou, Y., et al., Fine-mapping from summary data with the "Sum of Single Effects" model. PLoS Genet, 2022. 18(7): p. e1010299.
Selewa, A., et al., Single-cell genomics improves the discovery of risk variants and genes of atrial fibrillation. Nat Commun, 2023. 14(1): p. 4999.
Gokhan, I., T.S. Blum, and S.G. Campbell, Engineered heart tissue: Design considerations and the state of the art. Biophys Rev (Melville), 2024. 5(2): p. 021308.
Fullenkamp, D.E., et al., Simultaneous electromechanical monitoring in engineered heart tissues using a mesoscale framework. Sci Adv, 2024. 10(37): p. eado7089.
Wang, B.Z., et al., Cardiac fibroblast BAG3 regulates TGFBR2 signaling and fibrosis in dilated cardiomyopathy. J Clin Invest, 2025. 135(1).
Pietsch, N., et al., Chronic Activation of Tubulin Tyrosination Improves Heart Function. Circ Res, 2024. 135(9): p. 910-932.
Litvinukova, M., et al., Cells of the adult human heart. Nature, 2020. 588(7838): p. 466-472.
Xu, C., et al., Automatic cell-type harmonization and integration across Human Cell Atlas datasets. Cell, 2023. 186(26): p. 5876-5891 e20.
Dominguez Conde, C., et al., Cross-tissue immune cell analysis reveals tissue-specific features in humans. Science, 2022. 376(6594): p. eabl5197.
Yan, F., et al., From reads to insight: a hitchhiker's guide to ATAC-seq data analysis. Genome Biol, 2020. 21(1): p. 22.
Pirruccello, J.P., et al., Analysis of cardiac magnetic resonance imaging in 36,000 individuals yields genetic insights into dilated cardiomyopathy. Nat Commun, 2020. 11(1): p. 2254.
Locke, A.E., et al., Genetic studies of body mass index yield new insights for obesity biology. Nature, 2015. 518(7538): p. 197-206.
Trubetskoy, V., et al., Mapping genomic loci implicates genes and synaptic biology in schizophrenia. Nature, 2022. 604(7906): p. 502-508.
Wang, G., et al., A simple new approach to variable selection in regression, with application to genetic fine mapping. J R Stat Soc Series B Stat Methodol, 2020. 82(5): p. 1273-1300.
Watkins, H., Time to Think Differently About Sarcomere-Negative Hypertrophic Cardiomyopathy. Circulation, 2021. 143(25): p. 2415-2417.
Consortium, G.T., The Genotype-Tissue Expression (GTEx) project. Nat Genet, 2013. 45(6): p. 580-5.
Joslin, A.C., et al., A functional genomics pipeline identifies pleiotropy and cross-tissue effects within obesity-associated GWAS loci. Nat Commun, 2021. 12(1): p. 5253.
Sobreira, D.R., et al., Extensive pleiotropism and allelic heterogeneity mediate metabolic effects of IRX3 and IRX5. Science, 2021. 372(6546): p. 1085-1091.
Hou, N., et al., Transcription Factor 7-like 2 Mediates Canonical Wnt/beta-Catenin Signaling and c-Myc Upregulation in Heart Failure. Circ Heart Fail, 2016. 9(6).
Ang, Y.S., et al., Disease Model of GATA4 Mutation Reveals Transcription Factor Cooperativity in Human Cardiogenesis. Cell, 2016. 167(7): p. 1734-1749 e22.
Ansar, M., et al., Taurine treatment of retinal degeneration and cardiomyopathy in a consanguineous family with SLC6A6 taurine transporter deficiency. Hum Mol Genet, 2020. 29(4): p. 618-623.
McGurk, K.A., M. Kasapi, and J.S. Ware, Effect of taurine administration on symptoms, severity, or clinical outcome of dilated cardiomyopathy and heart failure in humans: a systematic review. Wellcome Open Res, 2022. 7: p. 9.
Bruckner, K., et al., EphrinB ligands recruit GRIP family PDZ adaptor proteins into raft membrane microdomains. Neuron, 1999. 22(3): p. 511-24.
Genet, G., et al., Ephrin-B1 is a novel specific component of the lateral membrane of the cardiomyocyte and is essential for the stability of cardiac tissue architecture cohesion. Circ Res, 2012. 110(5): p. 688-700.

The authors declare no competing interests.

SupplementalTables.xlsx
Supplemental Tables
SupplementalFigures.pdf
Supplemental Figures
DetailedMethods.pdf
Detailed Methods

Download PDF

Version 1

posted

You are reading this latest preprint version

Engineered heart tissues facilitate functional characterization of noncoding variants implicated in Hypertrophic and Dilated Cardiomyopathy

Status:

Version 1

Abstract

Background

Methods

Results

Conclusions

Figures

Introduction

Methods

Results

Discussion

Abbreviations

Declarations

References

Additional Declarations

Supplementary Files

Status:

Version 1