3D Biomechanics and Epigenomics Reveal Atypical Fibroblast Responses in Cardiometabolic Disease

Abstract

Cell behavior is influenced by substrate stiffness and cell-cell and cell-environment interactions. The limitations of two-dimensional (2D) culture, such as its inability to fully capture the complexity of cell interactions and tissue structure, highlights the necessity of three-dimensional (3D) cell culture. This explicitly applies to ‘disease modeling in a dish’ platforms for translational studies. 3D bioprinting demonstrates significant potential in recapitulating the intricate physiological environments of human tissues in both healthy and pathological states. With the alarming rise in obesity, addressing systemic pathophysiological dysfunction beyond adipose tissue itself, such as the heart, is inevitable. To capture cellular and tissue-level responses to overnutrition, we employed state-of-the-art 3D bioprinting technology to understand the acute response of 3D matrix-embedded human cardiac fibroblasts (HCFs) to a ‘high-fat diet’ mimic. Chromatin accessibility profiling revealed that excess fatty acid (FA) exposure in 2D induces a non-canonical extracellular matrix (ECM) gene program that is minimally expressed in healthy adult myocardium. In contrast, 3D cultures exhibited reduced fibroblast proliferation and blunted transcriptional responses to the impact of biomechanical cues under metabolic stress, reflecting a more quiescent and physiologically relevant phenotype. Furthermore, we incorporated human induced pluripotent stem cell-derived cardiac fibroblasts (iPSC-CFs), which mirrored key transcriptional changes, including sex-dependent gene regulation. Notably, male iPSC-CFs showed stronger fibrotic gene induction than females, reinforcing the need to account for biological sex in disease modeling. Together, our results highlight the limitations of 2D systems and demonstrate that 3D-bioprinted platforms provide a scalable, physiologically relevant tool for investigating cardiometabolic diseases and therapeutic targets.

New and Noteworthy

Biomechanical cues in conventional 2D systems can artificially prime fibroblasts toward activation, while 3D bioprinted hydrogels better preserve physiological phenotypes. The use of ATAC-seq to profile chromatin accessibility uncovered non-canonical epigenomic remodeling in response to FA overload, highlighting novel fibrotic gene programs not captured by traditional assays. The identification of sex-specific transcriptional responses in iPSC-derived fibroblasts highlights the importance of incorporating biological variables for personalized cardiometabolic disease modeling and accelerating therapeutic development.

Graphical Abstract

Introduction

Cardiometabolic diseases—including coronary heart disease, obesity, and diabetic cardiomyopathy—are among the leading global causes of morbidity and mortality, accounting for over 19 million deaths in just 2022 (1). These diseases are often exacerbated by obesity and poor dietary habits, which promote systemic inflammation, oxidative stress, and metabolic dysregulation that extend beyond adipose tissue to affect cardiac function and structure (2, 3). Among the cellular components of the heart, cardiac fibroblasts (CFs) play a pivotal role in maintaining myocardial structure and function. They are the most abundant non-myocyte cell type in the heart and are responsible for synthesizing and remodeling the extracellular matrix (ECM), providing both mechanical support and biochemical signaling cues to other cardiac cells (4, 5). Under homeostatic conditions, CFs remain quiescent; however, in response to stress signals such as those from hypertension or nutrient excess- they undergo phenotypic switching to activated myofibroblasts. This activation contributes to cardiac fibrosis, increased myocardial stiffness, and ultimately heart failure (6). Despite the critical role of fibroblasts in cardiometabolic pathophysiology, most in vitro models continue to rely on 2D culture systems that fail to capture the mechanical, spatial, and biochemical complexity of the cardiac microenvironment (7, 8). The traditional 2D models subject cells to supraphysiological stiffness and artificial polarization, often inducing stress-responsive gene expression even in the absence of exogenous stimuli (9, 10). Moreover, 2D culture lacks the capacity to mimic cell-ECM and cell-cell interactions that modulate fibroblast behavior in vivo (11).

3D culture platforms have emerged as promising tools to address these limitations, offering more physiologically relevant contexts for studying tissue-specific cell behavior, mechanotransduction, and drug responses (12, 13). Among these approaches, extrusion-based 3D bioprinting enables spatial control of cell placement within ECM-like hydrogels, providing a robust platform to model tissue architecture and mechanical cues (14, 15). In cardiac research, 3D systems more accurately recapitulate myocardial stiffness, topography, and signaling networks compared to 2D monolayers, making them valuable for disease modeling and therapeutic screening (16). Recent advances also point to the interplay between the physical microenvironment and the epigenome, a concept referred to as “mechanical epigenetics” This refers to the ability of ECM stiffness and geometry to influence chromatin organization and transcription factor accessibility, thereby modulating gene expression (17). Assays such as ATAC-seq have allowed unprecedented insight into how metabolic or mechanical cues reshape chromatin accessibility in stromal cells, revealing early regulatory changes that precede overt fibrotic remodeling (18). However, most of these studies remain confined to 2D systems, leaving an unmet need to evaluate epigenomic responses in 3D cardiac contexts.

In this study, we use state-of-the-art 3D bioprinting to encapsulate HCFs in collagen I matrices and expose them to fatty acid-enriched medium as a “high-fat diet” mimic. We compare these 3D cultures to traditional 2D monolayers to assess how dimensionality and matrix compliance affect cell proliferation and chromatin accessibility. To our knowledge, this is the first study to utilize chromatin accessibility profiling to assess human cardiac fibroblast behavior in a 2D platform, with follow up studies within a 3D-hydrogel platform, while also examining sex-based differences that may help explain variations in clinical susceptibility to cardiometabolic diseases.

Methods

Primary human cardiac fibroblasts culture and treatments

Primary male HCFs obtained commercially (30yo, Caucasian; Lonza Lot#21TL347547) were grown on 2D TCP or embedded in 3D hydrogels using bioprinting (Figure 1A). Cells were allowed to adhere and/or proliferate in fibroblast growth medium (FGM3; Lonza) supplemented with 10% FBS, insulin (5μg/mL), hFGF-B (1ng/mL), and Gentamicin-Amphotericin (GA-1000; G=30μg/mL, A=15ng/mL), in a humidified incubator at 37°C with 5% CO₂. HCFs were equilibrated in FGM3 medium (Lonza) supplemented with 0.1% FBS, for 24h prior to any treatments. Cells were then exposed to excess fatty acids in FGM3 (200 μM oleate:palmitate mix [2:1 ratio], bound to BSA at a molar ratio of 2.5:1), following standard protocols (19–21) for 48h. BSA was used as a vehicle for all experiments. HCFs were then fixed or flash-frozen for downstream assays as required. HCFs were used at passages 3–5 for experiments. Data for this study was generated using at least three different passage numbers of HCFs.

Figure 1: Human cardiac fibroblasts (HCFs) grown in 3D collagen hydrogels exhibit slow proliferation in response fatty acid overload.

(A) Schematic representation of the 2D and 3D HCF experimental culture systems are shown. HCFs from the same healthy male subject were used for both culture systems. (B) Schematic representation of the 3D bioprinting workflow using SolidWorks and Ultimaker Cura, a slicing software. (C) Brightfield images of HCFs cultured in 2D (TCP; tissue culture plastic) and 3D matrix; scale bar=100μm. Quantification of cell proliferation using EdU incorporation assay (D) and representative fluorescence microscopy images of EdU stained HCFs (E) in response to physiological ‘high-fat diet mimic’ medium (FA) are shown for each treatment group. Blue indicates nuclei (Hoechst 33342), red indicates EdU⁺ cells, pink indicates overlay of EdU⁺ cells over nuclei; scale bar=100μm. n=10–19; **p=0.0038, ****p<0.0001 vs respective groups as determined using one-way ANOVA and Tukey’s multiple comparisons test, as shown in the box and violin plot (D).

Generation of human iPSC-derived cardiac fibroblasts

Human iPSC-derived CFs were generated using a previously established stepwise differentiation protocol that mimics embryonic cardiac development, with minor modifications to optimize cell viability and yield (22) (Figure 4A). Briefly, hiPSCs (both male and female) maintained in mTESR plus medium (Stemcell technologies), at ~90% confluency was induced into cardiac mesoderm using RPMI/B27 minus insulin medium supplemented with CHIR99021 (6 μM), followed by treatment with IWP-2 (2.5 μM) to direct differentiation toward the cardiac lineage. By day 6, cardiac progenitors were transitioned to a proepicardial-like state using Y27632 (5 μM), retinoic acid (2 μM), and CHIR99021 (5 μM) in Advanced DMEM/GlutaMAX. Proepicardial cells were subsequently expanded and matured into quiescent cardiac fibroblasts by treatment with SB431542 (10 μM) and hFGF-B (1 ng/mL) in fibroblast growth medium on Matrigel-coated culture plates. Media were changed every 48 hours, and cells were passaged using TrypLE Express Enzyme upon reaching ~80–90% confluence. All cultures were maintained at 37°C in a humidified incubator with 5% CO₂. Prior to seeding, all tissue culture plates were pre-coated with Matrigel and incubated at 37°C for 1 hour to promote cell adhesion. Cell treatments were performed after low-serum supplemented FGM3 medium for 24h as mentioned previously. Exposure to fatty acid excess medium was performed in an identical manner to HCFs. iPSC-CFs were incubated with 10ng/mL of rhTGF-β₁ for 24h followed by downstream assays (Supplement Figure 3). 3D gel contraction assay was performed for functional validation of iPSC-CFs using standard protocol with minor modifications (23, 24). Briefly cells were embedded in Collagen type I matrix and seeded in 96-well plates. After polymerization, gels were released and exposed to rhTGF-β₁ or vehicle and gel surface area were measured for defined timepoints (Supplement Figure 3F). The human iPSC lines GSB-L2053 (male, 60yo, white) and GSB-L955 (female, 68yo, white) were obtained from Greenstone Biosciences Inc. (CA, USA).

Figure 4: Human iPSC-derived cardiac fibroblasts exhibit sex-dependent changes in ECM genes in response to FA overload.

(A) Schematic depiction of the directed differentiation protocol for generating cardiac fibroblasts (iPSC-CFs) from human induced pluripotent stem cells (iPSCs). The timeline shows the sequential stages of mesoderm induction, epicardial cell induction, and cardiac fibroblast maturation and expansion for experiments. (B-C) mRNA abundance of specific ECM genes was determined using qRT-PCR. Data were normalized to 18s and gene expression calculated using the 2^−ΔΔCt method. Data are presented as mean±SEM, n=3 per condition; *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001 vs. respective groups (as shown in the plots) as determined by one-way ANOVA with Tukey’s multiple comparisons test. (D) Summary schematic comparing 2D and 3D culture systems in their ability to model fibroblast behavior and mimic human myocardial tissue. Key metrics assessed include proliferation rate, baseline morphology, atypical ECM gene expression in response to fatty acid (FA) treatment, and translational relevance. The 3D hydrogel cell-encapsulation model more closely reflects in vivo conditions and offers improved fidelity for studying cardiometabolic disease processes in a dish. The schematic also illustrates the workflow for expansion and maintenance of iPSC-CFs for disease modeling applications.

3D- bioprinting using collagen hydrogel

Computer-aided design (CAD) models were created using SolidWorks and sliced using Ultimaker Cura for compatibility with the Tissue Scribe Bioprinter (3D Cultures) (Figure 1B and Supplemental Table 3). The prepared HCF-collagen suspension was loaded into a 3 mL syringe fitted with a 25G blunt-end extrusion nozzle. Constructs were printed layer-by-layer onto culture plates under sterile conditions. Following bioprinting, constructs were incubated at 37°C with 5% CO₂ for ~30 minutes to allow collagen polymerization. FGM-3 medium was then gently added to each well to fully submerge the constructs. 3D hydrogels were maintained for up to 5 days with media changes every 48 hours. iPSC-CFs were treated similarly. Fatty acid-rich medium incubations were performed as mentioned previously.

5-ethynyl-2’-deoxyuridine (EdU) incorporation assay for cell proliferation

Primary HCFs were grown on 2D TCP or embedded in 3D hydrogels as described above. After equilibration in low-serum medium for 24h, cells were incubated in excess fatty acid-rich FGM3 medium for 48h. EdU (10μM final concentration) was added in the last 24h of the experiments. Cells were then rinsed once with PBS and fixed with 4% PFA for 10 minutes at room temperature. After cell permeabilization using 0.5% Triton^® X-100 in PBS for 20 minutes, click chemistry reactions were performed using the Click-iT^™ EdU Cell Proliferation kit (Alexa Fluor^™ 555 dye) as per manufacturer’s instructions (Invitrogen, Supplemental Table 2). EdU-stained cells (EdU⁺; red) and nuclei (blue) were visualized using appropriate excitation/emission filter sets, (555/565nm for EdU and 350/461nm for Hoechst 33342 respectively) on a Zeiss LSM 880 laser-scanning confocal microscope. Data was recorded as percentage cell proliferation compared to 2D vehicle group from at least ten independent wells.

ATAC-Seq

2D-cultured HCFs were treated with vehicle or fatty acid-rich medium for 48h as described above. Multiple plates of cells were combined to yield sufficient material for the assay. Cells were then gently lifted and frozen for downstream processing for nuclei isolation, tagmentation, library preparation and sequencing at Novogene (USA) (Figure 2A). Briefly, cell viability was determined with trypan blue exclusion assay. ATAC-Seq was performed following established protocol with modifications as previously reported (18). Nuclei were extracted, and the pellets were used for transposition using the Vazyme Hyperactive ATAC-Seq Library Prep Kit for Illumina. The transposition reactions were incubated at 37°C for 30 min. Equimolar amounts of Adapter1 and Adapter2 were added after the transposition reaction, and library amplification was performed using PCR. The generated libraries were purified with the AMPure beads and library quality was assessed with Qubit fluoromoter. Libraries were diluted to 1 ng/μL and evaluated for insert size distribution using the NGS3K fragment analyzer. If libraries met expected size profiles, the accurate molar concentration was determined via qPCR. Libraries with effective concentrations >2 nM were pooled in equimolar amounts for sequencing. Pooled libraries were sequenced on an Illumina NovaSeq X Plus system using the 25B flow cell configuration. Sequencing was performed using paired-end chemistry based on sequencing-by-synthesis. Fluorescently labeled dNTPs, DNA polymerase, and adapter-specific primers were introduced to the flow cell for cluster amplification and sequencing.

Figure 2: HCFs grown on 2D substrate exhibit changes in chromatin accessibility in response to nutrient excess.

(A) Schematic representation of the assay for transposase-accessible chromatin with sequencing (ATAC-seq) of 2D cultured HCFs exposed to FA-rich medium or vehicle for 48h. (B) Heatmaps of the peaks mapped to TSS region (±3000bp relative to the position of the gene) for 2D HCFs exposed to FA or vehicle. (C) Venn diagram illustrating overlap of genes (±3000bp relative to TSS) between FA and vehicle-treated HCFs in 2D culture.

The raw sequence data (2X151bp) was assessed for quality using FastQC (v0.12.1), low quality reads and artificial sequences were removed using fastp (v0.23.4), and then quality assessment on the trimmed data was repeated. Preprocessing yielded approximately 63 to 69 million reads per sample. To align the raw ATAC-seq data to the reference genome, GRCh38 ‘no-alt’ assembly from NCBI, the most recent gene annotation file (v46) of GRCh38 annotations provided by ENCODE, and the aligner, bowtie2 (v2.4.2) with options that produce alignments only for pairs where both reads align properly as a pair, were used. For post-alignment quality control (QC), mitochondrial reads, duplicates, and ENCODE blacklist regions were removed. The number of uniquely mapped reads after post-alignment QC ranged from 44 to 48 million paired-end reads. Transcription start site (TSS) enrichment score was estimated at 12.2 for the FA and 9.9 for the Veh. Specifically to ATAC-seq, the fragment size distribution was checked, which is expected to correspond to the length of nucleosomes. MACS2 (v2.2.9.1) was used to perform peak calling and identify genome-wide regions of accessible chromatin with an option of the narrow peak. A total of 170,743 peaks were identified in the FA condition, with a fraction of reads in peak (FRiP) score of 53.7%. In the Veh condition, 171,336 peaks were detected, with a FRiP score of 48.6%. Peaks were annotated with TSS regions defined as ±3000 base pairs from the TSS sites using ChIPseeker R package. For downstream analysis, functional enrichment analysis against the Reactome pathway database, and motif analysis against JASPAR CORE motifs, using a suite of R packages, including ChIPpeakAnno, clusterProfiler, ReactomePA, JASPAR2022 were performed. The data have been deposited in NCBI’s Gene Expression Omnibus (GSE304619).

Quantitative real-time PCR (qRT-PCR)

Total RNA was isolated from 2D TCP and 3D-encapsulated cells using RNeasy Mini Kit as per manufacturer’s directions (QIAGEN). Equal amounts of RNA were subjected to cDNA synthesis using the Verso cDNA Synthesis Kit (Thermo Scientific^™). qPCR was performed on a QuantStudio^™ 7 Flex equipment (Applied Biosystems) using the TaqMan^™ Fast Advanced Master Mix (Applied Biosystems) and gene-specific TaqMan^™ assays. mRNA abundance was calculated using the 2^−ΔΔCt method and normalized to 18s ( Figure 3B). TaqMan^™ gene expression assay identifiers are listed in Supplemental Table 1.

Figure 3: Peak enrichment analysis reveals regulation of atypical ECM genes under FA-rich conditions.

(A) Pathway enrichment analysis of differential peak-associated genes using the Reactome database (Benjamini-Hochberg multiple testing correction and adjusted p-value cutoff of 0.01). Note that pathways with more than three gene differences between FA and vehicle are included. (B) Schematic depiction of the gene expression assay using qRT-PCR. (C-D) mRNA abundance was determined using qRT-PCR. Data were normalized to 18s and gene expression calculated using the 2^−ΔΔCt method. Data are presented as mean±SEM, n=3–6; *p<0.05, **p<0.01, ****p<0.0001 vs. respective groups (as shown in the plots) as determined by one-way ANOVA with Tukey’s multiple comparisons test.

Immunoblotting

Total protein from iPSC-CFs treated with TGF-β₁ or vehicle was isolated using RIPA Lysis and Extraction Buffer (Thermo Scientific) containing Halt^™ protease and phosphatase inhibitor cocktail (Thermo Scientific). Protein concentrations in the samples were determined using the Pierce BCA Protein Assay Kit as per manufacturer’s instructions (Thermo Scientific). Equal concentration of proteins (15 μg) was separated by SDS-PAGE using pre-cast gels (Bio-Rad) and then transferred to nitrocellulose membrane using Trans-Blot Turbo Transfer System (Bio-Rad). Membrane was blocked using EveryBlot (Bio-Rad) for 30 minutes followed by incubation in primary antibody overnight at 4°C (mouse anti-αSMA antibody; Sigma). Protein of interest was detected using an HRP-conjugated secondary antibody (Jackson ImmunoResearch Labs) and SuperSignal^™ West Pico PLUS Chemiluminescent Substrate (Thermo Scientific) on a ChemiDoc MP Imaging System (Bio-Rad). Relative band intensities were quantified using the Quantity One software (Bio-Rad; Supplemental Table 3). The intensity of the αSMA bands was normalized to GAPDH (loading). Reagent and antibody details are listed in Supplemental Tables 2 and 4.

Indirect immunofluorescence and neutral lipid staining

HCFs and iPSC-CFs were cultured on microscope cover glass (Fisherbrand 18CIR-1) in their respective media. After treatments, cells were fixed in 4% PFA in PBS for 10 minutes. Fixed cells were permeabilized in 0.03% Triton-X100 and blocked in PBS containing 0.1% Tween-20, 5% bovine serum albumin (BSA; Sigma) for 1 hr. Cells were incubated with the primary antibodies specific to vimentin (cell Signaling Technologies) or α-SMA (Abcam ab7817) overnight at 4°C. Cells were then rinsed and incubated with the either Alexa Fluor 488- or Alexa Fluor 594-conjugated secondary antibodies (Thermo Scientific) for 1 hour in the dark. Phalloidin (Alexa Fluor 647-conjugated) was used where applicable along with secondary antibody incubations. Coverslips were mounted using ProLong^™ Diamond Antifade Mountant with DAPI (Invitrogen) and allowed to cure overnight prior to imaging on a Nikon Model Eclipse Te2R-FL using a 10X objective. BODIPY 493/503 (MedChemExpress) staining was performed on PFA-fixed cells and imaged on a on a Zeiss LSM 880 laser-scanning confocal microscope to visualize lipid accumulation in cells (Supplement Figure 4). Reagent and antibody details are listed in Supplemental Tables 2 and 4 respectively.

Statistical analysis

GraphPad Prism 10 (Supplemental Table 3) was used for all statistical analyses except ATAC-Seq. All data are presented as mean±SEM. Unpaired t test with Welch’s correction (two-tailed) was used for comparisons between two groups. A one-way analysis of variance (ANOVA) followed by Tukey’s multiple comparison tests were used to determine statistical significance among multiple groups. A p-value <0.05 was considered statistically significant.

Results

Substrate stiffness induces changes in cell morphology and proliferation of human cardiac fibroblasts.

To model biomechanical cues reflective of the myocardial extracellular matrix (ECM), human cardiac fibroblasts (HCFs) were cultured on conventional 2D tissue culture plastic (TCP) or encapsulated in 3D collagen I hydrogels fabricated using extrusion-based bioprinting. These conditions simulated high-stiffness and physiologically relevant low-stiffness environments, respectively (Figure 1A–B). Brightfield microscopy revealed stark morphological differences between culture systems. HCFs in 2D adopted a flattened, polygonal morphology with stress fiber formation and increased nuclear area—hallmarks of myofibroblast-like activation (Figure 1C, upper panel). In contrast, 3D-embedded fibroblasts retained a more spindle-shaped morphology with reduced cell spreading, reminiscent of the native stromal fibroblast phenotype in non-injured myocardium (Figure 1C, lower panel). These observations suggest that 2D culture systems may impose mechanical signals that prime fibroblasts toward a pathologically activated state, even in the absence of external stimuli.

Quantification of proliferative capacity under metabolic stress was analyzed using EdU incorporation analysis. HCFs in 2D culture showed a statistically significant increase in EdU⁺ nuclei following FA exposure, consistent with activation and cell cycle entry (Figure 1D–E). In contrast, FA-treated HCFs in 3D culture showed no significant difference in proliferation relative to vehicle-treated controls. These data strongly suggest that the 3D hydrogel matrix environment not only restricts basal proliferation but also dampens the mitogenic effects of FA overload. This implies a role for ECM compliance and dimensionality in modulating metabolic responsiveness and cellular plasticity, including increased variance in the 2D group compared to the 3D-embedded cells.

Fatty acid overload induces genome-wide changes in chromatin accessibility in 2D-cultured human cardiac fibroblasts.

To examine epigenomic changes induced by FA overload, assay for transposase-accessible chromatin using ATAC-seq was performed (Figure 2A). While global accessibility patterns were largely similar, a subset of differentially accessible regions (DARs) near transcription start sites (±3000 bp) were identified, implicating key regulatory loci (Figure 2B–C, Supplement Figure 1). Pathway enrichment analysis of DAR-associated genes revealed significant upregulation of pathways involved in ECM organization, integrin signaling, and collagen biosynthesis (Figure 3A). Complementary qRT-PCR analysis (Figure 3B) was performed to confirm transcriptional changes of these same targets. Contrary to canonical models of fibrosis, accessibility and transcription of classical ECM remodeling genes remained unchanged following FA exposure. Instead, increased chromatin accessibility was observed in genes encoding atypical ECM components, including COL9A1, COL11A2, and PLOD3, a collagen cross-linking enzyme upregulated in several fibrotic and metabolic conditions (Figure 3C). COL9A1 and COL11A2 are primarily expressed during cartilage development and in embryonic mesenchyme (25, 26), with limited expression in the adult heart under homeostatic conditions. Their re-expression in cardiac fibroblasts under metabolic stress suggests a shift toward a developmental or non-canonical fibrotic program. It is intriguing to note that while COL9A1 gene expression was reduced at baseline in 3D- HCFs, this was not recapitulated by COL11A2. FA exposure in 3D setting blunted COL9A1 expression in contrast with 2D cultured HCFs. However, the opposite trend was observed in COL11A2 gene expression (Figure 3C). This suggests that selective effects of FA exposure on these collagen gene responses depend on substrate stiffness. PLOD3 is similarly expressed at low levels in healthy adult myocardium but is upregulated in systemic fibrotic conditions, including pulmonary fibrosis and metabolic liver disease (27, 28). The selective activation of these genes in both 2D and 3D cultured HCFs indicates that FA exposure triggers early transcriptional programs distinct from classical cardiac fibrosis, possibly reflecting alternative fibroblast states or adaptive remodeling in response to metabolic overload and substrate stiffness.

Additional FA-responsive loci were DMP1 (DMTF1), CLEC14A, and FKRP, which highlight the activation of non-canonical fibroblast programs (Figure 3D). DMTF1 (Cyclin D Binding Myb Like Transcription Factor 1) is a transcriptional factor known for its role in regulating cell cycle arrest, senescence, and apoptosis through activation of the ARF-p53 tumor suppressor pathway (29, 30). It is typically studied in the context of cancer biology and hematopoietic stem cell regulation (31, 32), but not commonly expressed in cardiac tissue. Its increased chromatin accessibility in response to FA exposure suggests early stress-induced reprogramming that links metabolic overload to fibroblast plasticity. CLEC14A, an endothelial adhesion molecule involved in angiogenesis and vascular remodeling, points to potential crosstalk between fibroblasts and vascular signaling networks (33, 34). FKRP (fukutin-related protein), a glycosyltransferase essential for α-dystroglycan modification and sarcolemma integrity, may reflect cytoskeletal or membrane-associated remodeling under metabolic stress (35, 36). Quite interestingly, 3D-cultured HCFs not only exhibited significantly lower expression of these genes at baseline compared to 2D-cultured cells, but FA exposure-induced upregulation of the genes in 2D HCFs were not recapitulated in 3D settings (Figure 3D).

Together, these data indicate that FA exposure promotes transcriptional reprogramming of fibroblasts, characterized by chromatin remodeling at non-canonical ECM-related genes. To further explore potential upstream transcriptional regulators driving these epigenomic changes, we performed motif enrichment analysis on differentially accessible peaks. The top 10 enriched transcription factor motifs in FA-treated HCFs included members of the GATA, TEAD, ETV, and IRF families (Supplement Figure 2). These motifs suggest activation of transcriptional networks associated with stress response, fibroblast plasticity, and metabolic signaling. Notably, the presence of GATA and TEAD motifs implicates regulatory programs commonly associated with cardiac development and fibrosis (37–40), while ETV and IRF family members may reflect broader chromatin remodeling under inflammatory or metabolic pressure (41, 42).

Response of 3D matrix encapsulated HCFs induces sex-dependent gene regulation variability under nutritional overload.

It has been well documented that there are sex-dependent changes in the heart when inflicted with cardiometabolic diseases (43–45). To expand the translational relevance of our platform, we incorporated human iPSC-CFs to assess the expression of ECM genes in response to FA exposure. These cells, generated from defined genetic backgrounds, offer a reproducible and scalable tool for modeling disease-relevant fibroblast behavior in vitro (Figure 4A). Validation of the differentiated cells was performed using immunofluorescence (Supplement Figure 3A) and gene expression assays (Supplement Figure 3B) for standard markers of pluripotency and/or fibroblast markers. To ensure that the derived iPSC-CFs were quiescent, they were exposed to the pro-fibrotic agent TGF-β₁ for 24 hours. Expression of α-smooth muscle actin, a fibroblast activation marker, was strongly upregulated as determined via immunofluorescence (Supplement Figure 3C) and immunoblotting (Supplement Figure 3D-E) assays. Functional characterization of these cells was performed using 3D collagen matrix contraction assay with measurements recorded every 24 hours for up to 120 hours (Supplement Figure 3F). As expected, TGF-β₁-treated iPSC-CFs contracted the gel strongly and significantly compared to vehicle-treated cells (Supplement Figure 3F). Following FA treatment, as anticipated, sex-based differences in transcriptional response of iPSC-CFs were observed. Quantitative RT-PCR analysis revealed that COL11A2 expression significantly increased in both male and female iPSC-CFs, but the magnitude was substantially greater in male cells. COL9A1 and PLOD3, which were previously identified in our ATAC-seq analysis of primary male HCFs, also showed FA-induced upregulation in both sexes, with male iPSC-CFs exhibiting a more robust upregulation in 2D (Figure 4B). In contrast, female iPSC-CFs demonstrated moderate increases, particularly in 3D cultures (Figure 4C). These results indicate that while fibrogenic gene induction occurs in both sexes, male cells display a heightened transcriptional response in 2D, whereas female cells appear more responsive in 3D matrix. These results are consistent with our findings in primary HCFs and further support the role of metabolic stress in initiating non-canonical ECM gene programs. The differential responses between sexes and across dimensional context highlights the importance of integrating both variables into disease modeling platforms. Incorporating iPSC-CFs into this framework enhances its utility for dissecting the molecular basis of sex differences observed in clinical outcomes of cardiometabolic disease.

Conclusions and Discussion

Our study highlights the critical importance of tissue dimensionality and biomechanical context in shaping fibroblast responses to metabolic stress. While 2D culture systems have long served as foundational tools in cell biology, they impose supraphysiological stiffness and spatial constraints that can artifactually activate stress-responsive and fibrogenic programs in otherwise quiescent cardiac fibroblasts. In contrast, 3D-bioprinted hydrogels provide a compliant, matrix-rich microenvironment that more closely resembles the human myocardium, thereby preserving key physiological features such as low basal proliferation, spindle-like morphology, and homeostatic signaling (Figure 4D).

Notably, FA overload in 2D cultures induced marked epigenomic remodeling, particularly at non-canonical ECM-related genes such as COL9A1, COL11A2, and PLOD3. These chromatin changes diverge from classical fibrotic signatures and may reflect an early or alternative activation state specific to metabolic stress (Figure 2B–E). Motif enrichment analysis of differentially accessible regions further revealed transcription factor binding signatures associated with GATA, TEAD, IRF, and ETV family members—transcriptional regulators implicated in fibrosis, development, and stress response(Supplement Figure 2). These data suggest that acute FA exposure activates a complex and non-traditional transcriptional network in fibroblasts that may be missed by conventional endpoint gene expression analysis.

Together, these findings support a paradigm shift toward 3D models for investigating cardiometabolic disease mechanisms, particularly in the context of fibrosis and ECM remodeling. While 2D systems remain useful for high-throughput screening and mechanistic reductionism, they fall short in capturing the nuanced interplay of mechanical, spatial, and paracrine cues that govern fibroblast fate and function in vivo.

Furthermore, our work sheds light on the underexplored concept of mechanical epigenetics—how the structural microenvironment shapes the chromatin landscape and transcriptional potential of stromal cells. The absence of overt fibrogenic gene upregulation in 3D cultures, even under FA stress, suggests that cell-matrix interactions play a critical regulatory role that could be exploited therapeutically to prevent or reverse maladaptive fibroblast activation in cardiometabolic diseases.

Importantly, a key limitation of this study is the absence of complementary chromatin immunoprecipitation sequencing (ChIP-seq) to directly assess histone modifications and transcription factor binding at the differentially accessible chromatin regions identified by ATAC-seq. While ATAC-seq provides a powerful assay of chromatin accessibility, integrating ChIP-seq data would strengthen conclusions regarding the epigenetic regulatory mechanisms and validate the functional relevance of candidate transcription factors suggested by motif enrichment. Future studies incorporating ChIP-seq and CUT&RUN (46) assays are necessary to precisely map the dynamic histone modification landscape and transcription factor occupancy under metabolic stress and different biomechanical conditions. To further enhance the translational potential of this 3D bioprinting model, future work will incorporate patient-derived HCFs and iPSC-CFs obtained from multiple individuals with varying degrees of metabolic dysfunction, including obesity, insulin resistance, and type 2 diabetes. This approach will allow for the interrogation of inter-individual variability in fibroblast activation, ECM remodeling, and responsiveness to pharmacological interventions using physiologically relevant model systems. By integrating clinical heterogeneity into the in vitro platform, our study provides the foundation to establish a personalized screening tool that captures the complex biological landscape underlying cardiometabolic disease progression and therapeutic response. Another critical advancement will involve the development of integrative co-culture systems that incorporate additional cardiac cell types, such as cardiomyocytes and endothelial cells. These studies are currently underway in our laboratory and have been explored by other groups as well (47–49). These multi-cellular 3D-bioprinted models will allow the study of intercellular crosstalk and paracrine signaling pathways that are fundamental to the pathophysiology of cardiac fibrosis and metabolic remodeling. Such systems would more faithfully mimic the cellular and structural complexity of myocardial tissue, enhancing our ability to model early pathogenic events and assess therapeutic efficacy in a more physiologically relevant context.

Ultimately, this work lays the groundwork for a modular, human-relevant platform to study fibroblast biology in metabolic disease. The strength of this study is two-fold. First, we present technologically advanced platform to study cardiac fibroblast form and function. Secondly, we validate this platform using iPSC-derived cells which has immense translational value, not just limited to fibroblast biology in cardiovascular or metabolic diseases. Although challenges remain, the results presented here offer a strong foundation upon which to build more predictive and mechanistically insightful models, such as 3D-bioprinted cardiac microtissues, for understanding and treating cardiac fibrosis in the context of metabolic dysfunction.

Supplementary Material

Supplemental data are available at https://doi.org/10.6084/m9.figshare.30048250.