The heart is a dynamic organ, and the cardiac tissue experiences changes in pressure and stretch during the cardiac cycle. Existing cell culture and animal models are limited in their capacity to decouple and tune specific hemodynamic stresses implicated in the development of physiological and pathophysiological cardiac tissue remodeling. This study focused on creating a system to subject engineered cardiac tissue to either pressure or stretch stimuli in isolation and the subsequent evaluation of acute tissue remodeling. We developed a cardiac tissue chip containing three-dimensional (3-D) cell-laden hydrogel constructs and cultured them within systems where we could expose them to either pressure changes or volume changes as seen in the left ventricle. Acute cellular remodeling with each condition was qualitatively and quantitatively assessed using histology, immunohistochemistry, gene expression studies, and soluble factor analysis. Using our unique model system, we isolated the effects of pressure and stretch on engineered cardiac tissue. Our results confirm that both pressure and stretch mediate acute stress responses in the engineered cardiac tissue. However, both experimental conditions elicited a similar acute phase injury response within this timeframe. This study demonstrates our ability to subject engineered cardiac tissue to either pressure or stretch stimuli in isolation, both of which elicited acute tissue remodeling responses.
The heart is a dynamic organ that behaves like a pulsatile pump, undergoing cyclical changes in intracardiac pressure and myocardial stretch due to the contraction and relaxation of cardiomyocytes. The magnitude of these mechanical stresses varies during developmental, healthy, and cardiovascular disease states, and alterations often result in physiological or pathological hypertrophic responses in the heart [Clark et al., 1986; Clark and Hu, 1990; Mann, 2004]. In general, cardiac hypertrophy is synonymous with increased cardiomyocyte cell size and thus heart mass rather than hyperplasia (i.e., an increase in cell number) [Diwan and Dorn, 2007]. Physiological hypertrophy is associated with development and exercise training, resulting in maintenance of normal cardiac function or increases in specific functional parameters [Donoghue et al., 2021]. In contrast, pathological hypertrophy occurs in the context of disease and is almost always associated with maladaptive myocardial remodeling and dysfunction. Notably, pathological hypertrophy is an adverse response to either pressure overload (PO) or volume overload (VO). In PO, elevated systemic resistance due to conditions such as hypertension or aortic stenosis results in high left ventricular systolic pressure, resulting in predominantly concentric hypertrophy [Grossman et al., 1975]. The hallmark PO phenotypes include increased extracellular matrix (ECM) deposition and myofibroblast differentiation, as well as cardiomyocytes growing in thickness due to parallel addition of sarcomeres, yielding myocardial wall thickening [Kong et al., 2014]. In VO, excess left ventricular filling volume due to conditions such as mitral and aortic valve regurgitation and septal defects results in eccentric hypertrophy [Carabello, 2002; Hutchinson et al., 2010]. This phenotype exhibits wall thinning due to ECM degradation and cardiomyocytes extending in length due to the addition of sarcomeres in series.
Tissue chip models containing engineered cardiac tissue offer advantages in screening cardiotoxic compounds and elucidating mechanisms and signaling pathways involved in disease development [Chan and Huang, 2021]. Since alterations in intraventricular pressure and myocardial stretch play significant roles in cardiovascular disease progression, incorporating these stresses is critical when constructing cardiac tissue chips aimed at modeling different cardiovascular disease states. Several groups have previously investigated the effects of stretch or pressure on cardiac myocytes or fibroblasts in 2D culture but have yet to perform side-by-side comparisons of the acute impact of these stimuli to examine their relative contributions to cardiac remodeling [Lee et al., 1999; De Jong et al., 2013; Nguyen et al., 2013, 2015; Herum et al., 2017]. Complex coupled (pressure and stretch) loadings that mimic the cardiac cycle have been investigated previously by our lab and were found to replicate characteristics of clinical pressure and volume overload states [Rogers et al., 2019]. Decoupling these stimuli with in vitro models could help delineate the relative contributions of pressure and stretch in mediating cardiac remodeling and allow for a greater understanding of the roles these stimuli play in normal and disease states. Unlike in animal models where the effects of pressure and stretch cannot be decoupled, in vitro models such as those described below allow for application of either stimulus in isolation. Thus, pressure-only or stretch-only cardiac tissue chips offer a unique ability to study the role of pressure and stretch in propagating physiological and pathophysiological cardiac remodeling.
Materials and Methods
Cardiac Cell Culture Chamber Fabrication
The cell culture chambers were fabricated using standard soft lithography methods established in our lab [Nguyen et al., 2022]. First, polydimethylsiloxane (PDMS) (QSil 216, Quantum Silicones, Richmond, VA, USA) with a base-to-crosslinker ratio of 10:1 was thoroughly mixed, desiccated for 45 min to remove any bubbles, and poured into a square mold of 5 cm in length, 5 cm in width, and 1.5 cm in height. The PDMS was cured overnight in a 70°C oven, and the inner section of the PDMS was cut away to yield a 3 cm length by 3 cm width square cavity surrounded by a 2 cm frame. This cavity, which is subsequently sealed from below by a flexible membrane, later served as the media reservoir. From a second PDMS slab, 6 cylindrical posts measuring 1 mm in diameter and 1 cm in height were excised using biopsy punches. For the base of the cardiac device, a flexible membrane was fabricated by spin-coating 10 mL of uncured PDMS on a 78.5 cm2 circular silicon wafer for about 3 s. The PDMS-coated wafer was then placed on a 70°C hotplate, and the 6 cylindrical posts were immediately placed on the uncured PDMS using a removable post stencil/guide to ensure proper positioning. Placing the posts before the PDMS membrane was cured allowed secure attachment and positioning. The wafer was then left on the hotplate overnight to cure the post-laden membrane. The final thickness of this PDMS base was 1 mm, and the 6 posts were oriented into 3 pairs (2 columns within each pair spaced 1 cm apart; each pair spaced 7.5 mm from the next). The post-laden membrane (plus underlying wafer) and square frame were then cleaned and bonded via oxygen plasma treatment (Harrick Plasma Systems, Ithaca, NY, USA) using 700 mTorr of pressure and 45 s of plasma exposure. The bonding created a tight seal to combine the 2 elements into 1 PDMS device. Before cell culture use, the underlying silicon wafer was carefully removed, excess PDMS was trimmed, and the device was autoclaved for sterilization. The inner chambers were then filled with 2% molten agarose up to 1 mm below the top of the posts, which was then allowed to gel. Dumbbell-shaped troughs were created using a 5 mm-diameter biopsy punch to remove agarose around the posts, and the agarose connecting the paired posts was cut with a scalpel and removed with tweezers. The resulting pattern was 3 dumbbell-shaped chambers, each serving as a mold for cell-hydrogel seeding. Finally, the PDMS-agarose device was washed with phosphate-buffered saline (PBS) and warmed in the 37°C incubator before seeding the cardiac tissue constructs.
Cell Culture and Collagen I-Matrigel Hydrogel Encapsulation
Rat myoblasts and mouse embryonic fibroblasts were purchased from ATCC [H9c2(2-1), ATCC; CRL-1446; MEF (CF-1) SCRC-1040] and cultured in Dulbecco’s Modified Eagle’s Medium (ATCC; 30-2002) supplemented with 10% Fetal Bovine Serum (FBS) and 2% non-essential amino acids (Thermo Fisher Scientific; 11140050) in 5% CO2 at 37°C. H9c2 and MEFs were dissociated once they reached ∼85% confluency using 0.25% trypsin/EDTA (Gibco; 25200-056). Each engineered tissue construct in the device contained approximately 2.8 × 106 H9c2 and 1.4 × 106 MEF resuspended in collagen I via the following method (Fig. 1a). First, 100 μL of 10× PBS and 20 μL 1 N NaOH were added to a small centrifuge tube, vortexed, and then placed on ice. While on the ice, 860 μL of collagen I (Corning; 354236) was added to the PBS-NaOH mixture, vortexed, and quickly returned to the ice. Then 100 μL of Matrigel® (Corning; 354230) was added to the mixture to create a collagen I-Matrigel hydrogel. Following trypsinization and combination of H9c2 and MEF, the cells were centrifuged for 30 s to form a pellet and the supernatant was aspirated. A volume of 1,080 μL of collagen I-Matrigel hydrogel was then added to the cells and pipetted thoroughly to homogenize cell distribution throughout the gel. Afterwards, 250 μL of this suspension was pipetted into each dumbbell chamber. The device was placed into the incubator for 45 min for gelation, and 3 mL of complete DMEM media was then added to the device’s inner reservoir and returned to the incubator overnight. The next day, the agarose mold was carefully removed, and the media was replenished.
Hemodynamic Experimental Setup
Two hemodynamic setups were designed to expose the engineered tissue constructs to stretch or pressure in isolation, both of which employed a modified version of the Biomimetic Cardiac Tissue Model (BCTM) developed by our lab [Rogers et al., 2018, 2019]. In the stretch experiments, the PDMS cardiac tissue chip was sandwiched between a top polycarbonate piece, which bore a 3 × 1 cm oval opening to allow for gas exchange and maintenance of constant atmospheric pressure in the media chamber, and a bottom polycarbonate piece, which had a 3 × 3 × 0.5 cm central cavity connected to a programmable vacuum pump (Fig. 1b). A cylindrical obstacle was also placed in the bottom chamber so that each time vacuum was applied, the flexible membrane would deform around the obstacle, exerting stretches of 0–5% upon the engineered tissue constructs during each cycle (Fig. 1b). The cardiac tissue chip was positioned between 2 non-deformable polycarbonate pieces for the pressure experiments, and “systolic” pressure (Fig. 1c) was created by connecting inlet and outlet ports on the top polycarbonate piece to a programmable pneumatic pump. Each time the pneumatic pump introduced positive pressure to the media chamber, the constructs remained static due to the non-deformable bottom polycarbonate piece. For static, pressure, and stretch conditions, the engineered tissue constructs were each given 3 mL of complete DMEM media contained soley in the PDMS′ reservoir.
Histology and Immunohistochemistry
On conclusion of the experiments at 24 h, engineered cardiac tissues were fixed with 1:10 buffered formalin (Fisherbrand; 245-684) for 24 h before processing. The formalin-fixed, paraffin-embedded engineered cardiac tissue sections were mounted on glass slides, stained with Hematoxylin and Eosin, Masson’s trichrome stain, labeled with Phalloidin (Thermo Fisher Scientific; A34055), or Wheat germ agglutinin (Thermo Fisher Scientific; W11261). Additionally, some slices were immunostained for collagen I using a mouse monoclonal IgG1 antibody (Santa Cruz Biotechnology; sc-59772) and a goat anti-mouse IgG-TR secondary antibody (Santa Cruz Biotechnology; sc-2781). Nuclei were stained using DAPI for all fluorescent stains.
After 24 h of stimulation, the engineered cardiac tissues were flash-frozen and maintained at –80°C until processing. Samples were first bead milled (BeadBug; D1030) in tubes containing 0.5 mm zirconium beads (Benchmark; D1032-05) for 40 s. The total RNA was isolated using the RNeasy® Plus Mini kit (Qiagen; 74134) as directed by the manufacturer’s protocol, and 2 μg of this total RNA was used to obtain cDNA for mRNAs using a High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems; 4368814) as directed by the manufacturer’s protocol. Mouse primers were designed to target the following genes: TIMP-1, TIMP-2, TIMP-3, Col1α1, αSMA, MMP-2, MMP-3, and MMP-9 with their primer sequences located in online supplementary Table 1 (for online suppl. material, see www.karger.com/doi/10.1159/525250). The quantitative real-time PCR utilized 5 ng synthesized cDNA in 1 μL, 5 μL of 2× Fast SYBR Green Master Mix (Applied Biosystems; 4385612), 0.5 μL forward and 0.5 μL reverse primers (both at 10 μ), and 3 μL nuclease free water were combined (total 10 μL) and pipetted into each well of a 384-well PCR plate. The plate was placed into a BioRad 4000 Real-Time PCR system using the following cycling conditions: 95°C denaturation for 10 min, 40 cycles of amplification (95°C denaturation for 15 s, and 60°C annealing for 30 s, 72°C extension for 30 s). Melting curves were used to monitor genomic contamination. For negative controls, SYBR Green-primer reactions were performed without cDNA. The obtained Cq values were normalized to mouse 18S rRNA expression levels and fold changes were calculated using the 2–∆∆Cr method [Livak and Schmittgen, 2001].
Soluble Factor Analysis
Media was collected after 24 h of hemodynamic stimulation, frozen at –20°C for storage and later shipped on dry ice to Eve Technologies (Calgary, AB, Canada). Two arrays were conducted: Mouse Cytokine Array, TGF-β 3-Plex (TGF β1-3) and Mouse MMP Discovery Array 5-plex for cell culture and non-blood samples (MDMMP-C,O). TGFβ 1–3 measured TGFβ 1, 2, and 3, and MDMMP-C,O measured MMP-2, MMP-3, MMP-8, proMMP-9, and MMP-12. Each experiment was independently completed 3 times to produce triplicates of each cell culture media condition. Only soluble factors with detectable concentration or statistical significance were illustrated and discussed.
All data are expressed as mean ± SEM. Statistical significance was evaluated with one-way ANOVA for comparison among ≥3 means with Prism 5 (GraphPad Software Inc). Values of p < 0.05 were considered to be statistically significant.
Generation of Decoupled Pressure and Stretch
Using our modified setups, we generated 3 unique experimental conditions (Fig. 1). As detailed in Table 1, the static controls were maintained at atmospheric pressure with 0% stretch. The pressure-only samples experienced cycles of pressure varying between peak “systolic” pressure of 140 mmHg and “diastolic” pressure of 0 mmHg and a stretch of 0% delivered at a frequency of 80 cycles per minute. The stretch-only samples experienced atmospheric pressure and time varying strain of 0–5% at an 80 cycles per minute frequency.
Tissue Morphology and Organization
Tissue constructs maintained under static, pressure-only, and stretch-only conditions were evaluated for cellular morphology, changes in cell numbers and cell size, organization, and alignment using fluorescence microscopy and histology. Immunofluorescent staining for collagen 1α1 showed similar upregulation in both the pressure and stretch conditions compared to static controls (Fig. 2a). Cytoskeletal organization of F-actin filaments was visualized using Phalloidin staining and showed that both the pressure and stretch conditions resulted in more stress fiber development than the static controls (Fig. 2b). Evaluation of cell size and shape using wheat germ agglutinin (WGA) suggested that acute exposure to pressure and stretch may not significantly increase cell size compared to static controls (Fig. 2c).
Visualization of cell numbers via staining of the nuclei using DAPI indicated that both pressure and stretch were associated with slightly higher cell densities in comparison to static controls (Fig. 2a–c). In these dynamic environments, the cells typically experience greater media perfusion, increasing viability and proliferation.
The H&E staining revealed cell alignment and homogeneous cell distributions in all 3 conditions without any distinguishable differences (Fig. 3a). In addition to immunofluorescent labeling, collagen deposition was determined using Masson’s trichrome staining (Fig. 3b). Our results indicated that both dynamic conditions were associated with higher collagen deposition levels than static controls, similar to that observed with Col1α1 immunohistochemistry.
Gene Expression Profiling
The choice of gene targets for this study was partially influenced by the results of previous studies in our lab, which showed that PO and VO led to upregulation of genes whose products mediate matrix remodeling [Rogers et al., 2019]. In the present study, no statistically significant alterations in gene expression were seen in either experimental group compared to the static controls, perhaps due to the brief time frame of the experiments. However, general trends were observed (Fig. 4).
First, both pressure and stretch conditions elicited an approximately 2-fold increase in Col1α1 expression. Evaluation of alpha-smooth muscle actin (αSMA) showed that stretch constructs had an approximately 5-fold increase in αSMA compared to static, whereas pressure led to a 2-fold increase. Evaluation of matrix metalloproteinases (MMPs) and tissue inhibitors of metalloproteinases (TIMPs) demonstrated an approximately 6-fold increase of MMP-3 and MMP-9 with stretch and 4-fold increase with pressure in comparison to the static controls. TIMP-1, TIMP-2, TIMP-3, and MMP levels were elevated in both pressure and stretch constructs compared to static controls. TIMP-3 expression levels were not discernible between pressure and stretch, while MMP-2 and TIMP-2 were slightly elevated in stretch versus pressure. Conversely, the pressure was associated with slightly greater upregulation of TIMP-1 than stretch.
Soluble Factor Analysis
Cytokine arrays that were used in this study enabled the detection of known mediators of matrix remodeling, inflammation, and angiogenesis in the culture media (Fig. 5). Regarding the matrix metalloproteinases, MMP-3 demonstrated statistically significant upregulation in constructs exposed to dynamic stretch compared to both static controls and pressure constructs, similar to the gene expression trend. MMP-12 was significantly decreased in pressure constructs compared to stretch, and TGFβ3 demonstrated a similar expression pattern. In general, inflammatory cytokine findings were inconsistent.
The heart is a dynamic organ that cyclically contracts and relaxes to pump blood throughout the body. With each cardiac cycle, the cells that make up the heart wall experience alterations in pressure and stretch. These mechanical stresses serve as signal inputs for feedback mechanisms that ensure proper heart function acutely through modulation of heart rate, contractile force, and vascular tone, and chronically through cardiac remodeling. Changes in these stimuli within physiological limits, such as those seen during exercise, can lead to physiological remodeling of the heart depending upon the frequency and duration of the increased stresses, resulting in a slight increase in organ size and improvement in cardiac function [Fagard, 2003; Vega et al., 2017]. In almost any type of cardiovascular dysfunction, extrinsic or intrinsic factors cause intraventricular pressure and/or cardiac wall stretch to shift from physiological to pathological levels, eventually leading to adverse, rather than compensatory, cardiac tissue remodeling. Specifically, in PO, cardiomyocytes become thicker via the addition of sarcomeres in parallel and cardiac fibrosis often develops, both of which contribute to increased wall thickness, oxidative stress, and arrhythmias, which can lead to sudden death [Spinale, 2007; Takimoto and Kass, 2007; Harvey and Leinwand, 2011; Vega et al., 2017]. In contrast, cardiomyocytes subjected to VO conditions become thinner and longer due to addition of sarcomeres in series, and chronically volume overloaded hearts demonstrate heightened ECM degradation and myofibrillar disorganization [Spinale, 2007]. Both PO and VO are also associated with contractile dysfunction, activation of fetal gene programs, metabolic switch from fatty acid oxidation to glycolysis, electrophysiological dysfunction, and increased oxidative stress [Mann, 2004].
In intact organisms, pressure and stretch in the heart are coupled and cannot be evaluated in isolation. However, as exemplified by the present study, in vitro models can study the effects of pressure or stretch alone. In this study, we engineered 2 unique platforms to study the impact of pressure or stretch in isolation on engineered cardiac tissue. In both dynamic conditions, the engineered tissues were subjected to 80 cycles per minute, as this falls within the normal heart rate range for adult humans [Bonnemeier et al., 2003]. 140 mmHg of pressure, which falls within the range of maximal left ventricular pressures in healthy adults, was applied to the engineered tissues in the pressure studies [Bahraseman et al., 2013]. Though the 5% strain achieved in the stretch condition falls below the estimated 19% experienced by myocytes in healthy adult hearts [Wang et al., 2016], the authors felt this was sufficient to elicit any stretch-induced cell remodeling phenomena. Using this platform, we sought to (1) characterize how engineered cardiac tissue responds to isolated acute stimulation with pressure or stretch, (2) determine if the responses were distinct, and (3) compare how the acute phase response to pressure or stretch in isolation compares to morphological and gene expression changes associated with chronic PO and VO.
As mentioned earlier, PO and VO are associated with adverse tissue remodeling. Specifically, PO results in increased fibrosis, whereas VO frequently results in excessive ECM degradation and tissue thinning [Eghbali, 1992; Nicoletti and Michel, 1999; MacKenna et al., 2000]. Since alterations to the ECM play significant roles in PO- and VO-associated pathological cardiac tissue remodeling, we tailored our evaluation of gene expression changes and analysis of soluble factors to assay proteins that either play direct roles in ECM remodeling or are indicative of a fibrotic phenotype. Both MMPs and TIMPs are widely recognized for their roles in mediating matrix remodeling in cardiac disease states [Spinale, 2007]. MMP classes with myocardial remodeling relevance include collagenases (MMP-1), stromelysins (MMP-3), gelatinases (MMP-9 and MMP-2), and the membrane-type MMPs [Mann and Spinale, 1998].
Our gene expression results showed that collagen 1α1 was upregulated in pressure and stretch compared to static controls, and this finding was also reflected by our histological and immunohistology analyses (Fig. 4). The progression of cardiac hypertrophy in response to chronic hemodynamic overload stimulates cardiac fibroblasts to undergo a phenotypic switch to become contractile myofibroblasts, cells that express abundant αSMA through the Rho signaling pathway [Black et al., 1991; Leslie et al., 1991; Zhao et al., 2007]. Similar to Col1α1, upregulation of αSMA expression was observed in stimulated constructs compared to static controls.
In summary, we were able to devise platforms that allowed for applying pressure or stretch in isolation to engineered cardiac tissue constructs and demonstrate that engineered tissue constructs respond to isolated pressure and stretch stimuli via changes in structure, gene expression, and soluble factor production. Though the experimental conditions led to unique responses, the majority were not statistically significant and differed from those seen in chronic PO or VO. Due to the short time course of these experiments and the uncoupled nature with which pressure and stretch were applied, it is not surprising that the findings do not align with those from clinical scenarios. These results suggest that acute exposure to pressure-only or stretch-only conditions elicit an acute injury response from the engineered cardiac tissue, which may not clearly distinguish pressure from stretch; however, we believe that chronic exposure to pressure-only or stretch-only conditions would enable adaptive changes past the acute injury response which would more closely resemble changes seen with chronic PO and VO.
Overall, the gene expression data suggest the cells within the engineered constructs are sensitive to the applied stimuli, as indicated by the upregulation of genes associated with ventricular remodeling in both experimental conditions compared to the static control. While both pressure and stretch stimuli elicited responses in similar genes in most cases, the magnitude of the gene expression changes and soluble factor concentrations were different.
Moving forward with these bioengineered constructs and experimental hemodynamic systems, it would be worthwhile to repeat this study over a more extended time period. Within 24 h, we observed morphological, gene expression, and soluble factor changes; however, many were not statistically relevant. Furthermore, many of the cardiac phenotypes associated with PO and VO result from prolonged cardiovascular disease. In addition to prolonging the duration of future studies, applying higher amounts of pressure and stretch to the engineered tissues would likely elicit more dramatic changes. Faithful recapitulation of the onset of cardiovascular disease is a fundamental challenge in in vitro models due to limitations in current cell culture and tissue engineering technologies. Alternatively, combining ex vivo cardiac tissue with the hemodynamic systems utilized here could provide another avenue for investigating the acute effects of pressure and volume overload on the healthy heart. In summary, our pressure-only and stretch-only cardiac tissue chips system was valuable in investigating the roles of pressure and stretch in mediating physiological and pathophysiological cardiac remodeling.
In conclusion, we developed a unique in vitro model system that allows for the evaluation of engineered tissue constructs under conditions of isolated pressure or stretch stimuli. Using this system, we showed that engineered cardiac tissue can be acutely exposed to pressure-only and stretch-only conditions. Both conditions elicited an acute phase injury response, which was similar between the 2 experimental states. Chronic exposure to pressure-only or stretch-only conditions may recreate pathophysiologic hallmarks of PO and VO.
The authors would like to thank Dr. Tariq Hamid for kindly providing primers for gene expression experiments and Chase Newton for his help with protocol and data generation.
Statement of Ethics
All authors have followed standard ethical procedures. The study did not require the use of animal or human subjects, therefore, ethical committee approval was not required.
Conflict of Interest Statement
The authors have no conflicts of interest to declare.
This work was supported in part by the National Institutes of Health R01 under Grant R01HL14862. The work of L.D. was supported by the NIH T32 Training Grant 1T32DK116672-01, NIH T32 Training Grant 1T32EB023872-02, and F31 1F31DK127809-01. The work of C.G. was supported by the NIH T32 Training Grant T32EB023872-03.
Palaniappan Sethu directed this study and was involved with the overall study design, experiments management, and manuscript editing. Leslie Donoghue was involved in experimental design, cell cultures, setup and running flow loops, imaging, cytokine analysis, and manuscript writing. Caleb Graham assisted with RT-qPCR and editing the manuscript.
Data Availability Statement
All data generated or analyzed during this study are included in this article. Further inquiries can be directed to the corresponding author.