Matrix-Guided Control of Mitochondrial Function in Cardiac Myocytes

Abstract

In ventricular myocardium, extracellular matrix (ECM) remodeling is a hallmark of physiological and pathological growth, coincident with metabolic rewiring of cardiac myocytes. However, the direct impact of the biochemical and mechanical properties of the ECM on the metabolic function of cardiac myocytes is mostly unknown. Furthermore, understanding the impact of distinct biomaterials on cardiac myocyte metabolism is critical for engineering physiologically-relevant models of healthy and diseased myocardium. For these reasons, we systematically measured morphological and metabolic responses of neonatal rat ventricular myocytes cultured on fibronectin- or gelatin-coated polydimethylsiloxane (PDMS) of three elastic moduli and gelatin hydrogels with four elastic moduli. On all substrates, total protein content, cell morphology, and the ratio of mitochondrial DNA to nuclear DNA were preserved. Cytotoxicity was low on all substrates, although slightly higher on PDMS compared to gelatin hydrogels. We also quantified oxygen consumption rates and extracellular acidification rates using a Seahorse extracellular flux analyzer. Our data indicate that several metrics associated with baseline glycolysis and baseline and maximum mitochondrial function are enhanced when cardiac myocytes are cultured on gelatin hydrogels compared to all PDMS substrates, irrespective of substrate rigidity. These results yield new insights into how mechanical and biochemical cues provided by the ECM impact mitochondrial function in cardiac myocytes.

Graphical Abstract

1. Introduction

In native myocardium, cardiac myocytes are embedded in extracellular matrix (ECM), a compliant, porous mesh of proteins, polysaccharides, and other macromolecules. The ECM provides mechanical support, resists contraction, and activates intracellular signaling pathways via integrin receptors [1, 2]. In fetal and neonatal hearts, the ECM is composed largely of glycoproteins, such as fibronectin, that are associated with an immature network of relatively thin collagen fibers [3]. As the myocardium matures, cardiac fibroblasts become more prevalent and increase ECM production and secretion [4, 5]. As a result, the ECM becomes dominated by thicker and denser collagen fibers that form a more rigid network, reducing overall tissue compliance [5–7]. Myocardial ECM also undergoes significant remodeling in many pathological settings. For example, after an infarction, necrotic cardiac myocytes are replaced by fibroblasts and myofibroblasts that deposit collagenous scar tissue, leading to local stiffening of the myocardium [8, 9]. In parallel to these developmental and pathological changes to the extracellular matrix, cardiac myocytes undergo distinct changes in their metabolism. At birth, oxygen availability increases and metabolism becomes preferentially oxidative [10–13]. Conversely, after injuries that deprive cells of oxygen, such as myocardial infarction [14–16] or ischemia/reperfusion [17, 18], cardiac myocytes increase their reliance on glycolysis and glucose oxidative phosphorylation. Thus, in both physiological and pathological settings, cardiac myocyte metabolism is rewired by mechanisms thought to be regulated predominantly by nutrient availability. However, because the ECM is a source of both biochemical signals and mechanical resistance, the diverse changes in the ECM during physiological and pathological growth could also contribute to alterations in cardiac myocyte metabolism. However, the direct impact of the biochemical and mechanical properties of the ECM on metabolic function in cardiac myocytes is poorly understood.

Tunable biomaterials offer flexibility in recapitulating the diverse mechanical and biochemical cues present in the native ECM in a controlled in vitro setting. For example, polydimethylsiloxane (PDMS), an organic silicone-based elastomeric polymer, is widely used as a substrate for culturing cardiac myocytes due to its optical clarity and ease of fabrication. The rigidity of PDMS can be easily tuned by altering the ratio of base to curing agent [19] or by blending with other silicone polymers [20, 21]. ECM molecules can also be attached to the surface of PDMS uniformly or via microcontact printing [22–26], allowing for independent modulation of bulk mechanical and surface chemical properties. Previously, we microcontact printed tunable PDMS substrates with fibronectin to determine if mitochondrial function in cardiac myocytes is regulated by ECM elasticity and/or tissue alignment. Our results demonstrated that baseline oxygen consumption rate (OCR) is increased when ECM rigidity increases, whereas the ability of tissues to respond to stress is regulated by both tissue alignment and ECM elasticity [20]. Although PDMS is advantageous due to its tunability, it is a highly synthetic material. For this reason, hydrogels derived from ECM molecules, such as Matrigel [27] or gelatin [28, 29], are another popular culture substrate because they more closely replicate the bulk biochemical, mechanical, and architectural properties of native cardiac ECM and generally have lower toxicity compared to synthetic materials [30–32]. Previously, we reported that certain aspects of metabolism in cardiac myocytes are upregulated when cells are cultured on gelatin hydrogels compared to fibronectin-coated PDMS [28]. However, this study was limited to a single formulation of gelatin hydrogel and fibronectin-coated PDMS, providing limited insight into which physical aspect of the substrate was responsible for the differences in mitochondrial function. Fibronectin-coated polyacrylamide hydrogels have also been used to correlate ECM elasticity to mitochondrial function in cardiac myocytes [33]. Thus, several studies have demonstrated that mitochondrial function in cardiac myocytes is modulated by the ECM. However, there are still many unanswered questions related to which specific physical features of the ECM dominate distinct mitochondrial functions.

In this study, we systematically tested the impact of multiple types of substrates on mitochondrial function in cardiac myocytes. To achieve this, we coated the wells of specialized XF24 cell culture microplates with fibronectin- or gelatin-coated PDMS of three distinct elastic moduli or gelatin hydrogels with four distinct elastic moduli. Next, we cultured neonatal rat ventricular myocytes on the functionalized microplates and characterized mitochondrial function by measuring OCR and extracellular acidification rate (ECAR) using a Seahorse Biosciences XFe24 Extracellular Flux Analyzer. In most comparisons, basal respiration, basal glycolytic activity, ATP production, and maximum respiration were higher in tissues on all gelatin hydrogels compared to those on all PDMS substrates, irrespective of rigidity or PDMS coating. PDMS substrates were slightly more cytotoxic than gelatin hydrogels, which likely contributed to the reduced metabolism on PDMS substrates compared to gelatin hydrogels. Cell size and mitochondrial content was preserved across all substrates, suggesting that any differences in metabolism were not caused by an increase in mitochondrial quantity. Collectively, these results suggest that diverse cues within the ECM impact the metabolism of cardiac myocytes, which is important for understanding mechanisms of cardiac remodeling in physiological and pathological settings. These data also have implications for selecting appropriate biomaterial scaffolds for engineering physiologically-relevant cardiac tissues for in vitro modeling and drug screening.

2. Methods

2.1. PDMS and Gelatin Hydrogel Substrate Preparation

Three formulations of PDMS with distinct elastic moduli were prepared, similar to previous studies [20, 21, 34]. Sylgard 184 silicone elastomer (Dow Corning) curing agent and base were combined at a 1:10 mass ratio and mixed and degassed for two minutes each using a planetary centrifugal mixer (Thinky AR-100 Conditioning Mixer). Sylgard 527 (Dow Corning) was prepared by mixing and degassing components A and B in a 1:1 mass ratio. Sylgard 184 and Sylgard 527 were also mixed and degassed in a 1:20 mass ratio.

Four formulations of gelatin hydrogels were prepared: 5% or 10% w/v solutions of gelatin from porcine skin (175 g bloom, Sigma Aldrich) with 2% or 4% w/v transglutaminase (TG, Ajinomoto) [28]. Gelatin was dissolved in ultrapure water at 65°C until homogenous. TG powder was then added to the gelatin solution and the mixture was vortexed.

For mitochondrial respirometry studies, 10 μL of either PDMS or gelatin hydrogel prepolymer solution was pipetted into wells of XF24 cell culture microplates, similar to previous studies [28]. For LDH cytotoxicity assay and mtDNA:nucDNA quantification, 120 μL of either PDMS or gelatin hydrogels were pipetted into the wells of 12-well plates. For both types of plates, PDMS-coated wells were incubated overnight at 65°C. Wells with cured PDMS were treated in a UVO cleaner (Jelight Company Inc.) for eight minutes to sterilize and oxidize the surface followed by incubation with 100 (for XF24 plates) or 500 (for 12-well plates) μL (for XF24 plates) of human fibronectin (Corning, 50 μg/mL) or porcine gelatin (Sigma Aldrich, 200 μg/mL) for at least five minutes. Wells were rinsed with sterile PBS and stored at 4°C until cell seeding (>24 hours). Hydrogel-coated wells were cured overnight at room temperature in a vacuum desiccator. These wells were then treated in a UVO cleaner for one minute for sterilization, rinsed with sterile PBS, and stored at 4°C until cell seeding.

For measuring cell size, PDMS-coated coverslips were prepared by spin-coating 25 mm diameter glass with the 1:20 blend of Sylgard 184:527 and curing overnight at 65°C, similar to previous protocols [35]. Next, coverslips were UVO-treated for 8 minutes, coated with either fibronectin (50 μg/mL) or gelatin (200 μg/mL) solutions, transferred to 6-well plates, rinsed with sterile PBS, and stored at 4°C until cell seeding. To fabricate gelatin hydrogel-coated coverslips, 25 mm diameter glass coverslips were covered with low-adhesive tape, leaving the center exposed. Next, they were treated with sodium hydroxide, (3-aminipropyl)triethoxysilane (APTES), and glutaraldehyde to increase gelatin hydrogel adherence [28]. Flat slabs of PDMS were then sonicated for at least 30 minutes in ethanol and blown dry. 5% gelatin/4% TG hydrogel was pipetted onto the center of the coverslips and compressed with the slab of PDMS. After curing overnight at room temperature, hydrogels were re-hydrated with sterile ultrapure water, stamps were removed using tweezers, and the outer-ring of tape was removed. Coverslips were rinsed with PBS, dried, sterilized for 1 minute with UVO treatment, rinsed with sterile PBS, and stored at 4°C until cell seeding.

2.2. Bulk Compressive Elastic Modulus Measurements

Gelatin hydrogels were prepared as described above, cast into 35 mm Petri dishes (6 mL/dish), and cured overnight in a vacuum desiccator. Cylinders of 6 mm diameter were cut and removed using a biopsy punch and mounted on an Instron 5942 Mechanical Testing System (Norwood, MA). Each cylinder was compressed to 40% of its initial height. The 9–19% range of compressive strain was used for elastic modulus calculations, accounting for cylinder height and radius. To determine if hydrogel elasticity changes in culture-like conditions, as observed previously [34], we prepared hydrogels in 35 mm Petri dishes as described above, added PBS to the Petri dishes, and incubated the samples overnight in a 37°C desktop incubator. Cylinders of 6 mm diameter were cut and removed, and elastic moduli were measured using the same procedure described above. For each type of hydrogel, at least four independent batches of hydrogel were fabricated and measured in triplicate.

2.3. Neonatal Rat Ventricular Myocyte Harvest and Culture

Neonatal rat ventricular myocytes were isolated from two-day old neonatal Sprague-Dawley rats, as previously described [20, 28, 36]. Euthanization, harvesting, and cell isolation procedures were approved by the University of Southern California Institutional Animal Care and Use Committee. Ventricular tissues were extracted from rat pups, incubated in Trypsin solution (1 mg/mL, Affymetrix) at 4°C for 11–13 hours, and subjected to four 1–2 minute collagenase (1 mg/mL, Worthington Biochemical Corp, in HBSS) digestions at 37°C, each followed by manual pipette agitation to dissociate tissues into a single cell suspension. Cells were strained using a 40 μm cell strainer, resuspended in cell culture media, and pre-plated twice for 45 minutes each to minimize fibroblast contamination.

For PDMS-coated microplates, 50,000 cells/50 μL/well were seeded in each well of the XF24 Cell Culture Microplates. For gelatin hydrogel microplates, 25,000 cells/50 μL/well were seeded in each well. These seeding densities were selected to maximize the range of measurements for mitochondrial respirometry without saturating the signal, based on calibration experiments. After seeding, plates were maintained in the biosafety cabinet for 30 minutes to prevent cell aggregation at the wells’ edges, as suggested by the manufacturer. Plates were then placed in a 37°C, 5% CO₂ incubator. After 1–4 h, an additional 450 μL of media was added to each well.

For 12-well plates, 500,000 cells/well were seeded in each PDMS-coated well and 250,000 cells/well were seeded in each gelatin hydrogel-coated well. These seeding densities were chosen to match the cell coverage of the XF24 cell culture microplates. Coverslips were seeded at a density of 150,000 cells/well in a 6-well plate so that cells would be sparsely distributed, enabling individual cell imaging.

Media consisted of M199 culture medium supplemented with 10% heat-inactivated FBS, 10 mM HEPES, 0.1 mM MEM nonessential amino acids, 20 mM glucose, 2 mM L-glutamine, 1.5 μM vitamin B-12, and 50 U/mL penicillin and was exchanged after one day. After two and four days, media was exchanged again, with FBS concentration reduced to 2%.

2.4. Immunostaining, Microscopy, and Image Analysis

After five days in culture, cells on coverslips or within the XF24 cell culture microplates wells were fixed with 4% paraformaldehyde and permeabilized with 0.1% Triton X-100 solution for ten minutes. Fixed tissues were incubated with monoclonal mouse anti-sarcomeric α-actinin (Sigma, 1:1000 for plates or 1:200 for coverslips) primary antibody for one to two hours at room temperature. After PBS rinsing, samples were incubated with chemical stains DAPI (1:1000 for plates or 1:200 for coverslips), Alexa Fluor 488 Phalloidin (Life Technologies, 1:1000 for plates or 1:200 for coverslips), and Alexa Fluor 546 goat anti-mouse secondary antibody (Life Technologies, 1:1000 for plates or 1:200 for coverslips) for one to two hours at room temperature. Cells in microplates were then coated with a drop of ProLong Gold Anti-Fade Mountant, sealed with Parafilm, and stored at room temperature. Coverslips were mounted on a glass slide with a drop of ProLong Gold Anti-Fade Mountant, sealed with nail-polish, and stored at −20°C. For cells in wells, a Nikon C2 point-scanning confocal microscope with a 20x air objective, using 2x digital zoom (total magnification of 40x), was used to acquire images of at least five locations dispersed across each well. ImageJ was used to count the total number of nuclei per field of view (based on DAPI stain) [20]. For coverslips, a Nikon C2 point-scanning confocal microscope with a 60x oil (n = 1.515) objective, using 2x digital zoom (total magnification of 120x), was used to acquire z-stacks (0.25 μm thick optical sections) of at least three cells dispersed across each coverslip. ImageJ, Cell Profiler, and MATLAB were used to measure individual cell cross-section area, height and volume based on the α-actinin signal, based on previously published protocols [37].

2.5. Pierce LDH Cytotoxicity Assay

Three, four (before media change), and five days after seeding, 150 μL of media was collected from each well of a 12-well plate. In addition, on the fifth day after seeding, 100 μL of lysis solution (Thermo Fisher Scientific) was added to some wells, followed by an incubation for 30 minutes in 37°C and collection of 150 μL of media. The samples were stored at room temperature for up to three days until a Pierce LDH assay (Thermo Fisher Scientific) was performed according to manufacturer’s instructions. The absorbance values at 490 nm and 680 nm were measured using a plater reader (Varioskan Lux, Thermo Fisher Scientific) and cytotoxicity was computed according to [38]:

$$Cytotoxicity(\%)=100\times \frac{ExperimentalLDHrelease}{MaximumLDHrelease}$$

The experimental LDH release is the difference of absorbance values (490 nm – 690 nm) measured from live samples after subtracting the absorbance of media alone. The maximum LDH release is the same measurement obtained from lysed samples.

2.6. Extracellular Flux Analysis

After five days in culture, OCR and ECAR was measured using a Seahorse Bioscience XFe24 Extracellular Flux Analyzer, as previously described [20, 28, 39]. Assay medium was prepared by supplementing XF Assay Medium (Seahorse Bioscience) with 10 mM glucose, 2 mM L-glutamine, and 1 mM sodium pyruvate (pH 7.4). After acquiring baseline measurements, the following drugs were used for a mitochondrial stress test: 2 μM oligomycin (port A), 1 μM FCCP (port B), and a mixture of 1 μM antimycin A and 1 μM rotenone (port C). Mitochondrial functional metrics were computed as previously described [20]. All OCR and ECAR measurements were normalized to total protein content, determined as described below. Lastly, we computed the bioenergetic health index (BHI) from the normalized OCR values according to the following formula [20, 40]:

$$BHI=\frac{(ATPproduction)\times (Sparerespiratorycapacity)}{(Protonleak)\times (Non-mitochondrialrespiration)}$$

2.7. Measurements of Protein Concentration

Tissues from duplicate plates or plates after OCR measurements were rinsed twice with PBS and incubated with 100 μL of radioimmunoprecipitation assay (RIPA) buffer and 400 μL of PBS to lyse the tissues. Protein lysates were stored at −20°C. BCA protein assay was performed per manufacturer instructions. Absorbance was measured using a plater reader (Varioskan Lux, Thermo Fisher Scientific) and values were fit to a BSA protein standard curve. Average total protein value per harvest for each condition was used to normalize the OCR and ECAR values.

2.8. mtDNA:nucDNA Quantification

After five days in culture, samples in 12-well plates were lysed and lysates were stored at −80°C until measurements. DNA was then isolated using a DNeasy Blood and Tissue kit (Qiagen). 260/280 ratios were measured using a Nanodrop (Thermo Fisher Scientific) to verify sample purity (all samples had ratios greater than 1.9). DNA samples were then diluted to 15 ng/μL. qPCR was performed by mixing SsoAdvanced Universal SYBR Green Supermix (Bio-Rad), DNA, and primers (Integrated DNA Technologies) into a 384-well PCR plate. A CFX384 Touch Real-Time PCR Detection System (Bio-Rad) was used to obtain the cycle threshold (Ct) values for each condition. PCR conditions were: 95°C for 10 minutes, followed by 40 cycles of 95°C for 15 sec, 55°C for 15 sec, 72°C for 20 sec. Housekeeping nuclear target was ApoB: forward 5’– CACGTGGGCTCCAGCATT – 3’, reverse 5’ – TCACCAGTCATTTCTGCCTTTG – 3’. Mitochondrial target was 235bp: forward 5’ – CCTCCCATTCATTATCGCCGCCCTTGC – 3’, reverse 5’ –GTCTGGGTCTCCTAGTAGGTCTGGGAA – 3’. mtDNA:nucDNA ratio was quantified using the formula [41, 42]:

$$mtDNA:nucDNA=2\times 2^{\mathrm{\Delta }Ct}$$

In which ΔCt is the difference between Ct values for housekeeping gene ApoB and the 235bp mitochondrial gene.

2.9. Statistical Analysis

Normality for all measurements was first validated using the Kolmogorov-Smirnov Test. Elastic moduli data were analyzed using a paired two sample student’s t-test, with α set to 0.05. The remaining data were analyzed using one-way and/or two-way ANOVA followed by Tukey’s test for multiple comparisons in MATLAB, with α set to 0.05. Each biological parameter was measured using cells from at least three independent harvests, and multiple wells per harvest per condition were used. Data is displayed as box plots, where the solid line indicates the median value and the bottom and top edges of the boxes indicate the 25th and 75th percentiles, respectively. Whiskers extend to the most extreme points not considered outliers, which are represented by crosses.

3. Results

3.1. Characterization and Preparation of Substrates

To determine the impact of different substrates on mitochondrial function in cardiac myocytes, we first coated the wells of XF24 cell culture microplates with three different blends of PDMS, as shown in Figure 1A. As reported previously, these three PDMS blends have elastic moduli of 1.6 kPa, 27 kPa, and 2.7 MPa [20], a range that corresponds to developing myocardium, healthy adult myocardium [6, 7], and the mechanical load applied to cardiac myocytes in pathological conditions [43]. These PDMS blends were coated with fibronectin or gelatin solution, which we will refer to as PDMS-fibronectin or PDMS-gelatin, respectively. These conditions were selected to roughly mimic developing myocardium, which is richer in fibronectin and other glycoproteins [3], and healthy [3] or fibrotic [44] myocardium, which is richer in collagen and its derivatives. We also coated microplate wells with four gelatin hydrogel formulations: 5% or 10% gelatin with 2% or 4% TG [28, 34], which we will refer to as 5%/2%, 5%/4%, 10%/2%, and 10%/4%. Because the elastic moduli of gelatin hydrogels can change in culture-like conditions [34], we fabricated gelatin hydrogels and measured their elastic moduli before and after overnight incubation in PBS at 37°C. As shown in Figure 1B, the elastic modulus increased for each condition by 5–15 kPa after overnight incubation in culture-like conditions, although this was not statistically significant for 10%/2%. Elastic modulus also increased as gelatin or TG concentration increased (5%/2%: 17 ± 2 kPa, 5%/4%: 27 ± 3 kPa, 10%/2%: 58 ± 6 kPa, and 10%/4%: 72 ± 4 kPa, all values after incubation in culture-like conditions), similar to previous measurements [28].

Figure 1: Characterization and Fabrication of Biomaterial Substrates.

A: XF24 cell culture microplates were coated with either 10 μL of PDMS (top row) or gelatin hydrogel (bottom row) solution (i). After overnight curing, the plates were treated with UVO (ii). PDMS-coated wells were coated with 100 μL of ECM protein solution (iii, top row). Both types of wells were rinsed with sterile PBS (iv, top row, iii, bottom row) and seeded with neonatal rat ventricular myocytes (v, top row, iv, bottom row). B: Average elastic moduli for the four different formulations of gelatin hydrogels, before and after incubation in culture-like conditions. Values are means ± SE; n = 4 for all conditions. * p<0.05. Refer to Table S1 for details related to statistical analysis.

3.2. Substrate Effects on Cardiac Myocyte Adhesion, Viability, and Morphology

To evaluate the effects of each substrate on cardiac myocyte adhesion and viability, we cultured neonatal rat ventricular myocytes within the wells of our modified microplates and immunostained tissues after five days. On all gelatin hydrogel (Figure 2A) and PDMS (Figure 2B), myocytes self-assembled into confluent, isotropic tissues. Tissues on all substrates were composed primarily of cardiac myocytes, as validated by positive sarcomeric α-actinin staining in the majority of cells. To quantify cell density, we counted the number of nuclei in our stained tissues (Figure 3A). Although we did not detect any statistical differences within PDMS substrates or gelatin hydrogels, select hydrogels had statistically lower cell density compared to select PDMS substrates. This is expected because gelatin hydrogels were seeded with half as many cells, which we chose based on the results of OCR calibration experiments. However, the final number of nuclei on PDMS was less than twice that measured for gelatin hydrogels, suggesting that myocyte adhesion and/or viability is higher on gelatin hydrogels. To further compare tissue composition, we quantified total protein content per well. As shown in Figure 3B, tissues on 72 kPa gelatin hydrogels and 1.6 kPa PDMS-fibronectin had higher protein content than those on 27 kPa PDMS-fibronectin and 2.7 MPa PDMS-gelatin. All other conditions were not statistically different. Thus, tissues on most substrates had similar protein content, although cell density was lower on select gelatin hydrogels compared to select PDMS substrates.

Figure 2: Cardiac tissues cultured on PDMS substrates within XF24 Cell Culture Microplates.

Composite images of neonatal rat ventricular myocyte tissues cultured on indicated (A) gelatin hydrogels or (B) PDMS. Blue: nuclei; white: α-actinin; scale bars: 25 μm.

Figure 3: Cell adhesion and viability.

(A) Nuclei per 0.1 mm2 and (B) total protein content per well in cell culture microplates. n indicated below each box. Cytotoxicity on day 4 (C) and day 5 (D) after seeding. n = 8 for all conditions. Letters above each box indicate a statistical difference (p<0.05) with the condition represented by the letter on the x-axis. For example, “a” indicates p<0.05 compared to gelatin hydrogel, 17 kPa. Refer to Tables S2–S4 for details related to statistical analysis.

Previous studies have reported that PDMS has some cytotoxic effects due to the leaching of uncured oligomers [30–32]. To account for this, we performed a series of LDH cytotoxicity assays for tissues on each substrate. On the third day post-seeding (Figure S1), cytotoxicity was, on average, lower than 2% on all substrates. The only statistical difference at this timepoint was between tissues on 17 kPa gelatin hydrogel and 2.7 MPa PDMS-fibronectin. On the fourth day post-seeding (Figure 3C), cytotoxicity measurements were higher compared to the third day due to the accumulation of LDH over this two-day period without media exchange. At this timepoint, cytotoxicity in tissues on all gelatin hydrogels were significantly lower than 2.7 MPa PDMS-gelatin. Additionally, cytotoxicity in tissues on 17 kPa and 58 kPa gelatin hydrogels was significantly lower than those on 27 kPa and 2.7 MPa PDMS-fibronectin. However, average cytotoxicity values were still below 5% on all substrates. On the next day (Figure 3D), which is the same timepoint when mitochondrial respirometry studies were performed, average cytotoxicity was below 2% in all conditions. These values are lower than the previous day due to media exchange. On this day, cytotoxicity was statistically similar between all conditions, except for lower values on gelatin hydrogels and 27 kPa PDMS-gelatin compared to 1.6 kPa PDMS-fibronectin. Thus, although cytotoxicity on all substrates was relatively low (<5% in most cases), PDMS substrates did have a more cytotoxic effect compared to gelatin hydrogels in many comparisons. This effect likely contributed to the lower number of nuclei detected on PDMS substrates relative to the initial seeding density compared to gelatin hydrogels.

Because of the observed substrate-dependent differences in cell density and protein content, we also measured differences in cardiac myocyte size to determine if cells were undergoing hypertrophy on gelatin hydrogels. To quantify and compare myocyte size, we sparsely seeded myocytes on coverslips with 27 kPa gelatin hydrogels, 27 kPa PDMS-fibronectin, and 27 kPa PDMS-gelatin (Figure 4A) to control for elastic modulus while altering the other properties of the substrates. We then immunostained cells for sarcomeric α-actinin, collected 3-D image stacks using confocal microscopy, and computed cell area, height, and volume (Figure 4B–D). No statistical differences were observed across the conditions for any of the parameters, indicating that cell size was mostly preserved across substrates. Although not statistically different, cardiac myocytes on gelatin hydrogels had a slightly higher area and lower height compared to PDMS, suggestive of more cell spreading. This subtle difference in cell spreading likely also contributed to the observation that nuclei number is lower, but protein content is equal, on gelatin hydrogels compared to PDMS substrates.

Figure 4: Cell geometry.

(A) Composite images of single cardiac myocytes seeded on coverslips. Blue: nuclei; white: α-actinin; scale bars: 10 μm. Myocyte area (B), height (C), and volume (D). n indicated below the boxes on (C). Refer to Table S5 for details related to statistical analysis.

3.3. Regulation of Mitochondrial Function by Biomaterial Substrate

To characterize mitochondrial function, we performed a standard mitochondrial stress test using a Seahorse Extracellular Flux Analyzer on tissues cultured on each substrate. In this assay, basal OCR is first measured. Then, OCR is measured after the addition of oligomycin, FCCP, and antimycin/rotenone in series to analyze specific mitochondrial and non-mitochondrial functions [20, 28, 39], as shown in Figure 5. The Seahorse Extracellular Flux Analyzer also simultaneously measures ECAR, which reflects glycolytic activity (Figures S2). All OCR and ECAR measurements were normalized to total protein content. Based on a one-way ANOVA and multiple comparisons, basal respiration was higher for tissues on 17 kPa, 27 kPa, and 58 kPa gelatin hydrogels compared to all PDMS substrates, while tissues on 72 kPa gelatin hydrogels had higher basal respiration than select PDMS substrates (Figure 6A, Table S6). Tissues on a few PDMS substrates (27 kPa PDMS-fibronectin, 1.6 kPa PDMS-gelatin, and 27 kPa PDMS-gelatin) also had higher basal respiration than those on 2.7 MPa PDMS-gelatin. Basal glycolytic activity was higher for 17 kPa, 27 kPa and 58 kPa gelatin hydrogels, 27 kPa PDMS-fibronectin, and 1.6 kPa PDMS-gelatin compared to 2.7 MPa PDMS-fibronectin. Additionally, ECAR was significantly lower for tissues on 2.7 MPa PDMS-gelatin compared to all other conditions, except for 2.7 MPa PDMS-fibronectin. To further compare these responses, we plotted basal ECAR as a function of basal OCR (Figure 6C). In this energy plot, tissues on gelatin hydrogels clustered in the energetic quadrant due to their high OCR and ECAR. Tissues on both 2.7 MPa PDMS conditions were in the quiescent quadrant due to their low OCR and ECAR. Tissues on 1.6 kPa and 27 kPa PDMS were in the glycolytic quadrant due to their relatively high ECAR but low OCR. Together, these data indicate that the metabolism of tissues on gelatin hydrogels is more energetic in terms of both mitochondrial respiration and glycolysis compared to those on PDMS substrates, especially those on the most rigid PDMS.

Figure 5: OCR measurements in cardiac tissues.

Average experimental OCR measurements for tissues on gelatin hydrogels (A), PDMS-fibronectin (B), and PDMS-gelatin (C) at baseline and after addition of oligomycin, FCCP, and antimycin and rotenone. Data are presented as mean v± s.e.m., n = 16 for all conditions.

Figure 6: Baseline oxidative and glycolytic activity.

Average OCR (A) and ECAR (B) associated with basal respiration. n = 16 for all conditions. Letters above each box indicate a statistical difference (p<0.05) with the condition represented by the letter on the x-axis. For example, “a” indicates p<0.05 compared to gelatin hydrogel, 17 kPa. (C) Energy map at baseline for cardiac tissues culture on the different substrates, colors correspond to the bars in (A) and (B). Values are presented as mean ± s.e.m., n = 16 for all. For details related to statistical analysis, refer to Tables S6–S7.

For the remainder of the mitochondrial stress test measurements, we focused only on the OCR measurements because the added compounds are mitochondrial inhibitors and thus less relevant for glycolysis. On all gelatin hydrogels, tissues had higher mitochondrial ATP production than those on all PDMS substrates (Figure 7A). Additionally, tissues on 27 kPa PDMS-fibronectin had higher mitochondrial ATP production than those on 2.7 MPa PDMS-gelatin. Proton leak was mostly preserved across all conditions (Figure 7B). The only statistical differences were a higher proton leak for tissues on 17 kPa gelatin hydrogels and 1.6 kPa PDMS-gelatin compared to those on 2.7 MPa PDMS-gelatin. Non-mitochondrial respiration was higher for select tissues on gelatin hydrogels compared to select PDMS substrates (Figure 7C). Thus, in most cases, basal respiration, ATP production, proton leak, and non-mitochondrial respiration were similar within tissues on gelatin hydrogels and within tissues on PDMS substrates. However, in many comparisons, tissues on gelatin hydrogels had higher values for baseline metabolic metrics compared to PDMS substrates, irrespective of substrate elasticity or ECM protein composition.

Figure 7: Baseline metabolic functions.

Average OCR associated with (A) ATP production, (B) proton leak, and (C) non-mitochondrial respiration. n = 16 for all conditions. Letters above each box indicate a statistical difference (p<0.05) with the condition represented by the letter on the x-axis. For example, “a” indicates p<0.05 compared to gelatin hydrogel, 17 kPa. For details related to statistical analysis, refer to Table S8–S10.

Next, we compared the response of tissues to metabolic stress. Maximum respiration followed a similar trend to basal respiration (Figure 8A). Specifically, tissues on the three softer gelatin hydrogels had higher OCR values than those on all PDMS substrates, whereas those on 72 kPa gelatin hydrogels were higher than select PDMS substrates. Because tissues on gelatin hydrogels generally had higher basal and maximum respiration, spare respiratory capacity was mostly preserved across the different conditions (Figure 8B). The only statistical differences were an increase in spare respiratory capacity in tissues on 17 kPa, 27 kPa, and 58 kPa gelatin hydrogels compared to those on 27 kPa PDMS-fibronectin, 27 kPa PDMS-gelatin, and 2.7 MPa PDMS-gelatin. Together, these data indicate that tissues on gelatin hydrogels generally have higher maximum respiration compared to those on all PDMS substrates, but spare respiratory capacity is mostly conserved due to parallel increases in basal and maximum respiration.

Figure 8: Metabolic stress responses.

Average OCR associated with (A) maximum respiration and (B) spare respiratory capacity. n = 16 for all conditions. Letters above each box indicate a statistical difference (p<0.05) with the condition represented by the same letter on the x-axis. For example, “a” indicates p<0.05 compared to gelatin hydrogel, 17 kPa. For details related to statistical analysis, refer to Tables S11–S12.

To determine if gelatin and/or TG concentrations have an independent effect on mitochondrial function, we next performed a two-way ANOVA analysis to determine if any OCR and ECAR measurements are regulated by one or both of these variables (Table 1). Based on these results, basal respiration, ATP production, proton leak, and maximum respiration were higher for tissues on 2% TG compared to 4% TG, whereas spare respiratory capacity, non-mitochondrial respiration, and basal glycolytic activity were preserved across the conditions (see Table 1). Gelatin concentration did not have a significant impact on any OCR or ECAR values. Thus, for the ranges we tested, TG concentration, but not gelatin concentration, impacted some metabolic parameters.

Table 1: Two-way ANOVA analysis of mitochondrial respirometry data for tissues on gelatin hydrogels.

Data for all conditions was normally distributed, as determined by the Kolmogorov-Smirnov test. p-values for each comparison are indicated, with * indicating p<0.05. n = 16 for all conditions.

	Comparison
Parameter	Gelatin Concentration	TG Concentration	Interaction
Basal Respiration	0.3454	0.0135 *	0.5784
ATP Production	0.6247	0.0434 *	0.7316
Proton Leak	0.1124	0.0322 *	0.4337
Maximum Respiration	0.0778	0.0240 *	0.2631
Spare Respiratory Capacity	0.0846	0.1378	0.2642
Non-mitochondrial Respiration	0.9539	0.3350	0.8474
BHI	0.6922	0.6009	0.6099
Basal Glycolytic Activity	0.0874	0.1931	0.8567

We also performed a similar two-way ANOVA analysis to investigate the independent effects of ECM elasticity and ECM protein ligand on mitochondrial function for tissues on PDMS substrates (Table 2). Based on these results, basal respiration, ATP production, proton leak, and basal glycolytic activity were regulated by ECM elasticity, but not ECM protein ligand. Specifically, basal respiration and basal glycolytic activity were higher in tissues on 1.6 kPa and 27 kPa PDMS compared to those on 2.7 MPa PDMS. ATP production was higher in tissues on 27 kPa compared to those on 2.7 MPa PDMS. Lastly, proton leak was higher in tissues on 1.6 kPa compared to those on 2.7 MPa PDMS.

Table 2: Two-way ANOVA analysis of mitochondrial respirometry data for tissues on PDMS substrates.

Data for all conditions was normally distributed, as determined by the Kolmogorov-Smirnov test. p-values for each comparison are indicated, with * indicating p<0.05. n = 16 for all conditions.

	Comparison
Parameter	PDMS Elasticity	Coating Protein	Interaction
Basal Respiration	0.0005 *	0.5510	0.0098 *
ATP Production	0.0065 *	0.4992	0.0381 *
Proton Leak	0.0052 *	0.9208	0.2948
Maximum Respiration	0.1367	0.2010	0.4865
Spare Respiratory Capacity	0.1188	0.2528	0.8952
Non-mitochondrial Respiration	0.2509	0.7911	0.4523
BHI	0.5137	0.0392 *	0.3072
Basal Glycolytic Activity	1.1056e-08*	0.2059	0.0023*

Next, we computed the BHI for tissues based on OCR measurements [20, 40] and compared values using a one-way ANOVA analysis (Figure 9A). The BHI combines positive and negative aspects of oxygen consumption into a single overall parameter by dividing the product of spare respiratory capacity and ATP production by the product of proton-leak and non-mitochondrial respiration. BHI was significantly higher for tissues on all gelatin hydrogels compared to those on all PDMS-gelatin substrates and 27 kPa PDMS-fibronectin. Additionally, tissues on 17 kPa hydrogels had higher BHI than tissues cultured on all PDMS-fibronectin substrates. No statistical differences were observed within hydrogels or within PDMS substrates. Two-way ANOVA analysis did not indicate independent regulation of BHI by gelatin or TG concentration for tissues on gelatin hydrogels (Table 1). Two-way ANOVA analysis also indicated that ECM ligand, but not PDMS elasticity, regulates BHI, with tissues on PDMS-fibronectin substrates having higher BHI than those on PDMS-gelatin. Collectively, these data indicate that BHI in cardiac tissues is regulated by multiple biomaterial properties, with tissues on gelatin hydrogels demonstrating the highest BHI values, followed by those on PDMS-fibronectin and finally PDMS-gelatin.

Figure 9: Bioenergetic Health Index and quantity of mitochondria.

(A) Average Bioenergetic Health Index for all conditions. n = 16 for all conditions. (B) Mitochondrial DNA to nuclear DNA copy number ratio. n indicated below each box. Letters above each box indicate a statistical difference (p<0.05) with the condition represented by the same letter on the x-axis. For example, “a” indicates p<0.05 compared to gelatin hydrogel, 17 kPa. For details related to statistical analysis, refer to Table S13–S14.

Because we observed substrate-dependent differences in OCR and ECAR, we next investigated if the quantity of mitochondria per cell was regulated by the substrate. To compare mitochondrial content, we quantified the ratio of mitochondrial DNA to nuclear DNA copy number (mtDNA:nucDNA). As seen in Figure 9B, there were no statistical differences in mtDNA:nucDNA across the conditions. Thus, the differences in metabolism that we observed were not caused by mitochondrial biogenesis, suggesting that other mechanisms, such as difference in mitochondrial efficiency, underlie the observed differences in OCR.

4. Discussion

Myocardial tissue remodeling during both development and disease is characterized by distinct changes in ECM composition and rigidity, as well as rewiring of cardiac myocyte metabolism. However, the direct impact of the ECM on cardiac myocyte metabolism is unclear. Additionally, the types of biomaterials used to model different stages of myocardial growth in vitro might have direct impacts on mitochondrial function that are important to characterize and consider for engineered tissue models. To address these questions, we cultured cardiac myocytes on a variety of gelatin hydrogel and PDMS substrates and evaluated their mitochondrial function with a Seahorse XFe24 Extracellular Flux Analyzer. Our results suggest that several metabolic parameters, including basal respiration, ATP production, and maximum respiration, are generally higher on gelatin hydrogels compared to PDMS-gelatin and PDMS-fibronectin substrates, independent of substrate rigidity. Thus, mitochondrial function in cardiac myocytes is sensitive to the biomaterial culture substrate, indicative of ECM regulation of metabolism.

For this study, one category of biomaterials that we fabricated were gelatin hydrogels cross-linked with TG for thermostability [45, 46]. Similar to previous studies, the elastic moduli of these hydrogels increased with increasing gelatin and/or TG concentration [28]. Importantly, we found that elastic moduli continued to increase during overnight incubation in culture-like conditions. Thus, the elastic moduli experienced by the cells was higher than that of the hydrogels as initially fabricated. Ultimately, gelatin hydrogels demonstrated a range of elastic moduli that mimic passive measurements of healthy adult myocardial tissue (17 kPa, 27 kPa) [6, 7] to fibrotic myocardial tissue (58 kPa, 72 kPa) [3, 5, 43]. One limitation of gelatin hydrogels is their limited range of elasticity. Fabricating gelatin hydrogels with substantially higher rigidity than these values is problematic due to the limits of gelatin solubility, while fabricating gelatin hydrogels with substantially lower rigidity is problematic due to the fragility of the hydrogel. This could potentially be overcome by adding other stabilizing proteins, such as tropoelastin [47]. Conversely, PDMS substrates, which are the second category of biomaterial that we fabricated, offer a much wider range of elastic moduli, spanning multiple orders of magnitude. By blending Sylgard 184 and Sylgard 527, we achieved a range of 1.6 kPa to 2.7 MPa, as reported in previous studies [20, 21, 48]. Depending on the research question, an additional benefit of PDMS is that it can be easily coated with specific ECM proteins to independently modulate the mechanical and biochemical properties of the substrate. However, the ECM protein forms a relatively thin layer on the surface, which can contribute to delamination over long-term culture due to the hydrophobic, non-fouling nature of the underlying PDMS [49, 50]. In contrast, gelatin hydrogels are more permissive for long-term culture, likely because the entire bulk of the substrate is comprised of cell-adhesive ECM proteins [28, 34]. In addition, we found in this study that PDMS substrates are slightly cytotoxic compared to gelatin hydrogels, another drawback of PDMS. Thus, gelatin hydrogels and ECM-coated PDMS substrates have distinct advantages and disadvantages in terms of fabrication, mechanical properties, and long-term biocompatibility. Our results here indicate that these biomaterial substrates also have unique impacts on the metabolic function of cardiac myocytes. These are important factors to weigh when selecting biomaterial substrates for engineering in vitro models of the myocardium or other muscle tissues.

To measure biomaterial-mediated differences in metabolic function, we coated XF24 cell culture microplate wells with gelatin hydrogels, PDMS-fibronectin, or PDMS-gelatin. On all substrates, neonatal rat ventricular myocytes attached and formed confluent tissues. Interestingly, although we seeded PDMS-coated wells with twice as many cells, we did not observe twice as many nuclei per field of view on PDMS substrates compared to gelatin hydrogels. This is likely because myocytes initially adhered more strongly to gelatin hydrogels and/or experienced more delamination or apoptosis on PDMS, consistent with the higher level of cytotoxicity that we measured on PDMS. We also found that tissues on all substrates had similar protein content. By very rough calculation (average protein content/average cell density), cardiac myocytes on gelatin hydrogels contained 2.1 ng of protein per cell while those on PDMS substrates contained 1.3 ng of protein per cell. These values are within the order of magnitude of more direct measurements of cell mass [51]. This increase in total protein content per cell could indicate that myocytes on gelatin hydrogels underwent hypertrophy. However, our analysis of the area and volume of isolated cardiac myocytes on gelatin hydrogels, PDMS-gelatin, and PDMS-fibronectin with preserved elastic modulus (27 kPa) revealed that cell volume was consistent across these substrates. Although not statistically significant, cells on gelatin hydrogels had a slightly higher area and lower height than those on PDMS, suggestive of more cell spreading on gelatin hydrogels. This phenotype could partially explain why protein content was similar, but cell number was lower, on gelatin hydrogels. Additionally, we did not quantify the quantity, length, etc. of sarcomeres or myofibrils, which could alter protein content, force generation [52, 53], and potentially metabolism by increasing energetic demands and/or rearranging mitochondria. For example, in our previous study, we found that maximum respiration was regulated by tissue alignment [20], which increases myofibril length [23, 35]. Our tissues also have a low level of fibroblast contamination. The adhesion or proliferation of fibroblasts could also be regulated by the substrate, but their contribution was neglected in this study. Overall, a more detailed analysis of proteins related to the cytoskeleton and other intercellular structures in cardiac myocytes could explain these differences in protein level per cell, which could contribute to the differences in mitochondrial function we observed.

Our OCR data revealed new relationships between mitochondrial function and biomaterial substrate. In most comparisons, basal respiration, ATP production, and maximum respiration were higher in tissues on gelatin hydrogels compared to those on PDMS-fibronectin and PDMS-gelatin substrates. Although cytotoxicity was slightly higher on PDMS, average cytotoxicity was below 4% for all conditions and thus cytotoxicity alone cannot explain these differences in metabolism. Furthermore, mtDNA:nucDNA was preserved across all conditions. Together, these results suggest that mitochondrial activity is generally higher on gelatin hydrogels. Because tissues on gelatin hydrogels exhibited higher basal and maximum respiration, spare respiratory capacity, which reflects respiratory capacity that is utilized only in times of increased metabolic demand, was mostly conserved across all substrates. Proton leak, a metric that can indicate mitochondrial damage [54, 55], was also preserved across all conditions. On average, non-mitochondrial respiration was higher for tissues on gelatin hydrogels compared to those on PDMS substrates. This could indicate increased rates of oxidative activity caused by the production of reactive oxygen species (ROS) by nitric oxide synthases or NADPH oxidases in the mitochondria [56, 57]. Additionally, higher values of non-mitochondrial respiration could represent upregulated enzymatic activities of many oxygenases, such as Heme oxygenase-1 [58, 59], or increased oxygen use in late stages of protein folding [60]. Quantifying ROS production as a function of biomaterial substrate could provide further insight into this relationship, especially considering that muscles with robust mitochondrial output have to develop equally robust systems to deal with oxidative stress [61, 62] and misfolding of proteins [63, 64]. The Bioenergetic Health Index (BHI), a parameter that combines positive and negative metrics of mitochondrial function into a single number to indicate overall mitochondrial health [40], was also significantly higher for tissues on gelatin hydrogels compared to all tissues on PDMS-gelatin and many tissues on PDMS-fibronectin. Collectively, these data indicate that gelatin hydrogels are, at least in the short-term, beneficial to overall mitochondrial function compared to PDMS-gelatin and PDMS-fibronectin substrates.

Similar to the OCR data, the basal ECAR values indicated that tissues on gelatin hydrogels were more metabolically active in glycolysis compared to those on PDMS substrates. The energetic map (OCR versus ECAR) generated for our tissues roughly clustered the tissues into three groups: tissues on hydrogels in the energetic quadrant, tissues on softer PDMS (1.6 kPa and 27 kPa) in the glycolytic quadrant, and tissues on 2.7 MPa PDMS in the quiescent quadrant. These results suggest that substrates with high elastic moduli reduce the metabolic activity of cardiac myocytes, which is in-line with other studies showing that stiff substrates reduce the ability of cardiac myocytes to shorten and contract [7, 36, 65, 66]. The high OCR and ECAR values measured from tissues on gelatin hydrogels compared to PDMS suggests that both oxidative and glycolytic activity is relatively high on gelatin hydrogels, which may explain why tissues on gelatin hydrogels maintain their contractility longer than on PDMS, as previously described [28].Among tissues on gelatin hydrogels, we did not observe any statistical differences in any metabolic metric based on elastic modulus. However, our two-way ANOVA analysis indicated that tissues on hydrogels with 4% TG had lower basal respiration, ATP production, protein leak, and maximum respiration compared to tissues on hydrogels with 2% TG. These differences cannot be explained solely by the increase in hydrogel rigidity due to TG because elastic modulus increased in this order: 5%/2%, 5%/4%, 10%/2%, 10%/4%. Thus, these differences are likely linked to TG-mediated changes in surface chemistry, porosity, toxicity, and/or other mechanisms that have yet to be established, which could limit the use of this platform for some applications. Alternative fabrication methods, such as UV-crosslinking of methacrylated gelatin [29], could potentially ameliorate any concerns related to TG toxicity.

Within tissues on PDMS substrates, we did not observe a clear relationship between ECM rigidity and OCRs associated with basal respiration and ATP production, as reported in our previous study with cardiac tissues cultured on micropatterned PDMS discs [20]. A possible explanation is related to differences in cell distribution on the substrates. In our previous study, we fabricated PDMS discs by first spin-coating, curing, and laser-engraving thin layers of PDMS. We then microcontact printed the PDMS and transferred the discs to the XF24 microplate wells. In the present study, uncured PDMS was pipetted directly into the bottom of the wells, cured, and coated with ECM proteins. Due to surface tension and interactions with the walls of the microplate, a meniscus formed in the PDMS. This non-uniform surface profile likely resulted in some subtle variation in the mechanical properties of the substrates as well as layering of cardiac myocytes. These confounding factors could have reduced the sensitivity of our mitochondrial respirometry measurements, which were more controlled in our previous report, leading to clearer trends in basal respiration. However, similar to our previous study [20] and others [33], spare respiratory capacity and non-mitochondrial respiration was independent of ECM elasticity for tissues on substrates with uniform coating of ECM protein.

Our data for tissues on PDMS-gelatin and PDMS-fibronectin demonstrate that, in some conditions, both ECM elasticity and ECM ligand can impact cardiac metabolism, which has implications for understanding physiological and pathological cardiac growth. As described above, ECM stiffening and increasing collagen fiber density are hallmarks of both cardiac development [3–7] and disease [8, 9]. Our softest PDMS substrate (1.6 kPa) resembles the rigidity of a developing heart, whereas our intermediate PDMS substrate (27 kPa) is closest in rigidity to an adult heart [6, 7]. For all OCR measurements, tissues on PDMS-gelatin and PDMS-fibronectin did not present any significant differences between the lowest and moderate rigidities, indicating that neither ECM ligand nor PDMS rigidity impacted mitochondrial function for these conditions. This is likely because neonatal rat cardiac myocytes express integrin receptors for both proteins (fibronectin and collagen) [67] and thus the cells could adhere and maintain a similar phenotype, irrespective of ECM ligand, on both the softest and intermediate PDMS.

Our stiffest PDMS substrate (2.7 MPa) roughly mimics the increase in mechanical load experienced by cardiac myocytes in certain pathological conditions, such as pressure overload [68, 69] or fibrosis [43, 70]. Interestingly, we observed in several instances that tissues on 2.7 MPa PDMS-gelatin, but not 2.7 MPa PDMS-fibronectin, had lower OCR and ECAR values compared to tissues on other PDMS substrates. This indicates that metabolic activity was compromised in tissues on 2.7 MPa PDMS-gelatin, but not 2.7 MPa PDMS-fibronectin, suggesting that the combination of a stiff ECM and collagen-based ligand could be deleterious to mitochondrial health. Hence, the biochemical cues provided by the ECM ligand could impact mitochondrial function when ECM stiffness is elevated, as in diseased tissues. Similar ECM ligand-mediated changes in cardiac myocyte function have previously been observed. For example, the viability of HL-1 cardiac myocytes is higher on glass coverslips coated with fibronectin compared to gelatin and collagen [71]. Additionally, neonatal rat cardiac myocytes express higher levels of Cx43 and N-cadherin on fibronectin compared to collagen [72]. In this same study, the expression of Cx43 and N-cadherin increased due to cyclic stretch only for myocytes on collagen, but not fibronectin. The electrophysiological maturation of cardiac myocytes differentiated from human induced pluripotent stem cells (iPSCs) has also been shown to be regulated by both substrate type (glass, PDMS) and coating (fibronectin, Matrigel) [73]. Together, these studies are indicative of ECM ligand-mediated changes in mechanosensitivity, similar to our results reported here.

Our study has some limitations that are important to note. First, we measured OCRs from isotropic cardiac tissues, which is more characteristic of diseased [74] or immature [75] cardiac tissues. As shown by our previous study, tissue alignment can impact mitochondrial function [20], but this was not tested in the current study. We also used neonatal rat cardiac myocytes, the gold standard for in vitro cardiac myocyte research due to their plasticity and longevity in culture. However, these cells are non-human and relatively immature. The immaturity of these cells likely impacts our results because integrin expression profiles are known to change with age. For example, studies have shown that neonatal rat cardiac myocytes can adhere to fibronectin, laminin, and collagen types I, II, III, IV, and V [67, 76]. Adult rat cardiac myocytes, however, adhere only to laminin and collagen type IV [67, 76], indicating that neonatal and adult cardiac myocyte express distinct integrin receptors. For these reasons, we would expect slightly different results from adult cardiac myocytes. In our experiments, we also minimized supporting cell populations, although cardiac myocytes in the heart are surrounded by fibroblasts, endothelial cells, and other supporting cell types. All of these features limit the translation of our results to cardiac myocytes in adult human myocardium. To increase human relevance, future work will focus on cardiac myocytes differentiated from human iPSCs. However, these cells are known to express immature phenotypes [77, 78], which could also limit translation. Another limitation is that the relatively short timescale of our experiments captured only the early responses of cardiac myocytes to ECM cues and thus do not reflect any long-term remodeling. However, long-term studies are difficult to perform in cardiac myocytes on PDMS-coated substrates due to cell delamination [28], limiting our experimental time frame. Furthermore, our OCR measurements were performed within intact cardiac tissues instead of isolated mitochondria. Although we verified that mitochondrial quantity per cell was preserved across the conditions, it is not clear if our observed results were caused by changes in mitochondrial morphology and/or mitochondrial structure within the cardiac myocytes [79, 80]. In the present study, we tested the impact of isolated proteins, specifically fibronectin and gelatin (denatured collagen). However, in the native heart, the ECM is comprised of proteins and macromolecules, including hyaluronans. Hyaluronans have also been shown to mediate cell adhesion and signaling in cardiac myocytes [81, 82], but these and other ECM molecules were excluded in this study. Lastly, we did not replicate the different nutrient sources that cardiac myocytes are exposed to in vivo. Our media was rich in glucose and amino acids, but lacked the fatty acids that are known to be the primary energy source for healthy adult cardiac myocytes in vivo [15, 83]. Thus, future studies will focus on determining how mitochondrial function is altered by the ECM in more human- and physiologically-relevant cardiac tissues in more native-like biochemical environments.

Due to recent breakthroughs in generating iPSC-derived cardiac myocytes from patient somatic cells [84, 85], there is now significant interest in engineering functional, micro-scale tissue constructs from iPSC-derived cardiac myocytes for personalized disease modeling and medium-throughput drug screening [86–88]. However, the success of these “Heart on a Chip” platforms is highly dependent on the long-term survival and health of the cardiac myocytes, especially for modeling slowly progressing diseases or screening the chronic cardiotoxic effects of drugs, including common chemotherapies such as doxorubicin [89]. Because contractility imposes high energetic demands [90], ensuring the metabolic health of cardiac myocytes in culture is critical for maintaining a stable, contractile phenotype. Here, our results indicate that gelatin hydrogels enhance the metabolic activity of cardiac myocytes compared to a variety of PDMS substrates, complementing our earlier report that gelatin hydrogels extend the culture lifetime of cardiac myocytes [28] and skeletal myotubes [34] compared to PDMS substrates. Previous studies from us and others have also demonstrated that gelatin hydrogels can be micromolded or photopatterned to induce tissue alignment and recapitulate the native architecture of striated muscle [28, 34, 91]. Additionally, we and others have shown that gelatin hydrogels are compatible with several assays for quantifying cardiac tissue function, including characterization of calcium wave propagation velocity [35], contractility [28], and extracellular field potentials [92]. Thus, gelatin hydrogels are a versatile substrate for engineering functional “Heart on a Chip” platforms that also enhance the metabolic activity of cardiac myocytes, as shown here, which helps maintain a healthy, robust phenotype for cardiac disease modeling and drug screening.

5. Conclusions

In conclusion, we characterized mitochondrial function in cardiac myocytes cultured on a variety of culture substrates to delineate the impact of diverse ECM cues on metabolism. Our results indicate that, in general, baseline mitochondrial function, baseline glycolysis, and maximum respiration are increased in tissues on gelatin hydrogels compared to PDMS substrates coated with gelatin or fibronectin, independent of elastic modulus. These differences could be partially, but not fully, attributed to slight increases in cytotoxicity on PDMS. We also found that ECM protein composition can reduce mitochondrial function in rigid microenvironments, mimicking those found in many disease settings. Collectively, our data demonstrate that the ECM can directly impact the metabolic phenotype of cardiac myocytes, which is important for understanding how ECM remodeling contributes to physiological and pathological cardiac growth. Our study also highlights how the properties of the biomaterial substrate can have a direct impact on the metabolism of cardiac myocytes, which is an important consideration for engineering in vitro cardiac tissue models.

Cardiac development and disease are associated with remodeling of the extracellular matrix coincident with metabolic rewiring of cardiac myocytes. However, little is known about the direct impact of the biochemical and mechanical properties of the extracellular matrix on the metabolic function of cardiac myocytes. In this study, oxygen consumption rates were measured in neonatal rat ventricular myocytes maintained on several commonly-used biomaterial substrates to reveal new relationships between the extracellular matrix and cardiac myocyte metabolism. Several mitochondrial parameters were enhanced on gelatin hydrogels compared to synthetic PDMS substrates. These data are important for comprehensively understanding matrix-regulation of cardiac myocyte physiology. Additionally, these data should be considered when selecting scaffolds for engineering in vitro cardiac tissue models.