In vitro sex-specific function–structure relationship in neonatal rat cardiac monolayers

Significance

The structure–function relationship is a central principle in cardiac biology, where coordination across multiple scales, from sarcomeres to tissue alignment to ventricle contraction, is essential for healthy function. While men and women differ in cardiac health and disease presentation, the extent to which biological sex shapes cardiac structure and, in turn, contractile function remains poorly understood. In this study, we reported sex-specific rat heart sheets showing structural and functional distinctions between male and female, and the differences only emerging when individual heart cells were organized into aligned, tissue-scale constructs. The ability to engineer and assess sex-specific confluent cardiac layer in controlled settings offers an approach for exploring the biological basis of sex disparities in cardiac structure and function.

Abstract

Even though in vivo rodent studies have been instrumental in investigating sex-specific differences in cardiac health, function, and pathology, they fall short in providing a fast and flexible platform for investigating sex differences of cardiac anisotropic monolayer in isolation. In vitro platforms offer an accessible and more controlled alternative to dissect and study the mechanisms by which male and female cardiac tissue sheets differ from one another. Here, we have shown on an in vitro heart-on-a-chip platform, primary neonatal rat ventricular myocytes can serve as a viable model showing sex chromosome-driven characteristics when presented with identical experimental conditions. With controlled experimental conditions, the self-assembly of isolated cardiomyocytes resulted in morphological differences in the structure of the contractile apparatus. More importantly, the assembly of cardiac cells into confluent monolayers had a sex chromosome-driven divergence in both structure and the corresponding function. This work reports the characterization of the difference between sex-specific neonatal rat ventricular myocytes in in vitro culture. Thus, this offers an avenue to investigate sex-based variations in cardiac function that are otherwise difficult to study.

Cardiovascular diseases, including heart failure, manifest differently in men and women, with distinct presentations and outcomes. As an example, women are more likely to develop heart failure with preserved ejection fraction, while men are more commonly affected by heart failure with reduced ejection fraction (1–3), highlighting the divergence in cardiac physiology between the sexes. There is a wealth of knowledge on the pathophysiological differences between male and female hearts, encompassing both anatomical and functional aspects; and these dissimilarities likely originate from the dimorphism in the structure and function of the healthy organ. Indeed, men generally have larger and thicker-walled hearts (4–6), and functionally, women’s hearts generate a smaller cardiac output at a faster rate and with larger ejection fraction (6–8). These dissimilarities could potentially be influenced by any of the factors differentiating men and women, such as hormone variations, differences in cardiac structure and function, and distinct molecular pathways activated (1). In vivo and ex vivo rodent studies have been instrumental in investigating sex-specific differences in heart failure, providing valuable insights into the physiological and pathological processes (9–12). To highlight a few discoveries, these studies have shown that male rat cardiomyocytes are larger than females in all dimensions (13, 14), cardiac function is sexually dimorphic down to the myofibril with females exhibiting weaker and slower contraction kinetics (14–17), and cardiac metabolism is a potential major contributor to sex dimorphism in heart failure pathogenesis. The divergence in metabolism occurs by differentially regulating the PKA pathway (18), the mitochondrial function (19), and metabolite utilization (20). However, these in vivo and ex vivo models fall short in providing a fast and flexible platform for investigating specific aspects of sex differences in cardiac function and pathology. This gap can be partially addressed by utilizing in vitro engineered systems.

While there are many successful, functional in vitro cardiac tissues (21–38), they often overlook sex as a biological variable or utilize induced pluripotent stem cells (iPSCs) derived cardiomyocytes, which are known to be immature and have high heterogeneity resulting from the differentiation process (39, 40). To engineer a more consistent cardiac tissue, some platforms use neonatal rodent ventricular myocytes, but many of these cardiac tissues have been generated using mixed-sex cells (22, 23, 27, 28). As a result, little is known about sex-specific structure and function of primary cardiomyocytes in engineered tissue, specifically of the intricate processes by which biological sex influences cellular organization and development, and the corresponding impact on overall cardiac function. This is a critical gap that needs to be addressed to make in vitro studies of sex differences more accessible, to develop more accurate models of heart disease, and to adapt existing in vitro platforms to study sex differences of the cardiac system.

Here, we utilized a previously developed heart-on-a-chip platform (41) to produce sex-specific, 2D laminar cardiac sheets. We characterized primary neonatal rats’ cell geometry and cytoskeletal self-assembly of their isolated cardiomyocytes at the single cell level without the addition of exogenous hormones. Furthermore, under the same media conditions, the cardiomyocyte monolayers were analyzed both for structure and contractile functions. These findings will allow for the augmentation of existing in vitro microphysiological systems, enabling them to support sex-specific ensembles, capturing their inherent differences, and providing a platform for testing targeted treatment strategies for cardiac injury as well as in vitro models for mechanistic studies.

Results

Classically, primary in vitro cardiac tissues have been generated from mixed-sex ventricular myocytes with microcontact-printed fibronectin patterns of parallel stripes to promote anisotropic organization in the cardiac layer (33, 41–44). To explore the possibility of using these in vitro platforms to study sex differences, we characterized cardiac sheets engineered from neonatal rat ventricular myocytes (NRVMs) isolated from either female or male pups.

First, we sought to compare the structure of individual NRVMs when they are seeded sparsely onto a 20 $\mathrm{\mu }$m-wide line pattern. There were no qualitative differences between male and female NRVMs (Fig. 1A). Additionally, no significant difference in thickness was observed between the cells isolated from the two sexes (Fig. 1B). Their cell areas and subsequent calculated cell volumes were also conserved (SI Appendix, Fig. S1). To quantify cytoskeletal architecture, cardiomyocytes were stained with $\alpha$-actinin and actin, then analyzed using parameters previously developed to evaluate engineered cardiac monolayer tissue quality (45). No statistically significant difference was found in the orientational order parameter (OOP) (Fig. 1C), implying that the self-organization of cardiomyocyte contractile apparatus is independent of biological sex. However, there was a statistically significant reduction in both the continuous z-line length and z-line fraction (Fig. 1 D and E) for the male cells compared to the female, suggesting a difference in the maturity level achieved by the self-assembly in isolation from other cell neighbors. One female cardiomyocyte exhibited markedly higher maturity metrics compared to the rest of its cohort, which was identified as a statistical outlier. However, the statistical significance of our findings remained unchanged upon its removal (SI Appendix, Fig. S2).

Fig. 1.

Single cell observations of sex-specific NRVMs. (A) Immunofluorescence of male and female individual cells stained for $\alpha$-actinin (red), actin (green), and nuclei (blue). (Scale bar $=$ 10 $\mathrm{\mu }$m.) (B) Thicknesses extracted from z-stacks of individual cells. (C–E) Sarcomeric structural metrics for isolated sex-specific cardiomyocytes. (C) Z-line orientational order parameter (OOP) is a metric of degree of organized z-lines. (D) Mean continuous z-line is a metric for the quality of cardiomyocytes. (E) Z-line fraction is a metric of cell maturity. Each data point represents a singular cardiomyocyte, seeded from three separate litters for both female and male (SI Appendix, Table S1). All data shown with Mean $\pm$ SD.

Next, the same metrics were used to characterized sex-specific self-assembly at a higher architectural level of confluent monolayers. Qualitatively, the myofibrils in female cultures were grouped in bundles, identifiable when they crossed (Fig. 2A, many white triangles), while flat and sheet-like myofibrils were more characteristic of male NRVM cultures (Fig. 2A, single white triangle). Corroborating with this observation, the self-assembly of these cardiomyocytes generated significantly thicker female NRVM sheets compared to their male counterparts (Fig. 2B). Z-line densities were then examined with respect to the total image area (Fig. 2C) and total cardiomyocytes area (Fig. 2D). There was a statistically significant difference between the two sexes for z-line density by image area (Fig. 2C), suggesting potential difference in the cell ensemble’s density given the same seeding density. However, neither sex had a significant advantage in cardiomyocyte z-line density (Fig. 2D). The sarcomeric architectures as described by previous metrics were not statistically different between male and female (Fig. 2E–G). Given all experimental conditions were controlled between the sexes, the observable differences suggested that the self-assembly of cardiac sheets was inherently sex chromosome-driven, but resulted in ensembles with similar z-line organization and maturation.

Fig. 2.

Sex-specific NRVM cardiac sarcomere structure at the mid seeding density (Table. 1). (A) Immunofluorescent staining of sex-specific NRVM sheets. Male NRVM cultures exhibited flat, sheet-like myofibril organization while female NRVMs form higher myofibril bundling (3D nature indicated by white arrows). (Scale bar $=$ 25 $\mathrm{\mu }$m.) (B) Measured monolayer thicknesses. (C) Z-line density as represented by normalizing the total z-line pixels over total image area. (D) Cardiomyocyte z-line density as represented by normalizing the total z-line pixels over the total of actin pixels belonging to cardiomyocytes. (E–G) Sarcomeric structure metrics of female and male NRVM cultures (46). (E) Z-line orientational order parameter (OOP) is a metric of degree of organized z-lines. (F) Mean continuous z-line is a metric for the quality of cardiomyocyte structure. (G) Z-line fraction is a metric of maturity. Each data point represents a heart chip. The data in its entirety were analyzed from three litters of male and female pups each, with each litter producing 2 to 3 heart chips (SI Appendix, Table S1). All data shown with Mean $\pm$ SD.

To explore functional differences between the sexes, NRVM sheets of varying densities (Table. 1) were generated on the heart-on-a-chip platform as previously described (41). The contraction of the cardiac films were recorded as stress traces over time, from which systolic, diastolic, and active stresses were extracted (Fig. 3A). As seeding density affected stress generation, systolic and active stresses were used to choose the optimal seeding density. Adjusted for the effects of the cells’ sex, there exists a trend of increasing systolic stress generation, statistically significant from low to mid seeding densities (generalized linear model, P$=$ 0.027, 0.022, respectively, SI Appendix, Table S2). In active stress, the overall model did not find a significant effect of seeding density, but post hoc contrasts showed marginal differences within each group (Fig. 3D, blue and green comparisons). Based on both findings, the mid density was chosen as a point of comparison for both sexes as it was the condition at which these sex-specific tissues performed best and the differences between them were not confounded by overcrowding effects (SI Appendix, section 2).

Table 1.

NRVM seeding density conditions used in heart-on-a-chip experiments

Density condition	# of cells seeded in 12- well plate	# of cells seeded per area	Sample sizes
	(cells per well)	(cells per cm2)	Male	Female
Low	$1.3\times {10}^6$	$3.7\times {10}^5$	24	9
Mid	$1.5\times {10}^6$	$4.3\times {10}^5$	29	41
High	$1.8\times {10}^6$	$5.2\times {10}^5$	9	34

The sample size refers to the number of contracting films on the heart chip. For detail descriptions and animal allocation, see SI Appendix, Table S1.

Fig. 3.

Contractility analyses. (A) Example heart-on-a-chip films at rest, at diastole, at systole, and the calculated stress trace over time. The width of the films do not influence stress calculation (32). (Scale bar $=$ 1 mm.) (B–D) Systolic, diastolic, and active stresses, respectively, of cardiac monolayers generated from sex-specific NRVMs at densities outlined in Table. 1. (E–G) Average sarcomeric forces for systole, diastole, and active, respectively, computed by combining structural (cardiomyocyte area) and functional (stress generation) analyses at the mid seeding density (all stresses and forces are log-normally distributed. *P$<$ 0.05, **P$<$ 0.01). Each data point represents a single film, with a range of 1 to 8 viable films per heart chip. The data were collected from 4 and 3 litters for male and female pups, respectively (SI Appendix, Table S1). All data shown with Mean $\pm$ SD.

At the lower seeding densities (low and mid), male NRVM sheets were significantly stronger than females in generating both systolic and diastolic stress (Fig. 3 B and C and SI Appendix, Fig. S3), showing a sex-dependent disparity in the ensembles’ abilities to contract and relax. However, no statistically significant sex-specific difference was observed in active stress generation (Fig. 3D), meaning the net stress produced was conserved between male and female cardiomyocyte ensembles. By combining the structural (Fig. 2D) and functional analyses, similar interpretations arose after calculating the stresses normalized by the nuclei count for each group (SI Appendix, section 6). Additionally, it was possible to estimate the average force for each sarcomere (Fig. 3E–G and SI Appendix, Fig. S5), which showed that at the micron scale, the two sexes were equivalent and matched previously estimated single sarcomere force (47, 48).

Discussion

Our observations highlight that when male and female cardiomyocytes are exposed to identical experimental conditions- such as the same extracellular matrix pattern, culture media, and cell seeding density- their biological sex drives distinct morphological outcomes and functional profiles. At the single cell level, male and female NRVMs exhibited differences in their sarcomeric architecture and maturity (Fig. 1 D and E), while maintaining comparable cellular height (Fig. 1B). These initial observations imply that female cardiomyocytes had the potential of outperforming male cells due to their superior maturity. However, the self-assembly of confluent monolayers allowed the male cells to match with the female cells in terms of the z-line maturity metric (Fig. 1E vs. Fig. 2G) and z-line density within the cardiomyocytes area (Fig. 2D). The continuous z-line length becomes particularly meaningful when cells assemble into monolayers, where continuous z-lines can span across multiple cells in highly organized and healthy tissue with aligned z-lines (45). The longer continuous z-line lengths observed in these cardiomyocyte sheets compared to single cells reflect this multicellular registry, where more organized and registered z-lines enable more efficient contraction (49). These observations dovetail to the contractility results- where the active stress was preserved between both sexes (Fig. 3D). Across the functional analyses, increasing cell seeding density from low to mid levels consistently enhanced contractile performance in both sexes. At high density, male tissues exhibited reduced stress generation and a reversal of the observed relationship to the females at low and mid seeding densities, suggesting that overcrowding disproportionately affects their function. Although complementary structural data showed comparable tissue organization across densities and between sexes (SI Appendix, Fig. S6), the selection of the mid seeding density was based on its functional stability rather than structural metrics. Additionally, finding the peak contractility output at the same seeding density is consistent with the consensus that the number of cardiac myocytes in the heart is the same in both sexes (5, 50). Yet, under these conditions, the male engineered NRVM sheets exhibited significantly elevated systolic and diastolic contractile forces compared to females at the same seeding density (Fig. 3 B and C, black), which is in accordance with ex vivo observations (14–17). Estimating the number of cells via the nuclei count and using that parameter to extract stress per nuclei from tissue data led to similar conclusions (SI Appendix, Fig. S4B). Based on our estimate for average sarcomere force (Fig. 3E–G), the contribution of the individual sarcomeres to generate contraction is consistent between the two sexes and is independent of z-line structure, organization, and cell seeding density (SI Appendix, Fig. S5). Thus, the difference in functionality is due to the difference in sarcomeric structure observed between the two sexes after assembling into confluent layer, which resulted in the disparity in thickness and z-line density (Fig. 2A–C) while maintaining comparable nuclei count, actin area per nuclei, and z-line density per nuclei between the male and female groups (SI Appendix, Fig. S4A). A fundamental principle in cardiac biology is the tight relationship between tissue structure and function wherein architectural features such as sarcomeric alignment and myofibril organization directly govern contractile performance. While hormones clearly influence cardiovascular outcomes, extensive animal studies show that gonadectomy–thus, removal of exogenous hormones–primarily alters remodeling processes, which are then correlated to the functional outcomes (51–55). In some of these cases, the structural changes can be attributed to other sex-mediated mechanisms such as modulating mitochondrial functions (51, 52). The reversal in the direction of sex differences between the two contexts observed in our experimentation indicates that the process of self-assembly into higher-order tissue constructs is itself sexually dimorphic and likely governed by intrinsic, chromosome-driven mechanisms, driving male and female cardiomyocytes to adopt distinct architectures and impacting the emergent force production.

In the future, to investigate the mechanism that leads to the differences in self-assembly, the cadherin protein family, particularly N-cadherin, could represent a compelling avenue to pursue. N-cadherin, a critical component of the intercalated disc that mediates cardiomyocyte-to-cardiomyocyte adhesion, has been established as essential for maintaining both structural and functional integrity of cardiac tissue (56–58), serving as a hub protein modulating both adhesion and myofibril formation (59), being capable of directing cell phenotype and alignment (60). As such, it could direct how monolayer thickness increased markedly compared to single cells (Fig. 2B vs. Fig. 1B), reflecting the transition from spread, flat isolated cells to thicker (61), more three-dimensional monolayer in the presence of neighbors (SI Appendix, Fig. S7), with female sheets showing particularly pronounced bundling (Fig. 2A). Additionally, recent work has identified differential expression patterns of cadherin family proteins between male and female iPSC-derived cardiomyocytes (62), suggesting that sex-specific variations in cell–cell junction composition may contribute to the distinct tissue architectures we observed (Fig. 2).

This work demonstrated that in vitro neonatal rat ventricular myocytes can serve as a potential model for probing sex differences in cardiac research (Figs. 1–3). While there has been no direct measurement of human cardiac force, women exhibit lower end-diastolic and systolic pressure and volume (7, 63), as reflected in our neonatal rat model (Fig. 3B–D). These observations between rat and human cardiac characteristics suggest that primary rat neonatal cardiomyocytes provide a promising and accessible platform to explore sex-specific cardiac function, offering insights that may be applicable to human health. While adult cardiac physiology is shaped by both chromosomal and hormonal factors, the understanding of the relative contributions of each remains incomplete (18). By isolating the effects of chromosomal sex in a simplified in vitro model, our work provides a foundational perspective that can inform and complement future studies incorporating hormonal influences or adult cardiomyocyte systems.

Indeed, even though NRVMs are developmentally immature at the time of isolation, engineered constructs similar to ours have been shown to have elongated morphology, aligned sarcomeres with appropriate lengths (64) (SI Appendix, Fig. S8), contractile function, inotropic response to changes in extracellular Ca²⁺ concentration, and gene expression profiles comparable to those of adult rat myocardium (65), as well as slower spontaneous kinetics in females (SI Appendix, Fig. S9) as seen in isolated cardiomyocytes (15, 16), supporting the physiological relevance of this approach. The sex chromosome-driven organizational differences we observe represent intrinsic biological programming, which should remain as a modulator of cardiac structure and function throughout development to adult myocardium. Furthermore, NRVMs’ accessibility, cost-effectiveness, and rapid experimental turnaround make them particularly valuable for studies requiring high-throughput tissue generation and functional assessment, which means they remain a widely utilized model in cardiac research (66–71). Nonetheless, NRVMs do not fully replicate the phenotypic complexity of adult human cardiomyocytes, and findings from this model should be interpreted within that limitation. Adult cardiomyocytes would likely show additional layers of complexity in their interaction with the sex axis. For example, while previous studies have demonstrated sex-specific differences in cardiomyocytes dimensions in adult rats, with males generally exhibiting larger cell size (13, 14), our data reveal no such differences at the neonatal stage. Specifically, we observed equivalent cell areas (SI Appendix, Fig. S1) and cell thicknesses (Fig. 1B), thus, their volumes are conserved between the sexes at this developmental stage. This suggests that cardiomyocyte size disparities may emerge during postnatal development, potentially influenced by larger body sizes in males, hormonal maturation, or other growth factors.

The adult myocardium presents layers of complexity that were not addressed in this model. The native myocardium is a three-dimensional structure with complex fiber orientations-spiraling and tangentially aligned layers-as well as contributions from multiple cardiac cell types. In that sense, our current model represents a simplified version of cardiac tissue, consisting of a single anisotropic monolayer composed solely of ventricular myocytes. However, we emphasize that the platform’s modularity allows for added complexity. Additionally, culture components such as fetal bovine serum (FBS), which is standard in NRVM studies, can provide background hormonal cues. However, when FBS is diluted to culture concentrations, the resulting estradiol concentration in the media is approximately 7 to 20 pM, which is below serum estradiol concentrations in premenopausal women (73 to 1,835 pM) (72–74) and age-matched men (36 to 300 pM) (74–77). Still, this consideration merits closer attention when modeling pediatric or geriatric populations, where endogenous estradiol levels are naturally lower (74, 75, 78). Further, phenol red has been employed as standard NRVM culture across studies investigating both general cardiac function (79–81) and estrogen-related mechanisms (82–85). While phenol red possesses weak estrogenic activity (86), its influence is cell type dependent and most pronounced in estrogen-sensitive cell lines such as breast cancer cells (87). Cardiomyocytes, in contrast, express comparatively low levels of estrogen receptors, even in adult rat hearts when compared to other tissue types (88), suggesting that the activation of estrogenic activity of phenol red is likely minimal. Nevertheless, these components represent potential background factors that could interact with sex-dependent effects, underscoring the importance of considering media composition when interpreting in vitro findings. Future studies could leverage genetic approaches such as CRISPR-mediated knockdown of hormone receptors in cardiomyocytes prior to harvest (89), enabling precise decoupling of hormonal and chromosomal axes to further dissect the mechanisms underlying sex disparities. Recent work from our group has extended this system to incorporate macrophages and hypoxic injury, enabling investigation of immune–cardiac interactions (90). These developments underscore the adaptability of the platform and its potential to model increasingly sophisticated aspects of cardiac physiology and disease, now with sex as a biological variable.

Many in vitro platforms have successfully used mixed-sex rat cardiomyocytes to generate functioning tissue (22, 23, 27–30), which have been contributory to cardiovascular research. In this study, we demonstrated that sex-separated NRVMs can also serve as a valuable model for investigating sex dimorphism in cardiac biology. Beyond identifying structural and functional differences between male and female confluent monolayers, we established a link between these differences, providing insight into their cause-and-effect relationship in the absence of supplement hormones. This finding opens the door for future studies to explore the specific biological pathways that drive sex-based disparities in cardiac tissue, including those governed by sex hormones. Notably, we observed that these sex chromosome-driven differences became evident when cardiomyocytes self-assembled into organized monolayers, underscoring the importance of studying cardiac function at the tissue level rather than just isolated single cells. Furthermore, existing tissue engineering techniques can be readily adapted to support sex-specific rodent cardiac sheet, ensuring their applicability in modeling sex-specific cardiac function. This adaptability strengthens the utility of rodent cardiomyocytes as a robust platform for drug testing, disease modeling, and the development of targeted therapeutic strategies.

Materials and Methods

Substrate Preparation.

The chips were fabricated as previously described and used (41), with some modifications. Briefly, a large glass coverslip (76 mm $\times$ 83 mm; Brain Research Laboratories, Newton, MA) was cleaned via sonication for 10 min in 200-proof Ethanol and air dried. Protective film strips were placed parallel to the edge of the glass, approximately 1 cm apart. Next, poly(N-isopropylacrylamide), (PIPPAAm, Sigma-Aldrich, Burlington, MA) was dissolved in 99.4% 1-butanol at 10%wt (weight/volume). An excess of the PIPPAAm solution (1 mL total) was deposited and spin-coated onto the glass surfaces not covered by the protective films at 6,000 RPM for 1 min. The protective film was removed. Then, Sylgard 184 polydimethylsiloxane (PDMS; Ellsworth Adhesives, Germantown, WI) elastomer was mixed at a 10:1 base to curing agent ratio and partially cured at 37 °C between 90 to 105 min prior to spin coating on the glass substrate, with a terminal velocity of 4,000 RPM for 1 min. Next, the glass substrate was left to cure overnight at 65 °C. After curing, the PDMS was scored lightly to create 4 strips of thin films without engraving the glass underneath, and the glass substrate was then cut into individual chips with a laser cutter (Trotec Speedy 360; Trotec Laser GmbH., Plymouth, MA). Three chips from each large glass coverslip were saved for thickness measurements. PDMS film thicknesses were measured to be within a range of 12 to 15 microns using a profilometer (Dektak 6M, Veeco Instruments Inc., Plainview, NY).

Prior to seeding cells, fibronectin (FN; Sigma-Aldrich, St. Louis, MO) was microcontact-printed onto the surface of the PDMS substrate. The stamps were patterned to be parallel lines, 22 $\mathrm{\mu }$m in width with 3 $\mathrm{\mu }$m gaps in between, made from PDMS. The stamps were sterilized in 200-proof Ethanol, and let dry in a biohood under sterile conditions. The patterned surface of the dry stamps was covered in 200 uL of 50 $\mathrm{\mu }$g/mL FN and incubated for 1 h. The surface of the cover slips was sterilized and functionalized by exposing them for 8 min to UV ozone (Model No. 342, Jetlight Company, Inc., Irvine, CA). The stamps were dried with compressed nitrogen and used to transfer the FN pattern to the coverslips with 4 min of contact. A 12-well plate was used to house the chips, initially coated with 1% Pluronics F127 (BASF Group, Parsippany, NJ) in DI water for ten minutes and washed three times with Phosphate Buffered Saline (PBS; ThermoFisher, Waltham, MA, Cat#10010049). The chips were transferred to the Pluronics-coated plate, submerged in PBS, and stored at 4 °C before use. All chips were used within 4 to 7 d after FN stamping. Chips used for structural assays were fabricated similarly (41), without coating PIPPAAm nor scoring PDMS into films.

Isolation and Culture of Neonatal Rat Ventricular Myocytes (NRVMs).

All animal procedures were performed with the approval of the Institutional Animal Care and Use Committee of University of California, Irvine (IACUC Protocol #2022-054). The data were collected over seven and six separate isolation procedures for male and female experiments, respectively. Each isolation procedure used a litter of ten rat pups. All procedures and data analysis methods in this work adhered to the ARRIVE and other relevant guidelines and regulations.

The isolation of primary ventricular cardiomyocytes was performed as previously described (41). Briefly, the provider Charles River Laboratories (Wilmington, MA) sexed two-day old neonatal Sprague-Dawley rats by the anogenital distance from multiple litters and provided groups of ten male-only or female-only pups with a random dam. Dams were euthanized with carbon dioxide asphyxiation followed by bilateral pneumothorax pneumothorax according to the approved procedure by the American Veterinary Medical Association Guidelines for the Euthanasia of Animals. The ventricles were extracted from only the pups, homogenized by washing in Hanks Balanced Salt Solution (HBSS; ThermoFisher, Waltham, MA, Cat#14170161), then incubated in 1 mg/mL trypsin solution (Sigma-Aldrich, St. Louis, MO) dissolved in HBSS for 12 h at 4 °C on a rocker. The trypsin was neutralized with M199 culture medium (Thermo Fisher, Waltham, MA, Cat#11150067, containing phenol red) supplemented with 10% fetal bovine serum (FBS; ThermoFisher, Waltham, MA, Cat#26140079) at 37 °C. The tissues were dissociated to single cells four times via digestion with 1 mg/mL collagenase type II solution (Worthington Biochemical Corporation, Lakewood, NJ) dissolved in HBSS at 37 °C. Cells were washed in chilled HBSS, centrifuged at 1,000 RPM for 6 min. Cells were then resuspended in warm M199 culture medium supplemented with 10% FBS, 10 nM HEPES, 3.5 g/L Glucose, 2 nM L-glutamine, 2 mg/L vitamin B-12, and 50 U/mL penicillin. The myocytes were purified by three repeats of differential plating at 37 °C for one hour each. Each harvest, consisting of a litter of 10 pups, produced roughly 10 heart chips, which were split between structural and functional chips. Purified cells were then seeded on the heart chips at densities of $1.3\times {10}^6$, $1.5\times {10}^6$, or $1.8\times {10}^6$ cells per well for confluent monolayers and at $1.5\times {10}^5$ cells per well for single cells. Chips used in structure analyses were seeded at $1.5\times {10}^6$ cells per well. The chips were cultured at 37 °C and 5% CO₂ over a period of four days prior to running the assays. During the initial 48 h, the cells were kept in previously described media, supplemented with 30% FBS, and media were refreshed every 24 h. After which the cells were incubated in 10% FBS supplemented media for the remaining 48 h. The FBS contained an average estradiol concentration of 18 ng/mL, which, when diluted, became 20 pM in the culture media at 30% and 7 pM at 10%.

Cardiac Structure Assay and Analysis.

On day 5 of culturing the cardiomyocytes, the chips were rinsed with PBS prior to being fixed and permeabilized in a solution of 4% paraformaldehyde (PFA; VWR, Radnow, PA) and 0.05% Triton-X (Sigma-Aldrich, St. Louis, MO) in PBS for 10 min. The NRVM cultures were stained for actin (Alexa 488 Phalloidin, ThermoFisher, Waltham, MA), nuclei (4’,6-diamidino-2- phenylindole hydrochloride, DAPI; Invitrogen, Waltham, MA), sarcomeric $\alpha$-actinin (clone EA-53, mouse monoclonal anti-$\alpha$-actinin, Sigma-Aldrich,St. Louis, MO), and fibronectin (rabbit anti-fibronectin; Invitrogen, Waltham, MA) at a dilution of 1:200 for each antibody for 1 h at room temperature. Secondary antibody solution was prepared with goat anti-mouse IgG (AlexaFluor594; ThermoFisher, Waltham, MA) and goat anti-rabbit IgG (Cy5; ThermoFisher, Waltham, MA) at 1:200 dilution, in which the chips were incubated for 1 h at room temperature. Finally, the chips were mounted on glass slides with ProLong Glass Anti-fade Mountant (ThermoFisher, Waltham, MA).

Image Acquisition.

For cardiac monolayer structure images, imaging was performed on a digital CCD camera ORCA-R2 C10600-10B (Hamamatsu Photonics, Shizuoka Prefecture, Japan) mounted on an IX-83 inverted motorized microscope (Olympus America, Center Valley, PA). For thickness measurements, a laser scanning confocal microscope (Olympus Fluoview FV3000; Olympus America, Center Valley, PA) was used. All images were taken at 40X magnification. Ten randomly selected fields of view for each chip were obtained. Z-stacks were obtained at the top, middle, and bottom slices for NRVM sheets at ten random fields of view. Z-stacks for single cells were obtained at a resolution of 0.1 $\mathrm{\mu }$m and processed in Fiji to obtain thicknesses.

Image Analyses.

The cardiomyocytes and cardiac monolayer structures were quantified using a custom MATLAB script, as previously developed (46). Binary skeletons of $\alpha$-actinin signals were extracted from microscopy images and analyzed for z-line orientational order parameter (OOP), z-line fraction, and mean continuous z-line lengths as averages from the ten fields of view per chip.

To quantify cardiomyocyte area from imaging data, a decision tree classifier was implemented using a custom MATLAB script, as previously described (91). A subset comprising approximately 1% of the dataset was manually labeled to identify cardiomyocyte regions, with a final distribution of 57% male and 43% female images. The classifier, optimized on this training set, achieved an accuracy of 98%. This classification reported the number of pixels of actin signals within the cardiomyocyte area for each sex.

Contractility Assay.

Contractility experiments were performed as previously described (41) with some modifications. The chips were transferred from the incubator to a stereoscope in a 60 mm petri dish filled with warm, 37 °C, media. To release the films, a razor blade was used to make two cuts in the middle of the chip, perpendicular to the PDMS scored lines, 0.5 to 1 mm apart. The resulting thin strip in between was peeled off with tweezers. Field stimulation electrodes were built from 1 mm diameter carbon rods (McMaster-Carr, Douglasville, GA) and attached to a PDMS mount in a parallel configuration where they were placed 1.5 cm apart. During the experiments, the PDMS mount was slotted into the edge of the 35 mm pertri dish containing the chip and warm, fresh media, with the electrodes pointing downward and submerged into the solution. The chip was imaged between the electrodes. The chips were paced at 2 Hz, 20 V using an external field stimulator (Myopacer, IonOptix Corp., Milton, MA), which applied a square wave pulse. Video recordings of the films contracting captured at least 5 cycles for 1 Hz and 10 cycles for 2 Hz. All images and movies were collected using a Basler camera (A602f Basler Inc, Exton, PA) controlled by MATLAB, with all movies containing 100 frames per second. An image of a scale ruler was taken afterward. Then, the media were removed from the petri dish, allowing the films to lay flat on the glass surface. An image of the films was taken for length measurements, and another image of the scale ruler was taken to account for changes in focus. Contractility videos were analyzed for stress produced by the cardiac films using custom Fiji and MATLAB scripts as described (41). Each chip produced a range of 1 to 8 viable films for analyses, with viable meaning the films were contracting spontaneously and were responding to the electrical stimulation.

The average sarcomeric force was determined by combining structural data of Normalized Sarcomeric Space and contractility data (${\overline{\sigma }}_{\mathit{monolayer}}$):

$$\begin{array}{c}\hfill {\overline{\sigma }}_{\mathit{sarcomere}}=\frac{{\sigma }_{\mathit{monolayer}}}{A_{\mathit{CM}}},\end{array}$$

where, $A_{\mathit{CM}}$ is the cardiomyocyte area postclassification (SI Appendix, Fig. S10) as represented in actin pixels normalized over the image are for each field of view.

Then, using previously published (48) estimate of sarcomere area, $A_{\mathit{sarcomere}}=\pi {(60)}^2{\text{nm}}^2$, the sarcomere force (${\overline{F}}_{\mathit{sarcomere}}$) was calculated as

$$\begin{array}{cc}\hfill {\overline{F}}_{\mathit{sarcomere}}& ={\overline{\sigma }}_{\mathit{sarcomere}}·A_{\mathit{sarcomere}}.\hfill \end{array}$$

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

Experimental data have been deposited in Dryad (DOI: 10.5061/dryad.rjdfn2zp2 and https://datadryad.org/share/0CVtsGUS9PKbp5cwfnhBSt10ATrArGIMO7r3vxgSpXs) (92).