Modulation of chromatin remodeling proteins SMYD1 and SMARCD1 promotes contractile function of human pluripotent stem cell-derived ventricular cardiomyocyte in 3D-engineered cardiac tissues

Human embryonic stem cells (hESCs) and induced pluripotent stem cells (iPSCs) have the ability of differentiating into functional cardiomyocytes (CMs) for cell replacement therapy, tissue engineering, drug discovery and toxicity screening. From a scale-free, co-expression network analysis of transcriptomic data that distinguished gene expression profiles of undifferentiated hESC, hESC-, fetal- and adult-ventricular(V) CM, two candidate chromatin remodeling proteins, SMYD1 and SMARCD1 were found to be differentially expressed. Using lentiviral transduction, SMYD1 and SMARCD1 were over-expressed and suppressed, respectively, in single hESC-VCMs as well as the 3D constructs Cardiac Micro Tissues (CMT) and Tissue Strips (CTS) to mirror the endogenous patterns, followed by dissection of their roles in controlling cardiac gene expression, contractility, Ca2+-handling, electrophysiological functions and in vitro maturation. Interestingly, compared to independent single transductions, simultaneous SMYD1 overexpression and SMARCD1 suppression in hESC-VCMs synergistically interacted to increase the contractile forces of CMTs and CTSs with up-regulated transcripts for cardiac contractile, Ca2+-handing, and ion channel proteins. Certain effects that were not detected at the single-cell level could be unleashed under 3D environments. The two chromatin remodelers SMYD1 and SMARCD1 play distinct roles in cardiac development and maturation, consistent with the notion that epigenetic priming requires triggering signals such as 3D environmental cues for pro-maturation effects.

www.nature.com/scientificreports www.nature.com/scientificreports/ Given the expression patterns SMYD1 and SMARCD1 as well as their known intriguing roles in the heart, the two chromatin remodelers were chosen for further experiments.
Validating lentiviral constructs for SMYD1 over-expression and SMARCD1 suppression in hESC-VCMs. To avoid ambiguities due to presence of non-cardiac cells or other chamber-specific types, a pure ventricular population was isolated by transduction with LV-MLC2v-tdTomato-t2A-zeo, followed by zeocin selection 3 . Lentiviral constructs were also designed to over-express and suppress SMYD1 and SMARCD1 in hESC-VCM (Fig. 3A,B) so as to mirror their endogenous developmental expression profiles. Over-expression and suppression were confirmed at the transcript and protein levels by quantitative real-time PCR (Fig. 3C,D) and Western blotting (Fig. 3E), respectively. eGFP over-expression and sh-scramble suppression constructs were used as controls. The localization of SMYD1 and SMARCD1 in hESC-VCMs transduced with the over-expression constructs were validated by immunofluorescence. While SMYD1 is expressed in both the nucleus and cytosol, the majority of SMARCD1 is expressed in the nucleus (Fig. S1). Furthermore, cell morphology and cell size, as determined by immunostaining as well as cell capacitance (Tables 1 and 2), cell nucleation and elongation, as determined by circularity index, were not different between different genetic manipulations (Fig. S2). SMYD1 over-expression enhanced the Ca 2+ transient amplitude without altering action potential of single hESC-VCMs. Given the fundamental importance of Ca 2+ handling and electrophysiology in the biology of hESC-VCMs, we first tested for the functional consequences of SMYD1 overexpression and suppression in single hESC-VCMs. Figure 4A and B shows that SMYD1-VCMs had a significantly larger Ca 2+  S.E.M. of three independent experiments are presented. *p < 0.05; **p < 0.01; ***p < 0.001 (Student's t-test). (E) hESC-VCMs were transduced with the over-expression lentiviral constructs and/or shRNA knockdown constructs and protein expressions were validated by Western blotting with either SMYD1 or SMARCD1 antibodies. GAPDH was used to demonstrate equivalent protein loading. www.nature.com/scientificreports www.nature.com/scientificreports/ time was faster in sh-SMYD1 when compared to sh-scramble control (p < 0.05) (Fig. 4E,F). Representative transient tracings are shown in Fig. 4G. The gene expression levels of several Ca 2+ -handling proteins were examined. While SMYD1-VCM did not show any statistically significant differences in all the genes tested (n = 6), CACNA1S (Ca v 1.1) (p < 0.05), ITPR3 (inositol 1, 4, 5-triphosphate receptor, type 3) (p < 0.01) and CACNA1C (Ca v 1.2) (p < 0.05) were increased in sh-SMYD1-VCM (n = 3) (Fig. S3A).
Spontaneously firing and electrically-elicited (1 Hz) action potentials (AP) of single hESC-VCMs were measured using the patch-clamp techniques. There were no significant differences in all the parameters of spontaneously firing APs measured between SMYD1-(n = 5) and sh-SMYD1-(n = 11) hESC-VCMs (Table 1). When paced at 1 Hz, time for APD 50 (action potential duration at 50% of repolarization) (p < 0.05) and APD 90 (action potential duration at 90% of repolarization) (p < 0.05) were significantly shortened in sh-SMYD1-VCMs (Table 1), indicating that rate adaptation remained unaltered. Representative AP tracings are shown in Fig. 4H. Transcript levels of potassium (K + ) and sodium (Na + ) ion channels were examined by quantitative real-time PCR (Fig, S3B,C). Despite the lack of functional changes, over-expression of SMYD1 (n = 6) did confer a statistically significant elevation of KCND3 (K v 4.3) expression (p < 0.01); on the other hand, expression of KCNA4 were up-regulated, and KCNH2 (hERG) (p < 0.01) and SCN5A (Na v 1.5) (p < 0.001) were down-regulated in sh-SMYD1-VCM (n = 3). Such lack of direct correlation between transcript levels and functional expression or consequences has been described previously 22 .
Combined SMYD1 over-expression and SMARCD1 suppression synergistically augmented maximum contractile force in hESC-VCM-CMTs and -CTSs. To further illustrate and validate the contractile effect of SMYD1 over-expression and SMARCD1 suppression, we performed the same experiments in a larger, more physiological 3D-engineered tissue model, the cardiac tissue strip (CTS), as we recently reported 26 . Unlike the microscales of CMTs, CTSs are approximately one centimeter in length, half a millimeter in diameter, and composed of ~10 6 cells 26 (Fig. S5). Qualitatively similar to our CMT results, both SMYD1 over-expression and SMARCD1 suppression led to increased contractile forces of CTS compared to controls (Fig. 6A,B).

Discussion
Epigenetic modulation has been well established as a transcriptional regulatory mechanism ubiquitously involved in development, maintenance, and disease 28 . Genes need to be physically accessible by the transcription machinery in order to be transcribed. Chromatin remodeling proteins confer chemical modifications, including acetylation, methylation and ubiquitination, to histone tails and physically rearrange genomic DNA to activate or repress transcription 29 . Indeed, histone modification enzymes, DNA methyl-transferases and chromatin binding proteins work in concert to remodel the chromatin into open or closed conformations. The temporal and dynamic epigenetic states are conferred by reversible chemical modifications to genomic DNA, typically cytosine, and to residues along the C-terminal tails of histone proteins. Distinct chromatin and gene expression patterns are associated with lineage and cell fate decisions. For instance, genes highly expressed in undifferentiated hESCs lose active H3K4me3 and gain repressive H3K27me3 histone marks over the course of cardiac differentiation; on the contrary, repressive H3K27me3 gradually decreases on genes involved in mesodermal differentiation and cardiac development as active H3K4me3, H3K36me3 and RNA expression appear [3][4][5] . Despite their importance, chromatin remodeling has not yet been extensively investigated in in vitro cardiac differentiation and maturation of hESC/iPSC-VCMs.
Using a data-driven approach, we have identified two chromatin remodeling proteins that are differentially expressed in human heart, SMYD1 and SMARCD1, and characterized their roles in the phenotypes of hESC-VCMs, CMTs as well as CTSs. In particular, we showed that expression levels of SMYD1 and SMARCD1 elicited different responses under 2D or 3D environment. At the single cell level, SMYD1 over-expression increased (2019) 9:7502 | https://doi.org/10.1038/s41598-019-42953-w www.nature.com/scientificreports www.nature.com/scientificreports/ www.nature.com/scientificreports www.nature.com/scientificreports/ the Ca 2+ transient amplitude but its suppression led to its decrease. Consistent with a positive inotropic effect as a result of the improved Ca 2+ handling, SMYD1 over-expression in CMTs also induced increased contractile forces. Furthermore, SMYD1 could also exert its effect on contractile proteins at the genetic level by upregulating myosin genes such as MYH6 and MYH7. Indeed, SMYD1 is very strongly correlated with MYH7b expression, which is the host gene of cardiogenic miR-499. In a high-throughput yeast-two-hybrid screen, SMYD1 was shown to directly www.nature.com/scientificreports www.nature.com/scientificreports/ bind MYH7b 30 . Additional experiments are required to confirm whether SMYD1 is directly involved in the regulation of such cardiac miRs as miR-499. Unlike SMYD1 over-expression, however, SMARCD1 suppression in single hESC-VCMs did not significantly promote CM function such as Ca 2+ transient properties. When Ca 2+ -handling genes were examined, we noticed a significant up-regulation of PLN (phospholamban), CASQ2 (calsequestrin2) and TRDN (triadin) gene expression in sh-SMARCD1 transduced hESC-VCMs. PLN and TRDN are important inhibitory regulators of ATP2A2 (SERCA2) and cardiac ryanodine receptor (RyR), respectively.
It has been suggested that a 3D culture environment in concert with appropriate temporal and micro-environmental niches would drive maturation of hESC-CMs to a more adult-like state 1,31,32 . Various 3D cell culture model systems have been established recently and have been shown to be pro-maturational 24,26,[33][34][35] . Unlike single hESC-VCMs, both SMYD1 over-expression and SMARCD1 suppression in the two 3D models, CMT and CTS, independently led to significant pro-maturational effects in the form of increased contractile forces and conduction velocities over the physiological range of electrical pacing. This effect was likewise accompanied by changes in a number of Ca 2+ -handling gene expression, including increased expression of L-type calcium channel (DHPR), ATP2A2 and RYR2, as well as their regulatory proteins including PLN, CASQ2, ASPH (junctin) and TRDN. Of note, these Ca 2+ -handling genes have been shown to be expressed at low levels in hESC-VCMs 3 . In addition, modulated expression of SMYD1 and SMARCD in CMT resulted in a decreased expression in NCX, which has been shown to be high in hESC-VCMs (compared to adult) due to the reliance of these immature CMs on NCX for Ca 2+ -exclusion 36 . Calreticulin (CALR), an embryonic cardiac substitute for CASQ2 37 , was also down-regulated in SMYD1-and sh-SMARCD1-CMTs, consistent with a more mature Ca 2+ -handling phenotype. In addition to Ca 2+ -handling genes, expression of various genes encoding for cardiac contractile proteins including myosin heavy chain -alpha (MYH6) and -beta (MYH7) as well as myosin light chain -a (MLC2a) and -v (MLC2v) were also up-regulated, contributing to the increased contractile force. The Na + channel subunit SCN5A and the K + channel subunit KCNA4 are also up-regulated in the CMTs transduced with the three different constructs. Therefore, certain effects that were not detected at the single-cell level could be unleashed under 3D environments. Interestingly, further maturity, as gauged by the improved contractile force and conduction velocity, was attained in CMTs and CTSs co-transduced for simultaneous SMYD1 over-expression and SMARCD1 suppression, hinting at a synergistic effect between SMYD1 and SMARCD1. Taken collectively, the overall results were consistent with the notion that epigenetic priming requires additional triggering signals such as 3D non-cell autonomous environmental cues to unleash the effects of these chromatin remodelers for pro-maturation effects.

Conclusion
Using transcriptomic and bioinformatics analysis as a guide, our functional experiments have revealed that the two chromatin remodelers SMYD1 and SMARCD1 play distinct roles in in vitro maturation of hESC-VCMs. SMYD1 overexpression and/or SMARCD1 suppression in single hESC-VCMs as well as the 3D constructs CMTs and CTSs led to improved contractile as well as electrophysiological phenotypes that signify partial maturation. Some of the effects could be unleashed only under 3D environments, underscoring the importance of non-cell autonomous environmental cues for maturation. Overall, these results were consistent with the notion epigenetic priming requires additional triggering signals for pro-maturation effects 3 .

Methods
Cell maintenance, cardiac differentiation and harvest. Undifferentiated human embryonic stem cell line -HES2 (ESI International, Singapore) was maintained at 37 °C with 5% CO 2 incubator in mTeSR ® 1 culture medium (Stem Cell Technologies). A directed differentiation protocol was used to differentiate HES2 into ventricular cardiomyocytes using the activin A, BMP-4, and IWR1, as reported 27 . The differentiation cultures were maintained in 5% CO 2 /5% O 2 /90% N 2 environment for the first 8 days, and then transferred into normal 5% CO 2 incubator. Fresh media is supplied every 3 to 4 days. Human fetal and adult cardiac tissues were isolated and experimented according to protocols approved by UC Davis International Union of Pure and Applied Chemistry (IUPAC) and Institutional Review Board (IRB) after informed consent were obtained from study participants (Protocol #200614787-1 and # 200614594-1). RNA was extracted with Trizol (Invitrogen) following manufacturer protocol.
Cardiac microtissue preparation. Cardiac microtissues (CMTs) were generated in microfabricated molds as previously described 24 . Briefly, transduced hESC-VCMs were passed through a 40 µm strainer (BD Biosciences) and resuspended in Collagen Solution Mix (1X Gibco ® Medium-199 (Life Technologies); 10 mM HEPES; 13 mM NaOH; 0.035% w/v NaHCO 3 ; 0.5 mg/mL Fibrinogen (Sigma); and 1.5 mg/mL Collagen I (BD Biosciences)) and seeded at 1.2 × 10 6 per mold. The cells were centrifuged into the microwells at the bottom of the mold and excess Collagen Solution Mix was aspirated. The cells were cultured in Gibco ® DMEM medium (Life Technologies) containing 10% FBS and 1% Chicken Embryo Extract (Gemini) for the first 48 hrs, followed by Gibco ® DMEM medium (Life Technologies) containing 5% FBS for 6 days. Culture medium was changed daily and compaction of cells was observed within 48 hours after seeding.
www.nature.com/scientificreports www.nature.com/scientificreports/ Gene expression microarrays. The Illumina bead array (Human Ref6 v3.0) was used to assay gene expression of 48,804 probes for each sample. Background subtraction and cubic spline normalization were performed on the array data using the Illumina BeadStudio 3.1 software suite. Annotation of the microarray probes was performed using the lumi package in R and the ReMOAT pipeline 38,39 . Bioinformatic analysis. WGCNA was performed on an input matrix of all normalized signal intensity values of 13,372 Illumina microarray probes that mapped to a validated RefSeq gene and were expressed over background signal intensity in at least one sample. The optimal β power coefficient of 9 was calculated to optimize scale free topology. An unsigned adjacency matrix was calculated using the WGCNA R package, using the equation: a(i, j) = |cor(x(i), x(j))| β . Dissimilarity is equal to the inverse of adjacency between two genes, or 1 − a(i, j). The modules were assigned by default Dynamic Tree Cut parameters 40 . Dissimilarity values between genes were used to construct hierarchical clustering of the genes and the topological overlap map. Hierarchical clustering was performed on log2 transformed and normalized data using Cluster 3.0, using uncentered correlation and average linkage 41 . Heatmaps were generated with Java TreeView v.1.1.6 42 . Biological process ontological classifications were assigned to groups of genes using the ToppGene suite 43 . P-values were calculated by comparing the number of genes within the test set belonging to a given GO identifier compared with an equivalent number of randomly selected genes from the genome. Bonferroni-corrected p-values < 0.05 were considered to be significant.

Lentiviral-mediated gene transfer.
Cardiospheres were dissociated and re-plated as single cells on Day14 and transduced with recombinant lentivirus particle LV-MLC2v-tdTomato-t2A-Zeocin R -EF1α-GOI or U6-shRN A(GOI)-hPGK-Puromycin R -t2A-GFP at MOI of 5 on Day15, for over-expression and knockdown, respectively. The genes of interest (GOI) were SMYD1, SMARCD1, and GFP or sh-Scramble as a control. Zeocin TM (300 µg/ mL) (Life Technologies) was added to the transduced cells from Day 19 to Day24 to eliminate non-ventricular CMs. The effectiveness using MLC2v-promoter lentivirus for VCM selection has been reported previously 36,44 . cDNA synthesis and quantitative real-time PCR. cDNA were prepared using QuantiTect Reverse Transcription Kit (Qiagen). Gene expressions were quantified using StepOnePlus TM Real-Time PCR System (Applied Biosystems). PCR amplifications were carried out in 96-well optical plates with 20 µL reaction volume, consisting of 100 ng of cDNA template, 4 pmol of forward and reverse primers, and 1X KAPA SYBR FAST qPCR Master Mix (KAPA Biosystems). The reactions were incubated at 95 °C for 3 min, and followed by 40 to 50 cycles of 95 °C for 3 sec, and 60 °C for 20 sec. Relative mRNA expressions were calculated as fold changes normalized by GAPDH. Primer sequences are listed in Supporting Information Table S1.
Western blotting. Protein extracts were loaded in 12% NuPAGE Bis-Tris Gel (Life Technologies) and separated by electrophoresis in 1X NuPAGE ® MES SDS Running Buffer (Life Technologies) at 100 V for 30 min, followed by 200 V for 2 hrs. The resolved proteins were transferred from the gel onto PVDF membrane in 1X NuPAGE ® Transfer Buffer (Life Technologies) containing 20% Methanol at 30 V for 1 hr at room temperature.
The PVDF membrane was blocked in 5% instant skim milk prepared in 0.1% v/v PBS + Tween20 (PBST) for 1 hour at room temperature to prevent non-specific binding of antibody. The membrane was probed with SMYD1 (ab32482), SMARCD1 (ab81621) or GAPDH (ab8245) primary antibodies overnight at 4 °C with agitation. After overnight incubation, the membrane was washed three times with PBST to remove excess primary antibody and was then incubated with appropriate secondary antibodies conjugated with horseradish peroxidase (HRP) for 1 hr at room temperature. Finally, the membrane was washed for another three times in PBST and was developed using the ECL Plus Western blotting detection system. Immunofluorescence staining of hESC-VCM. Lentiviral transduced hESC-VCMs were seeded in 96-well plate and cultured at 37 °C with 5% CO 2 in high-glucose Dulbecco's modified Eagle medium (DMEM) containing 10% fetal bovine serum (Life Technologies). After 48 hrs, hESC-VCMs were fixed with freshly prepared 4% paraformaldehyde for 20 min, washed three times with PBS, permeabilized with 0.5% Triton X-100 for 10 min, and washed three times with PBS. The fixed and permeabilized hESC-VCMs were then blocked in 1% BSA for 1 hour at room temperature to prevent non-specific binding of antibody. The hESC-VCMs were probed with SMYD1 (ab32482), SMARCD1 (ab81621) or cTNT (ab8295) primary antibodies overnight at 4 °C. After overnight incubation, the probed hESC-VCMs were washed three times with PBS to remove excess primary antibody and were then incubated with goat anti-rabbit or goat anti-mouse TRITC antibodies (Life Technologies) for 1 hr at room temperature. Finally, the stained hESC-VCMs were washed for another three times in PBS and ProLong ® Gold Antifade Mountant with DAPI (Life Technologies) was added for the staining of nucleus and protection from photobleaching. Immunofluorescence images were taken with 20X objective using Nikon DS-Qi2 camera.
Calcium imaging of hESC-VCM. The intracellular Ca 2+ ([Ca 2+ ] i ) transients were analyzed by loading the cells with 1.5 μM X-Rhod-1 (Invitrogen, Carlsbad, CA) for 10 minutes at 37 °C in Tyrode's solution containing: 140 mM NaCl, 5 mM KCl, 1 mM MgCl 2 , 1 mM CaCl 2 , 10 mM HEPES and 10 D-glucose at pH 7.4, followed by imaging with a spinning disc laser confocal microscope (PerkinElmer). The electrically-induced Ca 2+ transients (E[Ca 2+ ] i ) were triggered by pulses (40 ms pulse duration; 40 V/cm; 1 Hz) generated from a field generator. The amplitudes of Ca 2+ -transients are presented as the background corrected pseudoratio (ΔF/F) 1,2 = (F − F base )/ (F base − B) where F base and F is the measured fluorescence intensity before and after stimulation, respectively, and B is the average background signal from areas adjacent to the targeted cell. The transients rise (V upstroke ) and the transients decay (V decay ) were subsequently calculated and analyzed. www.nature.com/scientificreports www.nature.com/scientificreports/ Characterization of electrophysiological function. For electrical recording, the whole-cell con-Figureuration of the patch-clamp technique was used with an EPC-10 amplifier and Pulse software (Heka Elektronik, Lambrecht, Germany). Patch pipettes were prepared from 1.5 mm thin-walled borosilicate glass tubes using a Sutter micropipette puller P-97 and had typical resistances of 3-5MΩ when filled with an internal solution containing: 110 mM K + aspartate, 20 mM KCl, 1 mM MgCl 2 , 0.1 mM Na-GTP, 5 mM Mg-ATP, 5 mM Na 2 -phospocreatine, 1 mM EGTA, 10 mM HEPES, pH adjusted to 7.3 with KOH. The external Tyrode's bath solution consisted of: 140 mM NaCl, 5 mM KCl, 1 mM CaCl 2 , 1 mM MgCl 2 , 10 mM glucose, 10 mM HEPES, pH adjusted to 7.4 with NaOH. The tip potential was zeroed before the patch pipette contacted the cell. Upon seal formation and followed by patch break, the capacitance compensation was applied. Series resistance compensation was used up to 80%. Action potentials were recorded with the current-clamp mode. Experiments were performed at 37°C.
Optical mapping of CTS. CTSs were incubated with the calcium indicator X-Rhod-1 AM at 5 μM, (Life Technologies) for 45 min at 37 °C in NCS media. The staining solution was then removed and replaced by Tyrode's solution consisting of 140 mM NaCl, 5 mM KCl, 1 mM MgCl2, 1 mM CaCl2, 10 mM glucose, and 10 mM HEPES at pH 7.4. A MiCAM Ultima-L Dual Camera System (SciMedia, USA), was used to measure changes in dye fluorescence. The system consists of 1x objective and 1x condenser with a high-speed CMOS camera which gives a field of view of 10 mm × 10 mm at a resolution of 100 × 100 pixels. Imaging was performed at 5-sec intervals. Excitation of the dye was completed using a halogen light source (HL-151, Moritex Schott, Japan) mounted with emissions filters. Point electrical stimulations at different frequencies were applied at one end of the CTS at 10 V and 10 ms duration. Data collection and analysis were completed with BrainVision software (SciMedia, USA). Parameters including upstroke time, decay time and conduction velocity were calculated as previously described 45 .
Force measurements. CMTs within the microfabricated molds were imaged using Prosillica GX camera (Allied Vision). Movement of the cantilevers was tracked using SpotTracker plugin in ImageJ (National Institute of Health). The spring constant of the cantilevers was calculated empirically and was used to transform deformation of the cantilevers into μN of force 5 . For CTS, the deflections of the posts were captured in real time with a high-speed camera (100 frames/s) and LabView software (National Instruments, USA). The twitch force (mN) was calculated by applying a beam-bending equation from elasticity theory 26 .
The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.