Engineered Heart Tissue Model of Diabetic Myocardium

Abstract

Myocardial infarction resulting in irreversible loss of cardiomyocytes (CMs) is a leading cause of heart failure. Previously, we reported an in vitro test-bed for screening cell integration between injected test cells and host CM using the engineered heart tissue as a recipient. The objective of this study is to expand our system to diabetic cardiomyopathy conditions. Patients with diabetes show dysfunction of CMs independent of myocardial infarction, indicating that diabetes directly affects CMs. However, the underlying mechanisms are not fully understood, and developing a diabetic CM test-bed could enable drug screening studies specific to the diabetic heart. Diabetic cardiac conditions were mimicked by cultivating neonatal rat CMs seeded onto collagen scaffolds in normal or high glucose with or without insulin. Our results show that high glucose conditions, which mimic diabetic hearts, display poor electrical properties. Gene expression profiles from diabetic, adult, and neonatal rat hearts as well as engineered heart tissues under different conditions were compared. The diabetic rat heart and high glucose conditions increased the ratio of myosin heavy-chain isoform β to α indicative of diseased states; thus, this model system captures some molecular aspects of diabetic cardiomyopathy. Moreover, thiazolidinedione diabetic drug treatment improved electrical excitabilities and exhibited anti-apoptotic effects.

Introduction

Heart disease is the leading cause of death in the western world. Heart failure is caused by functional disorder of the contractile cells of the heart, cardiomyocytes (CMs). Coronary occlusion followed by ischemia results in CM death, and subsequent replacement of these cells by noncontracting fibroblasts leads to scar tissue formation. In 2010, 6.4% of world population had diabetes, which corresponds to 300 million people.¹ Patients with diabetes, of whom 65% die of cardiovascular complications, also exhibit cardiomyopathy, dysfunction of CMs independent of coronary artery disease, or hypertension, thus indicating that the diabetic condition itself affects CMs directly.² However, the underlying mechanisms are not fully understood.

Diabetic cardiomyopathy is defined as ventricular dysfunction in the absence of coronary artery disease. It starts with asymptomatic impaired diastolic function and then progresses to cardiac hypertrophy and fibrosis, ending up with both diastolic and systolic dysfunction. Hyperglycemic states change the cardiac metabolism increasing fatty acid oxidation and ultimately result in the re-establishment of a fetal gene program. Key contractile and metabolic proteins are changed shortly after birth to adjust to biomechanical requirements. In rodents, fetal CMs express higher levels of the myosin heavy chain (MHC) β isoform than the α isoform, whereas this ratio is reversed in the adult CMs. Also, heart failure due to diabetes or hypertrophy shows decreased contractile function and an increase in the slow-shortening-velocity -MHC β isoform, among other fetal genes.³

Engineered heart tissue (EHT) is a model system that recapitulates a number of structural and physiological characteristic of native cardiac muscle (reviewed in ref.⁴). Previously, we have shown that EHT, composed of collagen scaffolds seeded with neonatal CM, achieved a viable three-dimensional cardiac tissue with an ultra structural organization and functional properties similar to the neonatal heart.^5–8 In previous studies, we have also shown that EHT could be cultivated for a long-term, up to 4 weeks in order to study the integration of injected stem cells in the cardiac environment.⁹ In the present study, we hypothesized that EHT can also reproduce a condition of diabetic cardiomyopathy by varying culture conditions. These EHTs may be further used for cardiac cell injection studies or for drug screening under conditions that more closely mimic the diabetic heart. Here, we investigated electrical function, cellular metabolism, viability, and molecular expression of cardiac genes and proteins of EHTs under normal, diabetic, or therapeutic conditions. We selected genes and proteins that enable us to assess the cellular composition of the EHTs (cardiac troponin T [cTnT], a marker of CMs, vimentin, and prolyl 4-hydroxylase β polypeptide [p4Hb], a marker of nonmyocytes) coupling between the cells (connexin 43 [Cx43]) and the reestablishment of the fetal program common for pathological conditions (MHCα, MHCβ, natriuretic peptide precursor A [nppa]). Also, phosphatidylinositol-3-kinase (PI3K) was examined as a down-stream signaling molecule of insulin. Further, we showed that anti-diabetic thiazolidinedione drugs improved electrical properties of EHT and exhibited anti-apoptotic effects. Overall, our results demonstrate that EHT cultivated under high glucose conditions exhibit key markers of diabetic cardiomyopathy including diminishing electrical properties, increasing cell apoptosis, and changing the molecular contractile machinery.

Methods

Cell isolation

Neonatal (1–2 days old) Sprague-Dawley rats (Charles River, CA) were euthanized according to the procedure approved by the Institute's Committee on Animal Care. The hearts were removed, quartered, and enriched CMs were isolated by an overnight trypsin (6120 U/mL, Sigma-Aldrich) incubation in Hank's balanced salt solution (HBSS) at 4°C followed by a series of collagenase digestion (220 units/mL, Worthington Biochemical) in HBSS at 37°C for 8 min each as previously described.¹⁰ The supernatants from five collagenase tissue digestions were collected and centrifuged at 750 rpm (94 g) for 5 min; resuspended in culture medium (Dulbecco's modified Eagle's medium [DMEM], 4.5 g/L glucose, Gibco) supplemented with 4 mM L-glutamine, 10% certified fetal bovine serum (FBS; Gibco), 100 U/mL penicillin-streptomycin, and 10 mM 4-2-hydroxyethyl-1-piperazine-ethanesulphonic acid buffer (HEPES; Gibco); and pre-plated onto T75 flasks for one hour. The nonadhered cells from T75 flasks were collected and counted for live cells after trypan blue staining to determine the cell density.

Construct preparation

Dry Gelfoam collagen sponge scaffolds (Gelfoam, Delasco) (1 cm×1.5 cm×300 μm) were prewetted with culture media, and the enriched CMs (1.5 million cells per scaffold) were seeded using Matrigel (15 μL; Becton-Dickinson). Gelation was achieved within 30 min at 37°C. Subsequently, culture medium was added (2 mL/well in a 12-well plate) and cultured for 8 or 16 days under static condition. The following culture medium groups were included: normal glucose (1 g/L) DMEM without insulin (N), normal glucose DMEM with insulin (20 μU/mL) (NI), high glucose (4.5 g/L) DMEM without insulin (H), or high glucose DMEM with insulin (20 μU/mL) (HI). All the culture media were supplemented with 10% FBS (Gibco), 100 U/mL penicillin, 100 μg/mL streptomycin, and 10 mM HEPES (Gibco). The media was changed every 2 days and saved at −20°C for glucose and lactate assays. In some experiments, 5 μM Troglitazone or 10 μM Pioglitazone was added to H conditions.

Measurement of electrical properties

Electrical activity in constructs was evaluated at the end of cultivation (day 8 or 16) by using the pair of stimulating electrodes positioned 1 cm apart connected to an electrical stimulator (Grass 88×; AstroMed) as described.¹¹ The excitation threshold (ET), the minimum voltage at which the entire construct was observed to beat, and the maximum capture rate (MCR), defined as the maximum pacing frequency for synchronous construct contractions, were determined.

Glucose and lactate assays

The concentrations of glucose and lactate in culture medium were measured using colorimetric glucose (QuantiChrom™ Glucose Assay Kit; Bioassay Systems) and lactate (L-Lactate Assay Kit; University of Buffalo Biomedical Research Service Center) assay kits. The glucose assay is based on the formation of a colored complex between an aldohexose and o-toluidine-glacial acetic acid, and the lactate assay is based on the reduction of the tetrazolium salt INT in an NADH-coupled enzymatic reaction to formazan color product. The culture media were collected and frozen on the day of media exchange and assayed together at the end of the experiment.

Cell viability analysis

On day 8, the viability of cells in scaffolds was measured by ethidium monoazide bromide (EMA; Sigma-Aldrich) labeling followed by fluorescent-activated cell sorting analysis as described.¹¹ EMA diffuses into dead cells and covalently binds to the DNA on photolysis, thereby fluorescently labeling the DNA in dead cells. After the scaffolds were incubated in EMA (5 μg/mL) and activated under the fluorescent light, the scaffolds were then digested with 0.1% collagenase at 37°C for 15 min, followed by 15 min incubation on ice, with periodic pipetting to dissociate scaffolds and cell aggregates. 1 mL of 2% FBS in HBSS was added into the resulting cell suspension. Cell pellets were spun down and then resuspended in the same buffer. The cell suspensions were subjected to fluorescent-activated cell sorting to determine the percentage of EMA negative cells.

Immunohistochemistry

The EHTs were fixed in 10% neutral buffered formalin (Sigma-Aldrich) for 30 min, embedded in paraffin, and sectioned at 5 μm. Slides were deparaffinized, processed for antigen retrieval by heat treatment for 20 min at 95°C in a decloaking chamber (Biocare Medical), and then blocked with 10% donkey serum (Jackson Immuno Research) for 1 h at room temperature (RT). Antibodies used for immunohistochemsitry were as follows: primary mouse anti-cTnT (1:100; Labvision), secondary Alexa488 donkey anti-mouse (1:100; Invitrogen), and primary Cy3 conjugated mouse anti-vimentin (1:100; Sigma-Aldrich) in phosphate-buffered saline containing 0.5% Tween 20 (Sigma-Aldrich) and 1.5% donkey serum. The sections were counterstained with Hoechst (1:100; Sigma-Aldrich) and cover-slipped using mounting media (Sigma-Aldrich). All the primary antibodies were incubated overnight at 4°C, and secondary antibodies were incubated for 1 h at RT. A humidified chamber was used for all incubation steps. Cell death (apoptosis) analysis was performed using in situ cell death detection kit (TMR red) according to the manufacturer's instructions (Roche). Briefly, paraffin sections were pretreated as just described and incubated with TUNEL reaction mixture containing TdT and TMR-dUTP for 3 h at 37°C. For positive control, DNase (1 mg/mL) was treated for 10 min at RT before incubation with TUNEL kit. Slides were imaged using a confocal microscope at 20×, and three pictures per slide were quantified (Olympus, FV1000 laser scanning confocal).

Quantitative polymerase chain reaction

RNA was isolated by an RNeasy mini kit (Qiagen). Reverse transcriptase reaction was carried out by using superscriptaseIII (Invitrogen) with 0.5 μg of total RNA. Quantitative polymerase chain reactions (qPCRs) were performed at 50°C for 2 min, 95°C for 10 min, and then 40 cycles of 95°C for 15 s, 60°C for 1 min, and 72°C for 15 s followed by a dissociation step using 7900HT Fast Real-time PCR system (Applied BioSystems). The following primers were used in qPCR: cTnT- forward 5′-CCTGCAGGAAAAGTTCAAGC-3′, reverse 5′-GTGCCTGGCAAGACCTAGAG-3′; MHCα- forward 5′-TGATGACTCCGAGGAGCTTT-3′, reverse 5′-TGA CACAGACCCTTGAGCAG-3′; MHCβ - forward 5′-CCTC GCAATATCAAGGGAAA-3′, reverse 5′-TACAGGTGCATC AGCTCCAG-3′; p4Hb - forward 5′-GCAAAACTGAAGGC AGAAGG-3′, reverse 5′-TCACAATGTCGTCAGCTTCC-3′; Cx43 - forward 5′-TTCATCATCTTCATGCTGGT-3′ reverse 5′-ATCGCTTCTTCCCTTCAC-3′; PI3K - forward 5′-AGCC ACAGGTGAAAATACGG-3′, reverse 5′-TTTTCTTTCCGCA ACAGCTT-3′; nppa–forward 5′-GGGGGTAGGATTGACA GGAT-3′, reverse 5′-CTCCAGGAGGGTATTCACCA-3′; β-actin-forward 5′-TAAAGACCTCTATGCCAACAC-3′, reverse 5′-GATAGAGCCACCAATCCAC-3′; Caspase 3-forward 5′- AACCTCAGAGAGACATTCATGG-3′, reverse 5′-GAGT TTCGGCTTTCCAGTCA-3′; Caspase 9-forward 5′-CTCCTG CGGCGATGC-3′, reverse 5′-CCACTGGGGTGAGGTTTC-3′; Bax-forward 5′-TGCAGAGGATGATTGCTGAC-3′, reverse 5′- GATCAGCTCGGGCACTTTAG-3′; Bid-forward 5′-AGC AGGTGATGAACTGGACC-3′, reverse 5′-AGACGTCACGG AGCAGAGAT-3′; Bcl-XL-forward 5′-CATATAACCCCAGG GACAGC-3′, reverse 5′- GTCATGCCCGTCAGGAAC-3′.

Statistical analysis

Results are expressed as mean±standard error of the mean. Statistical significance was assessed using Tukey's test followed by one-way analysis of variance or Dunn's test for more than three groups (SigmaStat), or t-test for between two groups, and denoted with a * when the p-value for at least three independent experiments was less than 0.05.

Results

EHT constructs were cultivated for 8 days after primary neonatal CMs were seeded onto the collagen scaffolds under different media conditions. Normal glucose (5.5 mmol/L) and high glucose (25 mmol/L) media conditions were compared with or without the addition of insulin (20 μU/mL). Table 1 summarizes glucose and insulin concentration in the human or rat blood and in each media condition. Indicated glucose concentrations in the media were measured, and insulin concentration was calculated from obtained FCS profile (invitrogen) and by adding the known amount of insulin (20 μU/mL). Based on these values, we defined normal glucose with insulin treatment condition (NI) as healthy, high glucose without insulin (H) as hyperglycemic, and high glucose with insulin (HI) as hyperglycemic with insulin therapy. The group with normal glucose, but without insulin (N), served as a control.

Table 1.

Comparison of Glucose and Insulin Concentrations In Vivo and the Engineered Heart Tissue Model System

Subject	Glucose (mmol/mL)		Insulin (μU/mL)	Reference
Healthy human	S	3.6–6.0	2.3–23.9	44, 45
	NS	5–8	10–130
Diabetic human	S	>7	5–10	46
	NS	>11.1	10–30
Rat normal	S	5.6–6.1	14–24	45
	NS	5–7.2	20–80
Rat ZSF	12–37		130–150	47
EHT N	1.9–5.5		0.3–1.2	Control: Normal glucose, no insulin
EHT NI	1.4–5.5		20.3–21.2	Healthy
EHT H	18.2–25		0.3–1.2	Hyperglycemic
EHT HI	17.7–25		20.3–21.2	Hyperglycemic on insulin therapy

Ranges of glucose and insulin concentration in the blood of human and rat were obtained from the literature (S: starvation, NS: nonstarvation). Ranges of the glucose concentration were measured from the culture media for EHT. Ranges of insulin concentration for the EHT were calculated based on the fetal bovine serum content and added insulin. N, normal glucose (5.5 mM); NI, normal glucose (5.5 mM)+insulin (20 μU/mL); H, high glucose (25 mM); HI, high glucose (25 mM)+insulin (20 μU/mL); EHT, engineered heart tissue.

The EHT from all the groups started contracting in unison from day 2 onward. On day 8, the ET and MCR were assessed. The ET is the minimum electrical field voltage required for inducing synchronous contractions of the EHT, and it is the measure of the electrical excitability of the construct. The MCR, the maximum beating frequency attainable while maintaining synchronous contraction, is indicative of the beating capacity.¹⁰ Insulin treatment decreased the ET significantly under both normal (p<0.05) and high glucose conditions (p<0.05) (Fig. 1A); however, insulin treatment increased the MCR significantly only under the high glucose conditions (p<0.05) (Fig. 1B). These results indicate that insulin treatment improves the electrical properties of EHT.

FIG. 1.   .

Diabetic condition presented worsened electrical properties of EHT that could be reversed by insulin treatment. Excitation threshold (A) and maximum capture rate (B) were measured on day 8 after seeding. N, normal glucose (5.5 mM); NI, normal glucose (5.5 mM)+insulin (20 μU/mL); H, high glucose (25 mM); HI, high glucose (25 mM)+insulin (20 μU/mL) (n=8–10, *p<0.05). EHT, engineered heart tissue.

Next, we investigated the effect of glucose and insulin on cell viability. There was a trend of insulin treatment increasing the live cell number under both normal and hyperglycemic conditions, with the overall highest live cell number in the NI group (normal glucose with insulin treatment, that is, the group we defined as healthy, Table 1) (Fig. 2). This result indicates that insulin treatment may have effects on either cell proliferation or survival (anti-apoptosis) of the heart cells. Therefore, we stained the histological sections for Ki67 to assess cellular proliferation. There were very few positive cells detected, irrespective of condition (one or none per slide, data not shown); thus, we concluded that insulin treatment did not impact cellular proliferation in the EHTs. We also investigated cell apoptosis using TUNEL staining (Fig. 2B, C). DNAase treatment for 10 min before staining served as a positive control (Fig. 2C (d)). The percentage of the apoptotic cells in each group (Fig. 2B) correlated well with the live cell number (Fig. 2A). Insulin treatment clearly protected against apoptosis under both normal and high glucose conditions with significantly less apoptotic cells in the insulin treated groups compared with the insulin-free groups (NI vs. N; HI vs. H) (Fig. 2B). In addition, higher percentages of apoptotic cells (Fig. 2B) and lower live cell numbers (Fig. 2A) were observed in the high glucose group (H) compared with the normal glucose control (N), demonstrating the effect of hyperglycemia itself on cell death in EHTs. Overall, these results indicate that diabetic conditions (H) induce more cell death compared with the healthy conditions (NI).

FIG. 2.   .

Insulin treatment increased viable cell number. (A) On day 8, cellular viability was measured by ethidium monoazide bromide labeling followed by fluorescent-activated cell sorting analysis of the collagenase digested EHTs. Total number of cells in suspension from each scaffold was counted using a hemocytometer, and the live cell number was calculated. (B) and (C) paraffin sections were stained for apoptosis using a TUNEL kit (red) and nucleus (blue). Pink nucleus indicates an apoptotic cell (arrows). Three images at 200× were taken from each slide and counted for apoptotic (red) nuclei and total number of nuclei. (a) without labeling reagent, (b) N, (c) NI, (d) positive control involving DNase treatment before the staining, (e) H, (f ) HI (N=3–4, *p<0.05). Color images available online at www.liebertonline.com/tea

To evaluate cellular metabolism during the cultivation, we measured glucose and lactate concentrations from the conditioned media at 2-day intervals. The culture media changes ensured that the glucose concentrations were different and within the desired target range in the normal and high glucose conditions. The average glucose concentrations, which EHTs experienced during the cultivation, were 4.04 mmol/mL (N, range from 1.91 to 5.50), 3.59 mmol/mL (NI, range from 1.40 to 5.50), 22.53 mmol/mL (H, range from 18.16 to 25.5), and 21.99 mmol/mL (HI, range from 17.65 to 25.5). Both high glucose conditions (H and HI) maintained a significantly higher glucose level than normal glucose conditions (N and NI) throughout the culture period, indicating that EHTs were indeed exposed to high glucose diabetic media condition throughout the cultivation period. The average lactate concentrations were 10.99 mmol/mL (N, range from 3.09 to 22.06), 11.92 mmol/mL (NI, range from 3.97 to 23.15), 13.09 mmol/mL (H, range from 2.08 to 30.79), and 16.85 mmol/mL (HI, range from 4.61 to 32.28). We also calculated glucose consumption rate and lactate production rate. Average glucose consumption rates were 1.54 mmol/mL/day (N), 1.93 mmol/mL/day (NI), 2.98 mmol/mL/day (H), and 3.60 mmol/mL/day (HI); and average lactate production rates were 6.26 mmol/mL/day (N), 7.02 mmol/mL/day (NI), 9.39 mmol/mL/day (H), and 10.56 mmol/mL/day (HI).

Hematoxylin and eosin staining was used to determine the presence of cells and global morphology of constructs after 8 day of cultivation. Notably, in the insulin treated groups, the cells were present at higher densities forming a more compact tissue in comparison to their own appropriate glucose controls (Fig. 3, compare the left column with the right one).

FIG. 3.   .

Insulin treatment results in more compact tissue morphology. EHTs were fixed with 10% formalin, and paraffin sections were stained with hematoxylin and eosin after 8 days of cultivation. Color images available online at www.liebertonline.com/tea

Next, we examined the presence of vimentin (non-myocyte/fibroblast marker, red), cTnT (CM marker, green), and Cx43 (white) in the EHTs cultivated under different culture media conditions (Fig. 4A). Non-myocytes were found mostly at the periphery of the EHT and inter-dispersed among CMs. The CMs formed a syncytium with punctate distribution of gap junctions as evidenced by the pattern of Cx43 staining (Fig. 4B). The higher magnification images revealed the striations of troponin T staining (indicated in arrows in Fig. 4B and inset of Fig. 4A (b)). Under normal glucose with insulin condition (NI), the striation patterns were more apparent (Fig. 4B).

FIG. 4.   .

Immunostaining for cardiomyocyte and nonmyocyte markers. EHTs were fixed in 10% formalin after 8 days of cultivation, and paraffin-embedded sections were stained with mouse anti-cardiac troponin T (green), rabbit anti-connexin 43 (white), Cy3 conjugated anti-vimentin (red), and Hoechst (blue) for nucleus. (A) General morphology of EHTs show that majority of cells are cardiac troponin T positive (Inset in (b) shows striation of troponin T from a separate slide). (B) Enlarged pictures show striation pattern of cardiac troponin T (arrows), and connexin 43 presence (white dots). (a) N, normal glucose (5.5 mM); (b) NI, normal glucose (5.5 mM) + insulin (20 μU/mL); (c) H, high glucose (25 mM); (d) HI, high glucose (25 mM) + insulin (20 μU/mL). Color images available online at www.liebertonline.com/tea

To determine whether the increased live cell number under normal glucose with insulin condition (Fig. 2B) was due to the increase in number of fibroblasts, we stained the fixed slides from the EHTs with vimentin antibody and counted vimentin positive cells and the total cell numbers per slide. Supplementary Figure S1 (Supplementary Data are available online at www.liebertonline.com/tea) shows that there were no significant differences in percentage of vimentin positive cells among the different cultivation conditions, thus indicating that insulin treatment was not preferentially promoting the proliferation of non-myocytes (specifically fibroblasts). The percentage of vimentin positive cells in all groups at the end of cultivation was about 35%, which is consistent with the composition of the neonatal rat heart cell population after one preplating step that was used for construct seeding.¹²

Next, we hypothesized that the gene expression in EHT under the high glucose media condition should resemble diabetic cardiomyopathy. To investigate differential gene expression in CMs in a diabetic condition, we obtained ZSF (Zucker diabetic fatty [ZDF]×spontaneously hypertensive heart failure [SHHF]) obese rats as our positive control. These rats result from the breeding between SHHF rats and ZDF rats. They have been shown to develop cardiac hypertrophy due to diabetic complications from hyperglycemia and hyperinsulinemia.¹³ We measured glucose concentration from the hearts of the rats, and we confirmed that the nonfasting glucose concentration is twice higher than from normal adult Sprague-Dawley rats (Supplementary Fig. S2). Also, gene expression in neonatal heart and adult hearts of Sprague-Dawley rats were compared to distinguish differences in pathological and developmental states. qPCR on MHCα, MHCβ, cTnT, p4Hb, Cx43, PI3K, and nppa were compared. The increased ratio of MHCβ to MHCα and nppa expression indicates fetal or diseased states of CM. cTnT and p4Hb expression were used as markers for the cellular composition of EHTs, Cx43 indicates the level of cellular coupling, and PI3K is a down-stream signaling molecule of insulin.

Hearts from diabetic rats increased the ratio of MHCβ to MHCα and nppa expression compared with those of normal rats (Fig. 5); however, there was no difference in the expression of cTnT, p4Hb, or Cx43. The similar expression levels of cTnT (a marker of CMs, Fig. 5B) and p4Hb (a marker of fibroblasts, Fig. 5C) indicate that cell composition in the EHT did not change as the result of glucose or insulin treatment, consistent with the results of the immunofluorescence evaluation (Fig. 4 and Supplementary Fig. S1). Similar levels of Cx43 (Fig. 5D) may indicate that there were no differences in coupling through gap junctions as a result of glucose and insulin treatment.

FIG. 5.   .

Gene expression profile of EHTs cultivated under different glucose and insulin concentrations. Hearts from adult diabetic ZSF rats, Sprague-Dawley adult, and neonatal (2 day old) rats were used as controls. mRNA expression was normalized to β-actin and compared with either adult rat hearts or normal media condition (N). (A) β-MHC to α-MHC ratio, (B) cTnT, (C) P4Hb, (D) Cx43, (E) PI3K, (F) nppa. (n=3 each for rat hearts, n=4–7 for EHTs. *p<0.05). MHC, myosin heavy chain; cTnT, cardiac troponin T; P4Hb, prolyl 4-hydroxylase β polypeptide; Cx43, connexin 43; nppa, natriuretic peptide precursor A; PI3K, phosphatidylinositol-3-kinase.

Interestingly, EHT under insulin treatment condition decreased the ratio of MHCβ to MHCα, consistent with the trend observed in the normal versus diabetic rat heart. Thus, the EHT cultivated without insulin either at normal or high glucose concentrations shows similarities with the diabetic myocardium with regard to the ratio of MHC isoforms (Fig. 5A). In addition, normal glucose with insulin, the condition we define here as healthy, exhibited increased PI3K mRNA expression compared with other conditions (Fig. 5E). PI3K is a down-stream signaling molecule of insulin implicated in cell survival responses and prevention of cell death.¹⁴ Here, upregulation of PI3K mRNA (Fig. 5E) correlates with the increase in the live cell number in the NI condition (Fig. 2A). We also evaluated the nppa expression, which was increased in the diabetic and neonatal hearts compared with the normal adult hearts. Although the expression of nppa in the H condition, which we define here as diabetic, was increased compared with the other groups, the results were not statistically significant (Fig. 5F).

Further, we tested the effects of antidiabetic thiazolidinedione drugs by adding 5 μM Troglitazone or 10 μM Pioglitazone in H condition during 8 days or 16 days of cultivation. On day 8, ET was decreased with Troglitazone treatment (8.3 to 5.7 V/cm); and by day 16, Pioglitazone treated group exhibited improvement in electrical excitability parameters as evidenced by the lowering of ET and an increase in MCR (Supplementary Fig. S3). We also investigated gene expression on day 16 samples; and unlike insulin treatment on H condition (Fig. 5A; H and HI), thiazolidinedione drugs did not reverse MHC isoform conversion (Supplementary Fig. S3C). Rather, these drugs showed an anti-apoptotic effect by decreased expression of Caspase 9 and Bid compared to the H condition (Supplementary Fig. S3C).

Based on these results, EHTs under high glucose condition (H) recapitulate some characteristics of diabetic myocardium compared to normal glucose and insulin condition (NI); electrical properties measured by ET and MCR were diminished; and some aspects of the fetal gene program have emerged.

Discussion

Diabetic cardiomyopathy has been recognized as a distinct CM dysfunction apart from ischemic cardiac cell death due to coronary occlusion.¹⁵ Changes in cardiac metabolites seem to affect gene expression changes and function of CMs in the long term. Since insulin treatment for patients with diabetes does not address cardiac dysfunction, alternative treatments are necessary. There have been a number of animal models developed to study diabetes including streptozotocin-induced diabetic rats¹⁶ and mice,¹⁷ spontaneously mutated rats such as ZDF¹⁸ and ZSF,¹³ and genetically modified mice models such as db/db.¹⁹ However, it is still hard to distinguish from the results of these animal models whether the detrimental effect of diabetes on CMs is direct or indirect. Additionally, animal models are expensive, and the complexity of the in vivo environment and the heterogeneity of native cardiac tissues have made it difficult to isolate cell effects on heart function. Therefore, the objective of this study is to develop an in vitro tissue-engineering model to be used as a screening system for potential diabetic cardiomyopathy treatments.

Over the last 15 years, considerable progress has been made in the development of functional cardiac tissue substitutes to replace the damaged cardiac area with a viable cardiac tissue. There have been three main approaches to make cardiac tissue engineering substitutes: (1) creation of stackable cell sheets without using scaffolds,²⁰ (2) mechanical stimulation of CMs in collagen gel,²¹ and (3) cultivation of cell-seeded scaffolds in bioreactors (e.g., electrical stimulation of CMs cultured on porous scaffolds²²). Previously, we have successfully shown that the in vitro EHT injection model can serve as a test-bed for rapid screening of candidate cell types and potential survival or integration parameters for cell transplantation strategies.^6,23

Here, we show that EHT can also be used as a model for diabetic cardiomyopathy. In this study, cultivating EHT in high glucose conditions without insulin resulted in functional deterioration (Fig. 1), which may parallel a similar decrease in myocardial function seen in diabetes. Based on our results, this is not due to a decrease in coupling (e.g., less Cx-43 expression, Fig. 5D) or a change in cell composition (similar% vimentin positive cells, Supplementary Fig. S1). Instead, the effect was due to the lack of the insulin-induced cell survival (Fig. 2B) and a gene expression change to the fetal program including the MHC isoform switch (Fig. 5A) and an increase in nppa (Fig. 5F).

We have investigated the expression levels of several fetal genes in EHTs, which have been shown to be turned on by diabetes, and compared them with the expression levels in diabetic rat hearts. In CMs, myosin is a major contractile protein and is largely present in two isoforms, α-MHC and β-MHC. Although the two isoforms α- and β-MHC share 93% homology in amino acid composition, it has been shown that their functional properties are different.²⁴ α-MHC has been shown to have faster actin filament sliding velocity but generates half the force compared with β-MHC²⁵; however, β-MHC is more energy efficient.²⁶ In rodents, β-MHC is dominant prenatally and then it switches to α-MHC after birth; this switch is the opposite in humans. In rodents, fetal hearts beat slower than adult hearts, and it might be related to the dominant expression of slower β-MHC. On the other hand, in humans, the fetal heart beats faster than an adult heart, and the dominant isoform is α-MHC in the fetus. The reasons underlying the differences in MHC isoform expression in the different species and the activation of the fetal gene program during the disease process are not clear. In diabetic cardiomyopathy, streptozotocin-induced diabetic rat hearts showed an increased β-MHC expression in comparison to the α-MHC isoform.²⁷ Recently, the transcription factor Brg1 has been shown to regulate α-MHC downregulation during mice development and hypertrophic cardiomyopathy,²⁸ but whether or not diabetes induces the same factor is unknown. Although there is some evidence that cardiac disease process induces the isoform change, the switch alone does not seem to alter the cardiac function except for the diastolic relaxation.²⁶ In addition, the studies genetically replacing the isoforms did not induce heart failure itself.^25,29,30

nppa is another marker for cardiac myopathy. Nppa was shown to be increased in diabetes³⁰ and hypertrophy.³¹ Increased nppa inhibits cellular growth and proliferation and induces apoptosis in cultured CMs.^32,33 Thus, the increase of nppa in the EHTs cultivated under diabetic conditions (Table 1) mimics some aspects of the gene expression in diabetic animals.

We also examined cTnT gene expression in diabetic rat heart and neonatal heart compared with normal adult heart as well as our different EHT conditions. cTnT is the tropomyosin-binding subunit of the troponin complex that regulates striated muscle contraction. In the adult mouse, cTnT is only expressed in the heart,³⁴ and cTnT mutants have been shown to cause dilated cardiomyopathy.³⁵ Our results show that cTnT was upregulated during the maturation but did not change in diabetic conditions based on the animal heart studies (Fig. 5). However, MHC isoform switch and nppa upregulation have been detected in both neonatal hearts and diabetic hearts as well as in our diabetic condition (H) compared with the healthy condition (NI) (Fig. 5).

The PI3K pathway has been documented to regulate a number of physiological functions, including cell growth, survival, and actin cytoskeleton rearrangement.³⁶ It was documented to be involved in a physiological hypertrophy of CMs.^37,38 We have previously shown that inhibition of PI3K signaling partially inhibited the elongation and alignment of CMs.⁷ PI3-kinases take part in extracellular signal transduction by phosphorylating the hydroxyl group at positions 3 of membrane lipid phosphoinositides. The PI-3 phosphates then transduce signals downstream by acting as docking sites for a number of signal-transducing proteins (e.g., protein kinase B etc.) Insulin binding to the insulin receptor activates the PI3K-dependent pathways and the Ras-MAP (mitogen-associated protein)-kinase-dependent pathways. It has been shown that PI3K pathway activates the serine-theronin kinase Akt that may modulate contractile function³⁹ and cell survival.¹⁴ Interestingly, our qPCR results with PI3K expression (Fig. 5E) indicated that NI condition, here defined as healthy, had a significantly higher level of PI3K mRNA compared with the high glucose conditions. Consistent with the survival effects of the PI3K stimulation, the NI group supported the highest live cell number (Figs. 2 and 5). In addition, decreased PI3K expression seems to correlate with the worsening of contractile properties under diabetic conditions, consistent with the PI3K involvement in the cardiac cell contractility and cytoskeletal rearrangements (Figs. 1 and 5).

Cx43 is a major CM gap junctional protein. Nygren et al. found that propagation of cardiac impulse was decreased in hearts of STZ-diabetic rats due to Cx43 disorganization.⁴⁰ Our qPCR results show that diabetic rat hearts and diabetic EHT condition (H) showed a tendency for decrease in Cx43 (Fig. 5D).

Our molecular analysis in each test condition was compared with de novo isolated diabetic, neonatal, and adult heart tissue. Results showed that insulin treatment not only affected cell survival but also changed gene expression. These cellular and molecular changes collectively affect electrical properties of the whole tissue. Kim et al. explored the smooth muscle cell implantation in streptozotocin-induced diabetic rat heart and found that insulin treatment was necessary for cell injection therapy in diabetic rats.⁴¹ This study highlights a need for a diabetic cardiomypathy disease model to screen the drugs or cells for transplantation studies. Also, previous studies by us indicate that glucose and insulin states affect the integration of the injection of embryonic stem cell derived-CMs.¹⁰

Further, we investigated the effect of anti-diabetic drugs, Troglitazone and Pioglitazone, on H condition. These drugs are thiazolidinedione drugs (peroxisome proliferator-activated receptor γ agonists) that have been used as insulin sensitizers.⁴² Unlike insulin treatment on H condition, these did not affect MHC isoform conversion, but rather exhibited anti-apoptotic effects by decreased expression of Caspase 9 and Bid (Supplementary Fig. S3). Previously, Baraka and AbdelGawad showed that another peroxisome proliferator-activated receptor γ agonist, rosiglitazone, suppressed CM apoptosis in streptozotocin-induced diabetic rat model.⁴³

In conclusion, our data demonstrate that diabetes-induced gene expression changes cause contractile dysfunction and affect the electrical excitability in the EHT. The present study also provides a platform for in vitro functional screens. This platform could be used for drug screening, specifically in diabetic conditions, thus providing mechanistic insights into strategies to cure diabetic myopathy.

Acknowledgments

This work was supported by funding from Juvenile Diabetes Research Foundation (Innovative Grant to M.R. and P.W.Z.), Heart and Stroke Foundation of Ontario (M.R. and P.W.Z.), the Ontario Ministry of Research and Innovation Early Research Award (to M.R.), and NSERC Discovery Grant (M.R.).

Disclosure Statement

No competing financial interests exist.