Abstract

   Diabetic cardiomyopathy (DbCM) is increasingly prevalent, but
   intervention targets remain unclear due to the lack of appropriate
   models and the complexity of risk factors. Here, this work establishes
   an in vitro assessment system for DbCM function using cardiomyocytes
   derived from human pluripotent stem cells and engineered heart tissue.
   This work finds high‐fat status in complex diabetes risk factors
   majorly contributes most to cardiomyocyte death and contractile
   dysfunction. Notably, PA induced early electrophysiological
   abnormalities, and lately is associated with cardiac fibrosis,
   mitochondrial fission, and systolic and diastolic dysfunction at tissue
   level. Using this in vitro assessment system, this work finds that
   empagliflozin (EMPA), a first‐line glucose‐lowering drug, effectively
   alleviated early PA‐induced cardiomyocyte injury. Treatment with EMPA
   enhanced abnormal diastolic and electrophysiological functions in the
   PA‐hEHT model and significantly reduced endoplasmic reticulum stress,
   and apoptosis. Furthermore, these promising results are confirmed in a
   type 2 diabetes mellitus mouse model, reinforcing the potential of EMPA
   as a therapeutic option to alleviate cardiomyocyte injury under
   diabetic conditions. These findings suggest that this work has
   developed an engineered model of diabetic cardiomyopathy that mimics
   the various stages of lipotoxic myocardial injury and support the use
   of EMPA as a potential therapeutic option for diabetic or lipotoxic
   cardiomyopathy.

   Keywords: diabetic cardiomyopathy, diastolic dysfunction,
   empagliflozin, hEHT, in vitro model
     __________________________________________________________________

   This study develops a human engineered heart tissue‐derived model of
   diabetic cardiomyopathy that accurately replicates the dynamic changes
   in the structural, contractile, and electrophysiological properties of
   the myocardium. Furthermore, this model is used to confirm the direct
   protective effects of empagliflozin on the heart through an SGLT2
   off‐target mechanism, consistent with findings from mouse models.

   graphic file with name ADVS-12-e03173-g001.jpg

1. Introduction

   Diabetes has emerged as a significant global health challenge,^[ [42]^1
   ^] with China being particularly affected by this epidemic, where
   ≈11.2% of the adult population is diagnosed with diabetes.^[ [43]^2 ^]
   Cardiovascular diseases represent a leading cause of morbidity and
   mortality among individuals with type 2 diabetes mellitus (T2DM), with
   heart failure (HF) frequently manifesting as a common complication.^[
   [44]^3 , [45]^4 , [46]^5 ^] Among the various complications of
   diabetes, diabetic cardiomyopathy (DbCM) stands out as the most
   prevalent. Initially characterized as a distinct form of cardiomyopathy
   marked by cardiac hypertrophy, fibrosis, structural and functional
   impairment independent of preexisting cardiovascular risk factors such
   as hypertension or coronary artery disease,^[ [47]^6 ^] DbCM has
   recently been redefined to encompass systolic and/or diastolic
   myocardial dysfunction occurring in the context of diabetes—an entity
   referred to as diabetic myocardial disorder.^[ [48]^7 ^] This revised
   definition broadens the scope of DbCM, placing greater emphasis on
   myocardial function and highlighting the direct myocardial damage
   experienced by individuals with diabetes.

   DbCM/diabetic myocardial disorder presents a series of characteristics
   associated with multiple mechanisms that remain incompletely
   elucidated.^[ [49]^7 , [50]^8 ^] One significant challenge in
   elucidating these mechanisms is the involvement of complex systemic
   factors, including hyperglycemia, insulin resistance, hyperlipidemia,
   advanced glycation end products, renin‐angiotensin‐aldosterone system
   activation, and autonomic dysregulation, which interrelate to cause the
   development of diabetes, hypertension, accelerated coronary
   atherosclerosis, and microvascular dysfunction.^[ [51]^9 , [52]^10 ,
   [53]^11 ^] Numerous studies have shown that they jointly lead to
   abnormal myocardial metabolism, glucotoxicity, and lipotoxicity,
   ultimately leading to oxidative stress, fibrosis, hypertrophy,
   diastolic dysfunction, and systolic dysfunction.^[ [54]^12 , [55]^13 ^]
   However, it is challenging to analyze these factors separately.
   Additionally, a critical obstacle in managing DbCM is the early
   detection of diastolic dysfunction, which is often obscured by subtle
   symptoms.^[ [56]^14 , [57]^15 ^] By the time both systolic and
   diastolic dysfunction deteriorate, the complexity of intervention
   increases significantly and the prognosis becomes increasingly poor.^[
   [58]^16 ^] Therefore, it is essential to develop dynamic models that
   encompass the diverse subtypes of T2DM patients.^[ [59]^17 ^] These
   models should be based on stratifying risk factors and understanding
   the evolving pathophysiological changes from early to late stages. Such
   an approach is imperative for creating effective prevention and
   treatment strategies for heart failure within the T2DM population.

   Experimental models, including both animal models and in vitro cell
   models, have provided significant mechanistic and phenotypic insights
   into the pathogenesis underlying DbCM.^[ [60]^18 , [61]^19 ^]
   Transgenic, drug‐induced, and high‐glucose/high‐fat animal models have
   greatly enhanced our understanding of DbCM pathophysiology, including
   mitochondrial dysfunction, impaired Ca^2+ handling, ER stress and cell
   death.^[ [62]^17 , [63]^18 , [64]^19 , [65]^20 ^] However, limitations
   such as species‐specific genetic backgrounds and long experimental
   schedules persist in these models. Primary adult human cardiomyocytes
   are optimal for in vitro studies, but they face challenges related to
   sample scarcity and difficulties in culture. In this context, human
   induced pluripotent stem cell (hiPSC)‐derived cardiomyocytes
   (hiPSC‐CMs) present a promising alternative, providing a substantial
   number of seed cells for phenotypic modeling and risk factor assessment
   in DbCM research.^[ [66]^21 , [67]^22 , [68]^23 ^] However, iPSC‐CMs
   exhibit structural, metabolic, and electrophysiological immaturity
   compared to adult cardiomyocytes.^[ [69]^24 , [70]^25 ^] These immature
   properties can be improved through long‐term culture (up to 360
   days),^[ [71]^26 ^] biochemical factors, mechanical strain, or
   electrical stimulation.^[ [72]^24 , [73]^27 , [74]^28 , [75]^29 ^]
   Meanwhile, 3D culture systems such as engineered heart tissues (EHTs)
   and cardiac organoids have demonstrated superior maturation of
   hiPSC‐CMs compared to traditional 2D cultures,^[ [76]^30 , [77]^31 ^]
   and are now widely used to model various cardiomyopathies and for drug
   screening applications,^[ [78]^32 , [79]^33 , [80]^34 ^] including
   those related to DbCM.^[ [81]^35 , [82]^36 ^] Using an organoid system,
   researchers have successfully modeled embryonic heart development under
   pregestational diabetes‐like conditions, revealing ROS‐mediated ER
   stress and abnormal lipid metabolism.^[ [83]^37 ^] Another recently
   study evaluated the cardiotoxic effects induced by a high‐glucose and
   high‐lipid environment, including apoptosis, oxidative stress, and
   mitochondrial dysfunction, in spherical cardiac organoids.^[ [84]^38 ^]
   However, none of these current models comprehensively address the
   dynamic alterations in contractile and electrophysiological properties
   characteristic of diabetic hearts.^[ [85]^17 ^] Spheroid organoids rely
   on self‐aggregation through spontaneous co‐differentiation of cells in
   3D aggregates, while hEHTs are constructed by embedding hiPSC‐CMs
   within a hydrogel scaffold to form bundles, patches, or other
   structures according to different needs, which provides relatively
   better structural and/or electromechanical maturation compared to
   spheroids.^[ [86]^39 , [87]^40 , [88]^41 , [89]^42 ^] Although the
   long‐term stability of hEHTs can be further improved through more
   flexible technologies and compatible hydrogels, the hEHT‐based model
   offers a promising and adaptable platform for modeling the dynamic
   changes in contractile and electrical properties of DbCM, facilitating
   further drug screening, verifying the efficacy and mechanism of
   potential drugs, and serving as an important supplement to animal
   models.

   Sodium‐glucose cotransporter‐2 inhibitors (SGLT‐2is) are a class of
   prescription medications approved to lower blood sugar in adults with
   type 2 diabetes (T2DM) when combined with diet and exercise.^[ [90]^43
   ^] Beyond glycemic control, SGLT2is, including dapagliflozin and
   empagliflozin (EMPA), have demonstrated benefits in managing heart
   failure.^[ [91]^44 , [92]^45 ^] They reduce the risk of cardiovascular
   hospitalization and death in patients with HFrEF (heart failure with
   reduced ejection fraction)^[ [93]^46 ^] and lower the risk of
   cardiovascular and all‐cause mortality in those with HFpEF (heart
   failure with preserved ejection fraction).^[ [94]^47 , [95]^48 ^]
   However, whether SGLT2 inhibitors exert direct effects on the heart or
   act solely through systemic improvements remain controversial, and the
   underlying mechanisms of SGLT2i‐mediated cardioprotection continue to
   be a significant topic of interest in the field. On one hand, beyond
   their antihyperglycemic effects, SGLT2 inhibitors such as EMPA exhibits
   osmotic diuretic and natriuretic properties that contribute to
   improvements in cardiovascular and kidney diseases.^[ [96]^49 ^]
   Additionally, EMPA has been shown to reduce the formation of uremic
   toxins by modulating the microbiome, as demonstrated through
   comprehensive proteomics, phosphoproteomics, and metabolomics
   analyses.^[ [97]^50 ^] On the other hand, although SGLT2 is primarily
   expressed in the kidney and is virtually absent in the human heart,
   growing evidence indicates that SGLT2 inhibitors can directly inhibit
   sodium‐loaders such as sodium‐hydrogen antiporter 1 (NHE ‐1),
   voltage‐gated sodium channel Nav1.5, SGLT1, thereby exerting direct
   cardiac effects.^[ [98]^51 , [99]^52 , [100]^53 ^] Despite these
   promising findings, SGLT2 inhibitors remain underutilized in many
   eligible patients.^[ [101]^54 ^] Therefore, further studies employing
   diverse experimental systems or models are needed to clarify these
   controversies and provide new insights.

   To establish a research platform for investigating the dynamic changes
   in contractility associated with DbCM, this study employed hEHT to
   assess the detrimental effects of risk factors on the myocardium.
   Specifically, we examined the role of excessive palmitic acid (PA) in
   inducing DbCM. By integrating contractility, tissue structure,
   electrophysiology, and RNA sequencing analyses, we sought to elucidate
   the underlying biological processes contributing to the progression of
   DbCM from early diastolic impairment to systolic and diastolic
   dysfunction. Moreover, we investigated the therapeutic potential of
   empagliflozin, a first‐line diabetes medication, in iPSC‐CMs and hEHT
   models, providing direct evidence of its cardioprotective role and
   exploring the underlying mechanisms. Finally, we validated these
   findings in a T2DM mouse model.

2. Results

2.1. Identification of Palmitic Acid as a Representative Diabetic Pathogenic
Factor of Cardiac Injury

   Myocardial structural and functional damage in diabetic cardiomyopathy
   is primarily attributed to cardiomyocyte dysfunction and is influenced
   by various diabetes risk factors.^[ [102]^55 ^] To assess the relative
   impact of these risk factors, we utilized a CCK‐8 assay on iPSC‐CMs, as
   shown in the strategy flow chart in Figure [103]1A. The evaluated
   factors included high glucose, high fatty acid (palmitic acid, PA),
   endothelin 1 and cortisol^[ [104]^23 ^] alone or in combination. By
   implementing a step‐wise modulation of Wnt signaling, we achieved a
   high‐purity iPSC‐derived cardiomyocyte (iPSC‐CM) population exceeding
   80% purity (Figure [105]S1A–C, Supporting Information). Our findings
   revealed that PA exerted the most pronounced effect, significantly
   reducing cardiomyocyte viability (Figure [106]1B), particularly in
   synergy with varying glucose concentrations, mimicking distinct
   diabetic conditions (Figure [107]1C,D). Given the established
   correlation between serum free fatty acid levels and the severity of
   T2DM,^[ [108]^56 , [109]^57 ^] as well as our observed dose‐dependent
   results (Figure [110]S2, Supporting Information), we selected 500 µM PA
   as the stimulating condition, which induced a time‐dependent increase
   in the release of lactate dehydrogenase (LDH), a recognized marker of
   myocardial injury, as shown in Figure [111]1E. Moreover, calcium
   transient analysis was assessed to evaluate systolic and diastolic
   function. Compared to controls, PA‐treated cardiomyocytes exhibited a
   progressive decrease in amplitude starting at 24 h (Figure [112]1F,G),
   accompanied by reduced calcium release and reuptake rates, and
   shortened 50% of Ca^2+ transient duration (CaTD50) (Figure [113]1H–J),
   indicative of impaired systolic and diastolic function. Considering the
   well‐documented roles of mitochondrial dysfunction and ER stress in
   diabetic cardiomyopathy,^[ [114]^8 ^] we investigated mitochondrial and
   ER morphology within 48 h of PA treatment. Mitochondrial content
   transiently increased at 12 h before declining, coinciding with a
   gradual rise in mitochondrial fission (Figure [115]1K,L), resembling
   increased mitochondrial fission observed in DCM patients. Additionally,
   the endoplasmic reticulum exhibited abnormal aggregation and reduced
   overall quantity between 12‐ and 48‐h post‐PA treatment
   (Figure [116]1M). Altogether, these results indicated PA as a crucial
   pathogenic factor that impaired both cardiomyocyte function and
   viability.

Figure 1.

   Figure 1
   [117]Open in a new tab

   Palmitate acid represents a key diabetogenic risk factor that induces
   injury in iPSC‐CM. A) Scheme for evaluating the effects of diabetes
   risk factors on iPSC‐CMs. B–D) CCK8 assay showing survival of iPSC‐CM
   treated‐with high glucose (Glu), palmitic acid (PA), endothelin‐1(ET‐1)
   and cortisol (C) alone or in combination (B), or different
   concentrations of Glu or PA for 24 h (C) and 72 h (D) (n = 5). E)
   Lactate dehydrogenase (LDH) release from iPSC‐CM with PA treatment for
   the indicated time points. (n = 3). F–J) Representative images (F) and
   quantification (G) of Ca^2+ transient in PA‐treated iPSC‐CMs for
   24 h/48 h/72 h and the indicated control. iPSC‐CMs were loaded with
   Fluo‐4 AM and paced at 1 Hz, amplitude (H), maximum upstroke velocity
   (I), and duration at 50% repolarization (CaTD50) (J) were analyzed.
   K–M) Representative confocal images (K) and quantitative analysis of
   live Mito‐Tracker (L) and ER‐Tracker (M) staining in reseeded iPSC‐CMs
   at the indicated time points. (n = 8 images per group). ns p > 0.05, *
   p < 0.05, ** p < 0.01, *** p < 0.001, ^## p < 0.01 versus PA group
   in (B).

2.2. Palmitic Acid Induced Diastolic Dysfunction in Human Engineered Heart
Tissues in Short Term

   Human engineered heart tissue represents an ideal option for
   cardiomyopathy modeling.^[ [118]^58 ^] In this study, hiPSC‐derived
   EHTs were utilized for tissue‐level phenotype assessment (Figure
   [119]2A). As approximately one quarter of Type 2 diabetic individuals
   exhibit asymptomatic myocardial structural and functional
   abnormalities, with the majority presenting diastolic dysfunction.^[
   [120]^14 ^] Developing in vitro models to mimic diastolic dysfunction
   is crucial for studying diabetic cardiomyopathy. To dynamically
   evaluate changes in structure and contraction force, we designed a mold
   to construct 6 mm × 3 mm bundle‐shaped hEHTs (Figure [121]2B, Figure
   [122]S3A,B, Supporting Information). Using a custom‐built mechanical
   sensing and stretching platform, we investigated the early effects of
   PA on hEHT function while preserving normal systolic parameters.
   Although 24 h of PA stimulation did not significantly alter baseline or
   stretch‐induced active and passive forces (Figure [123]2C–E), it
   induced subtle reductions in contraction and relaxation Vmax (Figure
   [124]S4A,B, Supporting Information). Moreover, a notable increase in
   abnormal twin‐peak contractions under various stretch conditions in the
   PA group indicated impaired diastolic filling (Figure [125]2F),
   corroborating the development of early diastolic dysfunction.

Figure 2.

   Figure 2
   [126]Open in a new tab

   PA‐induced hEHT models of early diabetic cardiomyopathy injury. A)
   Scheme for evaluating the early and late effects of PA on hEHT. B)
   Physical image of hEHT on day 7. Scale bar = 1 mm. C–F) Schematic
   diagram (C) and statistical analysis of characteristic peaks, including
   active contractile force (D), passive contractile force (E), and the
   number of diastolic dysfunction peaks (F) under different stretching
   conditions of hEHT during progressive stretching with electrical
   stimulation (1.5 Hz) after 24 h of treatment in PA stimulation (n = 7
   per group). G) Immunofluorescence staining images (left) and
   quantification (right) of the percentage of Vimentin^+ cells in the
   hEHTs at 24 h post‐PA treatment. Scale bar = 100 µm. H) Lactate
   dehydrogenase (LDH) release from hEHTs following 24 h of PA treatment.
   (n = 4). I,J) Transmission electron microscopy (TEM) images showing the
   morphology of mitochondria and sarcomeres in hEHTs from both the PA and
   control groups (I), along with quantitative analysis of total
   mitochondrial area, individual mitochondrial area, and mitochondrial
   roundness (J). Scale bar = 500 nm. K–P) Optical mapping of hEHTs in the
   PA and control groups at 24 h, stimulated at a pacing frequency of
   1 Hz. The results included isochronal activation and propagation
   direction maps alongside calcium transient traces (K), as well as
   statistical analysis of Ca^2⁺ fluorescence signal conduction velocity
   (L), amplitude (M), maximum downstroke velocity (N), propagation
   direction dispersion (O), and the frequency distribution of two
   different types of diastolic dysfunction occurrences (P). n = 8 for the
   control group, n = 7 for the PA group. ns p > 0.05, * p < 0.05, ** p <
   0.01, *** p < 0.001, **** p < 0.0001.

   While immunofluorescence staining for vimentin did not reveal
   significant alterations in cell type and fibrosis (Figure [127]2G),
   increased LDH release (Figure [128]2H) indicated PA‐induced injury in
   hEHTs. Moreover, transmission electron microscopy (TEM) revealed an
   increase in individual mitochondrial area and a decrease in
   mitochondrial roundness (Figure [129]2I,J), indicative of morphological
   changes in mitochondrial fusion. We further assessed the
   electrophysiological characteristics of hEHTs using optical mapping. PA
   stimulation significantly reduced electrical conduction velocity,
   amplitude, and related parameters (Figure [130]2K–O). Moreover, the
   frequency of abnormal contraction peaks associated with diastolic
   dysfunction was markedly increased (Figure [131]2P). Altogether, these
   findings demonstrate that a short‐term exposure (24 h) to PA induces
   early diastolic dysfunction in hEHT without compromising systolic
   function. These changes are accompanied by subcellular structural
   alterations indicative of incipient hypertrophy and electrical
   conduction abnormalities.

2.3. Prolonged PA Stimulation Recapitulates Systolic Dysfunction Features of
Diabetic Cardiomyopathy in hEHTs

   Given the established presence of severe systolic dysfunction and
   pronounced cardiac symptoms in diabetic hearts, we investigated
   whether chronic PA exposure (for 72 h) induces apparent structural
   damage and impaired systolic function in hEHTs. Our findings revealed a
   significant decrease in the active force of hEHT in the PA group
   compared to the control (Figure [132]3A,B), indicative of reduced
   systolic function. Concurrently, a marked increase in passive force was
   observed (Figure [133]3C), suggesting augmented myocardial stiffness.
   Furthermore, the PA‐treated group exhibited significantly reduced
   contraction and relaxation rates (Figure [134]3D,E). Prolonged
   contraction and relaxation times, increased APD50/APD80 (Figure
   [135]S5A–D, Supporting Information), and an elevated proportion of
   abnormal diastolic peaks were also evident (Figure [136]3F).
   Collectively, these results demonstrated impaired systolic and
   diastolic function in hEHTs after 72 h of PA stimulation.

Figure 3.

   Figure 3
   [137]Open in a new tab

   PA‐induced hEHT models of later diabetic cardiomyopathy injury. A–F)
   Schematic diagram (A) and statistical analysis of characteristic peaks,
   including active contractile force (B), passive contractile force (C),
   maximum contraction velocity (D), maximum relaxation velocity (E), and
   the number of diastolic dysfunction peaks (F) under different
   stretching conditions of hEHT during progressive stretching with
   electrical stimulation (1.5 Hz) after 72 h of treatment in PA
   stimulation (n = 7 per group). G) Lactate dehydrogenase (LDH) release
   from hEHTs following 72 h of PA treatment. (n = 3). H)
   Immunofluorescence staining images (left) and quantification (right) of
   the percentage of Vimentin^+ cells in the hEHTs at 72 h post‐PA
   treatment. Scale bar = 100 µm. I) Representative TEM images and
   quantitative analysis of mitochondrial numbers in both groups. Scale
   bar = 500 nm. J–M) Optical mapping of hEHTs in the PA and control
   groups at 72 h, stimulated at a pacing frequency of 1 Hz. The results
   included isochronal activation and propagation direction maps alongside
   calcium transient traces (J), as well as statistical analysis of Ca^2⁺
   fluorescence signal conduction velocity (K), amplitude (L), and the
   frequency distribution of two different types of diastolic dysfunction
   occurrences (M). n = 8 for each group. * p < 0.05, ** p < 0.01, *** p <
   0.001, **** p < 0.0001.

   Furthermore, prolonged PA stimulation (72 h) resulted in a significant
   increase in LDH release (Figure [138]3G), indicative of severe hEHT
   injury. Immunofluorescence staining revealed a substantial increase in
   fibroblasts within the hEHT, predominantly in the inner region with
   concomitant cardiomyocyte loss (Figure [139]3H). Additionally, TEM
   demonstrated sarcomere disarray and a reduction in mitochondrial number
   (Figure [140]3I). Beyond structural damage, 72 h PA exposure led to a
   marked impairment of calcium signaling propagation in hEHTs, as
   evidenced by significantly decreased electrical conduction velocity,
   amplitude, and related parameters with optical mapping method
   (Figure [141]3J–L, Figure [142]S5E–G, Supporting Information).
   Moreover, the incidence of abnormal contraction peaks, particularly
   twin peaks, was dramatically increased (Figure [143]3M). Collectively,
   these findings demonstrate that prolonged PA stimulation recapitulates
   both systolic and diastolic dysfunction, as well as the structural
   hallmarks of diabetic cardiomyopathy in the hEHT model.

2.4. Short‐Term and Long‐Term PA Exposure Differentially Alters
DbCM‐Associated Gene Expression in hEHTs

   To elucidate the molecular mechanisms underlying short‐term and
   long‐term PA‐induced injury in hEHTs, we performed RNA‐sequencing
   analysis in each group at 24 and 72 h. Principal component analysis
   (PCA) demonstrated a clear separation between PA 24 h and control 24 h
   hEHT samples (Figure [144]4A), indicating substantial transcriptomic
   divergence. KEGG pathway analysis revealed significant downregulation
   of contractility‐related pathways, including cardiac muscle contraction
   and oxidative phosphorylation, coupled with upregulation of insulin
   resistance, fatty acid metabolism, and calcium signaling pathways in
   the PA group (Figure [145]4B–F). Notably, gene ontology (GO) analysis
   of biological processes corroborated the downregulation of
   mitochondrial translation (Figure [146]4G), supporting the notion of
   mitochondrial dysfunction during DbCM processing.

Figure 4.

   Figure 4
   [147]Open in a new tab

   Signaling pathways related to DCM in PA‐induced hEHT models. A)
   Principal component analysis of the global gene expression profiles
   between CON1d and PA1d as revealed by RNA‐seq. B–G) Kyoto Encyclopedia
   of Genes and Genomes (KEGG) pathway enrichment analysis was performed
   on up‐regulated genes and down‐regulated genes (p‐value < 0.05), and
   the top 10 KEGG pathways were listed with their p‐values and counts
   (B). Heatmaps showing the upregulation of pathways related to insulin
   resistance (C), fatty acid metabolism (D), and calcium signaling (E),
   as well as the downregulation of pathways related to oxidative
   phosphorylation (F) and mitochondrial translation (G) in the PA1d group
   compared to the CON1d group. H) Principal component analysis of global
   gene expression profiles between CON3d and PA3d. I–K) KEGG pathway
   enrichment analysis showed the top 10 pathways with upregulated and
   downregulated genes, along with their p‐values and counts (I). Heatmaps
   indicated that the TNF signaling pathway (J) was upregulated, while
   cardiac muscle contraction (K) was downregulated in the PA3d group. L)
   GSEA showed enrichment of the unfolded protein response and apoptosis
   pathways in the PA3d group. M) The top 10 biological process terms
   identified through Gene Ontology enrichment analysis were listed. N–P)
   Heatmaps displayed the DEGs in BP terms, including upregulation of
   apoptotic process (N), protein processing in endoplasmic reticulum (O),
   and downregulation of sarcomere organization (P), mitochondrial ATP
   synthesis coupled proton transport (Q) in the PA3d group compared with
   CON3d.

   Following 72 h of exposure, principal component analysis (PCA)
   effectively distinguished the PA group from the control group
   (Figure [148]4H). KEGG pathway analysis revealed significant
   upregulation of pathways associated with protein processing in the
   endoplasmic reticulum (ER), TNF signaling, and insulin resistance.
   Conversely, pathways related to cardiac muscle contraction and diabetic
   cardiomyopathy were significantly downregulated (Figure [149]4I–K).
   Consistently, GSEA of these genes revealed enrichment in unfolded
   protein response and apoptotic pathways (Figure [150]4L). Furthermore,
   GO analysis of biological processes identified upregulated genes
   involved in apoptosis and ER stress pathway (Figure [151]4M–O), while
   mitochondrial ATP synthesis coupled to proton transport and sarcomere
   organization were downregulated in the PA group (Figure [152]4P,Q).
   These findings collectively demonstrate that prolonged PA exposure
   predominantly elicits cell apoptosis and ER stress, contributing to
   subsequent structural and functional impairments. Given the critical
   roles of mitochondrial dysfunction and ER stress in diabetic
   cardiomyopathy,^[ [153]^8 ^] our model exhibits a high degree of
   functional and molecular similarity to this condition, suggesting its
   utility as a valuable in vitro platform for further investigations.

2.5. Empagliflozin Ameliorates PA‐Induced Diastolic Dysfunction in iPSC‐CM
and hEHT Models

   To investigate the potential therapeutic efficacy of empagliflozin, a
   frontline medication for diabetes, in the PA‐induced lipotoxic DbCM
   model, we initially assessed the effects of EMPA on cardiomyocyte
   viability using CCK‐8 assay. Our findings demonstrated that EMPA
   significantly improved cardiomyocyte viability at both 24‐ and 48‐h
   post‐PA exposure (Figure [154]5A), although LDH levels were only
   reduced at the 24‐h timepoint (Figure [155]5B). Furthermore,
   empagliflozin restored calcium handling capacity, including calcium
   transients, release rate, reuptake rate, as well as CaTD50 and CaTD 80,
   in iPSC‐CMs exposed to PA for 24 h (Figure [156]5C–H). Noting that
   SGLT2 inhibitors typically achieve relatively low serum concentrations,
   far below 22 µM (≈10 mg),^[ [157]^59 , [158]^60 ^] we conducted a
   dose‐response analysis which revealed that only high concentrations of
   empagliflozin (EMPA) significantly ameliorated PA‐induced cardiomyocyte
   injury (Figure [159]S6A–D, Supporting Information). These results
   suggest that high‐dose empagliflozin exhibits early cardioprotective
   effects in this PA‐induced cardiomyopathy model.

Figure 5.

   Figure 5
   [160]Open in a new tab

   Protective Role of Empagliflozin against PA‐Induced Injury in iPSC‐CMs
   and hEHTs. A,B) CCK8 (A, n = 5) and LDH (B, n = 5) assays of iPSC‐CM
   treated‐with PA, PA + empagliflozin (EMPA) and control for 24 h and
   48 h. C–H) Representative images (C) and quantification of Ca^2+
   transient in iPSC‐CMs from PA, PA+EMPA and control group after 24 h.
   iPSC‐CMs were loaded with Fluo‐4 AM and paced at 1 Hz, the parameters
   including amplitude (D), maximum upstroke velocity (E), maximum release
   velocity (F), the duration at 50% and 80% repolarization
   (CaTD50/CaTD80) (G,H) were analyzed. I,K) Statistical analysis of
   contraction force, including active contractile force (I), maximum
   relaxation velocity (J), and heatmaps showing the percentage of
   diastolic dysfunction peaks (K) after 24 h of treatment in PA, PA+EMPA
   and control group. n = 6 per group. L–P) Optical mapping of hEHTs in
   the PA, PA+EMPA and control groups at 24 h, stimulated at a pacing
   frequency of 1 Hz. The results included isochronal activation and
   propagation direction maps alongside calcium transient traces (L), as
   well as statistical analysis of Ca^2⁺ fluorescence signal conduction
   velocity (M), amplitude (N), maximum upstroke velocity (O), and the
   frequency distribution of two different types of diastolic dysfunction
   occurrences (P). n = 9 for the CON group, n = 9 for the PA group, n = 8
   for the PA+EMPA group. ns p > 0.05, * p < 0.05, ** p < 0.01, *** p <
   0.001.

   To further investigate the potential protective effects of
   empagliflozin in early‐stage diabetic cardiomyopathy, we focused on
   hEHTs exposed to short‐term PA stimulation. Empagliflozin significantly
   ameliorated early‐stage diastolic dysfunction in hEHTs, as evidenced by
   slight improvements in active and passive force, significantly
   recovered contraction and relaxation rates, and APD50
   (Figure [161]5I,J, Figure [162]S7A–C, Supporting Information). Notably,
   empagliflozin reduced the incidence of abnormal diastolic peaks
   compared to the PA group (Figure [163]5K). Furthermore, empagliflozin
   treatment attenuated PA‐induced impairments in electrical conduction
   velocity and amplitude, and markedly decreased the occurrence of
   abnormal contraction peaks (Figure [164]5L–P). These findings suggest
   that empagliflozin may mitigate lipotoxic myocardial injury,
   particularly during the early phases of diastolic dysfunction.

   To elucidate the mechanisms underlying empagliflozin's protective
   effects against PA‐induced diastolic dysfunction in hEHTs, RNA‐seq
   analysis was performed (Figure [165]6A). KEGG pathway analysis
   revealed that empagliflozin treatment upregulated pathways associated
   with cardiomyopathy, primarily involving genes related to cardiac
   muscle contraction, especially those involved in mitochondrial
   transcription and ATP synthesis (Figure [166]6B,C). Thus, JC‐1 staining
   was performed and revealed that EMPA ameliorates the PA‐induced
   decrease in mitochondrial membrane potential in iPSC‐CMs
   (Figure [167]6D). Meanwhile, both KEGG and GSEA analyses indicated
   downregulation of apoptosis, ER stress, and inflammatory pathways in
   the EMPA+PA group (Figure [168]6B,E, Figure [169]S8A–D, Supporting
   Information). Moreover, GO analysis identified a reduction in apoptotis
   and ER stress, ERAD, and unfolded protein response‐related genes in the
   PA + empagliflozin group (Figure [170]6F–I). To validate these
   sequencing results, additional experiments were conducted. The
   anti‐apoptotic effect of EMPA was confirmed by TUNEL staining in
   iPSC‐CMs (Figure [171]6J). Protein expression analysis of key pathway
   components revealed that ER stress‐related proteins were significantly
   upregulated by PA treatment but markedly reduced by EMPA after 1 day in
   hEHTs (Figure [172]6K,L). Collectively, these results suggest that
   empagliflozin ameliorates PA‐induced hEHT injury by suppressing
   apoptosis and ER stress, while concurrently promoting cardiac structure
   and function.

Figure 6.

   Figure 6
   [173]Open in a new tab

   Empagliflozin Protects hEHT and iPSC‐Derived Cardiomyocytes from
   Palmitate‐Induced Injury by Inhibiting Mitochondrial Dysfunction,
   Apoptosis, and ER Stress. A) Principal component analysis of global
   gene expression profiles between PA24h and PA+EMPA24h. B,C) KEGG
   pathway enrichment analysis showed the top 10 pathways with upregulated
   and downregulated genes, along with their p‐values and counts (B).
   Heatmaps showing the upregulation of pathways related to cardiac muscle
   contraction in the PA+EMPA group compared to the PA group (C). D)
   Representative fluorescent image (left) and quantification (right) of
   mitochondrial membrane potential measure by JC‐1 staining in iPSC‐CMs.
   Red fluorescence indicates the mitochondrial JC‐1 aggregate JC‐1, while
   green fluorescence represents the monomeric form of JC‐1. Bars = 50 µm.
   E) GSEA showed enrichment of apoptosis and unfolded protein response
   pathways in the PA group. F) The top 10 biological process terms
   identified through Gene Ontology enrichment analysis were listed. G–I)
   Heatmaps displayed the DEGs in BP terms, including downregulation of
   apoptotic process (G) and ER stress (H‐I) in the PA+EMPA group compared
   with the PA group. J) Representative fluorescent images (left) and
   quantification (right) of apoptotic cardiomyocyte rates in iPSC‐CMs
   were assessed by TUNEL assay in the CON, PA, and PA+EMPA groups at 24
   h. Bars = 50 µm. K‐L) Immunoblotting analyses of GRP78 and CHOP protein
   levels in hEHTs treated with CON, PA, or PA+EMPA for 24 h. β‐Actin was
   used as a loading control. n = 3 biological replicates. ns p > 0.05, *p
   < 0.05, **p < 0.01, ***p < 0.001.

   To investigate why EMPA exerts a direct protective effect against
   PA‐induced injury in hiPSC‐CMs and hEHTs despite the negligible
   expression of SGLT2, we analyzed the reported potential targets of
   SGLT2 inhibitors in our hEHT sequencing data. We found that the FPKM
   value of SGLT2 was extremely low, whereas its homolog SGLT1 was readily
   detectable, a finding further confirmed by western blotting (Figure
   [174]S9A,B,C Supporting Information). Additionally, other potential
   targets including Nav1.5, NHE1, and NCX1 were also exhibited relatively
   high FPKM values (Figure [175]S9A, Supporting Information).
   Importantly, protein levels of SGLT1 and NHE1 did not show significant
   changes after treating hEHTs with PA or PA+EMPA for 1 day (Figure
   [176]S9B,C Supporting Information), suggesting that EMPA likely
   inhibits channel activity rather than regulating protein expression
   under these conditions.

2.6. Empagliflozin Protects Mice from HFD + STZ Induced Functional Impairment
and Electrical Uncoupling

   To further validate the protective role of empagliflozin (EMPA) in a
   type 2 diabetes mellitus (T2DM) model, we employed a murine model
   induced by a high‐fat diet in conjunction with low‐dose streptozotocin
   (STZ) injections (Figure [177]7A). Hyperglycemic mice were subsequently
   treated with either daily gavage of EMPA or a saline control for 8
   weeks. Echocardiographic analysis revealed that T2DM mice exhibited
   systolic dysfunction, characterized by a decrease in ejection fraction
   and fractional shortening (Figure [178]7B–D). Doppler echocardiography
   indicated elevated E/e' ratios and decreased E/A and e“/a” ratios in
   T2DM mice compared to controls, suggesting the presence of diastolic
   dysfunction (Figure [179]7E–H). Notably, EMPA treatment significantly
   attenuated these functional impairments (both systolic and diastolic)
   and structural dysfunctions (Figure [180]7B–H). Ultrastructural
   analysis demonstrated significant mitochondrial injury and lipid
   droplet accumulation in the hearts of T2DM mice, which were markedly
   improved following EMPA treatment (Figure [181]7I–L). Furthermore,
   ventricular electrical conduction, assessed via activation mapping,
   revealed uniform electrical conduction with orderly propagation in
   control hearts. In contrast, T2DM hearts displayed disordered left
   ventricular conduction with aberrant conduction sites. The T2DM+EMPA
   group exhibited restored conduction patterns, correlating with improved
   conduction velocity, amplitude, and Vmax velocity (Figure [182]7M–P).
   Last, EMPA reduced the LDH content in the serum of T2DM mice heart
   (Figure 7Q), and the anti‐apoptotic effect of EMPA was further
   confirmed by TUNEL staining of mouse paraffin‐embedded heart sections
   (Figure [183]7R,S). Collectively, these results demonstrated that EMPA
   protects against diabetes‐induced cardiac dysfunction and abnormalities
   in electrical conduction.

Figure 7.

   Figure 7
   [184]Open in a new tab

   Empagliflozin protects against cardiac injury in T2DM mice. A)
   Schematic protocol of T2DM induction and EMPA administration in C57/BL6
   mice. B–H) Representative images of echocardiography (B, E) and
   quantification of parameters including left ventricular ejection
   fraction (LVEF, C) and left ventricular Fractional shortening (LVFS,
   D), Mitral Valve Velocity E/A Ratio (F), e“/a” Ratio (G) and E/a' Ratio
   (H) from the NC + vehicle group (n = 6), the HFD + Vehicle group (n =
   7), and the HFD + Empagliflozin group (n = 7). I,J) Representative TEM
   images (I) and quantitative analysis of mitochondrial roundedness (J)
   from each group. Yellow arrows indicated lipid droplets, orange circles
   indicated mild mitochondrial damage, and the red circles represented
   severe mitochondrial damage. K) The area of lipid droplets was
   quantified in the HFD+EMPA and HFD+veh groups. (n = 10 images per
   group). M–P) Optical mapping of hearts isolated from the NC+veh,
   HFD+Vehicle, and HFD+EMPA group stimulated at 8 Hz pacing. Statistical
   analysis of Ca^2⁺ fluorescence signal conduction velocity (N),
   amplitude (O), and maximum upstroke speed (P) were derived from over
   three recordings from one mouse per group. Q) Lactate dehydrogenase
   (LDH) levels were measured in the serum of mice from the NC‐Veh,
   HFD‐Veh, and HFD+EMPA groups (n = 6 mice per group). R,S)
   Representative fluorescent images (R) and quantification (S) of
   apoptotic cardiomyocyte rates in heart sections from each group are
   shown. Data were obtained from n = 3 mice per group, with at least
   three high‐power fields analyzed per mouse. Bars = 50 µm. ns p > 0.05,
   *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

3. Discussions

   Diabetes poses a significant public health threat due to its
   substantial contribution to cardiovascular disease (CVD) morbidity and
   mortality, particularly heart failure.^[ [185]^3 , [186]^4 , [187]^5 ^]
   However, the underlying mechanisms of diabetic myocardial
   disorder/diabetic cardiomyopathy remain elusive due to the complex
   interplay of risk factors and systemic effects. Traditional animal
   models often replicate the end‐stage events observed in clinical
   patients, making it challenging to distinguish primary from secondary
   factors and to establish causal relationships between risk factors and
   eventual outcomes. Therefore, the development of in vitro tissue models
   that accurately recapitulate these processes is essential for advancing
   effective prevention and treatment strategies for heart failure in the
   T2DM population. iPSCs‐related models have emerged as a robust
   substitute for animal testing in preclinical drug development, as
   validated by the FDA.^[ [188]^61 ^] However, an ideal DbCM model that
   can fully elucidate the complex interplay of risk factors leading to
   heart failure is still lacking. In this study, our hEHT model
   recapitulated the detrimental effects of excess saturated fatty acids
   under controlled high glucose culture conditions, leading to impaired
   staged myocardial diastolic and systolic function, coupled with
   all‐stagey abnormal electrical conduction, as well as myocardial
   structural damage, mitochondrial dysfunction, and cardiac fibrosis.
   Notably, our results first demonstrated that empagliflozin directly
   attenuated HFD/PA induced functional decrease and electrical
   abnormality both in mouse model and hEHT model, and reduced PA‐induced
   endoplasmic reticulum stress, calcium dysregulation, and apoptosis in
   hEHT, which are early‐stage contributors to diastolic dysfunction.
   These findings provide compelling evidence supporting the clinical
   application of empagliflozin for heart failure with preserved ejection
   fraction (HFpEF)‐related myocardial symptoms.

   Although the exact definition and clinical features of diabetic
   cardiomyopathy or diabetic myocardial disorder remain controversial,
   there is an undeniable association between T2DM and the development of
   HF, and it emphasizes that even patients with T2DM who have
   asymptomatic cardiac abnormalities are at increased risk for HF.^[
   [189]^62 ^] Notably, the detection of asymptomatic structural and
   functional cardiac abnormalities through imaging modalities, often
   manifested as early diastolic dysfunction,^[ [190]^14 , [191]^15 ^] is
   crucial for identifying individuals at high risk of HF. Previous study
   utilized murine myocardium‐derived EHTs treated with AGEs recapitulated
   the phenomenon of increased ROS, inflammatory pathway, cardiac
   hypertrophy and fibrosis by western blot and qPCR.^[ [192]^36 ^]
   Another study used heart organoid with pregestational diabetes‐like
   conditions to revealed the alteration of epicardial and cardiomyocyte
   dysfunction and lipid metabolism.^[ [193]^37 ^] However, these models
   lacked tissue‐level functional change as the new definition emphasized
   systolic and/or diastolic dysfunction. Our system supports the
   investigation of contraction force and electrophysiological function
   dynamically, which allows more choices for understanding the
   contributions of different diabetic risk factors for further study.

   Mitochondrial dysfunction, ER stress, and inflammation are key factors
   in the development and progression of diabetic cardiomyopathy.^[
   [194]^9 ^] Notably, human diabetic hearts exhibit an increased reliance
   on fatty acid‐based energy production^[ [195]^63 ^] yet show reduced
   ATP efficacy and heightened mitochondrial fragmentation.^[ [196]^64 ,
   [197]^65 , [198]^66 ^] Our hEHT model effectively recapitulated these
   characteristics, as evidenced by mRNA sequencing that revealed a
   downregulation of genes involved in mitochondrial translation
   (Figure [199]4G). Concurrently, there was an upregulation of genes
   associated with the ER unfolded protein response (UPR), TNF signaling,
   and insulin resistance pathways (Figure [200]4C,J,O). Transmission
   electron microscopy (TEM) analysis demonstrated changes in
   mitochondrial morphology, which aligned with elevated expression of
   fatty acid oxidation genes at 24 h (Figure [201]4D). However, this was
   accompanied by a decline in contractile function and electrical
   conduction capacity observed at 72 h (Figure [202]3A–M), indicating a
   dynamic transition from initial mitochondrial function impairment to
   subsequent dysfunction.

   In this study, we developed an in vitro tissue model to investigate
   lipotoxic myocardial injury. Our findings suggest that EMPA directly
   ameliorates abnormal diastolic peaks and Ca^2⁺ activity resulting from
   palmitate acid‐induced dysfunction. This protective effect is primarily
   mediated through the inhibition of apoptosis, mitochondrial injury, and
   ER stress, highlighting EMPA's direct impact on cardiac function. As a
   newly FDA‐approved combination treatment for heart failure in
   individuals with or without T2DM, EMPA's cardioprotective mechanisms
   remain incompletely understood, warranting further investigation.
   Although SGLT2 is virtually undetectable in the heart, its homolog
   SGLT1 is the predominant SGLT family member expressed in the myocardium
   and can be inhibited by SGLT2 inhibitors.^[ [203]^67 , [204]^68 ^]
   SGLT1 is recognized as an important potential target in cardiovascular
   disease.^[ [205]^68 , [206]^69 , [207]^70 , [208]^71 , [209]^72 ^]
   Overexpression of SGLT1 in mice leads to pathological hypertrophy and
   cardiac dysfunction, while knockdown attenuates these effects.^[
   [210]^73 ^] Moreover, SGLT1 plays a significant role in glucose
   fluctuation‐induced cardiac injury via oxidative stress and
   mitochondrial dysfunction.^[ [211]^74 ^] In addition, several studies
   have demonstrated that SGLT2is^[ [212]^51 ^] possess potential
   inhibitory activity against various sodium‐related channels, including
   sodium‐hydrogen antiporter (NHE1),^[ [213]^75 , [214]^76 , [215]^77 ,
   [216]^78 ^] Nav1.5,^[ [217]^52 , [218]^79 , [219]^80 ^] and NCX1.^[
   [220]^81 , [221]^82 ^] We analyzed the expression of these potential
   targets in our sequencing data and found that SGLT2 expression (FPKM)
   was nearly zero, while other genes were detectable (Figure [222]S9A,
   Supporting Information). Western blotting confirmed the protein levels
   of SGLT1 and NHE1, with no significant changes observed at either the
   transcriptional or translational level when hEHTs were treated with
   PA+EMPA compared to PA alone for 1 day (Figure [223]S9B–C, Supporting
   Information). Although EMPA exhibits over 2000‐fold higher selectivity
   for SGLT2 (IC[50]: 3.1 nM) than for SGLT1 (IC[50]: 8.3 µM),^[ [224]^83
   ^] our dose‐response assays revealed that only high concentrations of
   EMPA provided significant protection against PA‐induced cardiomyocyte
   injury (Figure [225]S6A–D, Supporting Information), preserved
   mitochondrial membrane potential (Figure [226]6D), and improved hEHT
   function (Figures [227]4, [228]5, [229]6). These results suggest that
   SGLT1 may be a key target through which EMPA exerts its protective
   effects in our system. However, we cannot exclude the involvement of
   other sodium‐related channels. For instance, NCX1 has been reported to
   participate in ER stress in neurons and islet β‐cells,^[ [230]^84 ,
   [231]^85 , [232]^86 ^] truncated variants of Nav1.5 can initiate the
   unfolded protein response (UPR) in the ER in systolic human heart
   failure,^[ [233]^87 ^] and NHE1 transgenic mice exhibit an increased ER
   stress response in the myocardium.^[ [234]^88 ^] In the future, the
   generation of SGLT1 and other candidate gene knockout iPSC‐derived
   cardiomyocytes and hEHTs could be utilized to specifically identify the
   primary targets of EMPA's protective mechanisms in this context.

   Our results support the notion that EMPA enhances both systolic and
   diastolic function, as well as electrophysiological function, in a
   diet‐ and drug‐induced DbCM model. This finding is consistent with
   observations from other animal models, such as db/db mice, where EMPA
   has been shown to alleviate oxidative stress and mitochondrial
   dysfunction,^[ [235]^89 ^] coronary microvascular dysfunction,^[
   [236]^90 ^] and improving diastolic function.^[ [237]^91 ^] Moreover,
   EMPA has been reported to improve endothelial function by reducing
   mitochondrial reactive oxygen species (ROS) in endothelial cells (ECs),
   which helps decrease frailty in diabetic and hypertensive patients.^[
   [238]^92 ^] Our results offer additional insights into the beneficial
   actions of EMPA in injured tissue models. However, the precise targets
   of EMPA and SGLT2 inhibitors within the myocardium, along with their
   underlying mechanisms, still require further clarification through both
   clinical evidence from heart samples and further experimental studies.
   This remains a critical area for future research.

   In summary, we have developed an engineered DbCM model that simulates
   lipotoxic myocardial injury, characterized by early‐stage diastolic
   impairment and late‐stage systolic dysfunction. Utilizing this model,
   we provide direct evidence that EMPA improves PA‐induced hEHT injury by
   inhibiting mitochondrial injury, ER stress, and cell death, which
   suggests that EMPA may help alleviate diabetic diastolic dysfunction.

4. Experimental Section

hiPSC Culture and Cardiomyocyte Differentiation

   Human iPSC line WTC (Gladstone Institute, derived from fibroblast) was
   used. Cardiac differentiation was conducted following the Wnt signaling
   protocol as previously described.^[ [239]^42 ^] Briefly, hiPSCs were
   dissociated to single cells (Versene, Thermo Fisher, 15 040 066) and
   reseeded onto Matrigel‐coated (Corning, 354 262) plates at a density of
   3.0 × 10^4 cells cm^−2. Cells were cultured in mTeSR1 medium (Stemcell,
   85 850) supplemented with Y‐27632 (Stem cell, 72 308) and incubated at
   37 °C with 5% CO[2] for 24 h. Then, the medium was replaced with a
   complete mTeSR1 medium without Y‐27632 and changed every 24 h. When the
   confluence reached 80%, the medium was switched to differentiation
   basal medium (RPMI 1640 containing B27‐insulin and 200 µg mL^−1
   ascorbic acid (Sigma A8960)) and treated with 7.5 µM CHIR99021 (Tocris,
   4423). After 48 h, the medium was replaced with differentiation basal
   medium for 24 h. Next, 5 µM IWR‐1 (Tocris, 3532) was added to the
   differentiation basal medium for 48 h. IWR‐1 was then removed, and
   differentiation basal medium was changed every 48 h until day 14 of
   differentiation. From day 14 onward, the cardiomyocytes were used
   either for generation of 3D hEHTs or reseeded onto plates for further
   assays.

Flow Cytometry

   Cardiomyocytes derived from hiPSCs were dissociated into single cells
   using 0.25% trypsin‐EDTA at room temperature for 5 min. Harvested cells
   were then washed twice with DPBS (containing 2% FBS) in a 15 mL
   centrifuge tube and fixed with 4% paraformaldehyde (Biosharp, BL539A)
   for 15 min. Following three washes with DPBS, the cells were blocked
   with blocking buffer (DPBS containing 5% donkey serum and 0.2% Triton
   X‐100) at 4 °C overnight. Next, the cells were incubated with PE‐CTNT
   (1:20, BD Pharmingen, 564 767) in blocking buffer at 4 °C for 1 h.
   After washing twice with DPBS, the cells were resuspended in 50–120 µL
   of DPBS for flow cytometry analysis.

CCK8 Assay

   Seed 14‐d differentiated iPSC‐CMs at a density of 0.75 million
   cells well^−1 in a 96‐well plate containing 200 µL of culture medium
   per well. Cells were cultivated for 3 days before starting the
   experiment. In the experimental groups with different factors (PA, Glu,
   and AGE), or in the PA experimental group, cells were exposed to
   different concentrations (200, 500, and 750 µmol L^−1) of Palmitic acid
   (PA, Sigma, P5585) and incubated for 24 or 72 h. Then, 10 µL well^−1 of
   CCK 8 reagent (Yeasen,40203ES76) was incubated with 90 µL well^−1 of
   cell culture medium mixture for 2 h, and the absorbance value (A) was
   measured at a wavelength of 450 nm using a Paradigm Molecular Devices
   (Gemini XPS). Cell viability was calculated as [(As‐Ab)/(Ac‐Ab)] ×100%,
   where “As” is the absorbance value of experimental well, “Ac” is the
   absorbance value of control well, and Ab is the absorbance value of
   blank well.

Lactate Dehydrogenase Release Assay

   Cell cytotoxicity was measured using an LDH assay kit (Nanjing
   Jiancheng Bioengineering Institute; A020‐2‐2) according to the
   manufacturer's protocols. Briefly, 20 µL of supernatant samples were
   collected from cardiomyocytes stimulated with different stimulies and
   loaded onto a 96‐well plate. Set up control and experimental groups,
   each with five replicates. Prepare the reaction system according to the
   experimental protocol (no coenzyme I in the control group, and addition
   of coenzyme I in the experimental group). Mix well and incubate at
   37 °C for 15 min, followed by the addition of
   2,4‐dinitrophenylhydrazine and further incubation at 37 °C for 15 min.
   The mixture was then incubated with NaOH solution room temperature for
   5 min. The absorbance value (A) of each well was measure at a
   wavelength of 450 nm.

Mitochondrial and Endoplasmic Reticulum Morphology Analysis

   HiPSC‐CMs were seeded on confocal dishes and treated with PA at
   different time points. To visualize mitochondrial or endoplasmic
   reticulum morphology changes, cardiomyocytes were incubated in a
   working solution containing 1 µM Mitotracker Red (Thermo Fisher,
   [240]M22425) or ER Tracker Green (Thermo Fisher, [241]E34251) at 37 °C
   with 5% CO[2] for 15 min. Subsequently, Hoechst staining was applied
   for 10 min, followed by washing twice. After that, the cells were
   incubated with 2.5 µM Blebbistatin (Sigma, B0560) for 3 min to suppress
   cardiomyocyte contractions. Imaging was performed using an LSM 980
   inverted laser scanning confocal microscope with a 63× oil immersion
   lens. Data analysis was conducted using the MINA plugin in Fiji‐ImageJ
   software.

Ca^2+ Transient and Intracellular Calcium Content Measurement

   HiPSC‐CMs were incubated with a working solution of 2.5 µM Fluo‐4 AM
   (Thermofisher, [242]F14201) for 20 min at 37 °C in a 5% CO[2]
   humidified incubator. The staining solution was then aspirated, and the
   cardiomyocytes were then incubated in a working solution of 2.5 µM
   Blebbistatin for 3 min to inhibit contractions. Calcium transients were
   recorded using line scan mode of LSM 980 microscopy at 1 Hz with a 63×
   oil immersion objective. Calcium activity changes within the CMs were
   recorded for 10 s. To assess intracellular calcium content,
   cardiomyocytes were placed in calcium‐free Tyrode's solution (140 mM
   NaCl, 5 mM KCl, 2 mM MgCl[2], 10 mM glucose and 10 mM HEPES at pH 7.4).
   Throughout the recordings, the cells were maintained in a heated
   chamber at 37 °C. Calcium transient data were analyzed using custom
   PyCharm and MATLAB programs.

Measurement of Mitochondrial Membrane Potential

   Mitochondrial membrane potential (MMP) was assessed using the JC‐1 dye
   (Thermo Fisher, T3168) following the manufacturer's instructions.
   Briefly, hiPSC‐CMs were washed and incubated with JC‐1 dye (1:1000
   dilution) and 1 mg mL^−1 Hoechst at 37 °C for 20 min, then washed again
   before imaging with a laser confocal microscope. The red fluorescence
   emitted by JC‐1 aggregates was detected at an emission wavelength of
   572 nm, while the green fluorescence from JC‐1 monomers was monitored
   at 517 nm. The ratio of red (aggregate) to green (monomer) fluorescence
   was quantified using ImageJ software to evaluate changes in
   mitochondrial membrane potential. A decreased JC‐1 ratio indicates
   mitochondrial depolarization, reflecting a loss of mitochondrial
   membrane potential.

hiPSC‐Derived 3D hEHTs Construction and Culture

   The hEHT was constructed as previously described.^[ [243]^40 ^] On day
   14 of differentiation, 1.2 × 10^6 cells were added into 58 µL 1×
   cardiac media consisting of RPMI 1640 medium (Gibco, C11875500BT),
   2 µg mL^−1 Vitamins B12 (Sigma‐Aldrich, V2876), 2 mg mL^−1
   6‐Aminocaproic acid (Sigma‐Aldrich, A2504), 1% Penicillin‐Streptomycin
   (Thermo Fisher, 15 140 122), 10% FBS and 2.4 µL 50 U mL^−1 thrombin
   (Sigma‐Aldrich, T7201). Then the cardiac medium was mixed with the
   hydrogel solution consisting of 24 µL 2× cardiac medium, 24 µL
   10 mg mL^−1 fibrinogen (Sigma‐Aldrich, F4883), and 12 µL Matrigel
   (Corning, 354 277), transferred to the PDMS model and incubated at
   37 °C for 30 min to form two bundle hEHTs. The hEHTs were subsequently
   placed in cardiac culture medium (containing differentiation basal
   medium with 1 mM sodium pyruvate (Gibco, 11360070), 2 mg mL^−1
   6‐Aminocaproic acid (Sigma, A2504), 0.1 mM non‐essential amino acids
   (Gibco, 11140050), 0.45 mM 1‐Thioglycerol (Sigma, M6145) and 2% FBS)
   and cultured on a rocker. During the initial 24 h, 10 µM Y‐27632 and 5%
   FBS were supplemented to the cardiac culture medium to optimize the
   viability of iPSC‐CMs. After removing 10 µM Y‐27632 and 5% FBS, the
   cardiac culture medium was changed every 48 h until day 9, when
   stimulations began. Palmitic acid (Sigma, P5585) and BSA (Yeasen,
   B9301110) were used as control treatments starting on the ninth day
   until harvest.

hEHT Contraction Force Measurement

   After treatment with 500 µM PA alone or in combination with 22 µM
   Empagliflozin (HY‐15409; MedChemExpress) for the indicated time points,
   hEHTs were placed in the customized force‐testing apparatus described
   previously.^[ [244]^40 ^] Briefly, hEHTs were subjected to electrical
   field stimulation at voltages of 10 V and different frequencies (0,
   0.5 Hz, 1 Hz, 1.5 Hz, 2 Hz, 2.5 Hz) in a bath solution containing
   1.8 mM CaCl[2]. Subsequently, the tissue strips were incrementally
   stretched to 12% of their original length at a rate of 2% per increment
   under conditions of 10 V and 1.5 Hz. Each stretch level was recorded
   for 30 s. Contraction force data were analyzed using custom MATLAB
   programs.

Animal Experiments

   8‐week‐old male C57BL/6J mice were purchased from Henan Skbes
   Biotechnology Co., Ltd. (Henan, China) and were maintained under a 12‐h
   light/dark cycle. All procedures were performed according to the NIH
   guidelines. The study protocols and use of animals were approved by the
   Hubei University Animal Ethics and Welfare Committee (20220012). Mice
   in the control group were allowed ad libitum access to water and a
   regular chow diet, with 14.5% fat, 27.38% protein, and 58.12%
   carbohydrate (cat. no.1035, BEIJING HFK BIOSCIENCE CO.LT, China).
   High‐fat diet (HFD) with 60 kcal% fat, 20 kcal% carbohydrate, 20 kcal%
   protein, was purchased from Research Diets, Inc. (D12492, USA).

   Mice were acclimated to the environment for 1 week and then randomly
   assigned to either the normal chow group (NC) or the HFD group. After 4
   weeks, the HFD‐fed mice were fasted overnight and received
   intraperitoneal injections of streptozotocin (50 mg kg^−1, S0130,
   Sigma) for four consecutive days. Mice with blood glucose levels
   ≥11.1 mmol L^−1 were selected and further divided into the
   empagliflozin group and the vehicle group. Empagliflozin was dissolved
   in DMSO (10 mg mL^−1) and administered to T2DM mice at a dosage of
   10 mg kg^−1 via oral gavage daily. Mice in NC group and the HFD‐vehicle
   group received an equal volume of saline and DMSO by gavage. After 8
   weeks, cardiac function was evaluated using echocardiography.
   Subsequently, the animals were sacrificed, and their hearts were
   excised for further analyses. n = 6 mice in NC + vehicle group, n = 7
   in HFD + Vehicle group, and n = 7 mice in HFD + Empagliflozin group.

Echocardiography

   Cardiac echocardiography was performed on mice anesthetized with 1–1.5%
   isoflurane using the VINNO V6 Vet system, the heart rate was maintained
   between 400 and 450 beats per minute to facilitate Doppler studies.
   Images were acquired with a X10‐23L transducer. Fractional shortening
   (FS), left ventricular ejection fraction (LVEF), ventricular
   dimensions, and volumes were assessed via a M‐mode. Passive left
   ventricular filling peak velocity (E, mm s^−1) and atrial contraction
   flow peak velocity (A, mm s^−1) were obtained by mitral valve Doppler
   flow. Early (e’, mm s^−1) and late (a′, mm s^−1) diastolic mitral
   annular motion velocity was obtained by tissue Doppler imaging from the
   apical 4 chamber view.

Transmission Electron Microscopy

   The EHT‐bundle and mice heart were directly fixed with a sufficient
   amount of pre‐cooling 2.5% glutaraldehyde and transferred to 1.5 mL EP
   tubes for fixation and stored in the refrigerator at 4 °C. This process
   should be gentle and quick to make sure not to damage cells or tissues.
   The hEHT samples were sent to Wuhan University People's Hospital for
   transmission electron microscopy preparation and imaging. The heart
   samples were sent to Wuhan Luochuang Biotechnology Co., Ltd. for
   subsequent preparation and imaging. Data analysis was performed using
   Fiji‐ImageJ software.

Optical Mapping in hEHT and Mice

   After 7 days of culture, hEHTs underwent optical mapping of
   transmembrane potentials using voltage‐sensitive fluorescent dyes and a
   high‐speed camera. hEHT were co‐incubated with Rhod‐2 AM (Thermo
   Fisher, R1245MP, 2.5 µM) at 37 °C for 25 min. Following two washes with
   DPBS, hEHT were cultured in tissue culture medium supplemented with 2.5
   µM Blebbistatin (Sigma, B0560) to suppress cardiomyocyte contraction
   and eliminate motion artifacts during recording. Electrical activity
   induced by stimulation with point electrodes was recorded for 10 s
   using a 594‐channel photodiode array in microscopy mode (16×).
   Conduction velocity (CV) data analysis was performed using BV Ana
   software. Membrane potential amplitude, APD50, APD80, maximum upstroke
   velocity, and maximum decay rate were analyzed using the PyCharm
   program. Rainbow mapping was performed for analysis using OMapScope.

   Similar to the methods described before,^[ [245]^42 ^] the mice were
   anesthetized and euthanized to remove the heart. The heart was placed
   on the Langendorff perfusion apparatus (Scope, SG‐LP011) and the aorta
   was retrogradely perfused with K‐H solution containing 119 mM NaCl,
   4 mM KCl, 1.8 mM CaCl[2]‐2H[2]O, 1 mM MgCl[2]‐6H[2]O, 1.2 mM
   KH[2]PO[4], 25 mM NaHCO[3] and 10 mM glucose at 37 °C and 10 mL min^−1
   flow rate. After 15 min of stabilization, 1 mg mL^−1 blebbistatin
   (Abcam, ab120425) was perfused into the heart to eliminate motion
   artifacts. Intracellular calcium changes were measured using 1 mg mL^−1
   RHOD‐2AM (Abcam, ab142780). The entire heart was illuminated with
   549 nm excitation light (MappingLab, LEDC‐2001), and CMOS cameras
   (MappingLab, OMS‐PCIE‐2002) were used to capture the fluorescent signal
   emitted by the heart. The intracellular calcium signal under 8 Hz
   stimulation (stimulation at the apex of the heart with twice the
   threshold current) was recorded. The commercially available analysis
   software OMapScope 4.0 (MappingLab, Version 5.7.8) and BV‐Ana Analyzer
   (BrainVision, Version 16.04.20) were applied for processing the optical
   mapping data from miceearts and hEHTs. Gaussian spatial filtering (3 ×
   3 pixels) and median filtering (3 × 3 pixels) were used for the spatial
   alignment and processing of the calcium signal.

Histological Analysis

   hEHTs were fixed with 4% paraformaldehyde at room temperature for 20
   min, then dehydrated, embedded in paraffin, and sectioned into
   6‐µm‐thick slices. Subsequently, tissue sections were deparaffinized
   and rehydrated for further staining. For immunostaining of tissue
   fibrosis, antigen retrieval was first performed using pH 6.0 citric
   acid solution, followed by blocking with blocking solution at 37 °C for
   1 h. Next, hEHT sections were incubated with rabbit anti‐ Vimentin (CST
   #5741, 1:1000 dilution) and goat anti‐cTnI (Thermo Fisher Scientific, #
   PA1‐86820, 1:1000 dilution) primary antibodies at 4 °C overnight. After
   washing away excess primary antibodies with DPBS, sections were
   incubated with secondary antibodies for 2 h, followed by
   counterstaining of cell nuclei with DAPI (Invitrogen, 62 248, 1:1000).
   The level of fibrosis was analyzed using Image J software to quantify
   and analyze the percentage of Vimentin positive area.

   The harvested mice hearts were fixed in 4% paraformaldehyde for 24–36 h
   at 4 °C, halved 1–2 mm above the ligature knot, dehydrated in an
   ethanol and xylene series, and embedded in paraffin. hearts were
   serially sectioned at 6 µm thickness from the heart, H&E was performed
   using standard procedures.

TUNEL Staining

   Apoptosis was detected using the In Situ Death Detection Kit (YEASEN,
   40306ES50). For cellular‐level analysis, iPSC‐derived cardiomyocytes
   (iPSC‐CMs) were cultured in confocal dishes. Cells were fixed with 4%
   paraformaldehyde (PFA) at room temperature for 15 min in the dark,
   followed by permeabilization and blocking with a solution containing
   0.2‰ Triton X‐100 and 5% donkey serum at 37 °C for 30 min. Samples were
   then incubated overnight at 4 °C with the SAA primary antibody
   (Sigma‐Aldrich, A7811). After washing with PBS, TUNEL staining was
   performed according to the manufacturer's instructions, followed by
   incubation with the appropriate fluorescent secondary antibody.

   For TUNEL staining of mouse heart tissue, 6 µm‐thick paraffin sections
   were first deparaffinized, and antigen retrieval was conducted using
   citrate buffer (pH 6.0). The subsequent staining procedure was
   identical to that used for cellular apoptosis detection. Finally,
   images were acquired using a confocal fluorescence microscope (Zeiss,
   LSM 980). The apoptosis rate was expressed as the proportion of
   apoptotic cardiomyocytes relative to the total number of cells.

Western Blot Analysis

   Briefly, hEHT samples from three experimental groups (CON, PA, and PA +
   Empagliflozin), each with three biological replicates, were lysed using
   RIPA buffer (Beyotime, China) supplemented with protease and
   phosphatase inhibitors (Roche, Switzerland). Protein concentrations
   were determined using the BCA protein assay kit (Thermo Fisher
   Scientific, USA). Equal amounts of protein (15–20 µg) were separated on
   4–20% gradient FuturePAGE precast gels (15‐well format, ET15420Gel; ACE
   Biotechnology, ChangZhou, China) and transferred onto PVDF membranes
   (Bio‐Rad, USA). Membranes were blocked with 5% non‐fat milk in TBST
   buffer (Tris‐buffered saline with 0.1% Tween‐20) for 2 h at room
   temperature, then incubated overnight at 4 °C with the following
   primary antibodies: GRP78 (ABclonal, A23453, Rb, 1:6000); CHOP
   (ABclonal, [246]A21902, Rb, 1:1000); SGLT1 (ABclonal, A26241, Rb,
   1:1000); NHE1 (Proteintech, 67363‐1‐ig, Ms, 1:5000); β‐actin
   (Servicebio, GB11001, Ms, 1:1000). After washing with TBST, membranes
   were incubated for 2 h at room temperature with HRP‐conjugated
   secondary antibodies against rabbit (Proteintech, SA00001‐2, 1:5000) or
   mouse (ABclonal, AS003, 1:5000). Following three washes, protein bands
   were visualized using ECL reagent (Immobilon Western Chemiluminescent
   HRP Substrate, WBKLS0100) and detected with a chemiluminescence imaging
   system (Bio‐Rad, USA). Results were expressed as density values
   normalized to β‐actin levels.

RNA Sequencing and Gene Expression Analysis

   The differential gene expression analysis of RNA‐seq data was conducted
   using the Npvaseq PE150 platform from GeekGene (Beijing, China) based
   on Illumina Novaseq technology with paired‐end sequencing. briefly,
   hEHTs were stimulated with PA, empagliflozin, or BSA control for 1 or 3
   days before harvesting using TRIzol (ThermoFisher Scientific, 15596026)
   according to the manufacturer's instructions. Each group included of 3
   biological replicates. For downstream analysis, FPKM lists were used.
   Principal component analysis (PCA) was conducted on the GeekGene
   platform ([247]http://www.geekgene.com.cn/). Gene Ontology (GO) and
   Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment
   analyses were performed by submitting gene lists to the DAVID
   bioinformatics tool., A cutoff of p‐value < 0.05 and an absolute
   fold‐change ≥ 1.5 was used to identify differentially upregulated or
   downregulated genes. The top 15 most significant gene entries were then
   analyzed further. Volcano plots and heatmaps were generated to
   visualize the significantly differentially expressed genes using the
   bioinformatics platform ([248]https://www.bioinformatics.com.cn/). For
   Gene Set Enrichment Analysis (GSEA), normalized FPKM lists were
   analyzed using GSEA_4.3.2 software for downstream analysis.

Statistical Analysis

   The data were examined to assess whether they met the assumptions of
   the selected statistical tests. Data analysis was performed using the
   Prism software package (GraphPad Software). Results were expressed as
   the mean ± standard error (SEM). An unpaired t test was used to compare
   two groups, whereas an ordinary ANOVA with Tukey's correction was used
   to compare data from more than two groups. Statistical significance was
   considered when p < 0.05.

Conflict of Interest

   The authors declare no conflict of interest.

Author Contributions

   L.C., Y.Z., and Z.L. contributed equally to this study. L.C. and D.Z.
   conceptualized the study. L.C., Y.Z., Z.L., and D.Z. designed the
   experiments; Y.Z., Z.L., L.X., and Y.W. conducted experimental studies
   in hPSC culture, model establishment, contraction force assessment,
   optical mapping, and sample collection. L.X. and Z.L. performed bulk
   RNA sequencing analysis. L.C., Z.L., and Y.W. conducted experiments on
   the T2DM model in mice. S.W. and Z.L. performed optical mapping in
   mice. Q.L. and Y.Y. provided some of the preliminary experimental
   results of cells and hEHTs before graduation. L.C., Y.Z., Z.L., L.X.,
   Y.W., S.W., Q.L., Y.Y., and D.Z. analyzed the experimental data and
   organized the figures. L.C., Y.Z., Z.L., L.X., Y.G., and D.Z. wrote or
   edited the manuscript.

Supporting information

   Supporting Information
   [249]ADVS-12-e03173-s001.docx^ (2MB, docx)

Acknowledgements