Diabetes in a Dish: Modeling and Phenotyping Acute and Chronic Type 2 Diabetes Mellitus In Vitro in Rodent Heart and Skeletal Muscle Cells

1. Introduction

Diabetes mellitus (DM) represents a large global disease burden with an ominous prognosis [1]. It is predicted that more than 550 million people worldwide will be diabetic by 2030, compared to 108 million in 1980 [2]. Diabetes is among diseases with the most rapidly increasing global prevalence and improving therapies and prevention methods to reduce diabetes-related complications and premature mortality is of utmost importance.

DM is characterized by a relative or absolute lack of insulin, resulting in elevated blood glucose levels. Type 2 diabetes (T2D) accounts for more than 90% of DM cases and is primarily caused by an inappropriate cellular response to insulin. In response to insulin resistance (IR) or intolerance, pancreatic beta cells hypertrophy and release more insulin, consequently leading to pancreatic exhaustion and failure in later stages of the disease [3]. The natural history of T2D is characterized by obesity in >80% of cases, which is further exacerbated by a Western diet and physical inactivity leading to IR, hyperglycemia, hyperlipidemia, and hyperinsulinemia [4]. Chronic hyperglycemia can lead to microvascular (e.g., nephropathy, retinopathy, and polyneuropathy) and macrovascular (e.g., stroke, cardiovascular disease) complications [5]. In T2D patients, cardiovascular disease is a major cause of morbidity and mortality, accounting for 68% of all diabetes-related deaths [6].

Given the physiological complexity of the disease, in vitro models that could provide a path to discovery and testing of novel therapeutics have been limited. By modeling this complex disease in vitro, we want to increase and accelerate the discovery of therapeutic targets, improve disease screening, and enable improved preclinical drug testing. In vitro cell models can provide an economically and ethically acceptable research tool and enable the targeting of specific processes linked to a single cell type of interest, without uncontrolled influences of the whole organism. In vitro cell models can be performed with high throughput and reproducibility and can be genetically modified by transfection to investigate specific genes of interest [7]. Thus, cell-based models, if developed properly, could serve as tools to accelerate therapeutic discovery.

In general, in vitro models of diabetes are derived from the tissues that are mainly involved in the disease pathophysiology, such as muscle, pancreas, adipose tissue, and liver [8]. Skeletal muscle is the primary recipient of postprandial glucose, and T2D patients often show significantly impaired blood glucose clearance [9]. Also, the higher flux of FFAs in patients with T2D leads to an increased FA-uptake into myocytes [10]. To assess the effect of T2D on skeletal muscle, we chose to use C2C12 cells, a mouse myoblast cell line, which is well-established and has been widely used in diabetes research [11]. In the setting of IR, the myocardium’s ability to utilize glucose as an energy source is reduced [12], whereas under physiological conditions energy utilization from FA and carbohydrates is based on metabolic demand and availability [13]. Even though FFAs are the major energy source for heart muscle, an excess of plasma FFAs may predispose the heart to ischemic damage and high oxidative stress [14]. These metabolic changes and the accumulation of lipids in the heart play a critical role in the development of diabetes-related cardiac complications, such as diabetic cardiomyopathy [15,16]. To further understand these mechanisms in heart muscle, we mimicked diabetic conditions in H9C2 cells, which are myocytes from embryonic rat ventricular tissue [17]. H9C2 myocytes are widely applied in studies addressing cardiac hypertrophy, metabolism, ischemic stress, and IR/diabetes [18,19].

To cause IR as it is occurring in T2D, most cell models with adipocytes and myocytes use chronic insulin exposure or fatty acid treatments (mainly palmitate) [20] (Table 1). The most commonly used immortalized muscle cell lines for in vitro diabetes models are L6 and C2C12 myoblasts [21].

To simulate the diabetic milieu in vitro, not only the cells themselves play a critical role, but also the conditions in which they are cultured. The main reason for metabolic syndrome and T2D is chronic overnutrition consisting mainly of fatty acids and carbohydrates in the developed world [32]. In vitro studies showed that once a cell becomes overwhelmed by the excess supply of FFAs, the accumulation of lipids such as diacylglyceride (DAG) and ceramide may contribute to mitochondrial dysfunction, generation of reactive oxygen species (ROS), low-grade inflammation, and finally IR and apoptosis [33]. Correspondingly, symptoms of T2D are caused by both hyperglycemia and hyperlipidemia, whereas hyperglycemia alone would more likely reflect uncontrolled T1D [34]. In cell culture research, it is a common practice to use culture media with supraphysiological glucose concentrations to promote and accelerate cellular growth. Previous studies addressing IR and diabetes in H9C2 and C2C12 muscle cells worked with 22-25 mM glucose as control and 33-40 mM glucose to mimic hyperglycemia, which is 8 times more than physiological levels [28,29]. In several studies investigating DM, glucose concentrations of cell culture media are not specified, leaving the reader uninformed (Table 1) [26,27,31]. The American Diabetes Association and WHO define normoglycemia as a fasting plasma glucose between 3.9 mmol/L and 5.6 mmol/L [35,36]. Diabetes is defined when the fasting plasma glucose reaches 7.0 mmol/L or higher [37]. It is recommend to first growing and differentiating cells in physiological glucose levels of 5 mM [25]. The simulation of hyperlipidemia requires the treatment with fatty acids (FA). Chavez and Summers showed that the saturated FA palmitate (the most dominant fat in the western fast food diet) causes IR, whereas unsaturated FAs such as oleate even reversed palmitate-induced IR in skeletal muscle [20]. This suggests using the widely applied FA palmitate for T2D-related hyperlipidemia conditions. High concentrations of palmitate (200-750 µM), which have previously been used in studies investigating IR, can cause apoptosis and high myotube loss in skeletal and cardiac muscle cells [26,38,39]. We therefore worked with a maximal concentration of 150 µM for C2C12 and 75 µM for H9C2, which still leads to IR but is less toxic and close to the physiological range (300-410 µM in humans [40] and 100-400 µM in rats [41]). In conclusion, exposure to high glucose, palmitate, and insulin together, may be closer to conditions observed in T2D patients compared to previously described cell-based approaches (Table 1). To achieve more similarity to human muscle tissue we worked with differentiated myotubes, whilst many research groups work with the undifferentiated myoblasts [25,42,43]. The insulin dependent glucose transporter 4 (GLUT4) is similarly expressed in C2C12 myotubes compared to human skeletal muscle cells [11]. Undifferentiated H9C2 myoblasts express little cardiac specific markers and can still differentiate into skeletal muscle instead of cardiac muscle. In 2017 Patten et al. stated that the combination of serum reduction with retinoic acid (RA) supplementation in the differentiation process did increase the expression of cardiac specific markers [44]. Lopashuk et al. identified that only differentiated H9C2 cells switch from glycolysis to oxidative phosphorylation, which is characteristic for heart tissue [45]. Also, insulin-signaling molecules such as insulin receptor substrate-1 and GLUT4 are significantly more expressed by differentiated H9C2 myotubes [46]. In conclusion, differentiation into myotubes to explore heart and skeletal muscle specific processes related to T2D is necessary. Various human studies suggested the existence of mitochondrial dysfunction in muscle cells of obese and insulin-resistant patients [47,48]. Comprehensive understanding of the mechanisms that relate mitochondrial function and insulin signaling is still lacking. Identifying the factors and mechanisms responsible for changes in mitochondrial energetics could make mitochondria to a potential target for treating diabetes. Therefore, we examined mitochondrial function in vitro in rodent skeletal and heart muscle utilizing the Agilent Seahorse XF Analyzer and utilized transmission electron microscopy (TEM) to examine mitochondrial morphometry.

Taken together, this study aimed to develop a novel model for simulating T2D in vitro in skeletal and cardiac muscle cells that reflects the cellular response typified by T2D patients.

2. Materials and Methods

2.1. Cell culture

C2C12 (CRL-1772™) and H9C2 (CRL-1446™) cell lines were purchased from American Type Culture Collection (ATCC, Manassas, VA, USA). C2C12 and H9C2 cells were cultured in Dulbecco’s Modified Eagle’s Medium (DMEM) (GIBCO/BRL, 11885092) containing 5 mM glucose, supplemented with 10% fetal bovine serum (FBS, Thermo Fisher Scientific, 16000044) and 1% penstrep (Penicillin/Streptomycin, GIBCO, 15140122). For differentiation of C2C12, DMEM (5 mM glucose) was supplemented with 2% horse serum (HS) (Thermo Fisher, 26050088) and 1% penstrep was refreshed daily. For differentiation of H9C2 cells, DMEM (5 mM glucose) was supplemented with 1% FBS, 1% penstrep and 10nM All-trans-retinoic acid (RA) (Sigma Aldrich, R2625-500MG). RA was diluted in DMSO and stored at -20°C in the dark. Medium change was performed daily in the dark. Cell treatments for assays were started 5-8 days after differentiation. C2C12 and H9C2 cell cultures were used in passages 2-6 in all experiments.

2.2. Treatments

DMEM containing 25 mM glucose (GIBCO/BRL, 11995073) was used as a hyperglycemic condition (HG). This hyperglycemic condition was also used in combination with insulin (HG-I) (1 nM, insulin glargine, Lantus, Sanofi), or palmitate (HG-P) (150 µM for C2C12, 75 µM for H9C2). A further hyperglycemic condition was combined with both insulin and palmitate (HG-PI). Groups treated with low glucose (LG, 5 mM) and the same combinations with palmitate and/or 1 nM insulin served as physiological controls (LG-I, LG-P, LG-PI). To exclude a possible hyperosmolar effect, cells treated with 20 mM mannitol plus 5 mM glucose served as control cultures (high mannitol, HM). The same treatment combinations were used (HM-I, HM-P, HM-PI). The cells were exposed to different conditions, either short-term (24h) or long-term (96h). Before all experiments, the culture medium was changed to serum-free medium for 4 hours. For long-term treatments, the concentration of palmitate was escalated every 24 hours to prevent myotube loss (Table 2).

Before adding sodium palmitate to cell differentiation medium, it was conjugated to fatty acid free bovine serum albumin (BSA). Transportation into cells is then enabled and cytotoxicity lowered. Briefly, ultra-fatty acid free BSA (Sigma Aldrich, A8806) was dissolved in the respective differentiation medium for each cell line, heating the solution to approximately 37°C. Sodium palmitate (Sigma Aldrich, P9767) was dissolved in a 150 mM NaCl solution, stirring at 70°C. The palmitate solution was added to the BSA solution while stirring at 37°C. After stirring for 1 h, pH was adjusted to 7.4 with NaOH. The conjugated palmitate-BSA solution was aliquoted in glass vials and frozen at -20°C. Stock solutions were made at 75 mM.

2.3. ³H-2-deoxy-glucose-uptake

³H-2-deoxy-glucose-uptake assays were performed using published methods [49]. Briefly, culture media were discarded 4 hours prior to the assay and changed to serum-free DMEM with 0.25% fatty acid-free BSA. Serum-Starvation was held in respective treatment media (serum-free DMEM with either high or low glucose (5 mM vs 25 mM) and supplemented with palmitate or 1 nM insulin respectively). After starvation cultures were washed twice with Hepes fortified Krebs-Ringer Bicarbonate buffer (HKRB), containing 10 mM Hepes, pH 7.4, NaH2PO4 (0.83 mM), Na2HPO4 (1.27 mM), NaHCO3 (15 mM), NaCl (120 mM), KCl (4.8 mM), calcium (1 mM), magnesium (1mM), pH 7.35 and 0.25% fatty acid-free BSA. Cultures were incubated in HKRB for 60 minutes at 37°C. Half of the cultures were pre-assigned for acute insulin treatment. These cultures were treated with 100 nM insulin from a 100x stock and incubated for 20 minutes at 37°C. After this incubation period ³H-2-deoxy-glucose (³H-2DOG, 2-[1,2-3H(N)]-, 250µCi (9.25MBq), Perkin Elmer, NET549A250UC) was added to each well for further 10 minutes incubation. ³H-2DOG-treatment was terminated by quick aspiration, followed by cold wash (2x) with ice-cold PBS. Next, we added 1N NaOH and swirled the plates slowly for 30 minutes to dissolve the cells. Aliquots of 20 µl were taken out of each well for protein measurements. The whole lysate of each well was transferred into a scintillation vial. After neutralizing NaOH with 1N HCl, scintillation cocktail (Ecoscint original, National Diagnostics, LS-271) was added and radioactivity was measured. Results were normalized with protein concentrations and presented as counts per minute (cpm) per mg protein. Alternately, the results were presented as cpm per million cells.

2.4. Mitochondrial function measurements

Mitochondrial function measurements were performed with the Agilent Seahorse XFe 96 Extracellular Flux Analyzer (Agilent Technologies, La Jolla, California). The analyzer measures real time extracellular acidification rate (ECAR) and oxygen consumption rate (OCR) of cells. C2C12 and H9C2 cells were seeded and differentiated in a density of ~1.2 x10⁴ cells per well, and the Seahorse XF Mito Stress Test assay was carried out as previously described [50]. During the assay, inhibitors of mitochondrial function are injected sequentially to distinguish the following parameters: Basal respiration: Oxygen consumption used to meet cellular ATP demand under baseline conditions. It can be set by the rate of ATP utilization and substrate availability and oxidation. ATP-linked respiration: Upon injection of the ATP synthase inhibitor oligomycin a decrease in OCR represents the part of basal respiration that accounts for ATP-linked respiration, meeting the energetic needs of the cell. It is largely set by the ATP demand of the cell and substrate oxidation, which can be decreased if there is mitochondrial dysfunction. Maximal respiration: By adding the uncoupler carbonyl cyanide p-(trifluoromethoxy) phenylhydrazone (FCCP) the maximal OCR can be attained. FCCP mimics an “energy demand” and stimulates the respiratory chain to operate at maximum capacity and substrates are rapidly oxidized. It is also set by substrate supply and oxidation including the functional substrate transport across the plasma and mitochondrial membranes. Changes may reflect membrane defects, altered mitochondrial biogenesis and function and decreased oxidation abilities. Spare respiratory capacity: This value can be calculated with basal and maximal respiration values. It indicates the cell’s capability to respond to an energetic demand and can be an indicator of cell fitness or flexibility. A decreased capacity can be linked to mitochondrial dysfunction but can also reflect an increased ATP demand/increased basal respiration (i.e., highly proliferative cells). Increases may reflect high substrate provision or enhanced oxidative capacity [51]. We plotted OCR versus ECAR values on an energy map to provide a qualitative measurement of the relative utilization of oxidative (OCR) versus glycolytic (ECAR) pathways for energy production.

2.5. Electron microscopy

C2C12 and H9C2 myotubes maintained short-term (24h) and long-term (96h) in different T2D-mimicking and control conditions (see: treatments) were fixed and embedded for electron microscopy (EM) as previously described [52]. Grids were viewed using a Jeol JEM1400-plus Transmission Electron Microscope and photographed with a Gatan OneView digital camera with 4k x 4k resolution.

2.6. Morphometric analyses of mitochondria

Using the free-hand tool of NIH ImageJ, we manually traced around the mitochondria to determine mitochondrial area and length as described previously [53]. Cytoplasmic area (excluding the area occupied by the nucleus) was also calculated using the ImageJ software. We then calculated mitochondrial density (area occupied by mitochondria in total area of cytoplasm). Mitochondrial numbers were calculated by placing the EM micrographs onto a grid with 10 horizontal x 10 vertical square grids followed by dividing the mitochondrial numbers with the area. The data were expressed as mitochondrial number/10 µm2 area.

2.7. Statistical analysis

All Data analyses were performed using GraphPad Prism 9.5.1 (GraphPad Software, La Jolla, CA). For glucose uptake assays the results are expressed as mean ± SD of 6 biological replicates for all LG and HG groups and 3 replicates for all HM groups. For Mito Stress Tests, all data shown are the means ± SD of 3-4 replicate plates with 8-15 wells per treatment group per plate. Relative frequencies of mitochondria area and length were analyzed by Kolmogorov-Smirnov test, mitochondrial number, and density were analyzed by one-way ANOVA, of 6-14 cells. For statistical analysis a p value <0.05 was considered significant. We used analysis of variance (one-way or two-way ANOVA), provided that the assumptions were fulfilled. For Agilent Seahorse measurements two-way ANOVA was used with row factor for treatment groups (Rf) and column factor for biological replicates/assays (Cf). F- and p values are provided in the legend of the figure panels. Significance of secondary pairwise comparisons among different treatments by Tukey’s test is indicated by asterisks over brackets in the figure panels (*: p<0.05; **: p<0.01; ***: p<0.001).

3. Results

3.1. ³H-2-deoxy-glucose-uptake

C2C12 myotubes: We found that treatment of C2C12 myotubes with T2D-mimicking conditions including insulin and palmitate treatments resulted in IR compared to controls. Control cultures preincubated with LG for 24h showed a significant increase in glucose uptake upon stimulation with 100 nM test insulin (Figure 1A) (LG +/- insulin). Short-term (24h) pretreatments with 150 µM palmitate either in presence or absence of 1 nM insulin impeded the stimulatory effect of insulin, indicating that palmitate caused IR in consistence with existing literature [54]. Similar effects were seen after 96h (Figure 1K). After short- and long-term treatment, basal glucose uptake was increased in LG-I treated groups compared to LG controls (Figure 1A,K). LG-P and LG-PI-treated myotubes displayed decreased glucose uptake rates compared to LG and LG-I treatments. Cultures treated short- or long-term in a HG condition did not increase glucose uptake upon insulin stimulus (Figure 1C,M) but showed elevated basal rates compared to LG. We observed no significant increase in insulin-dependent vs. basal glucose uptake in all T2D-mimicking treatment groups (HG, HG-I, HG-P, HG-PI). Myotubes treated with HG-P or HG-PI showed significantly lower basal and insulin-dependent glucose uptake rates compared to HG and HG-I groups (Figure 1C,M). Myotubes treated with HM or HM-I for 24h showed a significant decrease in basal glucose uptake compared to LG or HG and no significant up-regulation of glucose uptake upon insulin stimulation. Interestingly, C2C12 myotubes treated with HM and palmitate (HM-P and HM-PI) showed a marked increase in both basal and insulin-dependent glucose uptake (Figure 1G), compared to all other treatment groups. Looking at the half-violin plots with combined data of 24h and 96h treatments, treatments including mannitol show the widest spread in values due to the stimulatory effect on glucose uptake of palmitate in combination with HM (Figure 1E), whereas in cells exposed to HG and palmitate an inhibitory effect of palmitate is apparent (Figure 1I).

H9C2 cardiac myotubes: IR occurred in all T2D-mimicking treatment groups. LG or LG-I treated control cultures showed an increase in glucose uptake upon an acute insulin stimulus (100 nM). After 24h treatment, the increase was not significant, whereas, after 96h treatment, we observed a significant rise in insulin-dependent glucose uptake compared to basal uptake (Figure 1B,L). Basal glucose uptake rates in both LG and LG-I treatment groups were similar, but insulin-dependent rates were significantly higher in LG-I compared to basal rates in LG (Figure 1B). H9C2 myotubes treated short or long-term with LG and palmitate (LG-P and LG-PI) showed significantly decreased glucose uptake rates compared to LG and LG-I. We observed similar effects on glucose uptake after short- or long-term treatments with HG-P or HG-PI (Figure 1D,N), which indicates a massive suppression of basal glucose influx and insulin insensitivity by palmitate [31]. Groups treated short- or long-term with HG or HG-I were at least partially insulin-resistant and showed a significantly higher glucose uptake compared to groups treated with high glucose and palmitate (Figure 1D,N). In contrast, osmotic control cultures treated with HM-P or HM-PI showed a remarkable increase in basal glucose uptake compared to HM or HM-I treatment, but insulin response was suppressed (Figure 1H) even more marked than in C2C12 myotubes (Figure 1G). In summary, HM-treated groups showed markedly lower glucose uptake rates compared to LG groups (LG basal rate mean = 207.33; HM basal rate mean = 45.95). Cultures treated with HM or HM-I showed a non-significant trend towards an increase in glucose uptake upon an acute insulin stimulus (Figure 1H). Looking at the half-violin plots with combined data of 24h and 96h treatments, cells exposed to mannitol show the lowest glucose uptake rates of all treatments (Figure 1F). Overall, palmitate exerted a marked inhibitory effect on both basal and insulin-dependent glucose uptake, particularly in combination with HG treatment (Figure 1J,P).

3.2. Mitochondrial Function measurements using the Agilent Seahorse XFe 96 Extracellular Flux Analyzer

In all Mito Stress Test assays with C2C12 and H9C2 myotubes, baseline relative oxygen consumption rate percentage did not differ statistically (Figure 2D,H,L,P).

3.2.1. C2C12 Myotubes

24h treatments: C2C12 myotubes treated 24h with non-physiological glucose levels (HG, HG-I, HG-P, HG-PI) showed a significant decrease in ATP-linked respiration compared to LG and LG-I controls (Figure 2A). Osmotic control cultures (HM, HM-I) showed no difference to LG control groups in maximal respiration (Figure 2B) but a significant decrease in ATP-linked respiration (Figure 2A). In all groups including insulin treatment maximal respiration rates were higher compared to treatments without insulin, but no significant changes were detectable (Figure 2B). Spare respiratory capacity rates were significantly increased in HG-I and HM-I groups compared to LG control (Figure 2C). In short-term assays with C2C12 myotubes we observed a shift towards glycolysis in all HG treated groups. HM treated groups showed a metabolic shift towards glycolysis. LG groups showed a mainly aerobic metabolism and LG-I groups were in between an aerobic and high energetic phenotype (Figure 2Q).

96h treatments: C2C12 myotubes treated 96h with high glucose (HG, HG-I) or HM-I showed a significant decrease in maximal respiration compared to LG control (Figure 2F). ATP-linked respiration of myotubes treated with HG or HM was significantly decreased (Figure 2E). Spare respiratory capacity rates were significantly decreased in HG and HG-I treated groups compared to LG. HM and HM-I treated groups displayed a lower spare respiratory capacity compared to LG-I (Figure 2G). In the XF energy map HG and HG-I treated groups showed a more glycolytic phenotype, while LG and LG-I control groups were in between an aerobic (oxidative phosphorylation) and a high energetic phenotype. HM and HM-I groups were less energetic compared to LG and less glycolytic compared to HG (Figure 2R). All respiration rates were lower after 96h treatment compared to 24h treatment, with maximal respiration rates being remarkably lower in T2D-mimicking conditions after 96h compared to 24h treatment. After 96h treatment with HG-P and HG-PI, we didn’t observe significant difference to LG controls in one Mito Stress Test and therefore excluded these treatment groups in the following repetitions.

3.2.2. H9C2 Myotubes

24h treatments: H9C2 myotubes treated 24h with high glucose in combination with insulin or palmitate (HG-I, HG-P, HG-PI) showed a significant increase in ATP-linked respiration and maximal respiration compared to LG and LG-I controls, with HG-PI groups displaying the significantly highest ATP linked respiration and respiration rates (Figure 2I,J). Spare respiratory capacity was significantly increased in treatment groups including palmitate (HG-P and HG-PI) (Figure 2K). Plotting OCR vs ECAR showed a trend towards more energetic phenotypes in T2D-mimicking conditions compared to LG controls (Figure 2S). All treatment groups utilized both glycolysis and oxidative phosphorylation to generate energy. High mannitol control cultures showed no significant differences in all respiration rates compared to LG control groups in one Mito Stress Test assay and were therefore excluded in the following assay repetitions.

96h treatments: H9C2 myotubes treated 96h with high glucose in combination with insulin or palmitate (HG-I, HG-P, HG-PI) showed a significant increase in ATP linked respiration and maximal respiration compared to LG and LGI controls (Figure 2M,N). After 96h treatment no significant changes in spare respiratory capacity was detectable (Figure 2O). Plotting OCR vs ECAR respiration rates after 96h treatment revealed LG control cultures to be more quiescent compared to 24h treatment and compared to T2D-mimicking conditions. HG, HG-I, HG-P and HG-PI treated groups showed a phenotype in between glycolysis and oxidative phosphorylation, with HG-PI being the most energetic (Figure 2T).

3.3. Morphometric analyses of mitochondrial number, density, relative frequency of mitochondria with differing area and length

3.3.1. Electron Microscopy for Mitochondrial appearance.

The abnormalities caused by T2D conditions in mitochondrial function raised the question of accompanying structural abnormalities. For mitochondrial morphology analysis, the number of normal versus swollen/vacuolated mitochondria was compared after 24h treatment of C2C12 and H9C2 myotubes. Normal mitochondria were defined as indicated in Figure 3 A, swollen/vacuolated mitochondria as indicated in Figure 3B,C. HM, HG-P and HG-PI treated C2C12 myotubes showed a significantly higher fraction of swollen/vacuolated mitochondria compared to LG control (Figure 3K). The fraction of swollen/vacuolated mitochondria in HG compared to LG groups was not significantly different (Figure 3K). Compared to LG, the percentage of swollen/vacuolated mitochondria in H9C2 myotubes was higher in HM, HG and HG-PI groups (Figure 3S). Swollen mitochondria are marked as “sm” in Figure 3 in C2C12 HM, HG-P, and HG-PI treatments (Figure 3E,G,H) and H9C2 HM, HG and HG-PI treatments (Figure 3M,N,Q).

3.3.2. Mitochondrial number and density

Morphometric analyses revealed no significant changes in mitochondrial number per 10 µm2 area of cytoplasm after exposure to T2D conditions in both C2C12 (Figure 3I) and H9C2 (Figure 3O) cells.

Opposite changes in mitochondrial density between C2C12 and H9C2 cells: Morphometric analyses of mitochondrial density (area occupied by mitochondria in total area of cytoplasm) revealed a significant decrease in mitochondrial density when C2C12 cells were exposed to HG, HGP, or HGPI compared to LG (Figure 3J). In contrast, H9C2 cells showed significant increase in mitochondrial density when exposed to HGP or HGPI, compared to LG, HG (Figure 3R). These findings are consistent with the increase in ATP-linked respiration as described above.

3.3.3. Palmitate but not HG changed relative frequency of mitochondria with respect to area and length

Morphometric analyses revealed that HG alone was not sufficient to alter the relative frequency of mitochondria with respect to area and length in both C2C12 and H9C2 cells, compared to LG (Figure 4A,J). In contrast, palmitate with high glucose caused marked changes in relative frequency of mitochondria with respect to area and length in both C2C12 and H9C2 cells compared to LG and to HM controls (Figure 4B,K,F,O), indicating that palmitate is a stronger reagent than glucose to affect mitochondrial health.

3.3.4. Insulin failed to alter palmitate-induced changes in mitochondrial health

We found no significant changes in mitochondrial area and length between HGP and HGPI groups in both C2C12 and H9C2 cells (Figure 4H,Q), indicating that palmitate in combination with HG caused profound IR in muscle cells.

4. Discussion

We observed that complex T2D in vitro modeling has significant effects on the functionality of rodent skeletal and heart muscle cells, leading to severe IR, changes in ATP-linked respiration, shifts in energetic phenotypes, and mitochondrial morphology changes, all consistent with what has been observed in the muscle of patients suffering from T2D [55]. Pre-existing in vitro models of T2D often disregard many critical and complex aspects of the disease (Table 1). The current study shows that by following some essential steps lacking in the literature, a representative in vitro model of human T2D-related changes in muscle cells can be created.

Basal glucose uptake rates were increased in both cell lines after HG treatment compared to LG, especially after long-term exposure. The gradient-driven glucose uptake might have facilitated this via GLUT1 instead of insulin-dependent GLUT4. McMillin et al. showed that mGLUT4 knockout mice still showed increased basal glucose uptake after chronic exposure to HG. They also found that in mouse skeletal muscle cells, GLUT1, 3, 6 or 10 almost exclusively mediate glucose uptake after chronic glucose overload [56]. Gosmanov et al. observed that, compared to 5 mM glucose exposure, HG treatment (30mM, up to 48h) of aortic endothelial cells increased GLUT1 expression and GLUT4-dependent glucose uptake, but didn’t change baseline glucose uptake rates. Heilig et al. saw an 134% increase in GLUT1 mitochondrial RNA as well as a 50% increase in deoxy-glucose uptake in rat mesangial cells exposed to 20 mM glucose for 3 days, when compared to cells adapted to physiological glucose levels (8mM).

We examined mitochondrial function and metabolic phenotypes in rodent skeletal and heart muscle. Skeletal myotubes from insulin-sensitive subjects with T2D family history have decreased ATP content, consistent to previous studies showing decreased ATP and impaired mitochondrial activity in myotubes from lean offspring of T2D patients [57,58,59]. Mailloux et al. also observed that C2C12 myotubes exposed to 24 mM glucose for 24h showed a glycolytic phenotype and more ROS production than a low glucose control with an oxidative phenotype; however, ATP-linked respiration was not altered, indicating that those myotubes still were metabolically flexible and achieved the ATP demand via glycolysis [60]. Elkalaf et al. followed a similar approach and saw differences in maximal respiration in hyperglycemic C2C12 myotube cultures and a phenotype switch. In both studies, cells were differentiated for up to 7 days in media containing 5 mM or 25 mM glucose before any experiment [61]. Differentiating cells in high glucose might have caused an adaptation of the cells, leading to restored cellular flexibility. In the current study, cells were differentiated in low glucose levels and only exposed to high glucose for 24-96h when already differentiated, which more accurately represents the changes of T2D in vivo. Future experiments should include longer exposure to high glucose to investigate if, following differentiation, an adaptation to supraphysiological glucose and correspondingly non-altered ATP levels is possible.

Strongly increased respiration rates might be associated with elevated cellular stress since high glucose and palmitate levels lead to apoptosis in cardiomyocytes [62,63,64]. On the other hand, it can also reflect a highly energetic phenotype with increased oxidation abilities [51]. The increase in ATP-linked respiration might be associated with an increase of FA oxidation and decreased glucose utilization, which is assumed to be increased in the T2D heart in humans [65,66]. FA uptake into the heart is mainly driven by the availability in the blood stream [67]. With an oversupply in fatty acids, not only is FA oxidation increased but also detrimental lipid metabolites (i.e., ceramides) [63,68]. Reliance of the heart on FA oxidation to produce ATP might lead to oxidative stress and ischemic damage [14]. Therefore, losing the ability to switch to glycolysis combined to an increase in ROS due to the increased FA oxidation and increasing lipotoxicity contribute to both decreased ATP production and cardiac inefficiency.

In patients with T2D mitochondrial content is reduced, size and fusion are impaired, and endoplasmic reticulum stress occurs in different cell types [69]. Increased fission and impaired fusion was observed in human renal glomerular endothelial cells treated with high glucose (30 mM for 72h) [70]. A reduced number and fragmented mitochondria were found in skeletal muscle from T2D and obese subjects, as well as decreased electron transport chain activity [47]. In our study, we did observe a significant decrease in mitochondrial density in diabetic C2C12 cells, consistent with decreased ATP-linked respiration. We did not observe any significant changes in mitochondrial number per 10 µm² area of cytoplasm after 24h exposure to T2D mimicking conditions. Thus, it is likely that the reduction of mitochondrial number that was reported in patients suffering from T2D, is a chronic condition that will only occur in vitro after longer treatments.

H9C2 myotubes treated with T2D mimicking conditions including palmitate (HGP, HGPI) showed a significant increase in mitochondrial density compared to control, which can be a sign of increased fusion, allowing an enhanced transport of metabolites and enzymes. This observation is consistent to our Mito Stress Tests, where these cells showed increased ATP-linked respiration, confirming a better adaptability to an in vitro T2D environment of heart muscle cells.

Both C2C12 and H9C2 myotubes treated with HM and T2D mimicking conditions showed a higher fraction of swollen/vacuolated mitochondria compared to LG controls. Mitochondrial swelling can be caused by osmotic changes in cell culture media and FA treatments but is also a known sign of apoptosis and necrosis [71]. The question remains if whether the decreased ATP-linked respiration we observed in skeletal muscle cells in the T2D environment led to imminent apoptosis and then mitochondrial swelling, or if the T2D mimicking treatments directly caused mitochondrial swelling, subsequently leading to mitochondrial dysfunction and decreased ATP turnover.

There are several limitations to consider in our studies. We chose to use the saturated FA palmitate for our investigations, as it was previously used for studies addressing IR. Even when the Western diet is dominated by saturated FAs, using only one saturated FA is not physiologically accurate since in vivo circulating FFAs are a mixture of various saturated and unsaturated FAs. For future experiments, a mixture of the most common FAs in human plasma could be used (i.e. oleic, palmitic and stearic acids) [72].

High mannitol treatments served as osmotic controls for high glucose treatments in our study, as it has been widely applied in the literature. Yet we observed some hitherto undescribed effects, especially after 96h high mannitol treatments, such as decreased ATP-linked respiration, a shift to glycolysis but decreased basal glucose uptake, and mitochondrial swelling. The mechanistic background of the effects that we observed must still be explored. Still, it can be stated that the impact of hyperosmolarity and hyperlipidemia in a hyperglycemic T2D environment appears more complex than expected.

5. Conclusions

We used two well-established skeletal and cardiac myoblast cell lines for this model and conducted differentiation with close ‘normoglycemic and normoinsulinemic’ conditions thus avoiding diabetogenic preconditioning before use in experiments. In the experiments, we simulated, in addition to hyperglycemia and hyperinsulinemia also, other aspects of T2D, i.e., hyperlipemia and hyperosmolarity. Despite using lower concentrations than previous studies, closer to the range seen in patients with T2D, we demonstrated a very rapid and progressive development of insulin resistance, derangements in mitochondrial metabolism and morphology with specific differences between skeletal and cardiac myotubes that parallel findings in biopsies from T2D patients. The model could potentially be extended to other tissues impacted by diabetes. Our model should therefore help to spare animal experiments and studies requiring human biopsies and has the potential for exploring pathomechanisms, defining potential new targets, and screening candidate therapeutic compounds in T2D.