Physiological Genomics

Transcriptomic analysis of the cardiac left ventricle in a rodent model of diabetic cardiomyopathy: molecular snapshot of a severe myocardial disease

Sarah Glyn-Jones, Sarah Song, Michael A. Black, Anthony R. J. Phillips, Soon Y. Choong, Garth J. S. Cooper


Heart disease is the major cause of death in diabetes, a disorder characterized by chronic hyperglycemia and cardiovascular complications. Diabetic cardiomyopathy (DCM) is increasingly recognized as a major contributor to diastolic dysfunction and heart failure in diabetes, but its molecular basis has remained obscure, in part because of its multifactorial origins. Here we employed comparative transcriptomic methods with quantitative verification of selected transcripts by reverse transcriptase quantitative PCR to characterize the molecular basis of DCM in rats with streptozotocin-induced diabetes of 16-wk duration. Diabetes caused left ventricular disease that was accompanied by significant changes in the expression of 1,614 genes, 749 of which had functions assignable by Gene Ontology classification. Genes corresponding to proteins expressed in mitochondria accounted for a disproportionate number of those whose expression was significantly modified in DCM, consistent with the idea that the mitochondrion is a key target of the pathogenic processes that cause myocardial disease in diabetes. Diabetes also induced global perturbations in the expression of genes regulating cardiac fatty acid metabolism, whose dysfunction is likely to play a key role in the promotion of oxidative stress, thereby contributing to the pathogenesis of diabetic myocardial disease. In particular, these data point to impaired regulation of mitochondrial β-oxidation as central in the mechanisms that generate DCM pathogenesis. This study provides a comprehensive molecular snapshot of the processes leading to myocardial disease in diabetes.

  • fuel metabolism
  • gene expression analysis
  • mitochondria
  • diastolic heart failure

the risk of cardiovascular disease and heart failure is elevated in diabetes (1). In the Framingham study, for example, the incidence of heart failure in diabetes was elevated twofold in diabetic men and threefold in diabetic women compared with matched nondiabetic participants (24).

Diabetic cardiomyopathy (DCM) was described in 1972 on the basis of observations in four diabetic patients, who presented with heart failure without evidence of hypertension, coronary artery disease, or congenital heart disease (33). Since this publication, the existence of diabetic cardiomyopathy as a distinct clinical entity has been under debate; however, a recent review by Fang et al. (14) summarized evidence that strongly supports the existence of diabetic cardiomyopathy independent of comorbid conditions. DCM is now held to refer to a disease process that affects the myocardium in diabetic patients, causing a wide range of structural abnormalities including interstitial fibrosis, and eventually leading to left ventricular hypertrophy (LVH) and diastolic or systolic dysfunction or a combination of these. The concept of DCM is based upon the idea that diabetes is the antecedent factor that causes changes at the cellular level, leading to structural abnormalities (19).

In addition, there is increasing evidence that altered substrate supply and utilization by cardiac myocytes could be key to the tissue injury in the pathogenesis of this disease (14). The ability to switch from fatty acid β-oxidation to carbohydrate oxidation is impaired in diabetes, producing a state known as metabolic inflexibility (42); this is believed to lead to an increased reliance by the heart on fatty acids as its primary fuel supply and to decrease the ability of the heart to adapt to stress.

A third aspect of diabetic cardiomyopathy is the observed increase in reactive oxygen species (ROS) and the effect that this has on the cell (31). It has been hypothesized that excess ROS cause much of the cellular damage underpinning the downstream complications of this disease (6).

Until now, the multifactorial nature of DCM has made it difficult to examine all the concomitant tissue changes that interact to generate the observed pathophysiology. The advent of microarray technology, however, has enabled simultaneous investigation of all changes in gene expression that occur in DCM (18).

The aim of this study was to develop an improved understanding of the transcriptional changes that occur in the cardiac left ventricle (LV) following 16 wk of uncontrolled diabetes. Using this methodology, we have characterized the attendant molecular changes more completely than has previously been reported. The findings reported here have significance in further defining the molecular pathophysiology of diabetic cardiomyopathy, an increasingly recognized and potentially fatal disease of the heart.


Animal model.

Male Wistar rats (220–250 g, University of Auckland) were randomized into two groups prior to injection. They were anesthetized (halothane 2–5%) and then administered 60 mg/kg of freshly prepared streptozotocin (STZ, Sigma-Aldrich) via tail-vein injection; controls (normal) received the corresponding volume of saline (0.3–0.5 ml final). Animals were recovered and housed in pairs on FibreCycle bedding (12-h light-dark cycle, 50–70% humidity, 19–21°C). Initial [glucose]blood and body weight were measured at the time of injection and 3 days afterward and were thereafter monitored weekly. Diabetes was diagnosed as a sustained [glucose]blood >11 mM, and insulin was not administered. All procedures and protocols were approved by the University of Auckland Animal Ethics Committee.

Cardiac structure and function.

In a parallel study, similarly treated diabetic and sham rats were prepared, and cardiac performance was measured with an isolated-perfused working heart model to confirm cardiomyopathy had been established. Animals were anesthetized and heparinized (200 IU/kg iv), and hearts were excised and immersed in 4°C Krebs-Henseleit bicarbonate buffer (KHB). Retrograde (Langendorff) perfusion was established [KHB, 37°C, gassed with O2/CO2 95:5 (vol/vol)]. Working-mode perfusion was then established (preload, 10 cmH2O; afterload, 55.9 mmHg), with pacing (300 bpm, Digitimer) and intrachamber LV pressure (SP855, AD Instruments) and cardiac output (Transonic T206) measured. Pressure and flow data were recorded (Powerlab16s, ADI), and ventricular pressure development (+dPLV/dt) and relaxation (−dPLV/dt) were derived. Atrial filling pressure was decreased (5 cmH2O) and then increased [seven steps of 2.5–20 cmH2O (final)], and 1-min averages were extracted. Intergroup differences were contrasted (mixed models/repeated measures, SAS v. 8.1).

Tissue processing.

Sixteen weeks after diabetes induction, animals were anesthetized (halothane 2–5%) and given heparin (200 IU/kg iv, Mayne Pharma Pty). Animals were killed by cervical dislocation, and hearts were rapidly excised in an RNase-free environment. The aortic remnant was tied to a custom-made perfusion apparatus and perfused with 40- to 60-ml phosphate buffer solution (PBS: NaCl, 137 mM; KCl, 2.7 mM; Na2HPO4, 4.29 mM; KH2PO4, 1.46 mM) at 4°C, flow rate 15 ml/min (GENIE 220 pump) until no blood was evident in the eluent. The LV was then dissected from the rest of the heart and placed in RNAlater (Qiagen).


RNA from the LV of 14 animals was obtained with either the Qiagen MIDI RNeasy RNA extraction kit or the Ambion Mini RNAqueous RNA extraction kit, as per protocols. RNA was quantitated using a NanoDrop ND-1000 (NanoDrop Technologies), and the quality of RNA was determined (Bioanalyzer, Agilent Technologies). An RNA integrity number of ≥8.5 was considered adequate for analysis.

We used 5 μg from each RNA preparation to make a hybridization mixture containing 15 μg of biotinylated cRNA (Affymetrix protocol) and adjusted it for possible residual RNA carryover. Complementary RNA was hybridized overnight to a microarray chip (Affymetrix Rat GeneChip 230 2.0) before being scanned (Affymetrix Scanner 3000) and processed (GCOS, Affymetrix).

The data discussed in this publication have been deposited in the National Center for Biotechnology Information's Gene Expression Omnibus (GEO, and are accessible through GEO Series accession number GSE5606.

Statistical methods.

Microarray data (.CEL files) were analyzed by several statistical methods to identify differences in levels of RNA between the diabetic and normal animals. All analyses were performed in R (21) using the Bioconductor collection of analysis packages (17). Normalization was by the robust multichip averaging algorithm of Irizarry et al. (23) as implemented in the Affy package for Bioconductor (15) without background correction. Probe sets undergoing changes in expression level were detected with the limma package (35) with the moderated t-statistic of Smyth (36) used to assess the strength of differential expression (relative to variability) for each probe set. Benjamini and Hochberg's false discovery rate-controlling method was used to produce adjusted P values to limit the expected proportion of false positive results to <5% (3). Probe sets with adjusted P values <0.05 were considered to show statistically significant expression level changes.

Reverse transcriptase quantitative PCR.

Complementary DNA was synthesized using total RNA from each of the 14 samples used for the microarrays. Taq-Man Primers and reagents were from Applied Biosystems (Foster City, CA). Primers were: CPT1, CytB5 (mitochondrial), MYLDC, TFAM, TXN1, and PDK4 and 18s rRNA was used as the endogenous control for all experiments.

Briefly, RNA underwent a DNase-1 digestion and cDNA was synthesized using SuperScript III (Invitrogen) and random hexamer primers. We used 2 ng of cDNA per reaction with the Taq-Man primer MasterMix, as per the manufacturer's instructions. All RNA levels were measured in triplicate using an Applied Biosystems 7900HT Real Time PCR System and analyzed using the comparative CT method (ABI PRISM 7700 User Bulletin #2, P/N 4303859).


Diabetes and cardiomyopathy.

Diabetes was confirmed in animals 3 days after STZ injection. Diabetic animals had mean [glucose]blood of 27.7 ± 1.4 mM (mean ± SE, n = 22) compared with 4.8 ± 0.1 (n = 21 in controls, Table 1). Affected animals also displayed attenuated weight gain (diabetes: 307.3 ± 6.3 g, n = 21 vs. controls: 420.0 ± 20.2, n = 22, final Fig. 1B); polydipsia (diabetes: 443 ± 18 ml H2O/day vs. controls: 70.5 ± 3.1 ml, n = 17 for both groups); hyperphagia (diabetes: 649.3 ± 14.7 g food/wk vs. normal: 391.2 ± 6.6 g, n = 17, both groups); and a significant increase in circulating triglycerides and free fatty acids compared with controls (Table 1).

Fig. 1.

Diabetes caused impaired cardiac structure and function in rats. A: hearts were weighed prior to perfusion, and HW/BW was calculated based on the heart weight and the final weight of the animal (prior to death). The relative heart mass of the diabetic group is greater than that of the normal group indicating hypertrophy (*P < 0.0001). B: animals were weighed weekly; (▪) Control, (▴) Diabetes. The diabetic group gained weight for the first 3 wk before plateauing, whereas the normal group gained weight throughout (*P < 0.0001). C: and D: functional data derived from isolated perfused working hearts excised after 16 wk of diabetes; (▪) Control, (▴) Diabetes. C: +dPLV/dt with increasing preload (*P < 0.0001). D: −dPLV/dt with increasing preload (*P < 0.0001). HW, heart weight; BW, body weight.

View this table:
Table 1.

Concentrations (mM) of blood glucose, plasma triglycerides, and FFA in STZ-diabetic and control rats

The mean cardiac/body mass ratio was increased by 23% in diabetic compared with control animals, indicating an increase in the relative size of the diabetic heart despite there being significantly lower absolute cardiac (Fig. 1A) and body weights (data not shown). Diabetic animals gained weight more slowly than normal controls (Fig. 1B).

To identify that our model showed signs of established DCM, we performed an ex vivo perfused heart study. Isolated hearts from diabetic animals had impaired systolic and diastolic function compared with controls (Fig. 1, C and D), which was associated with deterioration in cardiac performance and is consistent with previous reports (7, 8, 11, 12).

Gene expression.

Analyses of changes in gene expression after 16 wk' diabetes were performed in LV RNA that was isolated from seven random animals in each of the STZ-diabetic and normal groups and then hybridized to 14 Affymetrix arrays. Statistical analysis of the expression data produced a list of genes whose expression was changed significantly in diabetes (see Tables 24). For each gene, the P value was used to determine significance rather than fold-change, to include a much larger proportion of candidate gene changes. By this criterion, 1,614 significant changes were determined (each P < 0.05) from a total of 31,100 spots on the array.

View this table:
Table 2.

Comparisons between fold-changes in gene expression in diabetic LV tissue as determined by RTqPCR and microarray methods

View this table:
Table 3.

Genes that play roles in carbohydrate metabolism (as defined by their GO annotation) whose expression in LV myocardium was significantly altered by diabetes

View this table:
Table 4.

Genes that play roles in lipid metabolism (as defined by their GO annotation) whose expression was significantly altered in diabetic LV

Changes in expression of genes that encode cardiac structural proteins.

As noted above, diabetic hearts had larger relative masses (Fig. 1A). Fang et al. (14) have identified three stages of diabetic cardiomyopathy, the middle and late stages of which are closely associated with myocellular enlargement and fibrosis.

The myosin heavy chain (MHC) is an integral part of the cardiac contractile apparatus. There are two myosin isoforms in rat heart, αMHC (adult) and βMHC (fetal). Here we show that diabetic hearts had increased βMHC (50% elevated), whereas αMHC was decreased (to 35% of normal). It has previously been reported in various other cardiac diseases that the MHC isoform expression pattern may strongly influence cardiac contractile properties and energetics (32). The amount of myocardial βMHC mRNA was minimal in normal rats but reportedly increased in several pathophysiological conditions, including cardiac hypertrophy or type II diabetes (12). The basis for such switching of MHC isoforms is not entirely clear; however, the transition from αMHC to βMHC is thought to induce a slowing of myocardial contraction and increased economy in force generation (32). This may be an adaptation of the heart to stress conditions induced by the diabetic state.

Disturbances in expression of genes whose products regulate energy metabolism.

The percentages of gene changes in specific cellular compartments were analyzed using the Gene Ontology (GO) cellular component classification system (1) (Fig. 2). The largest proportion of diabetes-modified genes detected was associated with “membrane” or “plasma membrane” components. The second was genes associated with the mitochondria, a feature that is noteworthy due to the mitochondrion's association with energy metabolism (38, 41, 42, 44). Our microarray results (discussed below) provide a unique global snapshot of the underlying molecular processes that until now have frequently been considered in various studies of isolated pathways. We now highlight a number of interesting points that relate to the changes in cardiac energy metabolism that occur in severe diabetes.

Fig. 2.

Pie chart illustrating the global functional distribution of genes in rat LV myocardium whose expression was significantly altered by diabetes. This graph is based on GO Cellular Component classification (1). The total number of genes with a GO Cellular Component Classification term and at least one other associated GO term were used in this analysis, yielding a total of n = 394 out of an overall total of n = 1,614 significantly modified genes. Genes were sorted according to their GO classification and the total in each category then divided by the total number of genes analyzed (n = 394) to generate respective percentages, as shown. LV, left ventricle; GO, gene ontology.

Glucose metabolism.

We found decreases in both GLUT1 and −4 mRNAs as well as changes in key metabolic regulatory enzymes such as pyruvate dehydrogenase phosphatase (PDP2) and pyruvate dehydrogenase kinase (PDK4), both of which play roles in the transport of pyruvate into the mitochondria (all genes involved in glucose metabolism found to be altered in this study are presented in Table 3). In accordance with previous reports (12), GLUT4 and GLUT1 expression were both decreased, by 25 and 50% respectively, consistent with a marked diminution in myocardial glucose uptake capacity. By contrast, PDK4 increased 659% in diabetic compared with normal individuals [confirmed by reverse transcriptase quantitative (RTq) PCR], whereas PDP2 was decreased 59% in diabetic tissue (Table 2).

In mitochondria, the kinase (PDK4) and phosphatase (PDP2) are simultaneously active and their relative activities determine the proportion of inactive/active pyruvate dehydrogenase (PDH) complex (25). It has been hypothesized that an increase in the amount of PDK4 protein leads to greater phosphorylation and lower activity of the PDH complex. This is said to suppress glucose and lactate oxidation and favor fatty acid oxidation (47). Such changes could be further enhanced by the increase in PDP2 expression seen here and by Huang et al. (20), who reported that STZ-induced diabetes (48-h duration) decreased both protein and mRNA levels of PDP2 in the heart. Together these findings emphasize that concurrent upregulation of PDK expression and downregulation of PDP expression likely contributes to hyperphosphorylation of PDH in diabetes (20). Inhibition of this complex can exacerbate the diabetic state by inappropriately sparing glucose and gluconeogenic substrates from complete oxidation despite abundant [glucose]blood (47).

Fatty acid metabolism.

In diabetes, the heart is exposed to an environment that is both hyperglycemic and hyperlipidemic (42). In contrast to glucose metabolism-related genes, however, we found that those involved in fatty acid uptake and oxidation were generally upregulated (Table 4). Here, CD36/fatty acid translocase (CD36/FAT) was increased 41% in the diabetic compared with the normal (Table 4), consistent with the hypothesis that there is increased overall fatty acid uptake in diabetic myocardium. CD36/FAT has been identified as one of three fatty acid transport proteins that increased fatty acid uptake when overexpressed in cell lines (30), and it is thought to be the primary fatty acid transporter in cardiac and skeletal muscle (30). This result and the disproportionate increase in lipid metabolism genes have been recently confirmed in LV tissue after only 2 wk of diabetes (27), indicating that this is an early and sustained change. Taegtmeyer et al. (42) have proposed that as diabetes progresses, the excessive availability of lipids and fatty acids exceed the maximal rate of oxidation in the heart, thereby causing lipids to accumulate within the cardiomyocyte. This theory is further supported with evidence from Su et al. (40), who reported a fourfold increase in acylcarnitine in diabetic myocardium (40).

In our study we report for the first time the upregulation of angiopoietin-like protein-4 (Angptl-4) in a diabetic heart (25-fold increase relative to normal tissue). Angptl4 is also known as PPARγ angiopoietin-related fasting-induced adipose factor or hepatic fibrinogen/angiopoietin-related protein (16) and has been shown to be a potent inhibitor of lipoprotein lipase (LPL) activity, leading to increased serum triglyceride levels (49). The highest expression of Angptl4 is found in white adipose tissue; however, there is significant expression previously reported in normal liver, heart, and skeletal muscle (26). As a transcriptional target of PPARγ, Angptl4 has been hypothesized to play a role in the modulation of adipogenesis, insulin sensitivity, or energy metabolism and is elevated in genetically obese mice (16).

A recent study published by Yu et al. (50), looking at heart-specific overexpression of Angptl-4 in mice, found that the presence of an Angptl-4 transgene in the heart caused an 80% reduction in cardiac LPL activity and a greater than twofold increase in fasted plasma triglyceride levels. Our current work supports this finding with a 25-fold upregulation of Angptl-4 and a 2.8-fold increase in serum triglycerides (see Table 1). Further studies by Yu et al. (50) found that overexpression of Angptl-4 led to altered substrate oxidation and impaired cardiac function similar to that seen in cardiomyopathy. They were able to show a 60% reduction in triglycerides (intralipid) utilization in the Angptl-4 transgenic mouse during isolated perfused heart studies.

The heart obtains fatty acid for use as an energy source through three different pathways: circulating free fatty acids bound to albumin, circulating chylomicrons and VLDL, which require LPL before utilization and hydrolysis of triglycerides present in the cardiomyocyte (50). Our current study is consistent with the view that the heart is not able to use the LPL pathway to obtain free fatty acids for energy use and it therefore this likely comes from one of the other two pathways.

The gene products that channel fatty acids into the mitochondria are: carnitine palmitoyl transferase-1α (CPT1α), which increased 116%; carnitine-acylcarnitine translocase (solute carrier family 25, member 20), which increased 30%; and carnitine palmitoyl transferase 2, which increased 40% in diabetic animals (Table 4). Fatty acid oxidation is limited by the transport of fatty acid residues into the mitochondria through CPT1 (12). The two isoforms of CPT1 (CPT1α and CPT1β) are both expressed in the rat heart. Cook et al. (10) reported that STZ-induced diabetes resulted in increased CPT1α but that CPT1β levels remained the same. CPT1α is predominantly expressed in the fetal heart and gradually reduces in expression as the rat develops. This increase in CPT1α observed in the current study indicates the heart may be increasingly reliant on fatty acids as a fuel source and provides evidence for the accumulation of acylcarnitine and intermediates in the heart during diabetes (40).

CPT1α is regulated mainly through inhibition by malonyl-CoA (34). Increased expression of malonyl CoA-decarboxylase (MCD), a negative regulator of malonyl-CoA production, could decrease the malonyl-CoA available to inhibit Cpt1. We found a significant increase (50%) in expression of MCD in the diabetic (Table 4), and this result was confirmed by RTqPCR (Table 2). These results indicate that there are at least two factors contributing to the reported increase (40) in transport of fatty acylcarnitine into the mitochondrion. One is decreased inhibition of CPT by malonyl-CoA (both isoforms), and the other is increased expression of the transporting proteins CPT1α and β, both of which potentially lead to excess fatty acids within the mitochondrion.

Once fatty acids have been transported into the mitochondrion, they are converted to acetyl-CoA through β-oxidation and the trifunctional β-oxidation complex (TOC) (48). Here, we found that three of the four genes whose products comprise the TOC: enoyl-coenzyme A, hydratase/3-hydroxyacyl coenzyme A dehydrogenase, and acyl-coenzyme A dehydrogenase very long chain, were increased in the diabetic state compared with normal. A proteomic analysis of changes in mitochondrial proteins with diabetes reported a similar array of β-oxidation protein changes (45). Therein, a number of proteins studied from the tricarboxcylic acid (TCA) cycle were unchanged in diabetes (45); that finding is consistent with our results and provides further evidence that, although the heart is taking up and processing more fatty acids, a similar increase is not observed in the TCA cycle, which could explain the increase in unused acyl-CoA. Furthermore, this result is also consistent with the hypothesis of Taegtmeyer et al. (42), by which fatty acid uptake exceeds the rate of β-oxidation leading to myocellular lipid accumulation and lipotoxicity. For an overview of the processes described in this section and in glucose metabolism, see Fig. 3.

Fig. 3.

Schematic representation of fuel-metabolic pathways affected by diabetes in the heart. Changes occurred in the expression of genes encoding enzymes that regulate pyruvate dehydrogenase (PDH), consistent with a decrease in its activity and therefore ability to catalyze use of pyruvate for energy generation. Concurrent increases in expression of genes encoding mitochondrial fatty acid uptake and β-oxidation are consistent with reliance of the heart on fatty acids as its main energy source in severe diabetes. CPT, carnitine palmitoyl transferase; TCA, tricarboxycylic acid.

In summary, much evidence supports the idea that increased fatty acid oxidation and accumulation in the heart substantively contribute to the damage seen in diabetes. Our results are also consistent with the hypothesis (42) that, when fatty acid availability exceeds its oxidation rate, intramyocardial lipid accumulation occurs that could lead to lipotoxicity, which in turn might induce ROS accumulation, inducible nitric oxide synthase induction, and apoptosis.

ROS generation: a hypothesis for changes in gene expression.

Linkages between PKC activation, advanced glycation endproduct (AGE) formation, increased flux through the polyol pathway, decreased GSH, and ROS formation have been reported by Nishikawa et al. (31), who suggested that, in cultured endothelial cells, the TCA cycle might be the source of increased ROS-generating substrate induced by hyperglycemia (31). As a result of the observed changes in genes involved in fuel metabolism we decided to assess expression of genes implicated in diabetic complications resulting from increased ROS, focusing specifically on the pathways outlined above (6). Glutathione (GSH) participates in the cellular defense against oxidative stress by reducing disulfide bonds in proteins and other intracellular molecules or by scavenging free radicals and reactive oxygen intermediates (46). Here, we found expression changes in several genes involved in GSH metabolism, such as members of the glutathione S-transferase family. This family of enzymes utilizes GSH in reactions contributing to the transformation and detoxification of a wide range of compounds, including carcinogens, therapeutic drugs, and products of oxidative stress (39). Expression of nine GSH S-transferase family members, including one specific to mitochondria, was significantly altered. Of these, six were increased in diabetes, which could explain decreased GSH availability for cellular defense and implicate this pathway as one that may contribute to ROS increases.

Another pathway previously implicated is the protein kinase C (PKC) pathway. Inoguchi et al. (22) reported increased membranous PKC activity and total diacylglycerol in the diabetic rat heart. In this study we found that PKC-δ was increased 30%. Others have suggested, based on immunohistochemical studies in aorta and heart, that PKC-β and PKC-δ appear to be preferentially activated in diabetic rats (28). Increased PKC activity in diabetic cardiovascular tissues may lead to alterations in expression of several growth factors, cytokines and vasoactive molecules, including endothelin-1 (ET-1). The endothelin family interacts with three receptor populations: ETA, ETB, and ETC. Endothelin and its receptor mRNAs were reportedly increased in retinas from diabetic rats (8). In our microarray study, ETA and ETB were both increased 59 and 92%, respectively, in diabetes, which implies that there is a pathway-based change occurring in this instance.

Of note, AGE formation and the hexosamine pathways were not found to have gene changes associated with them. This may be a result of nongene-related changes occurring (particularly in the case of AGE formation) or a dilution effect occurring due to the nature of the cell types within the myocardium. The myocardium is made up of cardiomyocytes, fibroblasts, and endothelial cells. Many of the results in the publications cited in our discussion were derived from cultured endothelial cells (6, 31); it may be that specific cell expression analysis is required to fully determine the contributions of each cell type to the overall damage seen in DCM.

The linking factor between ROS overproduction and the activation of the four pathways outlined above has been proposed to be activation of the nuclear DNA repair enzyme poly(ADP-ribose) polymerase (PARP) (13). Here, expression of PARP (TCDD-inducible PARP) was increased 28% in diabetic compared with control heart. Normally, PARP resides in the nucleus in an inactive form. Brownlee (6) reported that increased intracellular glucose generates increased mitochondrial ROS, thereby inducing DNA strand breaks and activating PARP. Once activated, PARP catalyzes polymerization of ADP-ribose, which reportedly accumulates on GAPDH and other nuclear proteins. PARP modification of GAPDH reportedly decreases its activity, leading to activation of the polyol pathway, increased intracellular AGE formation, and activation of PKC, NF-κB, and hexosamine pathway flux (6).

Relationship of fatty acid oxidation to ROS generation and mitochondrial function.

One prominent hypothesis, by which excessive ROS production might lead to diabetic complications, has in the main been derived from studies in endothelial cells (6). These have high, fixed glucose-uptake capacity and so are vulnerable to large glucose influxes driven by mass action in hyperglycemia (5). By contrast, glucose transport is decreased and fatty acid transport increased in other cardiac cell types, despite high circulating glucose. However, oxidation of fatty acids and acetyl-CoA generates the same electron donors as glucose oxidation. Thus, increased fatty acid oxidation can drive excess mitochondrial ROS production via the same downstream mechanism as glucose (6). St-Pierre et al. (37) reported that rat skeletal muscle and cardiac mitochondria produced measurable rates of H2O2 (ROS precursor) when respiring on palmitoyl carnitine but not when utilizing pyruvate or succinate (37), which is also consistent with this mechanism.

Partial evidence for increased ROS generation by fatty acid oxidation in this study comes from an observation that uncoupling protein (UCP)-3 mRNA levels were increased 150% in diabetic myocardial tissue (Table 4). UCP-3 mRNA expression is increased when fatty acids are high (37), as well as in skeletal muscle of db/db mice (9). UCP3's role is currently debated, but a recent review identified four distinct models for its possible function (4). One prominent hypothesis is that UCP family members attenuate mitochondrial free radical production, thereby protecting tissues against oxidative damage. Mitochondrial ROS production is sensitive to the proton-motive force established across the inner membrane by electron transport, so the mild uncoupling caused by activation of UCP3 might lower this force slightly, thereby attenuating ROS production and protecting against ROS-mediated cellular damage (4). An increase in UCP3 mRNA is consistent with the hypothesis of a diabetes-evoked increase in fatty acids leading to increased ROS production and a requirement for its protective upregulation. Further study is required to determine whether this effect occurs through attenuation of free radical production or export of fatty acids and their peroxides from the mitochondria (4).


We have shown here that genes corresponding to proteins expressed in mitochondria account for a disproportionate number of those whose expression was significantly modified in DCM. We interpret these findings as being consistent with the hypothesis that mitochondria act as key targets of the pathogenic processes that cause diabetic heart disease. Furthermore, our transcriptomic analysis is also consistent with the hypothesis that diabetes induces global perturbations in the expression of genes regulating fatty acid metabolism, whose dysfunction is likely to play a key role in the promotion of oxidative stress, thereby contributing to diabetic myocardial disease. In particular, our study points to impaired regulation of mitochondrial β-oxidation as central in the mechanisms contributing to DCM pathogenesis. We note that these conclusions, which are based on the number of gene changes we have documented herein, are consistent with other available lines of evidence (29, 34, 42, 43).


We thank L. Williams, V. Tintinger, and R. Smith for technical assistance.

We acknowledge support from the following sources: the Endocore Research Trust; the Health Research Council of New Zealand; the Foundation for Research, Science and Technology (FRST, New Zealand); the Maurice & Phyllis Paykel Trust; the NZ Lottery Grants Board; and The University of Auckland Research Fund.


  • Address for reprint requests and other correspondence: G. J. S. Cooper, Level 4, Thomas Bldg., School of Biological Sciences, Univ. of Auckland, Private Bag 92019, Auckland, New Zealand (e-mail: g.cooper{at}

    Article published online before print. See web site for date of publication (


View Abstract