Activation of the receptor for advanced glycation end products (RAGE) is associated with long-term complications in diabetes mellitus. In this study, we tested whether RAGE activation in the diabetic myocardium is implicated in the development of cardiac dysfunction.
Methods and results
Using MRI and conductance catheter techniques, we evaluated cardiac function in a type 2 diabetic mouse model (db/db), and assessed the effect of blocking RAGE with a RAGE antibody. Gene expressions were evaluated in samples of myocardial tissue. Diabetic db/db mice demonstrated an accelerated age-dependent deterioration in cardiac function associated with altered expression of genes related to cardiac structure and function. Blockage of RAGE signalling prevented the reduction in systolic function (preload recruitable stroke work: 109.8 ± 13.8 vs. 94.5 ± 14.9 mmHg/µL, P = 0.04) and development of increased LV diastolic chamber stiffness (0.18 ± 0.05 vs. 0.27 ± 0.07 mmHg, P = 0.01). The cardiac expression of collagen (col1a1) was reduced by approximately 45% and the expression of myosin was switched from the foetal isoform (MHCβ) to the adult isoform (MHCα).
Activation of RAGE is a significant pathogenetic mechanism for the development of cardiac dysfunction in type 2 diabetes. The underlying mechanisms involve not only the passive biophysical properties of the myocardium but also myocyte function.
The receptor for advanced glycation end products (RAGE) has gained attention as a modulator of complications associated with diabetes mellitus. Heart failure may develop in diabetic patients as a consequence of diabetic cardiomyopathy (DCM), independently of the high prevalence of coronary artery disease, and hypertension. However, the extent to which RAGE is involved in the pathogenesis of DCM is unknown.
The receptor for advanced glycation end products activation can initiate a number of mechanisms that may affect the function of the heart, including release of growths factors and additional RAGE ligands, formation of ROS, angiogenesis, and endothelial dysfunction.– In the vascular wall, RAGE activation has been associated with changes characteristic of diabetic microangiopathy, and in isolated cardiomyocytes changes in Ca2+ homeostasis that are consistent with cardiac dysfunction., The receptor for advanced glycation end products expression increases with ageing,– but expression in the diabetic heart is uncertain. The receptor for advanced glycation end products has multiple ligands including advanced glycation end products (AGEs) and proinflammatory ligands, which are formed at an accelerated rate in diabetes due to hyperglycaemia and increased oxidant stress., AGE crosslink formation may contribute to development of DCM by directly affecting the structure and function of proteins; and cardiac function seems to be improved by prevention of AGE cross-link formation or breakage of existing cross-links.,,, Whether reduced activation of RAGE prevents development of cardiac dysfunction has yet to be clarified.
The aim of the present study was to evaluate in vivo age-dependent alterations in cardiac function in type 2 diabetic mice and in healthy controls, and to assess the potential role of RAGE in the development of cardiac dysfunction. We used a validated mouse-model of type 2 diabetes mellitus to compare cardiac function at three time points during the progression of diabetes and to evaluate the effects of RAGE blockage with a RAGE antibody (RAGE-Ab). The influence of RAGE on underlying molecular mechanisms was evaluated by characterizing the expression of a number of genes related to intracrine growth factors, structural and functional components during the development of cardiac dysfunction before and after RAGE blockage.
Male db/db mice (C57BLKS/J-leprdb/leprdb) and age-matched db/+ littermates (C57BLKS/J-leprdb) (M&B, Ry, Denmark) were used. The mice were housed 4 to 10 per cage in a room with a 12:12 h light cycle and a temperature of 21°C. The mice were allowed free access to standard diet (Altromin, Lage, Germany) and tap water. The investigation conformed with the Guide for the Care and Use of Laboratory Animals published by the US National Institutes of Health (NIH Publication No. 85-23, revised 1996). All procedures were approved by the Danish Animal Experiments Inspectorate.
The haemodynamic, biochemical, and genetic changes in db/db mice (diabetic, n = 27) were compared with db/+ mice (controls, n = 27), when the mice were 8, 12, and 16 weeks of age. These ages correspond to early, manifest, and advanced stages of diabetes.
Receptor for advanced glycation end products blocking study
Diabetic and control mice (db/db n = 18, db/+ n = 16) were randomized at 7 weeks of age to receive intraperitoneal injections of a neutralizing murine RAGE-Ab or an isotype-matched placebo-Ab (placebo). An initial dose of 300 µg per animal was followed by 100 µg doses three times weekly for 5 weeks, then the mice underwent assessment of cardiac function and samples were collected for PCR and immunoblotting.
Non-fasting blood samples were collected from the tail vein 2 days before the in vivo evaluation of cardiac function. The plasma was used to measure glucose (Abbott Laboratories, Precision Xtra Plus, Bedford, MA, USA), non-esterified fatty acids (NEFA; Wako chemicals, no. 999-75406, Neuss, Germany), and total triglycerides (Sigma-Aldrich, TR0100, Saint Louis, MO, USA).
Preparation of receptor for advanced glycation end products blocking-Ab and placebo-Ab
The neutralizing monoclonal RAGE-Ab was prepared and characterized as previously described in detail. The placebo-Ab was a monoclonal anti-diphtheria toxoid IgG3 k-Ab produced by the cell line HYB 123-1 and purified by protein G chromatography. The antibodies were dissolved in NaCl (0.154 mmol/L) and injected in a volume of 0.5 mL.
In vivo evaluation of cardiac function
The animals were mechanically ventilated with 2.2% isoflurane in approximately 50% O2 and 50% room air during the evaluation of in vivo cardiac function using MRI and conductance catheter technique. A number of 1.0 mm thick adjacent short- and long-axis slices were acquired by MRI with a temporal resolution of 7 ms. The LV volumes were calculated according to the area-length method where the mid-LV cross-sectional area (A) and the length of the ventricle taken from the midpoint of the annulus to the apex (L) was used in the formula: 5/6 × A × L. The LV myocardial volumes were calculated by planimetry of slices in the short-axis view. Pressure–volume loops were acquired on open chest animals using a 1.4-F four electrode conductance catheter for mice (SPR-839, Millar Instruments Inc., Houston, TX, USA) inserted in the LV directly through the apex of the heart (Figure 1). In the age-dependant study, volumes were calibrated with a series of cylinders and injection of hypertonic saline in the pulmonary artery. In the RAGE blocking study, conductance catheter volumes at steady-state were calibrated according to the LV volumes obtained by MRI at steady state. The difference between the LV minimal pressure and the pressure at the peak of the atrial systole was used as ΔPdiastole. The LV diastolic chamber stiffness was calculated as ΔPdiastole/SV. The end-systolic elastance (Ees) was calculated as the slope of the end-systolic pressure–volume relation (ESPVR). Cardiac output divided by body weight was used to calculate cardiac index. The preload recruitable stroke work (PRSW) was calculated as the slope of the stroke work to end-diastolic volume relation.
Total RNA was extracted from 20 mg snap frozen LV tissue samples and subjected to DNase treatment to eliminate contaminating DNA. Only samples with an RNA integrity number >9.0 were accepted (Agilent 2100 bioanalyzer, Agilent, Santa Clara, CA, USA). First-strand cDNA was synthesized from 1 µg of total RNA. Primers and probes were designed using an online assay design centre (Roche, Universal Probelibrary, Version 2.20) and Oligo primer analysis software (Version 6.71) (Table 1). Primers were produced by a commercial supplier (DNA-technology, Aarhus, Denmark). All assays were evaluated and optimized for best specificity, sensitivity, and reproducibility using SYBR Green and melting curve analyses. Amplification efficiencies above 85% were accepted. Subsequently, fluorescent reporter probes were used for the real-time qPCR assays (Universal Probe Library, Roche, Hvidovre, Denmark). PCR reactions were setup on a MX3000P (Stratagene, Cedar Creek, TX, USA) and average values of duplicates were used for quantification. Standard curves were constructed for all genes. Based on evaluation of the expression stability of a group of potential reference genes, four were selected (Table 1) and measured in all samples. The Normfinder software (MDL, Aarhus, Denmark) was used to identify Actb and Ppia as the most stable combination of two reference genes. For each sample, a combined normalization factor of these genes was calculated and used to normalize the expression levels of target genes.
Table 1. List of evaluated genes
Product size (bp)
aReference genes are in boldface. Probe number refers to the LNA-based reporter probe.
Tissue samples of frozen LV (50 mg) were homogenized in extraction buffer. After centrifugation at 1000 g, the pellet was solubilized at 65°C for 15 min in Laemmli sample buffer containing 2% SDS and stored at −20°C.
Total protein concentration of the homogenate was measured using a Pierce bicinchoninic acid protein assay kit (Roche, Hvidovre, Denmark). The samples were separated on 26 lane 12% polyacrylamide gels (Bio-Rad Protean II, BIO-RAD, Hercules, CA, USA). For each gel, identical gels were subjected to Coomassie-staining which allowed for validation of equal loading. Proteins were transferred to PVDF membranes (Immobilon FL, Millipore, Copenhagen, Denmark). The samples were incubated overnight at 4°C with anti-RAGE rabbit antibody [(1:500). PA1-075, Affinity Bioreagents, Golden, CO, USA] and beta-actin antibody [(1:1000) 3597-100, Biovision, CA, USA]. After washing, blots were incubated with fluorescent labelled secondary-Ab (Alexa Flour 680, Molecular Probes, Invitrogen, Taastrup, Denmark) for 1 h at room temperature. All membranes were scanned in one session at 700 nm by an Odyssey infrared imaging system (Li-Cor Biosciences, Lincoln, NE, USA) and analysed using Odyssey application software (version 1.2.15). Samples from all groups were included on every gel and one sample from each group was repeated on all gels and used to normalize gels to each other. Furthermore, the expression of RAGE was normalized to the expression of beta-actin. The RAGE-Ab used for immunoblotting was evaluated on samples from lung tissue (positive control) and with a neutralizing peptide as negative control (PEP-194, Affinity Bioreagents, Golden, CO, USA) (Figure 3).
Student's t-test and one-way ANOVA with Bonferroni post hoc analysis were used to compare two or more groups. Gene expressions and protein content data were compared using Wilcoxon's rank-sum test or Kruskal–Wallis one-way ANOVA. P-values less than 0.05 were considered statistically significant.
The diabetic phenotype, characterized by elevated blood glucose and body weight, was present at all stages in the db/db mice (Table 2). The average maximal and minimal LV pressures in all groups were 104.8 ± 12.4 and 5.8 ± 2.5 mmHg, respectively. We found no differences between db/db and db/+ mice. The PV-loop-derived indices of LV contractility were reduced from 8 weeks (Ees) and 12 weeks (PRSW) in db/db mice and declined progressively during the observation period (Table 2 and Figure 1). Indices of diastolic function were reduced from 12 weeks (ΔPdiastole) and 16 weeks (time constant of isovolumic relaxation: Tauweiss) in db/db mice, which was reflected by a similar trend in LV diastolic chamber stiffness (Table 2 and Figure 1). Systolic volume was increased at 8 and 12 weeks in db/db mice and by 16 weeks, a generalized dilatation had developed (Figure 1). The cardiac index was significantly reduced in db/db mice at all ages (Figure 1).
Table 2. Characteristics of the diabetic and control mice in the age-dependent study at 8, 12, and 16 weeks (Means ± SD)
Diabetic (n = 10)
Control (n = 10)
Diabetic (n = 10)
Control (n = 10)
Diabetic (n = 7)
Control (n = 7)
aNS, not significant; Body temp., body temperature; EF, ejection fraction; Ees, end-systolic elastance; Tauweiss, time constant of isovolumic relaxation; ΔPdiastole, pressure increase during diastole.
The mRNA expression of Edn1, RAGE, and BNP in hearts from db/db mice differed from that in db/+ mice, but did not demonstrate age dependency (Figure 2). The mRNA level of RAGE and Edn1 was increased and BNP decreased, in db/db mice compared to db/+ mice. At the protein level, RAGE content was dependent on age. In db/db mice, RAGE was reduced by ageing and in db/+ mice the level was increased by ageing. Consequently, the level was significantly lower at 16 weeks in db/db mice compared with db/+ mice (Figure 3).
The mRNA expression of eNOS and ANP demonstrated moderate age dependency (Figure 2). At 8 and 12 weeks, the eNOS mRNA levels were identical in db/db and db/+ mice and at 16 weeks, the expression was reduced in db/db mice. At 8 weeks, ANP mRNA expression was reduced to approximately 25% in db/db mice compared with db/+ mice.
The mRNA expressions of VCAM-1 and MHCβ were highly dependent on age (Figure 2). The VCAM-1 mRNA level was reduced in db/db mice at 8 weeks and increased at 16 weeks compared with db/+ mice. The MHCβ mRNA expression was comparable at 8 and 12 weeks, but significantly increased by 3.5-fold at 16 weeks in db/db compared with db/+ mice. The expressions of connective tissue growth factor (CTGF), TGF-β1, phospholamban (PLB), SERCA2a, Col1a1, MHCα, and vascular endothelial growth factor A (VEGF-A) were unaffected at the mRNA level in db/db mice compared with controls.
The receptor for advanced glycation end products blockage study
The average non-fasting plasma-glucose in the db/db mice was 16.8 mmol/L, at inclusion. During the 5 weeks of treatment, one db/db mouse treated with placebo-Ab died. Body weight and blood glucose were unaffected by RAGE-Ab and placebo treatment (Table 3). The plasma concentration of NEFA was similar in all groups. Total triglyceride concentration tended to be increased in db/db mice compared with db/+ mice (RAGE-blocked: P = 0.01, placebo treated: P = 0.15) and reduced in RAGE-blocked mice compared with placebo-treated mice (db/db: P = 0.39, db/+: P = 0.02) (Table 3). The mean maximal and minimal LV pressures of all groups were 100.8 ± 12.1 and 5.6 ± 2.7 mmHg and were not different between groups.
Table 3. Characteristics of the diabetic and control mice in the receptor for advanced glycation end products blocking study after 5 weeks of treatment (means ± SD)
Diabetic vs. control
RAGE blocking (n = 9)
Placebo (n = 8)
RAGE blocking (n = 8)
Placebo (n = 8)
aNS, not significant; FS, fractional shortening. For other abbreviations, please see Table 2.
In contrast to placebo-treated db/db mice, RAGE-Ab treatment restored indices of LV contractility, PRSW, ejection fraction, and fractional shortening in db/db mice (Table 3 and Figure 4). Similarly, compared with placebo-treated db/db mice, indices of diastolic function (ΔPdiastole, LV stiffness) were improved in RAGE-Ab-treated db/db mice (Table 3 and Figure 4). Left ventricular systolic and diastolic volumes were increased in both groups of db/db mice compared with db/+ mice and were unaffected by the RAGE-Ab treatment (Figure 4). Left ventricular myocardial volumes were similar in all groups of mice (Table 3). The cardiac index was reduced in both groups of db/db mice compared with db/+ mice, but the reduction tended to be less in RAGE-Ab-treated db/db mice compared with placebo-treated db/db mice (P = 0.06)(Figure 4). No differences in dp/dtmax and dp/dtmin were observed between the two strains of mice and types of treatment (Table 3).
The mRNA level of RAGE, Col1a1, and MHCα was increased in placebo-treated db/db mice compared with placebo-treated db/+ mice (Figure 5). The increased mRNA expression of RAGE and Col1a1 was prevented in db/db mice by RAGE-Ab treatment. The mRNA expression of MHCα further increased in db/db as well as db/+ mice with RAGE-Ab treatment compared with placebo treatment. At the protein level, the RAGE content was elevated in db/db mice compared with db/+ mice and not significantly influenced by the RAGE-Ab treatment (Figure 3). The mRNA level of TGF-β1 and VEGF-A was similar in db/db and db/+ mice. However, in both strains, RAGE-Ab treatment reduced the expression (Figure 5). The mRNA expression of BNP, eNOS, VCAM-1, PLB, SERCA2a2, ANP, CTGF, MHCβ, and Edn1 were unaffected in RAGE-Ab-treated db/db mice compared with placebo-treated db/db mice.
The present study demonstrates that the blockage of RAGE at the onset of diabetes can prevent the progressive development of cardiac dysfunction in type 2 diabetic mice. The effect is achieved without any simultaneous modification of metabolic status or body weight.
The db/db mouse as a model of diabetic cardiomyopathy
We chose the db/db mouse model of type 2 diabetes because it presents with robust systolic and diastolic dysfunction., Our study validates the cardiac phenotype and further characterizes an in vivo age-dependent deterioration in cardiac performance. In type 2 diabetic patients, systolic and diastolic dysfunction precedes the development of clinical heart failure.– Because the diabetic heart may be characterized by a complex mixture of systolic and diastolic dysfunctions as well as remodelling, we used the conductance catheter technique and magnetic resonance imaging to obtain a detailed evaluation of cardiac function in vivo. The methods revealed LV systolic and diastolic myocardial dysfunction from an early stage of diabetes, evaluated by load-independent indices of contractility. Furthermore, the reduced systolic function was characterized by impaired contractility in the long axis rather than in the transverse plane, which is similar to type 2 diabetic patients.
Regulatory function of the receptor for advanced glycation end products
Increased expression of RAGE at the mRNA level without simultaneous changes in RAGE protein content has been observed in the hearts of rats with STZ-induced diabetes; however, data are not consistent since increased RAGE at protein level after 32 weeks has been demonstrated in diabetic hearts without any change in RAGE mRNA expression., We found that RAGE mRNA was uniformly expressed at an increased level through all phases of diabetes. However, the expression of RAGE protein was significantly decreased in advanced diabetes (week 16) compared to 8 and 12 weeks, suggesting that the expression of RAGE may be influenced by mechanisms other than transcriptional regulation, such as changed translation or degradation of RAGE. A similar pattern has been described in aortic smooth muscle cells of diabetic rats. These results do not add support to the hypothesis that the activation of RAGE initiates a positive feedback mechanism by generation of additional RAGE ligands and upregulation of RAGE.,
Effect of the receptor for advanced glycation end products blockage on cardiac function
We evaluated the heart using pre- and after-load-independent indices of LV function and found that the blockage of RAGE in diabetic mice from the age of 7 to 12 weeks prevented the development of cardiac dysfunction. dp/dt, which is a complex function influenced by pre- and after-load, was unaffected. Consequently, blockage of RAGE must improve cardiac function by a direct action in the heart.
The function of the heart is dependent on the biophysical properties of the myocardium. As a result, AGE cross-linking has specific consequences for cardiac function. In accordance with this, accumulation of collagen and alterations in myocardial collagen AGEs in the diabetic heart have been demonstrated to decrease myocardial compliance., We found that RAGE blockage normalized LV diastolic chamber stiffness. In addition, the mRNA expression of Col1a1 was normalized in RAGE-blocked diabetic mice. These data indicate that RAGE activation mediates increased collagen AGE formation, which leads to increased stiffness of the heart. However, in the RAGE blocking study, the LV myocardial volumes were equal in all groups, suggesting that generalized hypertrophy was not present in the diabetic mice or regulated by RAGE.
The indices of LV myocardial contractility were improved in RAGE-blocked diabetic mice, indicating that RAGE is involved in mechanisms related to the contractile properties of the myocytes. We found no evidence that this was caused by alterations in cytosolic calcium regulation, as the expression of SERCA2a and PLB was not affected in diabetic mice or after RAGE blockage. However, RAGE blockage modulated the expression at mRNA level of proteins responsible for the contractility of the heart muscles. MHCα was upregulated in RAGE-blocked diabetic mice while the expression of the foetal myosin isoform MHCβ was unaffected. The diabetic heart is characterized by re-activation of a foetal gene profile which includes a switch from expression of the adult isoform (MHCα) to the foetal isoform MHCβ.– The receptor for advanced glycation end products therefore appears to be an important regulator of the expression of myosin isoforms in the diabetic heart.
Left ventricular volumes were not affected by blockage of RAGE during our intervention period. Our data do not allow evaluation of an extended intervention against the generalized dilatation that occurred at a later stage in the diabetic mice.
Metabolic flexibility for substrate use, such as the reciprocal regulation of glucose and FFA utilization, is attenuated in type 2 diabetes. Preference, shifted almost exclusively towards β-oxidation of FFAs in diabetic hearts, has been suggested to be an important pathogenetic process of DCM., Although experimental animal studies have provided evidence for impaired contractility by increased myocardial FFA uptake and accumulation,,, recent clinical studies with therapies targeting specific metabolic modulation of myocardial substrate utilization towards enhanced glucose utilization have not yielded a similar recovery or improvement in morbidity and mortality., We did not evaluate the in vivo myocardial substrate utilization; instead we looked exclusively at levels of circulating substrates. In association with the deterioration in cardiac function in diabetic mice, we found an elevated plasma concentration of glucose and a trend towards elevated total triglycerides, while circulating free fatty acids were not significantly affected. The most important finding, however, was that the improvement in cardiac function after RAGE blockage in the diabetic mice was not associated with changes in circulating substrates. Consequently, RAGE blockage improves cardiac function without concomitant modulation of myocardial substrate supply.
Other effects of blocking the receptor for advanced glycation end products
We did not identify altered mRNA expression of the pro-fibrogenic factors in diabetic mice. However, in RAGE-blocked diabetic and non-diabetic mice, TGF-β1 was downregulated, indicating that a fibrogenic stimulus can be mediated through RAGE. A similar response was observed for VEGF-A. Vascular endothelial growth factor A has been demonstrated to function as a hypoxia-inducible factor promoting angiogenesis and compensatory development of collateral vessels. Consequently, RAGE blockage may prevent a possible beneficial effect of VEGF-A in diabetic patients with chronic ischaemic heart disease. The expression of genes related to endothelial function was affected towards increased vasoconstriction in the diabetic animals, but the expression was not changed by blockage of RAGE indicating that RAGE is not involved in the disturbed balance of vasoconstriction in the diabetic heart. However, in isolated coronary vessels from db/db mice, AGE–RAGE signalling has been shown to induce endothelial dysfunction resulting in reduced potential for dilation.
We did not find any indication that RAGE was involved in the regulation of ANP or BNP.
Tissue samples from the LV used for qPCR and immunoblotting contained a mixture of cell types including cardiomyocytes, endothelial cells, and fibroblasts, which prevented analysis of cell-specific abnormalities.
We did not provide confirmatory evidence of collagen accumulation or alterations in myocardial collagen AGEs in diabetic hearts, but these are previously well-characterized and in accordance with the upregulation of the structural proteins that we have observed.
Age-dependent changes in cardiac preference for fatty acid and carbohydrate oxidation associated with reduced cardiac function have been demonstrated in db/db mice. We did not quantify myocardial substrate utilization and were only able to conclude that the myocardial substrate supply did not differ between groups. Consequently, the effect of blocking RAGE may involve altered cardiac metabolism.
The specificity of the RAGE blocking activity of the monoclonal RAGE-Ab was not evaluated in this study, but has been tested previously in cell cultures. In addition, blockage of the receptor activity has been achieved with antibody dosages equivalent to ours in an identical mouse-model.
In conclusion, the present data support the hypothesis that RAGE is an important pathogenetic factor in the development of DCM in type 2 diabetes. Further studies are required to clarify the influence of the RAGE-pathway on structural and functional components and regulation of intracrine growth factors in the myocardium. Targeting RAGE activation may be a potential new therapeutic approach for the treatment of DCM.
This study was supported by the Danish Heart Foundation, the Danish Diabetes Association, the Danish Medical Research Foundation, the Eva and Henry Fraenkels Memorial Foundation, the Novo-Nordisk Foundation, the Leducq Foundation, and the Clinical Institute at the University of Aarhus.