Background
Cardiac hypertrophy is the adaptive enlargement of the myocardium in response to physical or neurohormonal stress. Type 2 diabetes is associated with a cardiac syndrome called diabetic cardiomyopathy, which is characterized by pathological hypertrophy, contractile dysfunction[
1,
2], and an intractable reliance on fatty acid oxidation[
3‐
5]. By contrast, chronic endurance exercise training causes physiological hypertrophy that improves contractile mechanics[
6,
7] and myocardial metabolism[
8,
9]. Exercise benefits the type 2 diabetic heart[
10‐
14], but the underlying mechanisms by which exercise and diabetes control cardiac hypertrophy are not well understood.
In non-diabetic hearts, activation of fetal genes is a protective mechanism[
15‐
17] that accompanies pathological hypertrophy[
17‐
21]. These fetal genes include fetal cytoskeletal proteins (skeletal α-actin, β-myosin heavy chain) and the atrial and brain natriuretic peptides (ANP and BNP)[
22‐
24]. Importantly, exercise and diabetes moderate these genes differently. Exercise increases adult cardiac α-actin[
25], but does not change fetal gene expression[
26], whereas type 2 diabetes actually reduces circulating natriuretic peptides[
27,
28], and blocks the activation of fetal genes by hypertrophic stimuli in vitro[
29]. This suggests that fetal gene regulation in diabetic hearts is different from that of exercised hearts and non-diabetic hearts, and therefore, may underlie the hypertrophic response to these conditions.
A potential mechanism for the effects of diabetes and exercise on fetal genes is through the post-translational modification of transcription factors by O-linked β-N-acetylglucosamine (O-GlcNAc). O-GlcNAc is a glucose derivative that post-translationally modifies serine/threonine residues[
30,
31]. O-GlcNAcylation modifies signal transduction in a manner analogous to phosphorylation; O-GlcNAc transferase (OGT) and O-GlcNAcase (OGA) add and remove the O-GlcNAc moiety, respectively. We have recently shown that O-GlcNAc modifies repressor element-1 silencing transcription factor (REST)[
32], a transcription factor that represses fetal genes by recruiting the corepressors mammalian switch-independent 3A (mSin3A) and histone deacetylases (HDACs) 1 and 2[
33]. The HDAC enzymes deacetylate histone tails, thus condensing euchromatin and silencing gene expression; HDAC1 and HDAC2 specifically mediate fetal gene regulation by REST and mSin3A, and have been repeatedly linked to hypertrophic growth of the heart[
34].
mSin3A and HDAC1 are O-GlcNAcylated in HepG2 liver carcinoma cells, and are recruited to gene loci by the OGT enzyme[
35], which also O-GlcNAcylates itself[
30]. The activity of OGT is regulated by cellular concentrations of UDP-GlcNAc substrate[
36,
37], which is increased in the diabetic heart[
38]. Indeed, total O-GlcNAc and protein O-GlcNAcylation are elevated in the diabetic heart[
4,
39], but the functional implications of this finding are unclear as elevated cardiac O-GlcNAc is implicated in heart failure and cardiac dysfunction[
39,
40], but also cardioprotection[
41,
42]. Nevertheless, two previous studies showed that total protein O-GlcNAcylation and O-GlcNAcylation of the SP1 transcription factor are lowered by swimming exercise in both non-diabetic and streptozotocin-induced type 1 diabetic hearts[
43,
44], and OGA overexpression directly ameliorates the cardiovascular complications of type 2 diabetes[
38].
We therefore hypothesized that a reduction in O-GlcNAc may be a mechanism by which exercise benefits the type 2 diabetic mouse heart. However, since these interactions have not been studied in the heart, the secondary purpose of this study was to characterize the effects of diabetes and moderate exercise on the mSin3A/HDAC1/2 complex. We used 4 weeks of moderate treadmill exercise training to investigate the early signaling mechanisms in the hypertrophic process. We show that exercise increases total protein O-GlcNAcylation in the type 2 diabetic db
+
/db
+
mouse heart, and that exercise and diabetes have reciprocal effects on the association of HDAC1 and HDAC2 with fetal gene-regulating transcription factors.
Methods
Animal care and facilities
The procedures in this study followed the guidelines of the Washington State University Institutional Animal Care and Use Committee and the Guide for the Care and Use of Laboratory Animals published by the National Institutes of Health (NIH publication no. 85–23, revised 1996). 8-week-old type 2 diabetic mice (B6.BKS(D)-Lepr
db
/J, db
+
/db
+
(db) and age-matched C57BL/6J non-diabetic lean db
+
/? background strain controls (C57) were purchased from Jackson Laboratories (Bar Harbor, ME). To control for activity, mice were singly housed without environmental enrichment in a climate-controlled vivarium on a 12:12 light:dark cycle. Mice consumed water and standard chow ad libitum, except for one overnight fast per week prior to blood glucose measurement.
Exercise training protocol
Mice ran on a 6-lane electric treadmill (Columbus Instruments, Columbus, OH) for 5 consecutive days a week with 2 days of rest. In week 0, all mice were acclimated to the treadmill by standing on the stationary belt for 10 min, then walking at 5 m/min for 20 min. Mice were then randomized to sedentary (n = 11) or exercise (n = 12) groups for 1 week, or sedentary (n = 16) or exercise (n = 15) groups for 4 weeks. Human patients with type 2 diabetes are commonly prescribed an exercise intensity of at least 40-60% of their aerobic capacity, but a higher intensity is recommended for maximum health benefits[
45]. Therefore, we exercised mice at 10 m/min, which corresponds to approximately 70% of maximal oxygen uptake (VO
2max) for the C57 strain[
46]. Mice ran at this speed at 0% grade for 10, 20, 30, or 40 min in weeks 1, 2, 3, and 4, respectively. O-GlcNAc is a highly dynamic stress response; therefore, to remove confounding effects of stress, we kept the treadmill covered, and used gentle tactile stimuli rather than the electroshock apparatus to keep mice running. Sedentary groups were handled identically and spent equal time in the same treadmill environment on a stationary belt.
Blood glucose and body weight measurements
Blood glucose and body weight were measured weekly after an overnight fast. Blood glucose was measured with a glucometer (ACCU-CHEK® Aviva, Roche Diagnostics, Indianopolis, IN) in a small sample of tail blood. Glucose readings that exceeded the accuracy limit of the calibrated meter (33.3 mmol/L) were imputed this value for statistical analysis.
Tissue harvesting and morphological measurements
Following an overnight fast, mice were anesthetized rapidly using an isoflurane vaporizer chamber with 2-4% isoflurane gas in 100% oxygen. Mice were immediately decapitated and blood glucose was measured in neck blood. Whole hearts were excised and the atria were removed. Ventricular tissue was immediately wet-weighed, cut into four equal tissue aliquots, snap-frozen in liquid nitrogen, and stored at −80°C. The left tibia was dissected from each animal and measured from the tibial plateau to the lateral malleolus.
Western blotting
Ventricular tissue was homogenized in Tissue Protein Extraction Reagent (Sigma-Aldrich, St. Louis, MO); 20 mM sodium fluoride; 1 mM sodium orthovanadate; 3% protease inhibitor cocktail (Sigma); and 0.02% PUGNAc, an OGA inhibitor (Sigma), to inhibit O-GlcNAc removal from proteins. Total protein was quantified with a modified Lowry assay (BioRad, Hercules, CA). Proteins were separated by SDS-PAGE and transferred onto PVDF membranes, which were probed overnight at 4°C with primary antibodies (anti-HDAC1 and -HDAC2, Cell Signaling, Beverly, MA; anti-mSin3A, -REST, -OGT, and -calsequestrin, Abcam, Cambridge, MA; anti-NCOAT/OGA, Santa Cruz Biotechnology, Santa Cruz, CA; anti-phospho-HDAC1 (Ser 421/423), Millipore, Billerica, MA), then probed with the appropriate secondary antibodies for 1 hour at room temperature (see Table
1 for antibody details). Chemiluminescent substrates (Thermo Fisher Scientific, Rockford, IL) were used to detect horseradish peroxidase activity on a ChemiDoc (BioRad). Protein levels were quantified on duplicate blots with standard densitometry using ImageJ software (National Institutes of Health, Bethesda, MD), and normalized to the loading control calsequestrin.
CTD 110.6 (anti-O-GlcNAc) | Gift from Mary-Ann Accavitti, University of Alabama at Birmingham | n/a |
RL2 (anti-O-GlcNAc) | Abcam, Cambridge, MA | 2739 |
anti-mSin3A | Abcam | 3479 |
anti-OGT | Abcam | 50271 |
anti-NCOAT/OGA | Santa Cruz Biotechnology, Santa Cruz, CA | sc-66612 |
anti-HDAC1 | Cell Signaling, Beverly, MA | 5356 |
anti-HDAC2 | Cell Signaling | 5113 |
anti-Calsequestrin | Abcam | 3516 |
anti-REST | Abcam | 21635 |
anti-phospho-HDAC1 (ser421/423) | Millipore, Billerica, MA | 07-1575 |
O-GlcNAc Western blotting
Ventricular lysates were separated with SDS-PAGE and transferred to nitrocellulose membranes. Membranes were blocked overnight at 4°C, probed with 1:5000 anti-O-GlcNAc antibody (CTD 110.6, generous gift of Mary-Ann Accavitti, University of Alabama at Birmingham), then probed with the appropriate secondary antibody. Horseradish peroxidase activity was detected on x-ray film with chemiluminescent substrate (Thermo Scientific) and quantified as described above for Western blotting.
Co-immunoprecipitation
Ventricular tissue was homogenized in Tissue Protein Extraction Reagent, 1% phosphatase inhibitor, 2% protease inhibitor (Sigma), and 0.02% PUGNAc. Lysates were assayed for total protein as described for Western blotting. Samples were diluted to equal protein concentrations and precleared over protein A/G agarose beads (Thermo Fisher Scientific) at 4°C for 4 hours. Precleared supernatants were then added to 25 ul of beads that had been incubated with primary antibody for 4 hours at 4°C. IP was performed overnight at 4°C; beads were then washed and eluted at 100°C for 5 min. The eluents were assayed for co-immunoprecipitated proteins using immunoblotting. The positive control was ventricular lysate; the negative control was ventricular lysate that was immunoprecipitated without antibody. Co-immunoprecipitated proteins were normalized to the level of captured target protein for analysis.
HDAC activity
HDAC enzyme activity was assayed with a colorimetric kit (Enzo Life Sciences, Farmingdale, NY) per the manufacturer’s instructions. The substrate for this assay is predominantly deacetylated by HDAC1/2 and sirtuin 1 (a class III HDAC), but not the class II HDACs. Optical density was read at 415 nm. Results are presented as fold changes from control absorbance.
RNA isolation and quantitative real-time PCR (qPCR)
RNA was isolated with an RNA isolation kit (Qiagen, Valencia, CA) and quantified on a NanoPhotometer™ spectrophotometer (Implen, Ontario, NY). Complementary DNA (cDNA) was generated using a cDNA synthesis kit (Thermo Fisher Scientific). qPCR was performed in triplicate with SYBR green fluorescence chemistry using a qPCR kit (Qiagen). The negative control contained water in place of cDNA template. Thermal cycling was performed on an iCycler iQTM Real Time PCR Detection System (Biorad) using the following cycle: 95°C for 10 min, and 40 cycles of 95°C for 30 sec and T
m for 10 sec. Primer specificity was confirmed by melting curve analysis. Amplification data were analyzed with the 2
-ΔΔCt method for normalization to the housekeeping gene GAPDH as previously described[
47]. See Table
2 for primer details.
Table 2
Real-time PCR primers
ANP | TTCCGGTACCGAAGATAACAGCCA | TGACACACCACAAGGGCTTAGGAT | 91 | 60 |
BNP | AGACAAGGGAGAACACGGCATCAT | ACAGAATCATCTGGGACAGCACCT | 85 | 60 |
HDAC1 | TTCCTGCGTTCTATTCGCCCAGAT | AACAAGCCATCAAACACCGGACAG | 98 | 60 |
HDAC2 | TACAACAGATCGCGTGATGACCGT | TCCCTTTCCAGCACCAATATCCCT | 94 | 62 |
α-skeletal actin | TTGTGCACCGCAAATGCTTCTAGG | GCAACCACAGCACGATTGTCGATT | 90 | 60 |
α-cardiac actin | TGTAGGTGATGAAGCCCAGAGCAA | TGGTGCCAGATCTTCTCCATGTCA | 105 | 60 |
β-myosin heavy chain | TGGCTGGTGAGGTCATTGACAGAA | TGGCTGGTGAGGTCATTGACAGAA | 104 | 60 |
GAPDH | TGTGATGGGTGTGAACCACGAGAA | CATGAGCCCTTCCACAATGCCAAA | 133 | Per plate |
Statistics
Longitudinal effects of genotype and exercise on body weight and blood glucose were analyzed with two-factor repeated measures ANOVA followed by Bonferroni post-hoc tests. Non-normal data were log transformed prior to analysis. Two-factor ANOVA with Bonferroni post-hoc tests were used to describe genotype and exercise effects on protein levels, HDAC activity, and gene expression. Cardiac hypertrophy, tibia length, and wet heart weight data were resistant to transformations of normality and were analyzed with Kruskal-Wallis analysis of variance followed by Dunn’s post-hoc test. Values are presented as mean ± SEM and significance was accepted at P < 0.05.
Discussion
While diabetes is a multifactorial cardiac insult, and diabetic cardiomyopathy is associated with multiple factors such as oxidative stress[
54], lipotoxicity and mitochondrial dysfunction[
55‐
57], and impaired calcium signaling[
13], O-GlcNAc is emerging as an important signalling mechanism in the development of diabetic cardiomyopathy. Total protein O-GlcNAcylation is chronically elevated in the type 1 and 2 diabetic heart[
4,
39], and reducing protein O-GlcNAcylation by adenoviral overexpression of OGA[
38] improves cardiac function. Similarly, lowering O-GlcNAc by intensive swim training[
43,
44] has been proposed as a mechanism by which exercise benefits the diabetic heart, and exercise lowers both the O-GlcNAc modification of the SP1 transcription factor and the OGT enzyme. O-GlcNAc directly mediates the expression of fetal genes in response to hypertrophic stimuli[
29], and O-GlcNAc modifies mSin3A and HDAC1[
35], which regulate cardiac hypertrophy[
20,
58]. Previously, we have shown that exercise lowers the O-GlcNAc modification of the OGT enzyme[
32], and others have shown that exercise lowers O-GlcNAcylation of the SP1 transcription factor[
43]. Moderate exercise improves cardiac structure and function in humans with type 2 diabetes[
59,
60]; we therefore tested the hypothesis that moderate exercise would reduce O-GlcNAc in the type 2 diabetic heart, and would be associated with changes in the O-GlcNAc modification and activity of the mSin3A/HDAC1/2 transcription factor complex, which regulates hypertrophic genes.
Surprisingly, and in contrast with the previous studies, we found that 4 weeks of moderate treadmill exercise increased total O-GlcNAc in type 2 diabetic
db mouse hearts. Also, while the previous studies showed that OGT was also reduced by exercise[
43,
44], we found that OGT and OGA expression was elevated in
db hearts and did not change with exercise. Such parallel regulation of OGT and OGA expression has been previously reported[
41], and may represent a compensatory relationship between these two opposing enzymes. The difference in our findings may be due to the use of type 2
db mice rather than streptozotocin-induced type 1 diabetic mice, and the use of moderate treadmill exercise rather than more intensive swimming exercise. However, other studies have shown that an upregulation of O-GlcNAc is essential in the cardiac stress response[
61,
62], is acutely cardioprotective[
63,
64], and is part of a constitutively active cardioprotection mechanism in the diabetic myocardium[
42]. Therefore, these data suggest that an increase in cardiac O-GlcNAc in the type 2 diabetic heart may be a beneficial effect of exercise.
In our study, mSin3A immunoprecipitation revealed that exercise increased the O-GlcNAc modification of mSin3A; however, this was not supported by reciprocal O-GlcNAc immunoprecipitation. It is possible that the large amount of protein captured in the O-GlcNAc immunoprecipitation masked the changes in mSin3A O-GlcNAcylation, which we observed in the more specific mSin3A immunoprecipitation. However, these data underscore the importance of verifying changes in O-GlcNAcylation of individual proteins with reciprocal assays, and suggest that moderate changes in protein O-GlcNAcylation – including those in the present study – should be interpreted cautiously and confirmed by additional studies.
Nevertheless, our data do suggest an alternate mechanism for the beneficial effect of exercise on the diabetic heart.
Db hearts showed lower protein levels of mSin3A, HDAC1, and HDAC2, and an increased association of mSin3A with REST, independent of exercise. Likewise, mRNA transcript levels of BNP and α-skeletal actin, which are typical markers of cardiac hypertrophy activated by HDAC1/2[
65] that are regulated via REST/mSin3A[
33], were significantly lower in
db hearts independent of exercise. The finding that blunted expression of fetal genes in diabetic hearts is not altered by exercise has been shown in previous studies[
27,
28]. Therefore, we suggest that the loss of HDAC1/2 and the increased association of the mSin3A corepressor with REST may underlie the blunted expression of fetal genes in the diabetic heart. Further, since the natriuretic peptides are both anti-hypertrophic and cardioprotective[
15,
17,
28], we suggest that this mechanism may be responsible for the increased vulnerability of the diabetic heart to stress and heart failure[
66,
67].
Although we did not measure the structural or hemodynamic effects of the exercise protocol in
db hearts, previous work has shown that the
db heart shows similar cardiomyopathy to humans with type 2 diabetes[
68,
69], which are improved by exercise[
13,
14]. We show additionally that even the low intensity of exercise used in this protocol was sufficient to elevate the expression of cardiac α-actin in C57 hearts (P = 0.050). Cardiac α-actin is a marker of cardiomyocyte differentiation and hypertrophy[
70], and is increased in physiologically hypertrophied hearts after chronic endurance exercise training[
25]. Additional transcriptional changes were observed in
db hearts, in which the exercise protocol significantly increased the association of mSin3A and OGT with HDAC1 and HDAC2, respectively. Therefore, although the exercise stimulus used in this study did not cause overt changes in cardiac mass, it induced transcriptional events consistent with the early stages of physiological cardiac remodelling.
Finally, these data show a potential interaction between HDAC1 and HDAC2 that has not previously been described in the heart. HDAC1 and HDAC2 regulate cardiac hypertrophy in a similar manner[
58], and HDAC1 deficiency induces HDAC2 expression in embryonic stem cells[
71]. In our study, the loss of HDAC1 protein preceded the loss of HDAC2 protein in
db hearts, and was similarly associated with an increase in HDAC2 gene expression in
db hearts. When HDAC2 deficiency was present at the 4 week time point, we observed an increase in the total activity of class I HDACs in
db hearts, which was verified by an increase in the phosphorylation status of HDAC1 at Ser421/423. Phosphorylation at these residues is specifically associated with HDAC1 activity[
52]. Therefore, these data suggest that the class I HDACs have compensatory effects on each other’s expression levels and activation by phosphorylation. Further, the reduction in HDAC2 protein levels in
db mouse hearts did not occur until 4 weeks, and was associated with overt cardiac hypertrophy. Thus, the loss of HDAC2 in the diabetic heart is associated with the progression of hypertrophy in the diabetic heart, and may be more specifically involved in hypertrophy than HDAC1.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
EJC performed the animal studies and laboratory experiments, and drafted the manuscript. SAM edited the manuscript and provided technical and intellectual guidance. Both authors read and approved the final manuscript.