Critical Role of Nuclear Calcium/Calmodulin-dependent Protein Kinase IIδB in Cardiomyocyte Survival in Cardiomyopathy*

Calcium/calmodulin-dependent protein kinase II (CaMKII) plays a central role in cardiac contractility and heart disease. However, the specific role of alternatively spliced variants of CaMKII in cardiac disease and apoptosis remains poorly explored. Here we report that the δB subunit of CaMKII (CaMKIIδB), which is the predominant nuclear isoform of calcium/calmodulin-dependent protein kinases in heart muscle, acts as an anti-apoptotic factor and is a novel target of the antineoplastic and cardiomyopathic drug doxorubicin (Dox (adriamycin)). Hearts of rats that develop cardiomyopathy following chronic treatment with Dox also show down-regulation of CaMKIIδB mRNA, which correlates with decreased cardiac function in vivo, reduced expression of sarcomeric proteins, and increased tissue damage associated with Dox cardiotoxicity. Overexpression of CaMKIIδB in primary cardiac cells inhibits Dox-mediated apoptosis and prevents the loss of the anti-apoptotic protein Bcl-2. Specific silencing of CaMKIIδB by small interfering RNA prevents the formation of organized sarcomeres and decreases the expression of Bcl-2, which all mimic the effect of Dox. CaMKIIδB is required for GATA-4-mediated co-activation and binding to the Bcl-2 promoter. These results reveal that CaMKIIδB plays an essential role in cardiomyocyte survival and provide a mechanism for the protective role of CaMKIIδB. These results suggest that selective targeting of CaMKII in the nuclear compartment might represent a strategy to regulate cardiac apoptosis and to reduce Dox-mediated cardiotoxicity.

cated in numerous cellular functions. The ␦ subunit of CaMKII predominates in the adult heart, and two isoforms generated by alternative splicing, ␦B and ␦C, are detected at the protein level in this organ (1)(2)(3)(4)(5). In contrast, the ␥ isoform is expressed at very low levels in heart muscle, whereas the ␣ and ␤ subunits are not detected at all (6,7). CaMKII␦ isoforms are highly homologous with the exception of a variable domain generated by alternative splicing (4,5). CaMKII␦B contains an 11-amino acid nuclear localization signal (NLS) not present in the ␦C, which directs the enzyme to the cell nucleus (8,9). The relative abundance of particular subunits dictates the subcellular localization of the enzyme (9).
CaMK signaling modulates gene expression in cardiac cells by increasing the activity of transcription factors such as the Mef2 family. CaMK regulates the activity of Mef2 members by controlling their interaction with class II histone deacetylase transcriptional repressors (HDACs; HDAC4 -7, -9, and -10) (for review see Refs. 10 -13). Interestingly, different CaMK isoforms phosphorylate different amino acid residues in class II HDACs. CaMKI and -IV phosphorylate two conserved serines located at the N terminus of these HDACs, Ser-246/467 in HDAC4, Ser-259/498 in HDAC5, and Ser-218/448 in HDAC9. Such phosphorylations lead to the dissociation of Mef2-HDAC complexes, binding to the chaperone protein 14-3-3 and subsequent nuclear export of HDACs leading to a relief of transcriptional repression (12,14). Recently, we and others have shown that the cardiac enzyme CaMKII␦B has characteristics distinct from CaMKI/IV. CaMKII␦B selectively transmits signals to HDAC4 and not to other class II HDACs, through phosphorylation of Ser-210, Ser-467, and Ser-632 (15,16).
CaMK signaling plays a significant role in cardiac disease (for review see Ref. 4). ␣-Adrenergic stimulation, endothelin-1, or leukemia inhibitory factor promote hypertrophic growth through activation of CaMK signaling in isolated cells. CaMKII inhibition in mice markedly inhibits cardiac hypertrophy and dysfunction after ␤-adrenergic stimulation or myocardial infarction (17). Increased CaMKII activity has been reported in several animal models of cardiac hypertrophy and heart failure. Decreased CaMKII activity and expression were observed in a number of animal models of myocardial infarction (18,19). Transgenic mice with high cardiac levels of CaMKII␦B or -␦C develop dilated cardiomyopathy (20,21). Recently, increased activity of both the ␦B and ␦C splice variants of CaMKII were reported in patients with end-stage idiopathic dilated cardiomyopathy and ischemic cardiomyopathy (22). Deletion of all CaMKII␦ isoforms in mouse heart decreases cardiac hypertrophy and remodeling induced by pressure overload (23). Despite clear evidence for a role of CaMKII signaling in cardiac diseases, the specific role and contribution of CaMKII␦ isoforms generated after alternative splicing still remain unclear.
Doxorubicin (Dox) (adriamycin)) is one of the most effective anti-cancer agents discovered so far. Despite its high efficacy in the treatment of many neoplastic diseases, chronic administration is limited because of severe side effects that lead to cardiomyopathy and congestive heart failure (for reviews see Refs. 24 -26). Dox cardiotoxicity is due in part to the down-regulation of contractile protein mRNAs in vivo and in primary cardiac cells (27). This effect is mediated by a loss of cardiac transcription factors such as Mef2C, NKX2.5 (28), and GATA-4 (29). Dox side effects are also due to the proteasome-mediated degradation of the co-activator p300 in primary cardiomyocytes (28), following activation of p38 mitogen-activated protein kinase (30). Cardiac apoptosis is a major factor in the development of the cardiomyopathy and heart failure induced by Dox (30 -32). There is some evidence that CaMK signaling plays a role in programmed cell death in the heart. Several studies have documented a pro-apoptotic role of CaMKII in cardiomyocyte apoptosis following ␤ 1 -adrenergic stimulation (33), ischemia-reperfusion injury (34), and UV light-induced DNA damage (35). Although two reports have demonstrated a pro-apoptotic role of the cytoplasmic ␦C isoform of CaMKII (36,37), the function of the major nuclear isoform, CaMKII␦B, in cardiac apoptosis remains unknown.
To further understand the molecular genetic mechanisms of Dox-mediated cardiotoxicity and apoptosis, we profiled gene expression in the heart of animals with degenerative cardiomyopathy after treatment with Dox. Here we report that chronic administration of Dox leads to the selective decrease of CaMKII␦B mRNA in vivo, which parallels the known side effects of Dox. In primary neonatal rat cardiomyocytes, siRNAmediated depletion of CaMKII␦B results in abnormal sarcomere organization and leads to a severe loss of the anti-apoptotic protein Bcl-2, suggesting a protective role of the kinase. CaMKII␦B exerts its protective effect by regulating GATA-4 binding to the Bcl-2 promoter. Importantly, forced expression of CaMKII␦B inhibits GATA-4 and Bcl-2 down-regulation by Dox. Collectively, our results demonstrate that nuclear CaMKII␦B is a novel major target of the cardiotonic agent Dox and that a persistent level of CaMKII␦B is required for cardiomyocyte integrity and survival.

MATERIALS AND METHODS
In Vivo Chronic Model of Doxorubicin-induced Cardiomyopathy-The chronic rat model of Dox-induced cardiomyopathy was generated as described before (38). For the microarray analysis and echocardiography, a total of five female Sprague-Dawley rats were used. Three rats received weekly intravenous injections of Dox in equally divided doses for a total cumulative dose of 17-20 mg/kg over 7-10 weeks. Two rats served as controls and received weekly saline injections. Trans-thoracic echocardiography was performed in all animals at the beginning of the treatment (base line) at 5 weeks and 2 weeks after discontinuation of Dox administration. After the animals were anesthetized with a mixture of ketamine and xylazine, a 7.5-MHz standard pediatric transducer was connected to an echocardiographic computer console (Hewlett Packard 1500, Andover, MA). Left ventricular end-diastolic and end-systolic diameters were measured using two-dimensional guided M-mode imaging. Fractional shortening was calculated from the mean value of three measurements and was used to evaluate cardiac function of the animals during the treatment. To validate the microarray and echocardiography, we added more animals to our study and performed transthoracic echocardiography in rats injected with saline (n ϭ 7) or Dox (n ϭ 8) as described before. Experiments were conducted in accordance with the institutional guidelines for the use and care of laboratory animals, which conforms to the Guide for Care and Use of Laboratory Animals (National Institutes of Health Publication 85-23).
Electron Microscopy-Electron microscopy of the left ventricle was performed in all five rats to confirm the degree of cardiac damage induced by Dox. Rat hearts were collected at the end of Dox treatment and were fixed in 0.1 M sodium phosphate buffer containing 2.5% glutaraldehyde in 0.1 M sodium cacodylate. After washing, the samples were post-fixed in 1% osmium tetroxide, 0.1 M cacodylate, prestained in 1% uranyl acetate, dehydrated in a graded ethanol series, and then embedded in 100% epoxy resin. After sectioning, the samples were mounted on parlodian-coated grids, stained with lead citrate, and examined with a Zeiss TEM electron microscope (Microscopy Core Facility of the Doheny Eye Institute, University of Southern California). The cardiomyopathy was scored on a 1-5 scale.
Expression Profiling and Statistical Analysis-Gene expression profiling was performed as described previously (39). Briefly, total RNA was extracted from each heart tissue using TRIzol reagent (Invitrogen). Biotin-labeled cRNA was prepared and used to hybridize GeneChip Rat U34 array set (Affymetrix). cRNA from each heart was hybridized to one array. Analysis of gene expression was done as described previously using Dchip software (39). Replicates for each condition (duplicates for controls and triplicates for Dox-treated rats) were used to estimate mean expression levels along with associated standard errors. Genes with a fold ratio above ϩ2 for up-regulated genes or below 2 for down-regulated genes were selected. The p value resulting from a t test for a change in the gene expression level was less than 0.05.
Real time PCRs were performed after RT-PCR (RETR-Oscript TM , Ambion) with various dilutions of cDNA template using IQ SYBR Green and an Opticon 2 reader (Bio-Rad). Primer sequences for both quantitative RT-PCR and real time PCR are as follows: ␣-actin sense 5Ј-TCTCTTCCAGCCC-TCTTTCA-3Ј and antisense 5Ј-CCCCCAATCCAGACAGA-GTA-3Ј; MLC2-a sense 5Ј-GCTGCATTGACCAGAAC-AGA-3Ј and antisense 5Ј-GCTGCTTGAACTCCTCCTTG-3Ј; MLC2-v sense 5Ј-AAAGAGGCTCCAGGTCCAAT-3Ј and antisense 5Ј-AAAAGCTGCGAACATCTGGT-3Ј; CaMKII␦B sense 5Ј-AGGAAGTCCAGTTCGAGTGTTC-3Ј and antisense 5Ј-CAGGATGATAGTGTGGATTG-3Ј; CaMKII␦C sense 5Ј-CCG-GATGGGGTAAAGGAG and antisense 5Ј-CAGGATGATAG-TGTGGATTG-3Ј. Each primer pair produced a single band after amplification, and melting curves were clean. Standard curves of reverse-transcribed RNA from heart tissue were generated with 18 S primer pairs and analyzed in triplicate by serial dilutions. Calibration curves were linear over the complete range for all primers (R 2 Ͼ 0.989). Concentrations were obtained by comparing Ct values of the samples to the standard curves using Opticon Monitor 2 software. Results are expressed as relative expression compared with our standard and are corrected for 18 S or GAPDH.
Isolation and Transfection of Primary Neonatal Rat Cardiomyocytes-Neonatal rat cardiomyocytes were prepared as described previously (30). Transfections were carried out by calcium phosphate precipitation or using Lipofectamine 2000 (Invitrogen) according to the manufacturer's recommendations. siRNA luciferase experiments were as described (16).
Western Blot Analysis-Nuclear and whole cell extracts were prepared from primary neonatal rat cardiomyocytes as described previously (30). Briefly, extracts were resuspended in a buffer containing 50 mM Tris, pH 7.6, 150 or 500 mM NaCl, 1 mM EDTA, 1% Triton X-100, 1 mM phenylmethylsulfonyl fluoride, a phosphatase and a protease inhibitor mixture (Sigma). After centrifugation 30 g of the supernatant was separated by SDS-PAGE on 4 -12% Tris-glycine gradient gels. Western blotting was described previously (16), Detection was performed with chemifluorescence (ECF reagent, Amersham Biosciences) and a Storm scanner.
siRNA Experiments-To establish the specificity of siCaMKII␦B, recombinant adenoviruses expressing CaMKII␦B (Ad-CaMKII␦B) and CaMKII␦C (Ad-CaMKII␦C) carrying HA tags were produced and amplified as described previously (16). Primary cardiomyocytes were transfected with increasing concentrations of siCaMKII␦B or siControl using Lipofectamine. 24 h later, the cells were infected with Ad-CaMKII␦B or Ad-CaMKII␦C, and exogenous CaMKII␦B and CaMKII␦C was measured by Western blot using an HA antibody. For all other experiments, siCaMKII␦B or siControl was transfected in primary cardiomyocytes, and endogenous proteins were measured by Western blot 72 h later using the indicated specific primary antibodies. Apoptosis was measured in cardiomyocytes transfected with siCaMKII␦B or siControl and treated with Dox for 6, 12, and 24 h using the Cell Death Detection ELISA PLUS (Roche Applied Science).
Immunofluorescence-Immunofluorescence was performed as described previously (16,30). For siRNA experiments, primary cardiomyocytes were transfected with 300 pmol of either siCtrl or siCaMKII␦B oligonucleotides 72 h before fixing or were maintained in media supplemented with Dox for the indicated times. For overexpression of CaMKII␦B, primary cardiomyocytes were infected with a control Ad-GFP or Ad-CaMKII␦B. 12 h later, half of the cells were maintained in normal media or in media supplemented with Dox for the indicated times. Sarcomeric structures were analyzed by indirect immunofluorescence using an antibody specific for ␣-actinin. Images were visualized by confocal microscopy.
Apoptosis Assay-Apoptosis assays were carried out using the Cell Death ELISA PLUS kit from Roche Applied Science according to the manufacturer's protocol. Photometric analysis was carried out using Victor 3 (GE Healthcare).
Electrophoretic Mobility Shift Assay-Electrophoretic mobility shift assays (EMSAs) and probes were as described (41), with the exception that 10 g of nuclear extract prepared from primary cardiomyocytes was used in each reaction (30).

In Vivo Chronic Model of Dox-induced Cardiomyopathy-To
identify Dox-sensitive genes in heart muscle in vivo, we analyzed gene expression profiling in a chronic rat model of Doxinduced cardiomyopathy (38). Rats treated with Dox (animals 3-5) received weekly intravenous injections over 7-10 weeks, and control rats (animals 1 and 2) received weekly saline injections. To monitor the effect of Dox on cardiac contractility, transthoracic echocardiography was performed in all animals at the beginning of the treatment (base line) and at 5 and 2 weeks after the last dose (Fig. 1A). Fractional shortening (FS) was calculated as a reliable indicator of cardiac function. At 5 weeks, FS was similar to base line in control animals (supplemental Table S1). However, in the three animals that received Dox, FS decreased incrementally with time to 54, 75, and 71% of base line in animals 3-5, respectively. At the end of the treatment, FS decreased to 48 and 33% of base line for animals 3 and 4. Final functional parameters could not be measured in animal 5 which died of heart failure at week 7. Heart tissue from this animal was recovered immediately after death and was kept frozen until further analysis. Together, our results indicate a significant deterioration of cardiac function in animals 3-5.
To evaluate the degree of cardiac damage exerted by Dox, heart sections from each animal were examined by electron microscopy ( Fig. 1B and supplemental Table S2). The analysis revealed a normal ultrastructure of the two control animals. Animal 3 displayed a mostly intact myocardium with only slight sarcoplasmic edema, whereas animal 4 revealed sarcoplasmic and interstitial swelling characteristic of Dox cardiotoxicity. Animal 5 showed, in addition to intense sarcoplasmic and interstitial swelling, severe mitochondrial damage suggesting a pronounced cardiomyopathy (Fig. 1B and supplemental Table  S2). Together, these data indicate different degrees of cardiac damage in the three animals chronically treated with Dox, animal 3 displaying almost no signs of cardiac damage, and animals 4 and 5 revealing a more profound cardiomyopathy.
Genes Down-regulated in the Heart of Rats Treated with Dox in Vivo-We next compared changes in gene expression between the hearts of control animals and the hearts of Doxtreated animals using Affymetrix Gene Chips (rat gene array U34A). cRNA from each heart was hybridized to two arrays twice, so that a total of four chips were used per animal. As expected, chronic exposure to Dox caused a decreased expression of structural proteins (27) such as myosin light chain-2a (MLC2-a), atrial myosin light chain 1, SM22, and ␣-actin. We also observed a significant decrease of mitochondrial genes such as cytochrome c and acetyl-CoA acetyltransferase ( Table 1). The results of the gene chip analysis were then validated by quantitative RT-PCR analysis performed from the same RNA samples. Dox-sensitive mRNAs such as cardiac ␣-actin (CAA), cardiac troponin I, MLC2-a, and Mef2C were slightly decreased in the heart of animal 3 and showed a stronger decrease in animals 4 and 5, which displayed a severe cardiomyopathy (see supplemental Fig. S1). These results establish different degrees of cardiac damage in the three animals chronically treated with Dox, animal 3 displaying a mostly preserved myocardium and animals 4 and 5 revealing a pronounced cardiomyopathy. A, M mode echocardiography of control heart #1 and Dox-treated heart #4 is shown. The inward excursion of the posterior wall (arrowhead) and anteroseptal wall (arrow) is brisk in the control rat (animal 1). The motion of the posterior wall is blunted, and the anteroseptal wall is nearly motionless in the Dox-treated rat, suggesting a severe reduction of global cardiac function. B, electron microscopy of control (heart 1) and Dox-treated rat hearts (hearts 4 and 5) is shown.

TABLE 1 Genes repressed in rat hearts after chronic treatment with Dox
Affymetrix Gene Chips were used to analyze changes in gene expression of the heart of two control rats and three rats treated with Dox. Statistical analysis was done using Dchip software, and changes with a fold ratio Ͼ2.0 were selected. S.D. mean standard deviation.

Genes repressed by Dox
GenBank TM accession no. CaMKII␦B Is a Target of Dox in Vivo and in Primary Neonatal Rat Cardiomyocytes-Of particular interest, transcripts for the ␦B subunit of CaMKII (CaMKII␦B) were among the genes down-regulated by Dox (Table 1 and Fig. 2A). This CaMKII isoform is the predominant nuclear isoform of CaMK in the heart (4). The down-regulation of CaMKII␦B was confirmed by quantitative RT-PCR analysis (Fig. 2B). Next, we analyzed CaMKII␦B mRNA expression in the heart of a larger number of animals treated with Dox. Animals exposed to Dox chronically developed cardiomyopathy as evidenced by a decrease in cardiac performances (Fig. 2C). Relative CaMKII␦B transcript levels decreased from 290 Ϯ 13 in control hearts (n ϭ 7) to 180 Ϯ 9 in Dox-treated hearts (n ϭ 8) (Fig. 2D). The adult heart is enriched in CaMKII␦B and also in CaMKII␦C. Other CaMK isoforms or other CaMKII␦ splice variants are also expressed in the normal adult myocardium but at very low levels (levels 1-3). CaMKII␦B and -␦C are generated by alternative splicing of the same gene and differ in sequence by an NLS present only in the ␦B isoform (8,9). To determine whether the effect of Dox is specific to the CaMKII␦B subunit, we measured CaMKII␦C transcript levels by real time PCR in the hearts of rats treated with Dox. CaMKII␦C mRNA slightly increased in the heart of Dox-treated animals, although the difference did not reach statistical significance (Fig. 2E). Next, we attempted to measure CaMKII␦ protein isoforms in saline and Dox-treated hearts using an anti-CaMKII␦ specific antibody. We detected several CaMKII␦ isoforms in adult heart tissue, but were unfortunately unable to resolve the ␦B and ␦C splice variants by SDS-PAGE.
Next, we sought to determine whether the depletion of CaMKII␦B we observed in our in vivo model could be recapitulated in primary neonatal rat cardiomyocytes to allow for a more detailed analysis. Indeed, CaMKII␦B transcripts were severely decreased in cardiomyocytes treated with Dox for 48 h (Fig. 3A). The down-regulation of CaMKII␦B paralleled the decreased expression of mRNAs long known to be Dox-sensitive such as cardiac troponin I and the transcription factors Mef2C and NKX2.5. p300 mRNA expression was not affected by Dox, as we reported previously (28). Next we measured CaMKII␦B mRNA expression over time in Dox-treated cells using real time PCR. We found that CaMKII␦B expression rapidly decreases in cardiomyocytes treated with Dox (Fig. 3B). Expression was ϳ50% that in untreated cells after 2 h of Dox treatment. Down-regulation of ␤-myosin heavy chain and CAA mRNAs was also detectable as early as 2 h but was not as striking as the reduction in CaMKII␦B expression. Cardiac muscle creatine kinase expression was reduced at a much slower rate than myosin heavy chain, CAA, or CaMKII␦B. CaMKII␦B protein levels were also measured in control and Dox-treated cardiac nuclei by Western blot analysis using a specific anti-CaMKII␦ antibody. One major band migrating at ϳ55 kDa, which corresponds to the molecular weight of CaMKII␦B, was detected in Western blot and was decreased after 12 and 24 h of Dox treatment (Fig. 3C). To establish the specificity of this band, primary cardiomyocytes were transfected with siCtr or siCaMKII␦B, and Western blot analysis was performed using the anti-CaMKII␦-specific antibody. The band migrating at 55 kDa was significantly decreased in siCaMKII␦B-transfected cells, whereas other bands remained similar (Fig. 3D). This result demonstrates that the protein that migrates at 55 kDa is indeed CaMKII␦B. Together, these results identify nuclear CaMKII␦B as a new, and early, Dox gene target in primary neonatal rat cardiomyocytes and in vivo. This led us to hypothesize that CaMKII␦B may be a key mediator of Dox cardiotoxicity.  1 and 2) and Dox-treated rats (lanes 3-5) is shown. C, left ventricular dimensions were measured by transthoracic echocardiography at base line and at 9 -12 weeks; D, real time PCR for CaMKII␦B is shown; E, CaMKII␦C in saline-treated rat hearts (3 from our first study and 4 from our second study, total control (Ctr) number n ϭ 7) and in Dox-treated hearts is shown (2 from our first study and 6 from our second study, total Dox number n ϭ 8). DD2-DD1, difference between diastolic diameter at the end of treatment and at base line; SD2-SD1, difference between systolic diameter at the end of treatment and at base line; FS, fractional shortening as percentage of base line. Quantitative RT-PCR and real time PCR values are corrected for 18 S. Asterisks indicate that differences were statistically significant by Student's t test (*, p Ͻ 0.05; ***, p Ͻ 0.001). SEPTEMBER 11, 2009 • VOLUME 284 • NUMBER 37

JOURNAL OF BIOLOGICAL CHEMISTRY 24861
CaMKII␦B Is Required for the Structural Integrity of Cardiomyocytes-The cardiomyopathy induced by Dox is mediated by an alteration of gene expression of transcripts implicated in the structural integrity of cardiomyocytes such as sarcomeric proteins (27). To determine whether CaMKII␦B plays a role in Dox-mediated cardiomyopathy, we tested whether loss of CaMKII␦B activity could by itself engender the cellular hallmarks of Dox cardiotoxicity. For this, we reduced CaMKII␦B levels in primary cardiomyocytes using siRNA oligonucleotides against CaMKII␦B (siCaMKII␦B) or control siRNA (siControl) and analyzed the effect on myofibril organization. siCaMKII␦B was designed within the NLS of CaMKII␦B (Fig. 4A) (16) to specifically reduce expression of the ␦B isoform and not affect expression of other isoforms expressed in the adult myocardium such as CaMKII␦C. The specificity of siCaMKII␦B toward CaMKII␦B was established by testing the effect of siCaMKII␦B on exogenous CaMKII␦B and CaMKII␦C delivered as recombinant adenoviruses to primary cardiomyocytes. siCaMKII␦B only silenced CaMKII␦B but not CaMKII␦C, whereas siControl had no effect at all, thus establishing the specificity of siCaMKII␦B for CaMKII␦B (Fig. 4B).
The efficiency of siCaMKII␦B was demonstrated by quantitative RT-PCR and Western blot analysis performed in parallel experiments in cardiomyocytes transfected with siCaMKII␦B or siCtr. These showed down-regulation of endogenous CaMKII␦B by siCaMKII␦B by at least 80% consistently (data not shown). The majority of cardiomyocytes transfected with siControl (69.5 Ϯ 0.5%) displayed normal well organized sarcomeres as demonstrated by ␣-actinin staining (Fig. 4, C and  D). In contrast, less than 30% of cells transfected with siCaMKII␦B showed a normal ␣-actinin staining (29.25 Ϯ 3.75%), and these were also smaller in size (Fig. 4, C and D). In addition, the percentage of nonviable cells, which displayed a total loss of ␣-actinin staining, was significantly higher in cells expressing reduced levels of CaMKII␦B (31.5 Ϯ 1.5%) compared with cells transfected with a control siRNA (19.25 Ϯ 0.75%) (Fig. 4E). The dramatic reduction in cell size and ␣-actinin staining observed in siCaMKII␦B transfected cardiomyocytes was comparable with the decrease observed in cardiomyocytes treated with Dox for 36 h (Fig. 4F). These results demonstrate that cardiomyocytes with reduced CaMKII␦B expression fail to develop fully organized sarcomeres and show intense signs of cardiotoxicity. Together, these data suggest that a decrease in CaMKII␦B expression recapitulates, to some extent, cellular abnormalities observed in Dox-induced cardiomyopathy.
CaMKII␦B Greatly Attenuates the Apoptotic Effect of Dox on Cardiac Cells-Cardiac apoptosis is a major contributor to the development of the cardiomyopathy induced by Dox (30 -32, 42). If a decrease in CaMKII␦B activity is the mediator of Dox cardiotoxicity, then preservation of CaMKII␦B levels should prevent or ameliorate the effect of Dox on apoptosis. Accordingly, we tested whether forced overexpression of exogenously delivered CaMKII␦B can attenuate Dox-mediated apoptosis. We evaluated the ability of CaMKII␦B to reduce apoptosis in Dox-treated cells by measuring the enrichment of monoand oligonucleosomes leaking to the cytoplasm of apoptotic cells using the Cell Death Detection ELISA Plus (Roche Applied Science). CaMKII␦B was overexpressed in primary neonatal rat cardiomyocytes using a recombinant adenovirus (Fig. 5, A and B). We used viral titers so that ϳ90% of the cells expressed exogenous CaMKII␦B or, as a control, GFP. Overexpression of CaMKII␦B in cardiomyocytes treated with Dox for 24 h significantly attenuated Dox-induced apoptosis (Fig. 5C) suggesting that CaMKII␦B has an anti-apoptotic role in primary cardiomyocytes.
To identify the specific target(s) of CaMKII␦B in the apoptotic pathways, we analyzed changes in the expression of genes belonging to the mitochondrial pathway, which is affected by Dox (29,41,43). For this, we compared levels of the anti-apoptotic protein Bcl-2 in cardiomyocytes treated with Dox and infected with either a control Ad-GFP or Ad-CaMKII␦B. Western blot analysis showed that Dox treatment reduced Bcl-2 protein expression and that overexpression of CaMKII␦B significantly attenuated the down-regulation of Bcl-2 protein by Dox. Expression of the housekeeping gene ␤-actin remained similar (Fig. 5D). Next we measured levels of the pro-apoptotic protein Bax to estimate the Bcl-2/Bax ratio, which is an established indicator of Dox-induced apoptosis in cardiomyocytes (44 -46). Whereas Bax levels remained unchanged with Dox treatment (Fig. 5D), the Bcl-2/Bax ratio significantly decreased in cardiomyocytes after Dox treatment, indicating an increase in apoptosis. Importantly, forced expression of CaMKII␦B prevented this effect (Fig. 5F). To show that this effect is specific to nuclear CaMKII␦, we also infected cardiomyocytes with Ad-CaMKII␦C to increase cytoplasmic CaMKII␦. CaMKII␦C was expressed at a similar level to CaMKII␦B (Fig. 5A) but was unable to prevent a reduction in the Bcl-2/Bax ratio in Doxtreated cells (Fig. 5G) indicating that the rescue of apoptosis is specific to nuclear CaMKII␦B.
The GATA-4 transcription factor plays an anti-apoptotic role in the heart by regulating Bcl-X L and Bcl-2 genes. Dox exposure leads to a loss of GATA-4 factors in primary cardiomyocytes and in vivo. Forced expression of GATA-4 partially inhibits Dox-mediated down-regulation of Bcl-X L and Bcl-2 proteins (29,41). Thus, we tested whether CaMKII␦B can pre-vent GATA-4 depletion by Dox. Overexpression of CaMKII␦B partially rescued the down-regulation of GATA-4 mRNA induced by Dox (Fig. 5H). GATA-4 protein levels were also significantly higher in cardiomyocytes treated with Dox and expressing high levels of CaMKII␦B compared with cells only treated with Dox (Fig. 5I). Furthermore, the well described Dox-mediated upregulation of pro-apoptotic p53 protein was attenuated by overexpression of CaMKII␦B (Fig. 5I). We also tested whether forced expression of CaMKII␦B can rescue the Dox effect on sarcomere organization. Primary cardiomyocytes treated with Dox showed a progressive loss of sarcomere organization, which was partially rescued by forced expression of CaMKII␦B (Fig. 5J). Overall, these results demonstrate that Dox-induced apoptosis is mediated significantly by the loss of CaMKII␦B in primary cardiomyocytes. They also demonstrate that maintaining CaMKII␦B levels in cardiomyocytes treated with Dox significantly attenuates GATA-4 loss and myofiber degeneration by the anthracycline.

Specific Silencing of CaMKII␦B Enhances Apoptosis in Primary
Cardiomyocytes-To fully establish the direct link between CaMKII␦B and apoptosis, we measured apoptosis in cardiomyocytes transfected with siControl or siCaMKII␦B by Cell Death Detection ELISA Plus. Importantly, reduced expression of CaMKII␦B significantly enhanced apoptosis after 24 h of Dox exposure (Fig. 6A). Next, we asked whether CaMKII␦B is required for Bcl-2 expression in cardiomyocytes. For this, we measured Bcl-2 protein levels in cells transfected with siCaMKII␦B or siControl. Strikingly, specific elimination of CaMKII␦B decreased levels of Bcl-2, whereas Bax protein levels remained unchanged (Fig. 6, B and C). These results indicate that CaMKII␦B is required for constitutive Bcl-2 expression in primary cardiomyocytes. They also show that CaMKII␦B plays an anti-apoptotic role in cardiomyocytes.
CaMKII␦B Is Required for GATA-4 Co-activation of the Bcl-2 Promoter-To dissect the mechanisms whereby CaMKII␦B exerts its protective role in cardiomyocytes and to define the mechanisms by which reduction of CaMKII␦B abolishes Bcl-2 expression, we tested whether CaMKII␦B directly regulates the Bcl-2 gene because the kinase is expressed predominantly in the nuclear compartment. Primary cardiomyocytes were transiently transfected with a 1281-bp fragment of the Bcl-2 promoter upstream of the luciferase gene (Bcl-2-luc) (Fig. 7A). This promoter fragment contains two consensus GATA-4-binding sites at positions Ϫ1025 and Ϫ266 bp and is activated by GATA-4 mainly by the Ϫ266 GATA-4-binding site (41). Basal Bcl-2 promoter activity was minimal in primary cardiac cells (Fig. 7B). Forced expression of CaMKII␦B only slightly activated the Bcl-2-luciferase reporter suggesting that endogenous levels of the kinase are sufficient for maximal Bcl-2 expression. Co-transfection of the Bcl-2 promoter with a GATA-4 expression vector resulted as expected in a striking increase in Bcl-2luciferase activity. This activation was greatly decreased with co-transfection of siCaMKII␦B but not siControl (Fig. 7C). Collectively, these results demonstrate that CaMKII␦B is required for GATA-4-mediated activation of the Bcl-2 promoter.
CaMKII␦B Is Necessary for GATA-4 Binding to the Bcl-2 Promoter-Next we further investigated the mechanism whereby silencing of CaMKII␦B inhibits GATA-4 co-activation of the Bcl-2 promoter. We hypothesized that reduction of CaMKII␦B might decrease GATA-4 protein levels in cardiomyocytes. To test this, we measured GATA-4 protein levels in Infected cells also express GFP, and nuclei are stained with 4Ј,6-diamidino-2phenylindole (DAPI). C, apoptosis was measured by Cell Death Detection ELISA PLUS (Roche Applied Science). Ad-CaMKII␦B or Ad-GFP infected cardiomyocytes were treated with Dox for the indicated times. The data were calculated from three separate cardiomyocyte preparations. D, levels of Bcl-2 were measured by Western blot in primary cardiomyocytes infected with Ad-CaMKII␦B or Ad-GFP and either left untreated or Dox-treated for 24 h. Bax protein level was also measured. ␤-Actin and GAPDH were used as loading controls. Quantification of Bcl-2 is shown corrected for GAPDH in E and Bcl-2/Bax ratio in F. Bcl-2/Bax ratios were also calculated from cardiomyocytes infected with CaMKII␦C in G.   (22). GAPDH expression is also shown as loading control. C, quantification of the Bcl-2/Bax ratio is the average Ϯ S.E. of three independent experiments from separate cardiomyocyte preparations. Asterisks indicate that the difference is statistically significant by Student's t test; *, p Ͻ 0.05; **, p Ͻ 0.01. cardiomyocytes transfected with siCaMKII␦B or siControl. Surprisingly, we found that silencing of CaMKII␦B minimally affected GATA-4 protein expression (Fig. 7, D and E). We then asked whether CaMKII␦B might regulate GATA-4 binding activity. Since GATA-4 binding to the Ϫ266 conserved site on the Bcl-2 promoter partially contributes to Bcl-2 promoter activity (41), we studied endogenous GATA-4 binding to the Bcl-2 promoter by EMSA. We used nuclear extracts prepared from primary neonatal rat cardiomyocytes and a probe containing the Ϫ266 GATA-4-binding site. A strong band shift was detected, which was GATA-4-specific as it could be fully competed with a cold Ϫ266 probe but not with a probe Ϫ266m containing a mutated GATA-4-binding site (Fig. 7F). To study the role of CaMKII␦B in GATA-4 binding to the Bcl-2 promoter, we carried out EMSA experiments using nuclear extracts from cardiomyocytes transfected with siControl or siCaMKII␦B. Silencing of CaMKII␦B resulted in reduced binding of GATA-4 to Bcl-2 promoter at the Ϫ266 consensus site (Fig. 7G). We verified that protein concentrations were not modified in nuclear extracts expressing reduced levels of CaMKII␦B. Indeed, phosphorylated Akt was identical in siControl and siCaMKII␦B cardiac nuclear extracts (Fig.  7H). Overexpression of CaMKII␦B did not increase GATA-4 DNA binding over control cells (data not shown) suggesting that endogenous levels of the kinase are sufficient for full GATA-4 binding activity. These data suggest that nuclear CaMKII␦B is required for GATA-4 binding to the Bcl-2 promoter. Collectively, the results suggest a mechanism for Dox induction of apoptosis through loss of CaMKII␦B-mediated GATA-4 binding to the Bcl-2 promoter.

DISCUSSION
In this study, we report for the first time that nuclear CaMKII␦B plays a critical role in cardiomyocyte survival and in Dox-induced cardiomyopathy and provides a mechanism for the anti-apoptotic role of the kinase in cardiac cells. Our results also demonstrate selective targeting of CaMKII␦B, but not the ␦C variant, by the anthracycline Dox. This novel observation establishes specific signaling transmitted to selective CaMKII isoforms in the heart during degenerative cardiomyopathy. These results are supported by the observation that levels of CaMKII␦B are severely depleted in the adult heart after chronic treatment with the cardiotoxic agent Dox, whereas the ␦C variant remained unaffected. Furthermore, specific elimination of CaMKII␦B in primary cardiomyocytes alters the structural integrity of the cells, decreases the expression of the anti-apoptotic factor Bcl-2, and exaggerates Dox-induced apoptosis. Mechanistically, our data show that CaMKII␦B is required for GATA-4 activation of the Bcl-2 gene. Maintenance of CaMKII␦B protein levels by forced overexpression partially rescues Dox-mediated down-regulation of GATA-4 and Bcl-2 proteins. These results have important implications for both the cardiotoxic effects of Dox and for mechanisms regulating apoptosis.
Selective Targeting of CaMKII␦B by Dox-CaMKII is a serine/threonine kinase regulated by calcium and plays a significant role in cardiac disease (4). However, little is known about the contribution of specific CaMKII isoforms to the pathophysiology of heart disease and on the mechanism involved. The ␦B and ␦C subunits of CaMKII are predominant enzymes in the adult heart (1-4). Studies performed in human and rodent Cold competitor was at 50-and 250-fold molar excess. When EMSAs were carried out using nuclear extracts from cardiomyocytes transfected with siCaMKII␦B, there was a significant reduction in intensity of the GATA-4-shifted band compared with extracts transfected with siControl. G, representative data from two independent experiments in separate cardiomyocyte preparations. H, shown are the phosphorylated Akt in cardiac nuclear extracts prepared from siControl and the siCaMKII␦B-transfected cardiomyocytes used in G.
hearts with hypertrophic and ischemic cardiomyopathy demonstrated increased activity in both splice variants (22). Despite the high similarity of their amino acid sequences and the ability of both isoforms to induce cardiac hypertrophy in mice (20,21), the ␦B and ␦C variants have unique characteristics. CaMKII␦B is expressed in the nucleus via its nuclear localization signal and regulates transcriptional events in the heart (15,16,47,48). The ␦C variant is enriched in the cytoplasm of cells and controls excitation-contraction events (8,9). Although the ␦B and ␦C isoforms both regulate Mef2 activity, they differentially phosphorylate proteins involved in calcium handling (48). Thus, the different subcellular localization of the two enzymes together with differences in amino acid sequence suggest that they regulate separate cellular functions or even play antagonistic roles. Our results demonstrate that Dox exposure depletes CaMKII␦B mRNA in vivo, whereas in contrast CaMKII␦C transcript levels are not significantly affected. Previous studies demonstrated a pro-apoptotic role of CaMKII␦C in cardiomyocytes after ␤ 1 -adrenergic stimulation and in response to oxygen free radicals (36,49). Overexpression of constitutively active CaMKII␦C in the cytoplasm is sufficient to trigger apoptosis, and repression of CaMKII␦C protects against cardiac apoptosis (33,36,37). In our experiments, CaMKII␦C mRNA levels were not significantly affected by Dox treatment in vivo. This result establishes that Dox selectively targets specific CaMKII isoforms. We cannot exclude the possibility that Dox cardiotoxicity is not only because of decreased CaMKII␦B but also to impaired activity of CaMKII isoforms other than CaMKII␦C. These possibilities not necessarily exclusive are currently under investigation.
Critical Role of CaMKII␦B in Cardiomyocyte Survival-The anthracycline antibiotic Dox is widely and effectively used in the treatment of many cancers. Unfortunately, its use in clinics remains limited by life-threatening dose-dependent cardiac side effects that lead to degenerative cardiomyopathy and congestive heart failure (for reviews see Refs. 24 -26). Mechanisms of Dox cardiotoxicity include the generation of free radicals leading to oxidative stress, a down-regulation of cardiac muscle genes, and cardiac transcription factors and calcium mishandling (24, 25, 27-29, 31, 42, 50). Increased apoptosis is also a key factor in Dox-mediated cardiomyopathy (30 -32, 42). Our findings provide the first evidence that the ␦B isoform of CaMKII has an anti-apoptotic function in the heart. This conclusion is based on the observation that levels of the kinase are severely depleted in the heart of animals with degenerative cardiomyopathy following chronic exposure with Dox, as well as in cardiomyocytes treated with the drug. Specific reduction of CaMKII␦B in primary cardiomyocytes by siRNA induces structural damage similar to that induced by Dox. Most importantly, specific elimination of CaMKII␦B increases cell death by apoptosis and decreases levels of the anti-apoptotic protein Bcl-2 in primary cardiomyocytes. Ectopic expression of CaMKII␦B prevents the loss of Bcl-2 and GATA-4 protein after Dox treatment, both of which are key mediators of apoptosis in cardiomyocytes (41,44,51). The observation that Dox treatment does not affect the expression of Bax suggests that Dox-mediated apoptosis may be due to a loss of survival factors rather than an increase in the expression of pro-apoptotic genes. A detailed analysis of the expression of additional apoptotic factors will help identify the specific apoptotic pathways activated in response to Dox and the ones regulated by CaMKII␦B.
Overexpression of CaMKII␦B in transgenic heart leads to dilated cardiomyopathy (20). Thus, it may seem as a paradox that CaMKII␦B has protective characteristics. One possible explanation is that constitutive CaMKII␦B is anti-apoptotic, whereas supra-physiological levels of the kinase activate additional pathways that lead to dilated cardiomyopathy and decrease hemodynamic function. It will be of interest to investigate whether CaMKII␦B can protect against Dox cardiotoxic effects in the animal. Since overexpression of nuclear CaMKII␦B in mouse heart results in cardiac hypertrophy (20), a transient moderate overexpression coinciding with Dox treatment might minimize any hypertrophic side effects. Alternatively, a better strategy might be to develop transgenic mice where CaMKII␦B is specifically reduced and to test whether these mice are hypersensitive to Dox-mediated apoptosis. Also, the neuregulin/ErbB pathway plays a critical role during cardiac development and is essential to maintain cardiomyocyte structure and function (for review see Ref. 52). Patients undergoing cancer therapy with the monoclonal antibody Herceptin that targets neuregulin/ErbB2 receptors develop cardiac dysfunction. Most importantly enhanced cardiotoxicity by concomitant treatment with Dox and Herceptin has been reported in several studies (for review see Refs. 53,54). These observations suggest that there could be a cross-talk between neuregulin/ ErbB and CaMK pathways in the heart.
CaMKII␦B Is Required for GATA-4 Regulation of the Bcl-2 Gene-The GATA-4 transcription factor plays an anti-apoptotic role in isolated cardiomyocytes and in vivo (29,51). Because GATA-4 has been shown to directly regulate the transcription of the anti-apoptotic Bcl-x L protein and activates the Bcl-2 promoter through direct binding (29,41), we hypothesized that CaMKII␦B may exert its anti-apoptotic function by regulating GATA-4 activity and by preventing the mis-regulation of Bcl-2. Indeed, we show that CaMKII␦B is necessary for binding of GATA-4 to the Bcl-2 promoter. We propose a model where CaMKII␦B allows recruitment or stabilization of GATA-4 at the Bcl-2 promoter (Fig. 8). Since GATA-4-binding sites are CaMKII␦B is required for GATA-4 recruitment to the Bcl-2 gene and promotes cardiomyocyte survival under basal conditions. Dox treatment down-regulates CaMKII␦B and GATA-4 protein by independent mechanisms, which both contribute to increase apoptosis in cardiac cells. Specific activation of CaMKII␦B might represent a therapeutic strategy to enhance the GATA-4-Bcl-2 pathway and to protect against Dox-mediated apoptosis. present in numerous cardiac specific genes that are critical for contractile function, it is tempting to speculate that CaMKII␦B may well be important for cardiomyocyte integrity. Consistent with this idea, our data clearly show that reduction of CaMKII␦B by siRNA impairs sarcomere assembly. Thus, our data suggest that CaMKII␦B is both necessary and sufficient for proper regulation of cardiac muscle genes and for cardiomyocyte survival.
Dox treatment decreases both CaMKII␦B and GATA-4 expression whereas reduction of CaMKII␦B minimally affects GATA-4 protein level but rather reduces its recruitment to the Bcl-2 promoter. This suggests that CaMKII␦B is required for Bcl-2 but not for GATA-4 basal gene expression and that Doxmediated GATA-4 depletion involves pathways independent of CaMKII␦B. However, the observation that forced expression of CaMKII␦B rescues, at least in part, Dox-mediated down-regulation of GATA-4 and Bcl-2 indicates that CaMKII␦B can upregulate both genes. Phosphorylation of GATA-4 plays a role in its ability to rescue anthracycline cardiotoxicity (55), and this can be induced through hypertrophic stimuli (56), which are known to be mediated through CaMKII␦. Studies underway are designed to determine whether GATA-4 is a direct target of CaMKII␦B phosphorylation.
Our microarray data revealed that levels of residual CaMKII␦B expression are inversely related to the degree of the pathology observed after Dox treatment in the animal. This result, together with the observation that Dox treatment of isolated cardiomyocytes leads to the rapid loss of CaMKII␦B, suggests that the kinase is an early target of Dox and that its loss is likely a cause rather than a consequence of the cardiomyopathy engendered by Dox. Thus, the ability of the kinase to rescue Dox-mediated cell death in cardiomyocytes identifies CaMKII␦B as an attractive cellular target for the development of novel therapeutic strategies for the prevention of Dox cardiotoxic effects.
In conclusion, our studies report the selective loss of CaMKII␦B expression after treatment with Dox in vivo and reveal a novel function of the kinase in preventing cell death by regulating GATA-4 binding to the Bcl-2 gene in cardiac cells. Specific activation of CaMKII signaling in the cell nucleus might be an important therapeutic opportunity to prevent cardiomyocyte loss and cell death following anthracycline chemotherapy. This could also provide a potential new therapeutic approach for the treatment of heart failure.