|Home | About | Journals | Submit | Contact Us | Français|
Human heart failure is accompanied by repression of genes such as α myosin heavy chain (αMyHC) and SERCA2A and the induction of fetal genes such as βMyHC and atrial natriuretic factor. It seems likely that changes in MyHC isoforms contribute to the poor contractility seen in heart failure, because small changes in isoform composition can have a major effect on the contractility of cardiac myocytes and the heart. Our laboratory has recently shown that YY1 protein levels are increased in human heart failure and that YY1 represses the activity of the human αMyHC promoter. We have now identified a region of the αMyHC promoter that binds a factor whose expression is increased sixfold in failing human hearts. Through peptide mass spectrometry, we identified this binding activity to be a heterodimer of Ku70 and Ku80. Expression of Ku represses the human αMyHC promoter in neonatal rat ventricular myocytes. Moreover, overexpression of Ku70/80 decreases αMyHC mRNA expression and increases skeletal α-actin. Interestingly, YY1 interacts with Ku70 and Ku80 in HeLa cells. Together, YY1, Ku70, and Ku80 repress the αMyHC promoter to an extent that is greater than that with YY1 or Ku70/80 alone. Our results suggest that Ku is an important factor in the repression of the human αMyHC promoter during heart failure.
Cardiovascular disease, including heart failure, is the leading cause of death in the United States. Heart failure is characterized at the molecular level by changes in gene expression that result in the repression of adult genes, such as α myosin heavy chain (αMyHC) and SERCA2A, and the induction of fetal genes (e.g., βMyHC, atrial natriuretic factor [ANF], and skeletal α-actin ). The change in MyHC expression is a logical candidate for affecting cardiac contractility, since small changes in isoform composition have been shown to affect contractility of cardiac myocytes and the heart. Despite sharing 93% amino acid identity, αMyHC has a higher rate of ATP hydrolysis than βMyHC. Transgenic mice with cardiac myocytes expressing only 12% of total MyHC as the β isoform have decreased contractility and a 22% decrease in myofibrillar ATPase (40). In single-cell experiments, cardiac myocytes expressing only 12% of their MyHC as α had 52% greater power output than those expressing only the β isoform (12).
In the nonfailing, nonhypertrophied human heart, approximately 20 to 30% of total MyHC mRNA consists of αMyHC mRNA, whereas in the failing heart, αMyHC expression represents less than 2% of total MyHC mRNA (19, 24). At the protein level, αMyHC in the normal heart constitutes 7 to 11% of total MyHC, but it is undetectable in the failing heart (21). The importance of αMyHC expression in the human heart has been recently emphasized by the finding that mutations in the αMyHC gene can cause both hypertrophic and dilated cardiomyopathies (7, 25). The decrease in αMyHC in human heart failure may play an important role in the well-established reduction of cardiac contractility. The goal of this study was to determine the role of transcriptional control in the decrease of αMyHC in heart failure and to identify potential transcriptional repressors.
The promoter-proximal region of human αMyHC is not as well characterized as the rat αMyHC promoter, but they share ~80% sequence similarity over a 400-bp region. Based on sequence comparison, the human αMyHC promoter has putative binding sites for transcription factors that have been shown to be important for the positive regulation of the rat αMyHC promoter, e.g., GATA4, NFAT3, TEF, thyroid response, MEF2, and SRF, among others (23). Our laboratory has recently shown that the YY1 transcription factor acts as a repressor of the human αMyHC promoter in cardiac cells and that YY1 levels and DNA binding activity are increased in human heart failure (38). This was the first report of a transcriptional repressor of the human αMyHC promoter. YY1 is a 414-amino-acid protein that has been shown to activate or repress transcription depending on promoter context or protein interaction (41). It has also been shown to function as a polycomb group protein responsible for repression of certain developmentally regulated genes in Drosophila melanogaster (3).
In the present study we show that in addition to binding to YY1, the −370 to −350 region of the human αMyHC promoter binds a second protein complex with sixfold-higher activity in failing human hearts. Through DNA affinity column purification followed by mass spectrometry, we have identified this complex as the heterodimer Ku70/Ku80. Mutations in the binding site for Ku70/80 result in up-regulation of the αMyHC promoter. Coexpression of Ku70 and Ku80 together in neonatal rat ventricular myocytes (NRVM) results in repression of αMyHC promoter activity. Western blotting of normal and failing human heart extracts showed that Ku70 protein levels are increased sixfold in the failing heart. Immunoprecipitation experiments showed that YY1 interacts with Ku70 and Ku80 and that coexpression of YY1 and Ku70 proteins in NRVM resulted in increased repression of αMyHC promoter activity. Moreover, infection of NRVM with adenovirus constructs expressing Ku70 and Ku80 resulted in a selective repression of endogenous αMyHC mRNA expression and up-regulation of skeletal α-actin mRNA, suggesting that the Ku proteins play an important role in the regulation of the fetal gene program in cardiac disease.
YY1 (SC-7341X) and Ku80 (SC-5280) antibodies (Abs) were purchased from Santa Cruz Biotechnology. Ku70 Ab (NB 100-102) was purchased from Novus Biological. For immunoprecipitation experiments, the YY1 and Ku Abs were purchased from Santa Cruz Biotechnology. The alkaline phosphatase-conjugated (115-053-146) and horseradish peroxidase (HRP)-conjugated (115-035-146) anti-mouse Abs and HRP-conjugated anti-rabbit Ab (711-035-152) were purchased from Jackson Laboratories.
The −454/+32 bp fragment of the human αMyHC promoter was cloned into the pGL3 basic vector (Promega). Mutations were created in the Ku binding site by generating oligonucleotides (see description of EMSAs, below) containing the mutation of interest and amplifying the fragments containing the mutation by PCR. The YY1 expression construct was a gift from Michael Atchison (University of Pennsylvania), and the Ku expression construct was a gift from William Dynan (Medical College of Georgia). The DNA constructs were purified by using the QIAGEN method.
NRVM were prepared according to the method described by Waspe et al. (44). Briefly, 150,000 cells/well were plated in 12-well tissue culture plates coated with gelatin. Eighteen hours later, the medium was changed to minimal essential medium supplemented with Hank's salt and l-glutamine. HEPES (20 mM; pH 7.5), penicillin, vitamin B12, bovine serum albumin, insulin, and transferrin were added to the medium. Transfections were carried out by the Fugene 6 (Roche) method according to the manufacturer's recommendations; 0.75 μl of Fugene per 0.25 μg of plasmid DNA was transfected in each well. In the cotransfection experiments, the total amount of DNA was kept constant by the addition of a plasmid containing the cytomegalovirus (CMV) promoter not driving the expression of any gene.
Protein extracts from human normal and failing (idiopathic dilated cardiomyopathy) left ventricles were prepared according to the methods of Molkentin et al. with minor modifications (22). A 0.5-g aliquot of tissue was homogenized with a Teflon homogenizer attached to a drill (SKIL PA6-GF30) at 50% power. The resulting preparation was sonicated in a cell disruptor (Ultrasonics W185F) at 50% power for 15 s. Following cell lysis, the proteins were precipitated by the slow addition of an equal volume of 4 M NH4SO4. This resulted in a 50% final concentration of NH4SO4, allowing a greater protein recovery compared to the 30% final concentration of NH4SO4 described in the original method. HeLa cell nuclear extracts were prepared by the method of Dignam et al. (8).
Electrophoretic mobility shift assays (EMSAs) were carried out as described elsewhere (37). Double-stranded oligonucleotides of 30 bp were labeled by Klenow fill-in using [32P]dCTP. The reaction was performed using 100,000 cpm of wild-type or mutant probe in a 30-μl binding reaction mixture containing 10 mM HEPES (pH 7.9), 100 mM KCl, 4% glycerol, 1 mM EDTA, 0.1% Nonidet P-40, 1 μg of poly(dI-dC) (Pharmacia Biotech), and 10 μg of extract. The resulting complex was resolved in a nondenaturing 4% acrylamide gel in 0.5× Tris-borate-EDTA.
EMSA using circular DNA was done essentially as described elsewhere (11). An oligonucleotide containing three copies of the above fragment was cloned into the mini circle essentially as described previously (11).
HeLa cell pellets were bought from the Computer Cell Culture Center, Seneffe, Belgium, and nuclear extract was prepared as described above. The double-stranded oligonucleotide used in these experiments had the YY1 binding site mutated. A multimerized oligonucleotide containing three copies of the Ku binding site was labeled with biotin on the 5′ end. A 20-mg aliquot of Dynabeads (Dynal) coupled to 3 nmol of the double-stranded oligonucleotide was washed three times in binding buffer (10 mM HEPES [pH 7.9], 4% glycerol, 1 mM EDTA, 0.1% Nonidet P-40, and 200 μg of insulin/ml, supplemented with 1× protease inhibitors [Boehringer-Mannheim]). One milligram of HeLa cell nuclear extract was incubated with the oligonucleotide-coupled magnetic beads for 30 min at room temperature. The beads containing bound proteins were washed three times, for 5 min each, with 500 μl of binding buffer containing 100 mM KCl. For the last two washes, 15 μg of poly(dI-dC) was added. Proteins were step eluted with 100 μl of binding buffer containing 0.2, 0.3, or 1 M KCl for 10 min each. A 1-μl aliquot of each fraction and 5 μl of each wash were assayed by EMSA, without poly(dI-dC). The peak fraction (1 M KCl) of the purification was submitted to denaturing polyacrylamide gel electrophoresis, stained with colloidal Coomassie (Invitrogen), and destained with water. Images of the gels were captured with a Umax scanner.
Protein bands were excised from the gels with a scalpel. Gel pieces were washed twice for 10 min with 200 μl of 50% acetonitrile (CH3CN)-25 mM ammonium bicarbonate (NH4HCO3) and then once with 200 μl of 100% CH3CN. The gel pieces were dried in a Speed-vac for 15 min. To the dried gel pieces was added 20 μl of a 20-μg/ml trypsin (Promega sequencing grade) solution in 50 mM NH4HCO3. The gel pieces were allowed to rehydrate on ice for 20 min and incubated overnight at 37°C. The next day the peptides were extracted into 200 μl of 50% CH3CN-0.1% trifluoroacetic acid (TFA) with vigorous mixing. The extracts were transferred to fresh tubes and taken to dryness (Speed-vac). The peptides were then redissolved overnight in 20 μl of 0.1% TFA. ZipTips (C18; 0.6-μl bed volume; Millipore) were wetted with 50% CH3CN-0.1% TFA and equilibrated in 0.1% TFA. Peptides were bound by pipetting 10 times through the bed. The ZipTips were then washed three times with 0.1% TFA, and the last wash was completely expelled. For each sample, a 2-μl aliquot of 80% CH3CN-0.1% TFA was placed on a stainless steel MALDI-TOF (matrix-assisted laser desorption-ionization time of flight) plate. The ZipTip was brought into contact with this solution, which was pipetted five times through the bed to elute the peptides. Immediately, a 1-μl aliquot of matrix solution was spotted on top of the eluate and allowed to dry. The matrix solution consisted of 10 mg of recrystallized α-cyano-4-hydroxycinnamic acid/ml dissolved in 80% CH3CN-0.1% TFA. A standard mix of des-Arg1 bradykinin, angiotensin I, glu1-fibrino-peptide B, and adrenocorticotropin (18-39) in matrix was also applied to an adjacent region of the plate. MALDI-TOF spectra were acquired on a Voyager-DE PRO (PerSeptive Biosystems) instrument operating in reflector mode. Following initial data collection using external calibration on the standard mix, the spectra of unknowns were recalibrated using autolytic tryptic fragments as internal mass markers. Monoisotopic peptide masses with a signal-to-noise ratio of greater than 5:1 were entered into the Mascot program (www.matrixscience.com). This is a peptide mass fingerprinting program that finds the best fit of observed tryptic peptides to predicted tryptic peptides from database proteins. These masses were searched at a tolerance of 50 ppm against the NCBInr database. One missed cleavage was allowed. Oxidation of methionine was allowed as a variable modification, because this has been observed after colloidal Coomassie staining.
Western blotting assays were performed essentially as described elsewhere (38). Ku Ab was diluted 1:1,000 in 1× phosphate-buffered saline containing 3% bovine serum albumin, 2% normal goat serum, and 0.1% Tween and incubated with the blot for 1 h at room temperature. The mouse secondary Ab conjugated to alkaline phosphatase was diluted 1:2,500 in 1× phosphate-buffered saline containing 5% low-fat dry milk and 0.1% Tween and incubated with the blot for 1 h at room temperature. We used alkaline phosphatase-conjugated secondary Abs, the Vistra (Amersham) enhanced chemiluminescence reagent, an ABI STORM PhosphorImager, and ImageQuant software to visualize and quantify bound Ab.
Immunoprecipitation experiments were done using YY1, Ku70, Ku80, and hemagglutinin probe Abs. Experiments were done according to Santa Cruz Biotechnology recommendations with minor modifications. After four washes with 1× radioimmunoprecipitation assay buffer, the sample was incubated with two to three packed volumes with 2× sample buffer (Bio-Rad) and incubated at room temperature for 30 min. β-Mercaptoethanol was added to the supernatant after centrifugation, and samples were loaded without boiling. Ethidium bromide was added to HeLa nuclear extracts as described previously (17), and immunoprecipitation experiments were performed as described above. Western blotting experiments were done using the above Abs, and the anti-mouse or anti-rabbit HRP-conjugated Abs were used for detection.
Coomassie staining of polyvinylidene difluoride membranes was performed essentially as described elsewhere (38). Following Western blotting experiments, the membrane was stripped of the Abs and stained with Coomassie for 1 h. The membrane was then destained for at least 12 h. The proteins on the membrane were quantified on a LiCor scanner.
Methods for generating recombinant adenovirus expression vectors have been described previously (2, 33). The Ku70 and Ku80 adenovirus expression vectors, AdCMV-Ku70 and AdCMV-Ku80, are based on the Cre-dependent luciferase expression vector AdCUL (18). AdCUL consists of oppositely oriented mutant lox sites, lox71 and lox66, flanking an antisense firefly luciferase reporter gene downstream of the CMV immediate-early promoter. Cre-mediated recombination between lox71 and lox66 inverts the floxed cassette into the sense orientation, resulting in luciferase gene expression (2). Substitution of the Ku70 and Ku80 cDNAs for the luciferase gene in AdCUL generates the antisense expression vectors, AdCMV-αKu70 and AdCMV-αKu80, respectively. AdCMV-αKu70 and AdCMV-αKu80 were subsequently passaged through the Cre-expressing cell line HEK293Cre57 (18) to facilitate generation of the sense-oriented expression vectors, AdCMV-Ku70 and AdCMV-Ku80. The emergent virus was plaque purified, and the sense orientation of the floxed cassettes was verified by PCR.
Total RNA was extracted with TRIzol (Invitrogen) and used in RNase protection assays (RPAs). RPAs were performed essentially as described elsewhere (14, 28). Briefly, 5 μg of total RNA was hybridized against probes specific to skeletal α-actin, SERCA2a, αMyHC, βMyHC, and glyceraldehyde-3-phosphate dehydrogenase (14, 28). RNase protection experiments were performed using the RPA II kit (Ambion).
In an attempt to identify transcription factors that could be responsible for the down-regulation of the αMyHC promoter, several regions of the promoter were chosen for further study based on deletion-promoter activity studies (data not shown) and used as probes in EMSAs against normal and failing human heart extracts. Some of these regions contained consensus binding sites for known transcriptional activators (e.g., GATA4, MEF2, and TEF-1). One site (Fig. (Fig.1)1) was chosen for further studies because its deletion resulted in a twofold increase in αMyHC promoter activity in NRVM. This region contains a putative binding site for the YY1 transcription factor, which we have shown to function as a repressor (see below and reference 38). As shown in Fig. Fig.1A,1A, this region also binds a factor which is distinct from YY1, with fivefold more DNA binding activity in extracts from failing human hearts. As a control, a probe containing an Sp1 binding site was used in an EMSA against the same extracts shown in Fig. Fig.1.1. As seen in Fig. Fig.1D,1D, there was no difference on Sp1 binding activity between failing and normal human hearts, consistent with the specificity of the increase in binding activity. The fact that we did not detect YY1 binding to this site from failing heart extracts is not understood at the present time, particularly since HeLa cell nuclear extracts form two complexes with this DNA sequence: one that comigrates with the complex seen in human failing hearts and another faster-migrating complex (Fig. (Fig.1B).1B). The faster-migrating complex contains YY1, as shown by the ability of an anti-YY1 antibody to block its formation (see Fig. Fig.3A3A).
In order to identify the slower-migrating proteins present in the complex from failing human heart and HeLa cells, the oligonucleotide containing the binding site for this complex but with the YY1 site mutated was multimerized, labeled with biotin, and incubated with HeLa cell nuclear extracts. The complex was then bound to Dynabeads (Dynal). The eluted fractions (see Materials and Methods) were analyzed by EMSA. As seen in Fig. Fig.2A,2A, lane 3, the 1 M elution fraction contained an enriched protein with similar migration to the slower-migrating complex in the HeLa cell nuclear extract (Fig. (Fig.2,2, lane 4). UV cross-linking experiments suggested that the molecular mass of the protein of interest was approximately 70 kDa (Fig. (Fig.2B,2B, lane 2). The 1 M elution fraction was resolved on a denaturing polyacrylamide gel. Five major bands were obtained (Fig. (Fig.2C,2C, lane 1), and all of them were analyzed by mass spectrometry. Of the five bands, two bands that showed the greatest intensities (Fig. (Fig.2C,2C, bands 1 and 2) were identified as Ku 70 (band 1), with 15 out of 21 peptides matching, 23% sequence coverage, and a Mascot probability-based MOWSE score of 187, and Ku 80 (band 2), with 18 out of 24 peptides matching, 17% sequence coverage, and a Mascot probability-based MOWSE score of 186. The molecular masses of the bands (70 and 85 kDa) corresponded to the molecular masses of Ku70 and Ku80, respectively. The bands were of similar intensity, which is consistent with the known heterodimerization of these proteins.
Since Ku was identified as binding to the αMyHC promoter by mass spectrometry, an antibody against Ku was used in EMSA experiments to determine if Ku is in fact part of the complex that binds to this sequence in HeLa cells and is increased in failing heart extracts. As seen in Fig. Fig.3A,3A, addition of the Ku70 Ab supershifted the slower-migrating complex in a HeLa cell nuclear extract (lane 3). Addition of the YY1 Ab prevented YY1 from binding to the DNA but had no effect on Ku binding (lane 2). Similar results were obtained with purified Ku proteins (lanes 4 to 6). Interestingly, the addition of purified Ku proteins to the binding experiment generated two complexes. Even though the migration of the faster complex was similar to that of YY1, it did not contain YY1, since it was not supershifted by the YY1 Ab (lane 6). It seems likely that this band corresponds to Ku70 only, since it is supershifted by the Ku70 Ab and the slower-migrating band contains Ku70 and Ku80. Addition of Ku70 or Ku80 Abs also supershifted the complex in binding experiments with human failing heart extracts (Fig. (Fig.3B),3B), while the YY1 Ab failed to supershift the complex (data not shown). To determine the sequences in the αMyHC promoter to which Ku is binding, a series of mutants was created. Since Ku binding sites are quite variable, different mutations were designed. Sequence-specific binding of Ku to DNA has been documented in the context of the long terminal repeat of the mouse mammary tumor virus (10). Ku has been shown to bind a sequence containing a direct repeat whose monomer is GAGAAAGA. However, Ku is capable of interacting with a truncated version of this sequence that contains only the monomer. Transversion mutation of the fifth nucleotide to a T in the first repeat prevents binding of Ku (10). Ku has also been shown to bind the sequence GAGAGGGGTCGG in the human CD34 promoter (39). The human αMyHC promoter contains two potential Ku binding sites. The first potential Ku binding site contains a sequence similar to the CD34 promoter and to the mouse mammary tumor virus long terminal repeat: GAGAGG. The second potential Ku binding site in the human αMyHC promoter is GTAAG followed by two purines. Camara-Clayette et al. have proposed a consensus site for Ku that is SHBAGAYAS (S is for G or C; H is for A, T, or C; B is for T, C, or G; and Y is for C or T) (6). Although not all Ku binding promoters contain this binding site, many of the promoters contain at least five of the nine nucleotides proposed in the consensus binding site (6, 10, 13, 20, 39, 45). Both of the potential Ku binding sites in the human αMyHC promoter region fall into this category. Therefore, mutations were created in both regions. As seen in Fig. Fig.3C,3C, mutations in the two potential Ku binding sites completely abolished binding of Ku but maintained binding of YY1 (lane 3), and mutations in the YY1 site prevented binding of YY1 but allowed binding of Ku (lane 2). Probes containing mutations in each of the Ku binding sites were created and used in an EMSA. Ku retained the ability to bind to the DNA if either one of the binding sites was present (data not shown). In order to test binding specificity, the YY1 mutant probe was used in EMSA experiments and either the same oligonucleotide or the Ku doubly mutant oligonucleotide was used as cold competitor. As seen in Fig. Fig.3D,3D, the wild-type oligonucleotides competed for binding to Ku (lanes 2 and 3), while the mutant competitor could not (lanes 4 and 5). Competition with higher amounts of Ku mutant oligonucleotide led to the appearance of a slower-migrating complex. This complex was only seen under these conditions, and its identity is not known. These results suggest that the binding of Ku to the αMyHC promoter is sequence specific.
The Ku70/80 complex has been shown to bind DNA ends and to be involved in DNA repair (for review, see reference 42). The complex has also been shown to bind to DNA in a sequence-specific manner and function as a transcription factor (for review, see reference 9). In order to determine whether the Ku70/80 complex recognizes the human αMyHC promoter independently of the presence of DNA ends, an oligonucleotide containing three copies of the Ku binding region was labeled and cloned in a mini circle (see Materials and Methods). As seen in Fig. Fig.4,4, Ku was capable of binding the human αMyHC promoter even when the DNA was in a circular form. Figure Figure4A4A shows that exonuclease III (Exo III) digested the linear probe to completion but did not affect the circular probe. Figure Figure4B4B shows that the linear probe was capable of binding to Ku70/80 but that, upon digestion with Exo III, the complex was not present. The binding of Ku to the circular probe, however, was stable even in the presence of Exo III, showing that binding of Ku to the human αMyHC promoter is independent of the presence of DNA ends.
EMSA experiments showed that Ku binding activity was increased in human failing heart extracts (Fig. (Fig.1).1). To test if protein levels were also increased in human heart extracts, Western blot experiments were performed using the Ku70 and Ku80 Abs. As shown in Fig. Fig.5,5, Ku70 protein levels were increased in human failing heart extracts compared to extracts from normal hearts. The increase in the levels of Ku protein was comparable to the increase in the binding activity seen in the EMSA experiments. Thus, there is an inverse relationship between αMyHC expression (mRNA) and Ku protein in myocardial pathological hypertrophy and failure. Interestingly, Ku80 protein levels were not increased in human failing hearts (data not shown). This finding was surprising, since under in vitro conditions up-regulation of either of these proteins requires its counterpart (W. Dynan, personal communication). However, under pathological conditions (e.g., in response to radiation treatment ) the levels of Ku70 are increased, but those of Ku80 are not. We next tested the hypothesis that the increase in Ku levels was directly related to the repression of αMyHC.
In order to test the effect of Ku on the activity of the human αMyHC promoter, cotransfection experiments with Ku70 and Ku80 cDNA were carried out in NRVM. As seen in Fig. Fig.6A,6A, expression of Ku70 and Ku80 repressed the activity of the αMyHC promoter. Transient-transfection experiments were done using both Ku70 and Ku80, since overexpression of either one did not result in increased protein levels (see below). To test if the repression of the promoter was specific to the Ku binding site, an αMyHC construct containing the two putative Ku binding sites mutated was transfected into NRVM. The mutant construct showed twofold-higher activity than the wild type (Fig. (Fig.6B),6B), suggesting that Ku repression of the αMyHC promoter is dependent on a direct interaction with the promoter. To test if the Ku-mediated repression was specific for the αMyHC promoter, cotransfection experiments with Ku70 and Ku80 cDNA and the ANF promoter were carried out in NRVM (Fig. (Fig.6C).6C). ANF expression is up-regulated in hypertrophy and failure. Ku70 and Ku80 together do not repress the activity of the ANF promoter. These results suggest that Ku-mediated repression in cardiac hypertrophy and failure is specific for the adult gene program and does not affect the fetal gene program, as exemplified by the ANF promoter.
Ku has been shown to interact with various transcription factors (15, 30, 34, 35, 45), and we hypothesized that one or more of its interactions might facilitate the interaction of Ku with the αMyHC promoter. Recently, Ku has been shown to be important for the transcription reinitiation process (46) and to bind directly to RNA polymerase II (42). It has been suggested that Ku interacts with TATA binding protein and, most interestingly, Ku has been shown to be a promiscuous activator of transcription in yeast (4). In addition to its repressive effect on the αMyHC promoter, YY1 has also been shown to interact with the basic transcription machinery (41) and, interestingly, recent reports have implicated YY1 as being involved in DNA repair through interaction with poly(ADP-ribose) (PARP) (26, 27). Since Ku is part of the DNA repair complex and also interacts with PARP (29, 30), we tested the hypothesis that Ku and YY1 interact by performing coimmunoprecipitation experiments in HeLa cells. As seen in Fig. 7A and B, an Ab against YY1 brought down Ku70 and Ku80; in Fig. 7B and C, anti-Ku70 brought down YY1 and Ku80; in Fig. 7A and C, anti-Ku 80 brought down Ku70 and YY1. As a control, we tested if an unrelated Ab could bring down any of those proteins. As seen in Fig. 7A, B, and C, the hemagglutinin Ab failed to bring down YY1, Ku70, or Ku80. All coimmunoprecipitation experiments were also done in the presence of ethidium bromide to rule out the possibility that their interaction was dependent on the presence of DNA (Fig. (Fig.7)7) and, as shown in the figure, the presence of ethidium bromide did not change the abilities of YY1 and Ku to interact, showing that their interaction is independent of the presence of DNA. Although our EMSA experiments suggested that Ku can interact with the αMyHC promoter in the absence of YY1 in vitro, interaction of YY1 with Ku might facilitate the binding of Ku to the promoter in vivo.
Cotransfection experiments were done in NRVM using YY1, Ku70, and Ku 80 cDNAs with the αMyHC promoter linked to luciferase. As seen in Fig. Fig.8,8, the Ku complex and YY1 were capable of repressing the promoter independently. Addition of YY1 increased the repression mediated by Ku70/Ku80. The YY1 and Ku binding sites in the αMyHC promoter are adjacent. One possibility is that if they do not interact, they might compete for binding to the promoter. Since we did not see competition in the cotransfection experiments, we speculate that the interaction between the three proteins enhances their function. Moreover, cotransfection experiments using Ku70 and Ku80 and an αMyHC promoter construct with the three YY1 sites mutated (38) resulted in lower repression levels than with the wild-type promoter construct. We know from the experiments shown in Fig. Fig.44 that Ku70/80 can bind to the αMyHC promoter independently of YY1, but this experiment suggested that YY1 facilitates the recruitment of the Ku complex to the promoter.
In order to analyze the effect of Ku70 and Ku80 overexpression on endogenous αMyHC gene expression, virus constructs were created that encoded either Ku70 or Ku80. As mentioned above, under in vitro conditions, overexpression of Ku70 was not stable in the absence of Ku80, and vice versa. As seen in Fig. Fig.9A,9A, overexpression of either protein singly did not result in an increase in the protein levels for either protein. Overexpression of Ku70 and Ku80 together, however, resulted in increased levels of both proteins. Increased levels of Ku70 and Ku80 resulted in repression of endogenous αMyHC gene expression (Fig. (Fig.9C).9C). This repression was specific, since increased levels of these proteins did not change endogenous βMyHC gene expression levels, but skeletal α actin levels were increased. Skeletal α actin gene expression is up-regulated during cardiac hypertrophy and failure as part of the up-regulation of the fetal gene program. Our data are consistent with the notion that the increase in Ku levels in human heart failure could result in repression of αMyHC expression.
The down-regulation of the αMyHC gene during heart failure seems likely to contribute to the pathogenesis of the disease (see reference 1). We have recently shown that YY1 is increased in heart failure and represses the activity of the αMyHC promoter. In this report we have identified the Ku factors as additional repressors of the activity of the αMyHC promoter. We show that Ku70/80 protein levels and binding activity are increased sixfold in failing human heart extracts and that their overexpression decreases the activity of the αMyHC promoter and endogenous αMyHC gene expression in NRVM.
There has been controversy surrounding the function of Ku. Ku was first identified as a DNA repair protein that recognizes DNA ends without any preferences for the nature of the ends (16). Once bound to DNA, Ku can interact with the catalytic subunit of the DNA-dependent protein kinase (DNA-PK), and together they constitute the active kinase (9). This complex can phosphorylate several nuclear proteins in vitro, e.g., p53, c-fos, Sp1, XRCC4, DNA-PKcs, or Ku itself, and it is involved in nonhomologous DNA end-joining repair and V(D)J recombination (16). Recently, Ku has been shown to be important in preventing apoptosis by interacting with Bax and preventing it from entering the mitochondria (31, 32).
At the same time, various reports have shown that Ku can bind DNA in a sequence-specific manner, as shown in the cases of several genes (6, 10, 13, 20, 39, 45). Recent reports showed that the Ku70/80 complex interacts with different transcription factors, e.g., heat shock factor, DNA binding domain of the progesterone receptors, and homeodomain proteins (15, 30, 34, 35, 45), suggesting that this could be a mechanism by which Ku is recruited to specific regions of promoters and enhancers. Ku has also been shown to interact with RNA polymerase II and with TATA binding protein (42). In this report we have shown that Ku interaction with the αMyHC promoter is sequence specific. As shown in Fig. Fig.3,3, Ku and YY1 bind to adjacent sites in the promoter. Point mutations created in the proposed Ku binding sites abolished binding of Ku but retained YY1 binding. Point mutations in the YY1 binding site prevented binding of YY1, allowing binding of Ku. Finally, point mutations generated in both YY1 and Ku binding sites prevented binding of either protein. It has been proposed that mutations in the Ku binding site that abolish binding are not enough evidence for sequence-specific binding, due to the preference that Ku has for certain nucleotides at DNA ends (5). Another approach to show specificity of binding is the generation of a circular DNA radiolabeled probe. As shown in Fig. Fig.4,4, Ku bound to the circular probe even after digestion with Exo III. Ku was also capable of binding to the linear probe, but its binding was inhibited following Exo III digestion. These experiments suggested that binding of Ku to the αMyHC promoter is sequence specific.
As discussed above, Ku has been recently shown to interact with different transcription factors. At the same time, YY1 has been shown to be involved in DNA repair though the interaction with PARP-1 (26, 27). Ku is also known to interact with PARP-1, which led us to test the hypothesis that YY1 and Ku interact. As shown in Fig. Fig.7,7, immunoprecipitation experiments using YY1, Ku70, or Ku80 Abs suggested that these proteins interact in cells. The consequences of this interaction are important to the understanding of Ku function as a transcription factor and of YY1 function as a protein involved in repair, and vice versa. It has been recently proposed that Ku is involved in the reinitiation phase of transcription and that there would be an equilibrium between reinitiation and repair (47). In this model, Ku would be either sequestered in a Ku-dependent reinitiation complex, where it would not be capable of interacting with DNA ends or, in the presence of DNA damage signals, Ku and DNA-PKcs would be released from this complex and become active for repair. This would, in turn, disrupt the transcription apparatus, preventing reinitiation from occurring. YY1 has, in turn, been shown to be important for transcription initiation, and Usheva and Shenk (43) have shown in vitro that YY1, TFIIB, and RNA polymerase II are sufficient to initiate transcription of the adeno-associated virus P5 promoter. We propose that YY1 and Ku are part of a transcription complex that can be involved in DNA repair or transcription and that their function will vary according to the integrity of the DNA.
Ku has been shown to function either as a repressor (5, 8, 11) or as an activator (36) in transient-transfection experiments. Here, we have shown that Ku functions as a repressor of the αMyHC promoter in NRVM cells and that coexpression of YY1 and Ku increases the repressive effect. Moreover, overexpression of Ku70 and Ku80 result in repression of endogenous αMyHC and up-regulation of skeletal α actin mRNA levels, indicating that up-regulation of Ku70 and Ku80 in the failing heart plays an important role in the regulation of components of the fetal gene program.
These results are consistent with our hypothesis that Ku is a repressor of αMyHC and that the increase in Ku protein levels is at least partially responsible for the down-regulation of the αMyHC gene observed in human heart failure. It is extremely interesting that both YY1 and Ku levels are increased in heart failure. Based on our results, the increase in the levels of these two proteins has a dramatic effect in the activity of the αMyHC promoter and most likely on the levels of αMyHC gene expression.
Finally, Ku has been shown to be expressed in various organisms besides humans, including monkey, Xenopus laevis, yeast, Drosophila, and rodents (5). Interestingly, Ku levels in rodents are decreased in comparison to humans (5). At the same time, αMyHC mRNA levels are increased in rodents compared to levels in humans. One can speculate that the difference in the levels of αMyHC and Ku are related and part of an evolutionary process. Further studies will allow us to elucidate the mechanism by which YY1 and Ku repress the activity of the αMyHC promoter in cardiac cells.
We thank Michael Atchison for critically reviewing the manuscript and for providing YY1 cDNA. We also thank William Dynan for providing Ku70 and Ku80 cDNAs and recombinant protein. We thank Karin Nunley for performing the RPA experiments.
This work was supported by National Institutes of Health grant RO1 HL56510 to L.L.