Molecular mechanisms of doxorubicin-induced cardiomyopathy. Selective suppression of Reiske iron-sulfur protein, ADP/ATP translocase, and phosphofructokinase genes is associated with ATP depletion in rat cardiomyocytes.

Doxorubicin, a cardiotoxic antineoplastic, disrupts the cardiac-specific program of gene expression (Kurabayashi, M., Dutta, S., Jeyaseelan, R., and Kedes, L. (1995) Mol. Cell. Biol. 15, 6386-6397). We have now identified neonatal rat cardiomyocyte mRNAs rapidly sensitive to doxorubicin, or its congener daunomycin, including transcripts of nuclear genes encoding enzymes critical in production of energy in cardiomyocytes: ADP/ATP translocase, a heart- and muscle-specific isoform; Reiske iron-sulfur protein (RISP), a ubiquitously expressed electron transport chain component; and a muscle isozyme of phosphofructokinase. Loss of these mRNAs following doxorubicin or daunomycin is evident as early as 2 h and precedes significant reduction of intracellular ATP. ATP levels in control cardiomyocytes (17.9 ± 2.9 nM/mg of protein) fall only after 14 h and reach residual levels of 10.4 ± 0.9 nM (doxorubicin; p = <0.006) and 6.7 ± 1.9 nM (daunomycin; p = <0.001) by 24 h. Loss of mRNAs generating ATP was highly selective since mRNAs for other energy production enzymes, (cytochrome c, cytochrome b, and malate dehydrogenase), and genes important in glycolysis (pyruvate kinase and glyceraldehyde-3-phosphate dehydrogenase) were unaffected even at 24 and 48 h. The drugs had no effect on levels of ubiquitously expressed RISP mRNA in fibroblasts. These findings could link doxorubicin-induced damage to membranes and signaling pathways with 1) suppression of transcripts encoding myofibrillar proteins and proteins of energy production pathways and 2) depletion of intracellular ATP stores, myofibrillar degeneration, and related cardiotoxic effects.

. We have now identified neonatal rat cardiomyocyte mRNAs rapidly sensitive to doxorubicin, or its congener daunomycin, including transcripts of nuclear genes encoding enzymes critical in production of energy in cardiomyocytes: ADP/ ATP translocase, a heart-and muscle-specific isoform; Reiske iron-sulfur protein (RISP), a ubiquitously expressed electron transport chain component; and a muscle isozyme of phosphofructokinase. Loss of these mRNAs following doxorubicin or daunomycin is evident as early as 2 h and precedes significant reduction of intracellular ATP. ATP levels in control cardiomyocytes (17.9 ؎ 2.9 nM/mg of protein) fall only after 14 h and reach residual levels of 10.4 ؎ 0.9 nM (doxorubicin; p ‫؍‬ <0.006) and 6.7 ؎ 1.9 nM (daunomycin; p ‫؍‬ <0.001) by 24 h. Loss of mRNAs generating ATP was highly selective since mRNAs for other energy production enzymes, (cytochrome c, cytochrome b, and malate dehydrogenase), and genes important in glycolysis (pyruvate kinase and glyceraldehyde-3-phosphate dehydrogenase) were unaffected even at 24 and 48 h. The drugs had no effect on levels of ubiquitously expressed RISP mRNA in fibroblasts. These findings could link doxorubicin-induced damage to membranes and signaling pathways with 1) suppression of transcripts encoding myofibrillar proteins and proteins of energy production pathways and 2) depletion of intracellular ATP stores, myofibrillar degeneration, and related cardiotoxic effects.
We have sought to explain the selective and cumulative dose-dependent cardiac toxicity of doxorubicin (Dox), 1 an anthracycline drug widely used for the treatment of various cancers. Biochemical and physiological data favor the hypothesis that the general antineoplastic cytotoxicity of anthracyclines stems from the formation of reactive free radical species whose intermediates interact with and damage cellular membranes (1). The interactions of free radicals, including Dox-Fe 3ϩ , with mitochondrial membranes and proteins of the respiratory chain (2, 3) may also account for the observed depletion of ATP in the hearts of Dox-treated animals (4,5). While such events also clearly progress in myocardium, they do not explain the cardiospecific tissue toxicity of the anthracycline drugs that leads to the irreversible degenerative cardiomyopathy (6) and limits the amount of the drug that can be administered.
Our previous work (7-9) has focused on alternative mechanisms that might explain the cardioselective toxicity. Dox alters transcriptional events specific to the myocardium. Initially we demonstrated that Dox selectively inhibits the accumulation of mRNA of heart-specific genes in cardiomyocytes (10) and inhibits the transcription of muscle-specific genes in skeletal muscle cells without affecting the transcription of housekeeping genes (7). Our subsequent studies showed that Dox blocks such transcription at least in part due to the transcriptional induction of a dominant negative regulator, Id (11), and possibly also by inhibiting the DNA binding capacity of transcription factors active in cardiomyocytes such as MEF2 (11). Id can form heterodimers through its helix-loop-helix motif with E proteins (which are ubiquitously expressed basic helix-loop-helix proteins) and thus prevent the formation of functional E proteins heterodimers (12). More recently we demonstrated by several criteria that Doxinduced protein kinase-mediated phosphorylation is involved in myofibrillar degeneration (13) and that the activation of specific regulatory elements in the Id promoter by Dox is mediated through novel Dox-induced kinase signaling pathways independent of Ca 2ϩ /phospholipid-dependent protein kinase (protein kinase C), cyclic AMP-dependent protein kinase (protein kinase A) and tyrosine kinase (9).
Although the generation of free radicals by Dox occurs in all cell types, it might yet play a critical role in initiating these cardiac-sensitive transcriptional events. This form of oxidative stress alters the structure and function of lipid components that act as precursors of second messengers (14). In this way Dox could indirectly effect the activation of Ca 2ϩ /phospholipiddependent protein kinase C and the regulation of the concentration of free cytosolic Ca 2ϩ (15)(16)(17), thus interfering with many inositol phospholipid-mediated signaling events including the alteration of tissue-specific transcriptional programs. In our search for additional cardiocyte expression programs that might be selectively disrupted by Dox, we have discovered that within several hours Dox dramatically and selectively suppresses the expression of several nuclear genes encoding enzymes that play central roles in normal energy production. One of these genes, ADP/ATP translocase, is a heart-and muscle-specific isoform. A second encodes the Reiske iron-sulfur protein, a ubiquitously expressed component of the electron transport chain. A third mRNA, encoding the critical metabolic enzyme 6-phosphofructo-1-kinase, is a muscle isozyme (M), which is the most abundant isoform in heart muscle (18).

MATERIALS AND METHODS
Cell Culture-Neonatal rat cardiac myocytes were prepared and cultured using a previously described protocol (19). After the cells started beating 24 h post-culture, they were treated either with 1 M Dox or 1 M daunomycin (Dn) for 24 h.
mRNA Differential Display-The differential display was carried out using an RNA map TM kit (GeneHunter Corporation) following the supplied protocol (20). The PCR reactions were performed for 40 cycles of 30 s at 94°C, 40°C for 2 min, 72°C for 30 s, followed by 5 min of a 72°C elongation step. The PCR products were analyzed on a 6% sequencing gel. The differentially expressed bands were cut from the dried gel and eluted. The eluted DNAs were amplified using the same primers and PCR conditions used for initial amplification except the dNTP concentration was higher (20 M). The PCR products were cloned using the TA cloning vector (Invitrogen), and the inserts from the colonies thus obtained were used as probes in Northern blot analysis.
Northern Blot Analysis-Total RNAs isolated using TRIzol reagent (Life Technologies, Inc.) from drug-treated and untreated cardiomyocytes were separated in 1.2% formaldehyde-agarose gels and transferred to Hybond N ϩ nylon membranes (Amersham Corp.) as described previously (7). The probes for cardiac ␣-actin, cardiac troponin I, cardiac troponin T, and pyruvate kinase were from our laboratory (7). A plasmid containing a partial cDNA for cytochrome c (21) was provided by Dr. R. C. Scarpulla (Northwestern University Medical School, Chicago). The 1.9-kilobase pair cDNA was excised using BamHI and KpnI and used as for radiolabeling. Malate dehydrogenase cDNA (22) was provided by Dr. A. W. Strauss (Washington University, St. Louis). An antisense oligonucleotide (ATTA-ATTTCATTTTAACATTTTTGTGTTCAACAAT) of 35 bases was synthesized spanning the last 11 C-terminal amino acids of cytochrome b (23). This oligo was end-labeled using polynucleotide kinase and [␥-32 P]ATP. The end-labeled probe was used for Northern blot hybridization. Phosphofructokinase cDNA (24) was provided by H. Nakajima (Osaka University Medical School). A 2.7-kilobase pair BamHI-XbaI fragment was used as a probe. The cDNA inserts excised from the TA clones were labeled with [␣-32 P]dCTP by random primer DNA-labeling kit (Boehringer Mannheim). The membranes were prehybridized and hybridized as described earlier (7), washed in 1% SDS, 50 mM NaCl, and 1 mM EDTA three times at 55°C for 15 min each before drying and exposure to x-ray film for autoradiography.
Nucleotide Sequencing and Similarity Searches-Double-stranded sequencing of the positive clones were carried out using Sequenase 2.0 kit (U. S. Biochemical Corp.). Nucleotide and predicted amino acid sequence searches were performed using BLAST searches (GenBank TM , at NCBI).
Measurement of ATP Concentrations-Control cells and Dox-or Dntreated primary rat neonatal cardiomyocytes were harvested in cold 5% trichloroacetic acid. The cell extract was used to measure the ATP content by the luciferin-luciferase bioluminescence kit (Sigma) following the manufacturer's protocol. The values are expressed in nanomoles/mg of protein.

Inhibition ADP/ATP Translocase and Reiske Iron-Sulfur
Protein (RISP) Genes-Based on our hypothesis that Dox may inhibit genes vital for the myocardium we used an mRNA differential display technique to identify Dox-repressed gene transcripts in cardiomyocytes. We compared the RNAs derived from control neonatal rat cardiomyocytes with those after exposure to 1 M Dox for 24 h. We also extracted RNA from cells that had been exposed to 1 M Dn, a Dox derivative also known for its cardiotoxicity. Reverse transcription was done using oligo(dT) primers followed by PCR using arbitrary oligos containing 10 nucleotides. Many different combinations of oligo(dT) and arbitrary primers were used to cover a large number of messenger RNAs. Since Dox and Dn are likely to have an identical mechanisms of cardiotoxicity, we scored as positive only those PCR products that were similarly altered in the Doxand Dn-treated cells. We were able to isolate PCR products from eight bands in the control cardiomyocyte sample that were differentially absent or significantly repressed in the Doxand Dn-treated samples. A representative differential display is shown in Fig. 1A. The bands were cut out, reamplified using the same set of primers, and subcloned. Seven of the eight clones showed repressed or absent mRNAs in RNA from Doxand Dn-treated cells. Three of the partial cDNAs were overlapping segments and contained nucleotide sequences identical to a heart-specific isoform of ADP/ATP translocase (ANT1) (25). One other cDNA was identical with RISP (26). The arrows in Fig. 1A point to the PCR bands that gave rise to ANT1 and RISP cDNAs. Three other PCR products did not share significant homology with any sequences present in the DNA data base (data not shown). The ANT1 and RISP mRNAs were severely reduced in the anthracycline-treated cells (Fig. 1, B and C). Northern blot analysis of RISP, using the 3Ј-UTR of the partial cDNA, consistently revealed two different bands, both of which are repressed during Dox treatment. The lower band (1.2 kilobase pairs) corresponds in size to the published rat RISP mRNA (26). The upper band was not further investigated but could be an alternatively spliced isoform of RISP.
Time Course Analysis of Dox-inhibited ANT1 and RISP Expression Levels-Our initial RNA analyses examined mRNAs after 24 h of Dox/Dn exposure. To further understand the kinetics of the effect of Dox on the expression of ANT1 and RISP, we isolated myocardiocyte RNA at frequent intervals of anthracycline exposure. Northern blot analysis (Fig. 2) shows that both ANT1 and RISP transcripts rapidly fall beginning as early as 2 h after Dox exposure. Low level expression of ANT1 mRNA persists but by 24 h it is barely detectable. The upper band of RISP mRNA is not detectable after 4 h of Dox treatment. A housekeeping gene, cyclophilin, known not to be affected by Dox, was used as an internal control for the amount of RNA loaded. In Fig. 2, differences in RNA loading or transfer account for the apparent drop of cyclophilin mRNA in lanes representing the 24-and 48-h Dox treatment. The relative levels of ANT1 and RISP mRNAs is presented in Fig. 3 as a function of time of exposure to Dox. These results strongly suggest that ANT1 and RISP gene expression are highly sensitive to Dox treatment. If the down-regulation of these mRNAs is the result of transcriptional modulation, then we can also infer that both genes are early rather than late targets of Dox.
It was perhaps not unexpected that the ADP/ATP translocase gene, a cardiac-specific isoform, was down-regulated since all cardiac-specific genes (but not housekeeping genes) that we have examined to date are down-regulated by Dox (7, 9 -11). The reduced expression in cardiomyocytes of RISP, a ubiquitously expressed gene, represents, however, the initial exception to this rule. To determine whether the Dox sensitivity of RISP is also manifest in noncardiac cells, we carried out Northern blot analyses on RISP mRNA from both Hela and 10T1/2 cells with and without exposure to 1 M Dox. Neither RISP not ANT2 mRNAs (the cross-hybridizing, non-muscle equivalent of ANT1) are effected by Dox in these cell types (data not shown). We conclude that the cellular mechanisms responsible for Dox modulation of RISP mRNA expression is myocardiocyte-specific.

Effect of Dox on the Total Cellular ATP Content and Contraction of Cardiomyocytes-Previously published reports have
shown that the ATP content in the heart is depleted after treatment of animals with Dox (6,27). One simple observation suggests that the depletion of mRNAs precedes physiologically significant depletion of ATP stores. Under the experimental conditions and drug concentrations we used, cardiomyocytes continued to beat for 16 -20 h after drug exposure suggesting that there were adequate intracellular energy stores. To more accurately compare the time courses of mRNA fall and ATP content, we measured the total ATP content of the cardiomyocytes after exposure of 1 M Dox or 1 M Dn at various time points up to 48 h. The result of three separate experiments (Fig. 3B) shows that ATP levels remain elevated despite drug exposure up to 14 h, well past the onset of significant reductions of mRNA levels at 2 h (Fig. 3A). ATP levels appear to stabilize after 24 h. Since the fall in specific transcript levels is detected almost immediately following anthracycline drug exposure, and since intracellular energy stores continue to be adequate for some time, it is likely that the repressed expression of key enzymes involved in ATP synthesis such as RISP, phosphofructokinase (PFK), and ANT1 may contribute to this reduction in ATP levels.

Effect of Dox on Other Members of Cytochrome bc 1 Electron Transport Complex and Other Cytoskeletal and Mitochondrial
Genes-The RISP is a crucial component of the cytochrome bc 1 complex made up of 10 -11 subunits (28): cytochrome c, cytochrome b, two core proteins, RISP, a ubiquinone-binding protein, and four to five low molecular weight proteins. The ironsulfur centers of RISP are essential for the electron transport phenomenon (28,29). Although only the RISP moiety of the cytochrome bc 1 complex was identified on our differential display screen it remains possible that mRNAs for the other members of this complex or mRNAs for proteins that play key roles in ATP production are similarly effected by anthracycline drugs. Accordingly we analyzed the expression of a number of other mRNAs including cytochrome c, cytochrome b, and malate dehydrogenase and genes important in glycolysis (phosphofructokinase, pyruvate kinase, and glyceraldehyde-3-phosphate dehydrogenase). The results (Fig. 4A) show that cytochrome b and cytochrome c are not affected by Dox or Dn, nor are pyruvate kinase, glyceraldehyde-3-phosphate dehydrogenase, or malate dehydrogenase. However, PFK mRNA appears to be inhibited by both Dox and Dn. The previously reported suppressive effects of Dox on cardiac cytoskeletal mRNAs, such as cardiac ␣-actin, cardiac troponin I, and cardiac troponin T, are included in the figure for comparison. PFK mRNA is highly sensitive to Dox. As early as 2 h after Dox exposure, PFK mRNA levels are highly repressed (Fig.  4B). The band intensities in Fig. 4B were scanned and plotted as a function of hours of exposure to Dox (Fig. 3). RISP and PFK mRNAs have an apparent half-life of less than 5 h and for RISP mRNA the half-life is less than 15 h. DISCUSSION Doxorubicin is a highly effective cytotoxic antineoplastic agent used in a wide variety of malignancies. Its use is limited because of a cumulative, dose-dependent irreversible cardiomyopathy observed in patients who receive a dose of greater than 550 mg/m 2 (6). Previous reports revealed that the level of ATP in the perfused heart is reduced following Dox treatment (4, 27, 30 -33). Isolated mitochondria from Dox-treated hearts also showed uncoupling of oxidative phosphorylation (6). However, the molecular mechanisms that lead to reduced levels of ATP are unclear.
Following exposure to Dox there are rapidly reduced levels of mRNAs for key enzymes responsible for fundamental energy production in myocardiocytes, including ANT1, RISP, and PFK. Although the loss of templates for sarcomeric proteins that we have reported earlier may well contribute to the myofibrillar degeneration that is a hallmark of Dox-induced cardiotoxicity (7,9,11), the rapid changes in ATP production that would follow from reduction of these key metabolic enzymatic activities might well be more central to the effects of the anthracycline drugs. The depletion of cytosolic ATP would likely lead to more serious consequences in the heart compared to most other organs since the heart needs a continuous supply of ATP for its uninterrupted contraction-relaxation cycle. This could be one of the reasons why Dox damages the heart while other organs are relatively unaffected.
A less likely possibility is that Dox damages the mitochondria and depletes ATP stores and subsequent feedback mechanisms repress expression of the nuclear genes we have evaluated. We are unaware of any studies that demonstrate a negative feedback by depletion of mitochondrial energy stores on nuclear-encoded genes involved in energy production. Furthermore, the kinetics of our observations strongly suggests that the transcriptional effects of the drugs precedes significant effects on intracellular energy stores. Whether we consider as evidence the persistence of cell beating or of ATP levels, it is most likely that the effects of the drugs on intracellular energy stores at least follow their impact on transcription of specific genes. Finally, the transcriptional inhibition appears to be highly selective, since other nuclear genes that encode components of the electron transport chain are unaffected. If the mechanism of transcriptional repression were a feedback secondary to general depletion of energy stores or nonselective mitochondrial damage, a more general effect on gene expression might be expected.
Several alternative mechanisms may explain the selective effects of Dox on gene expression in the heart. Dox generates free radicals that lead to lipid peroxidation and membrane instability including leakage of calcium by the sarcoplasmic reticulum (1, 2, 14, 34 -36). Such events may directly impact intracellular signaling pathways that initiate in the membrane and also involve intracellular calcium homeostasis (1, 30, 37), but these would not necessarily be cardiac-specific effects. However, the up-regulation of dominant transcriptional repressors such as Id or interference with the function of critical cardiac transcriptional activators such as MEF2 (1,7,9,11) might play a key role in the myocardiocyte response to Dox. Since ANT1 is a cardiac-specific isoform, its repression would be cardiac-limited.
A number of genes expressed both in the myocardium and in skeletal muscle have cardiac-and muscle-specific transcription regulatory elements (38,39). It will be of interest to determine whether the RISP or PFK promoters contain two sets of cisacting regulatory elements, those that interact with ubiquitous transcription factors and those that are active only in myocardium. In the event, interference by Dox with such cardiac regulatory effectors would selectively reduce only cardiac expression of the RISP or PFK genes. Alternatively we have not ruled out the possibility that the RISP or PFK mRNAs represent cardiac specific transcripts.
ADP/ATP translocase, a key protein involved in aerobic energy metabolism in every cell type, is located in the inner mitochondrial membrane. There are three known isoforms, ANT1, ANT2, and ANT3, each encoded by a nuclear gene (40,41). Based on the energy requirements, the expression patterns of the different isoforms varies leading to alterations in the ADP/ATP transport functions. For example, the ANT expression patterns are different in liver and heart mitochondria, as are the kinetics of ADP/ATP exchange, and also vary under changing conditions of energy demand (42,43) or at different stages of human muscle development (40 -43).
ANT1 is a heart-and skeletal muscle-specific (40,41) nuclear-encoded mitochondrial isoform. ANT1 translocates newly synthesized ATP from inner side of the mitochondria to the cytosol, and ADP from cytosol to the inner side of the mitochondria (44). This polypeptide constitutes almost 10% of the total inner mitochondrial membrane protein, and its function is  (13) involved in the induction of gene expression including Id (8,9). Id induction is likely responsible for the Dox-induced suppression of tissue-specific cardiac genes including those encoding contractile proteins (7,10). Since the ANT1 and PFK genes are cardiac-specific isoforms (40,41,58), we speculate that Id may directly squelch their expression as well. The mechanism suppressing RISP gene expression is unknown at this time. Suppression of cardiac-specific contractile protein genes would lead to defective formation or maintenance of the contractile apparatus. Alternatively or simultaneously, Dox can inhibit ATP production by destruction of the electron transport system, which affects virtually all energy-dependent cellular functions. Both pathways would lead to an impaired contraction-relaxation function of cardiomyocytes. Open arrows represent events associated with suppression of activities. critical in maintenance of cytosolic ATP concentration (44).
RISP is also a nuclear-encoded mitochondrial protein (29,45). RISP contains a [2Fe-2S] cluster, an integral part of the cytochrome bc 1 complex, essential for the electron transport system in mitochondria and for the synthesis of ATP (29,45). The cytochrome bc 1 complex could not perform its function without RISP because of the lack of iron-sulfur reaction centers (28,(45)(46)(47)(48). Repression of RISP would disrupt the electron transport chain and drastically reduce ATP production. Within a matter of a few hours Dox rapidly depletes the myocardiocyte, but not HeLa or 10T1/2 cells, of the nuclear-encoded RISP mRNA, and by 24 h there is no trace of residual messenger. It appears that only the transcriptional program that regulates RISP in cardiomyocytes but not in non-muscle cells is sensitive to Dox.
PFK, the third mRNA we found to be selectively depressed by Dox treatment of cardiocytes, encodes a key enzyme in the glycolytic pathway (49). Clinical PFK deficiency is often associated with severe myopathic symptoms attesting to its role in maintaining energy stores (50 -52). There are PFK isoforms in red blood cells, brain, and muscle (50,(53)(54)(55). The PFK muscle M isoform examined here is the most abundant isoform in heart (18) but is also expressed in non-muscle tissues (56,57).
Depletion of any one of these three proteins would likely have severe consequences on ATP production in cardiomyocytes. Dox exposure clearly led to a significant reduction of ATP content (Fig. 3B). As argued earlier, we infer that the downregulation of these three mRNAs and, presumably, the proteins they encode, is responsible, at least in part, for the subsequent depletion of ATP. The heart in particular has a continuous need for ATP and energy. Depletion of cytosolic ATP eventually leads to diminution of the strength of cardiac contraction-relaxation. Moreover, the assembly and maintenance of myofilaments is dependent on ATP. Thus, in addition to the downregulation of contractile protein gene expression by Dox, depletion of cellular ATP stores may also directly contribute to the myofibrillar degeneration associated with Dox cytotoxicity.
Although it has been known that the ATP content in the whole heart is depleted after treatment of animals with Dox (6,27), no study has yet distinguished whether the effect of Dox is on cardiomyocytes or cardiac fibroblasts. One outcome of our studies clearly establishes that Dox or Dn treatment is associated with a decrease of ATP in cardiomyocytes.
Do these observations allow us to propose a more unified hypothesis regarding the mechanisms of Dox cardiotoxicity? In Fig. 5 we have attempted to represent some testable connections between these complex observation. Generation of free radicals and subsequent peroxidation of lipid membranes by Dox could be responsible for most if not all subsequent events. The link between membrane damage and subsequent perturbations of intracellular second messenger signaling pathways remains a likely possibility. Such events could well be responsible for the transcriptional events we have observed including the induction of the negative regulator Id and the subsequent suppression of tissue-specific transcript including those encoding myofibrillar proteins and proteins of the key energy production pathways described here. These events as well as direct membrane injury to mitochondria could both contribute to the depletion of intracellular ATP stores, myofibrillar degeneration, and the subsequent cardiotoxic effects.