Cardiovascular Toxicology (2005) 05 269–283 $30.00 (http://www.cardiotox.com) Mechanisms of Doxorubicin Cardiotoxicity Humana Press © Copyright 2005 by Humana Press Inc. All rights of any nature whatsoever reserved. 1530-7905/01

269

Doxorubicin Cardiac Dysfunction Effects on Calcium Regulatory Proteins, Sarcoplasmic Reticulum, and Triiodothyronine Richard D. Olson,1,2,3,* Hervé A. Gambliel,1,2 Robert E. Vestal,1,2,3 Susan E. Shadle,2,3,4 Henry A. Charlier, Jr.,2,4 and Barry J. Cusack1,2,3 1

Pharmacology and Gerontology Research Unit, Veteran’s Affairs Medical Center, 500 W. Fort St., Boise, ID 83702; 2Mountain States Tumor and Medical Research Institute, 100 E. Idaho, Boise, ID 83712; 3 School of Medicine, University of Washington, Seattle, WA 98195; 4 Department of Chemistry, Boise State University, Boise ID 83702

Abstract

*Author to whom all correspondence and reprint requests should be addressed: Richard D. Olson, PhD, Research Service (151), VA Medical Center, 500 West Fort Street, Boise, ID 83702. E-mail: [email protected] Cardiovascular Toxicology, vol. 5, no. 3, 269–283, 2005

Cardiovascular Toxicology

Utilizing a model of chronic doxorubicin cardiomyopathy, this study examines the relationship between changes in expression and function of calcium handling proteins and contractile dysfunction. A possible mechanism to account for this relationship is suggested. New Zealand white rabbits were injected with either doxorubicin (1 mg/kg, twice weekly for 8 wk) or 0.9% NaCl. Gene transcript, protein levels, and the function of several proteins from the left ventricle were assessed. Protein levels of sarcoplasmic reticulum (SR) Ca2+ transporting ATPase (SERCA2a and b), Ca2+ release channel (RYR2), calsequestrin, Na/Ca exchanger, mRNA levels of RYR2, and [3H]-ryanodine binding (Bmax) to RYR2 were significantly decreased in doxorubicin-treated rabbits; protein levels of phospholamban, dihydropyridine receptor α2 subunit, and SR Ca2+ loading rates were not decreased. However, only protein levels of SERCA2 and RYR2, mRNA levels of RYR2, and Bmax of RYR2 significantly regressed with left-ventricular fractional shortening. Analysis of contractile function of atrial preparations isolated from doxorubicin-treated rabbits revealed that doxorubicin diminished contractility (dF/dt) of rest-potentiated contractions consistent with SR dysfunction. Serum concentrations of free triiodothyronine (T3) decreased in doxorubicin-treated rabbits. Our results suggest that chronic doxorubicin administration in the rabbit causes a SR-dependent contractile dysfunction that may result, in part, from decreased T3. Key Words: Doxorubicin; cardiotoxicity; echocardiography; calcium; triiodothyronine; rested contractions.

Introduction Chemotherapy of neoplastic disease with doxorubicin incurs a significant risk of cumulative dose-related cardiotoxicity (1,2). This serious side effect limits the clinical utility of doxorubicin and has spurred numerous investigations designed to

269 Press Humana

Volume 5, 2005

270

Olson et al.

determine the mechanism. Although several hypotheses, including impairment in Fe metabolism and reactive oxygen injury, have been suggested to mediate the cardiotoxicity (3,4), additional data suggests that doxorubicin administration also perturbs cardiac Ca2+ homeostasis (5–10). However, the primary intracellular target(s) responsible for impaired calcium handling remain enigmatic. Olson and co-workers (5) and Wang and co-workers (10) demonstrated that the primary metabolite of doxorubicin (doxorubicinol) impaired Ca ATPase activity (SERCA2) and sarcoplasmic reticulum (SR) Ca2+ handling in isolated cardiac membrane preparations or patch clamped myocytes isolated from guinea pig heart. Dodd et al. (6) found that the density of the cardiac SR Ca release channel (ryanodine receptor; RYR2) was reduced by 50% in a chronic rabbit model of doxorubicin cardiotoxicity, suggesting that the SR might be an important target in doxorubicin-induced disruption of cardiac Ca2+ homeostasis. Arai and coworkers (7) demonstrated that chronic doxorubicin administration in rabbits downregulated the gene expression of cardiac RYR2, SERCA2, phospholamban, and calsequestrin. Additional studies in-dicated that acute exposure to doxorubicin can induce transcriptional repression of SERCA2 within 24–48 h in rat neonatal myocytes, activate MAP kinases, and elicit repression via early growth response factor-1 (Egr-1) binding to its cognate promoter element (11). Although impairment of Ca2+ handling is implicated in doxorubicin cardiotoxicity, the mechanisms are unknown and it is still unclear which of the doxorubicin-induced changes in SR and sarcolemmal (SL) Ca2+ regulatory proteins best correlate with the development of left-ventricular dysfunction. To address these questions, in this study we used a chronic rabbit model of doxorubicin cardiomyopathy that exhibits a range of dysfunction. Our results indicate that the degree of doxorubicin cardiotoxicity was better predicted by RYR2 and SERCA2 than by other Ca2+ regulatory proteins, including phospholamban pentamer, dihydropyridine α2 subunit, calsequestrin, and Na/Ca exchanger. The best predictor of degree of cardiac dysfunction was the density of SR calcium release channels. Our data also demonstrate a decrease in free triiodothyronine (T3), a hormone that regulates expression and function of Ca2+ regulatory proteins (12–17).

Cardiovascular Toxicology

Materials and Methods Rabbit Model of Chronic Doxorubicin Cardiomyopathy Young adult male New Zealand white rabbits (3– 4 kg; Western Oregon Rabbit Company, Philomath, OR) were observed closely for 1 wk to ensure that they were in good health. The rabbits were randomly assigned to a doxorubicin-treated, or a pair-fed control group. Doxorubicin (1 mg/kg) was infused into a marginal ear vein over 5 min twice weekly for a total of 16 doses. Control animals received equivalent volumes of 0.9% NaCl (intravenously) according to the same schedule. Body weight (twice weekly) and food intake (daily) were monitored throughout the experiment. The weight of food consumed daily by each animal in the doxorubicin-treated group determined the amount of food given to paired control rabbits at the next feeding. Pair-feeding was carefully maintained to avoid effects due to nutritional differences between the two groups (18). Echocardiography was performed every 2 wk during the 8wk course of treatment and weekly thereafter. The endpoint of the study occurred when left-ventricular fractional shortening (LVFS) decreased below 30% or 12 wk after the last doxorubicin or vehicle injection (20 wk after beginning the study). Twelve weeks after the last doxorubicin or vehicle injection was chosen as the endpoint because LVFS stabilized and was no longer decreasing. In addition, previous experience demonstrated that LVFS 30%. Rabbits were sacrificed by captive bolt discharge in accordance with the Declaration of Helsinki and the Guide for the Care and Use of Laboratory Animals as adopted by the US National Institutes of Health (NIH Publication No. 85-23, revised 1996), and all animal experiments were approved by the Boise Veterans’ Affairs (VA) Institutional Animal Care and Use Committee (IACUC). Control rabbits were euthanized at the same time interval after the last saline injection as their paired doxorubicin-treated counterparts. Immediately, a median sternotomy was performed and the heart was removed. The heart was placed in Krebs bicarbonate buffer (25 mM NaHCO3, 127 mM NaCl, 2.5 mM CaCl2, 2.3 mM KCl, 1 mM KH2PO4, 5.6 mM glucose, pH 7.4; Sigma Chemical Company, St. Louis, MO) and the atria were removed

Humana Press

Volume 5, 2005

Mechanisms of Doxorubicin Cardiotoxicity

271

for functional studies. A section containing left- and right-ventricular free wall and septum was removed and processed for preparation of SR vesicles within 3 min after sacrifice. Fifty to one hundred milligrams of left-ventricular free wall was processed for RNA isolation. The rest of the left-ventricular free wall was frozen at −80°C until needed to assess protein levels.

Echocardiography Each rabbit was restrained supine on the counter and the chest area was shaved. The shaved area was coated with transducer gel to enhance ultrasound conduction. With the animal resting calmly, echocardiography was performed using a 7.5-MHz probe over the right lower parasternal area to permit visualization of the left ventricle. Using an Advanced Technology Laboratories (Bothell, WA) Ultramark 4 Ultrasound system, the left ventricle was imaged by two-dimensional echocardiography in planes parallel to the long axis and the minor (short) axis at a level just proximal to the mitral valve. The leftventricular dimensions from the left-ventricular free wall to the septal endocardium were measured in Mmode during systole and diastole. Three measurements (two in the short axis and one in the long axis) at peak systole and diastole were obtained to calculate LVFS. Reported individual values represent the average of these three measurements.

RNAs and RNase Protection Assays Fifty to one hundred milligrams of left-ventricular myocardium per animal was homogenized (3 × 30 s) in Chaosolv reagent (Ultraspec RNA reagent, Biotecx, Houston, TX) using a Brinkman polytron homogenizer and processed according to the protocol provided by the manufacturer (19). Total RNA was precipitated in ethanol and stored at −80°C until needed. Prior to RNase protection analysis, RNA was quantified spectrophotometrically at 260 nm. For each sample, the integrity of 18S and 28S RNA bands was verified by agarose electrophoresis. The rabbit sarcoplasmic reticulum Ca2+ transporting ATPase 2 (SERCA2) cRNA was derived from a subclone (pSercaBP800, +1925 to +2750, codon 782877) of a plasmid provided by Dr. D.A. McLellan (Charles H. Best Institute, University of Toronto,

Cardiovascular Toxicology

Ontario, Canada). Digestion of this template with Bsty 1 and in vitro synthesis from the T3 promoter yielded a 260-nt cRNA probe that is detected as a 190-nt (codon 782- 877) fragment in RNA protection experiments. The rabbit cardiac calcium channel (RYR2) cRNA was synthesized from a partial cDNA (+8286 to +10376), also provided by Dr. McLellan, that had been linearized at Nco 1. In vitro transcription of this template yields a 138-nt radiolabeled cRNA probe using the T7 promoter and an 84-nt protected cRNAs (codon 3334-3349) in protection experiments. 32P-cRNA probes were synthesized from linearized plasmids using an in vitro transcription reagent kit from Stratagene (La Jolla, CA). The probes were then treated with 1 U RQ1 DNAse (Promega, Madison, WI), extracted with phenol and chloroform, ethanol precipitated, electrophoresed, and eluted from a nondenaturing 6% polyacrylamide gel according to standard procedures (20). RNase protection assays were performed using kit reagents from Boehringer Mannheim (Indianapolis, IN). 32P-cRNAs (300,000 cpm/probe) were hybridized for 16 h at 45°C with indicated amounts of total RNA. RNase digestion conditions were as recommended by the manufacturer. Protected RNA fragments were ethanol precipitated, denatured, and resolved on a 6% polyacrylamide gel containing 8 M urea. After drying under vacuum, the gel was subjected to autoradiography. Protected cRNA fragments were then quantified by densitometry and normalized to 28S RNA using ImageQuant software on a Molecular Dynamics (Sunnyvale, CA) scanning instrument.

Sarcoplasmic Reticulum Vesicles SR membranes were prepared according to a modification of Harigaya and Schwartz (21). Tissue was homogenized once in 10 mM Tris-Maleate pH 6.8, 137 mM NaCl using a Polytron (Brinkman). The homogenate was centrifuged at 4000g. The supernatant was passed through four layers of cheesecloth and centrifuged again at 8000g for 20 min at 4°C. This supernatant was then centrifuged at 45,000g for 30 min at 4°C. Microsomes were resuspended lightly in a 1-mL glass homogenizer in 10 mM Tris-Maleate pH 6.8, 137 mM NaCl containing 0.3 M sucrose and frozen in aliquots in liquid N2 until needed.

Humana Press

Volume 5, 2005

272

Olson et al.

Measurement of [3H] Ryanodine Binding [3H]-ryanodine

SR membranes (50 µg) and (0, 0.11, 0.47, 0.94, 1.25, 2.5, 5, 7.5, 10, 15, 20, 25, or 30 nM; Perkin-Elmer, Boston, MA) were incubated in duplicate with shaking at 37°C for 2.5 h in buffer containing 40 mM MOPS pH 7.4, 0.4 mM CaCl2, 1 M NaCl, 0.2 mg/mL bovine serum albumin (BSA), and 10 mM caffeine (total volume of 225 µL). The samples were filtered on Whatman GF/F membranes in a Brandel harvester (Gaithesburg, MD), rinsed four times with 5 mL of ice-cold H2O, dried, and counted in the presence of 4.5 mL Beckman Ready Protein Scintillation cocktail (Beckman Coulter, Fullerton, CA) in a Wallac 1410 liquid scintillation counter. Nonspecific binding was determined in the presence of 100-fold excess of unlabeled ryanodine. Estimates for Bmax and Kd were obtained from fits of the data to a single site hyperbolic binding model using Prism 3.0 (San Diego, CA).

Calcium Uptake Assays Calcium transport across SR membranes vesicles were assessed using a UV-Vis photodiode array detector (Hewlett Packard model 8453, Palo Alto CA) as described by Palade (22). Three hundred eighty micrograms of SR membranes were incubated at 32°C in 50 mM KH2PO4, 20 mM MOPS, 5 mM KCl, 2 mM MgCl2 and 0.3 mM of antipyrylazo III in the presence of 10 µM ruthenium red. The reaction was started by addition of 25 nmoles of calcium chloride followed within 15–20 s by 1 mM MgATP (pH 7.0). [Ca2+] outside the microsomes was monitored by measuring the absorbance difference (A710–A790) of antipyrylazo III every 0.5 s.

Protein Lysates and Western Analysis For western analysis, left-ventricular samples (0.3– 0.5 g) were homogenized with a polytron in 4 mL 10 mM NaHCO3 containing 10 µg/mL leupeptin and 1 mM phenylmethylsulfonyl fluoride (PMSF). One milliliter of extract was solubilized by addition of 2 mL 4X sodium dodecyl sulfate (SDS) polyacrylamide gel electrophoresis (PAGE) protein sample buffer (250 mM Tris, pH 6.8, 8% SDS [weight by volume], 40% glycerol, 40% β-mercaptoethanol, and 0.4% bromphenol blue) and centrifuged at 12,000g for 10 min at 4°C. Protein concentrations were determined using the BCA protein assay kit (Pierce Biochemicals, Rockford, IL). Cardiovascular Toxicology

Lysates (30 µg) were electrophoresed on 4% (RYR2) or 10% (SERCA2) denaturing gels (Mini Protean II apparatus, Bio-Rad, Hercules, CA) and proteins were either transferred to PVDF membranes (Semi-Dry transfer blot, Bio-Rad) or stained with Coomassie blue. For Western analysis, membranes were blocked in Tris-buffered saline (TBS) containing 0.5% Tween 20 and 5% fish gelatin (Sigma). After incubation for 1 h at 4°C with the appropriate primary antibody, the membranes were washed out three times for 15 min in wash buffer (TBS containing 0.2% Tween 20 and 0.2% fish gelatin). The membranes were then incubated with horseradish peroxidase-conjugated goat anti-mouse antibody (Sigma Corp., St. Louis, MO) at 1:30,000 dilution for 1 h at 4°C and then washed twice for 5 min and twice for 15 min with agitation at room temperature prior to chemiluminescent detection using the ECL Western blotting detection reagents (Amersham Bioscience, Buckinghamshire, England). Quantitative estimates of protein levels were performed using densitometric scanning of appropriate film exposures.

Antibodies Mouse monoclonal antibodies anti-SERCA2 (clone 2A7-A1), anti-RYR2 (Clone 34C), anti-dihydropyridine receptor α2 subunit (clone 20A), anti- Na/Ca exchanger (cloneC2C12), and anti-phospholamban (clone D2D12, MA3-922) were purchased from Affinity Bioreagents (Golden, CO).

Triiodothyronine and Thyroid-Stimulating Hormone Assays Free T3 and thyroid-stimulating hormone (TSH) levels in rabbit serum were determined by Anilytics, Inc. (Gaithesburg, MD) using solid phase radioimmunoassays. The TSH assay uses a polyclonal Ab and calibrators from Diagnostics Products Corp (Los Angeles, CA).

Atrial Contractility The left atrium of each rabbit was cut in half at sacrifice. The atrial strips were attached to an isometric force transducer in a thermojacketed (30°C) muscle bath containing 25 mL of Krebs-bicarbonate buffer continuously bubbled with 95% oxygen, 5% carbon dioxide. Atrial strips were then electrically stimulated (S88 stimulator, Grass Medical Instruments, Quincy, MA, USA) at 1 Hz with square wave

Humana Press

Volume 5, 2005

Mechanisms of Doxorubicin Cardiotoxicity

273

pulses (3 ms in duration) at a voltage 10% above the threshold voltage of the muscle. Preparations were allowed to stabilize for 90–180 min. Atria were stretched to a resting tension of 0.5 g. After the muscles had stabilized (i.e., developed force and resting force had not changed for 30 min), left-atrial functional parameters were obtained. Variables measured at 1 Hz included the maximum rate of rise in force (dF/dt, g/sec) and 90% relaxation time (90% RT, ms). These parameters were recorded using a highspeed (100 mm/s) oscillographic recorder (Gould 4200S) and the data was analyzed with a pulsatile analyzer (Buxco Electronics Inc., Troy, NY). Cardiac variables were also determined on the first contraction following a 20-, 30-, and 60-s rest interval (PRP). Our data indicate isolated atria and papillary preps provide similar myopathic-state assessment in anthracycline cardiotoxicity.

Statistics The reported data are expressed as means ± S.E.M. unless otherwise indicated. The significance of differences in mean values between control and doxorubicin-treated rabbits was assessed using a two-tailed Student’s unpaired t-test correcting for the new α when multiple t-test were performed. Linear regression analysis was used to assess the possible relationships between mRNA expression, protein concentration, Bmax for ryanodine receptors, and left-ventricular fractional shortening at sacrifice. Doxorubicin-induced effects on rest-potentiated contractions were analyzed by one-way analysis of variance (ANOVA) using the Tukey post hoc analysis. p < 0.05 was considered statistically significant.

Results In this study, rabbits were injected with a chronic regimen of doxorubicin (1 mg/kg, twice a week into the marginal ear vein for 8 weeks) or matched to a pair-fed control group injected with 0.9% saline on the same schedule. LVFS was monitored every other week during the first 8 wk and weekly after the last doxorubicin or saline injection. At 20 wk after initiation of doxorubicin injection or when LVFS declined below 30% (whichever came first), each rabbit and its paired control were sacrificed for further studies. This dosing regimen produced a wide range of cardiac dysfunction in doxorubicin-treated rabbits. LVFS Cardiovascular Toxicology

ranged from 20% to 38% in the doxorubicin-treated rabbits at sacrifice with a mean LVFS of 29 ± 2% (Fig. 1 and Table 1). The decline in mean LVFS in doxorubicin-treated rabbits occurred between 2 and 6 wk after the last doxorubicin injection, and did not decline at all during the period of infusion, demonstrating the chronic nature of the doxorubicin effects on LVFS (Fig. 1). In contrast, pair-fed control rabbits receiving the doxorubicin vehicle did not exhibit a decrease in mean LVFS throughout the duration of the study (mean LVFS of 37 ± 1% at sacrifice; Table 1 and Fig. 1). Thus, this model produced a relatively mild cardiotoxic lesion, allowing detection of molecular and biochemical changes related more to doxorubicin administration than to generalized debilitation and end-stage cardiac failure. RNase protection analysis was used to measure mRNA levels of SR calcium release channel (RYR2) and SR calcium transporting ATPase (SERCA2) in the left ventricle of rabbits (Fig. 2). Doxorubicin treatment caused a significant decrease in RYR2 mRNA (145 ± 10 to 103 ± 15; p < 0.05) but did not significantly decrease SERCA2 mRNAs levels relative to control (103 ± 14 vs 83 ± 15; Fig. 2B,C and Table 1; p = 0.06). Further studies using larger sample sizes may help clarify whether there is a statistically significant decrease in SERCA2 mRNA by doxorubicin by providing greater statistical power. Western analysis of left-ventricular myocardial samples (Fig. 3) revealed that doxorubicin caused significant reductions in RYR2, SERCA2a, and SERCA2b proteins (471 ± 26 to 231 ± 58, 402 ± 25 to 266 ± 22, and 503 ± 21 to 298 ± 24, respectively; p < 0.01, Table 1). SERCA2a and SERCA2b are cardiac specific isoforms of the Ca-ATPase of SR (23). Chronic doxorubicin administration also significantly decreased protein levels of calsequestrin, the SR Ca+2 binding protein (791 ± 52 to 411 ± 77; p < 0.01), and the SL Na/Ca exchanger (NCX1; 107 ± 9 to 80 ± 6; p < 0.05), but not phospholamban pentamer or dihydropyridine α2 subunit (Fig. 4 and Table 1). Both mRNA and protein levels of SR targets were useful in predicting the degree of cardiac dysfunction. Protein levels of both SERCA2a and SERCA2b (r = 0.87, p < 0.025 and r = 0.70, p < 0.05, respectively) and RYR2 (r = 0.68, p < 0.05), and mRNA levels of RYR2 but not SERCA2 (r = 0.66, p < 0.05 and r = 0.36, NS, respectively) correlated significantly with the degree of cardiac dysfunction (measured by LVFS) induced

Humana Press

Volume 5, 2005

274

Olson et al.

Fig. 1. (A) Left-ventricular fractional shortening (LVFS) up to 20 wk after doxorubicin treatment in age-matched and pair-fed rabbits infused with vehicle (normal saline, N = 8, open circles) or doxorubicin (1 mg/kg, intravenously twice weekly for 8 wk [Doxirubicin infusion period], N = 8, closed circles). If rabbits were sacrificed before 20 wk, their values at sacrifice were retained in subsequent mean calculations of the group so that all values contained a mean of eight individual values in each group at each time. Values are means ± SEM. *p < 0.05 compared with controls. (B) Representative M-mode echocardiograms from a doxorubicin-treated rabbit (A-24) and the age-matched pair-fed control rabbit (C-24) 10 wk after initiation of doxorubicin or saline infusion. End-diastolic and end-systolic dimensions were 15 and 10 mm, respectively (LVFS = 33%) for C-24 and 19 and 15 mm, respectively (LVFS = 21%) for A-24. LVFS below 30% is abnormal in New Zealand white rabbits.

by doxorubicin (Fig. 5) in the doxorubicin treated rabbits. In contrast, protein levels of calsequestrin and Na/Ca exchanger did not significantly regress with LVFS (r = 0.53 and 0.62, respectively, NS). Studies using isolated SR vesicles in the rabbit (6) have suggested that chronic doxorubicin treatment elicits a persistent decrease of [3H]-ryanodine binding (Bmax) to cardiac SR Ca+2 release channels. We also found a significant decrease (p < 0.02) in the density of the ryanodine binding sites in doxorubicin Cardiovascular Toxicology

treated rabbits when compared to controls (Bmax of 0.29 ± 0.04 in the doxorubicin group vs 0.48 ± 0.06 pmol/mg in the control group; Table 1 and Fig. 6) . However, the Kd for ryanodine binding remained unchanged (1.26 ± 0.10 nM vs 1.29 ± 0.07 nM in the control group), suggesting that the decrease in ryanodine binding likely resulted from downregulation of RYR2 protein rather than an inherent change in the ability to bind ryanodine. The level of reduction of Bmax for RYR2 (59%, Table 1) is similar to that ob-

Humana Press

Volume 5, 2005

Mechanisms of Doxorubicin Cardiotoxicity

275

Table 1 Effects of Chronic Doxorubicin and Saline Administration on Molecular and Functional Cardiac Parameters in Rabbits at Sacrifice. Values Are Mean ± S.E.M. Treatment parameter

Control

Doxorubicin

Doxorubicin/control

P