点击显示 收起
School of Molecular Biosciences, Washington State University, Pullman, Washington (I.Y.P., E.J.K., H.P., K.F., C.K.)
Center for Computational Biology and Informatics and the Department of Molecular Biology and Biochemistry, Indiana University School of Medicine, Indianapolis, Indiana (K.A.D.)
Abstract
Ca2+ regulation is coupled to critical signals in eucaryotic cells, and calsequestrin is one of the crucial components for this calcium regulation. Our previous observations of calsequestrins revealed the existence of three thioredoxin-like folds, a basic motif that often provides the platform for small molecule binding. Therefore, we have examined the previously reported trifluoperazine and other pharmaceuticals that have similar heart-related side effects (such as tachycardia; bradycardia; palpitation; changing PR, QRS, QTc intervals in electrocardiogram; heart failure) for their binding affinity to cardiac calsequestrin (cCSQ) using isothermal titration calorimetry. Our results showed that several antipsychotic phenothiazine derivatives, tricyclic antidepressants, and anthracycline derivatives bind cCSQ with Kd in the micromolar range. For these compounds that have a significantly low Kd, their effect on Ca2+ binding capacity of cCSQ was checked using equilibrium dialysis and atomic absorption spectroscopy, which clearly showed a significant reduction in Ca2+ binding capacity of cCSQ as a result of this interaction. Furthermore, 8-anilino-1-naphthalene sulfonate (ANS) binding to cCSQ closely resembles ANS binding to flavine or nucleotide binding sites. The combination of this information with the high abundance of CSQ in SR and the high membrane permeability of those drugs led us to the specific hypothesis that there are undesirable and damaging interactions between cCSQ and tricyclic antidepressants, phenothiazine derivatives, anthracyclines, and many other pharmaceutical compounds and to the corollary hypothesis that better understanding of the molecular details of cCSQ-drug interactions could lead to modified drug molecules with reduced heart-related side effects.
Calsequestrin (CSQ) is a major Ca2+ storage protein within the sarcoplasmic reticulum (SR) of both cardiac and skeletal muscle. Muscle CSQ has high capacity and low affinity for Ca2+, binding about 40 to 80 ions per molecule with a binding constant of about 1 mM under physiological conditions (MacLennan and Wong, 1971; Park et al., 2004). Not only does CSQ act as a Ca2+ buffer inside the SR, lowering free Ca2+ concentrations and thereby facilitating further uptake by the Ca2+-ATPases, but also CSQ actively participates in muscle contraction by localizing Ca2+ at the release site and regulating the amount of Ca2+ released through the ryanodine receptor (MacLennan et al., 2002). Therefore, CSQ has been thought to be essential for a short contraction/relaxation cycle that can be completed within a time interval of less than 100 ms (Mitchell et al., 1988; Collins et al., 1990; Damiani and Margreth, 1990; Yano and Zarain-Herzberg, 1994; Jones et al., 1995, 1998; Zhang et al., 1997; Ohkura et al., 1998; Kobayashi et al., 2000; Terentyev et al., 2003; Tijskens et al., 2003).
The overall structures of both cardiac and skeletal CSQs (cCSQ, sCSQ) are composed of three thioredoxin-like domains, even though no tandem repeats were evident in their primary sequences (Wang et al., 1998; Park et al., 2004). Individual thioredoxin-like domains have a five-stranded -sheet sandwiched by four -helices, composed of 100 residues. Almost without exception, thioredoxin-like folds bind small molecular ligands or have a redox site at a specific locus (Branden, 1980; Katti et al., 1990). These sites occur at chain reversals that generate a crevice defined by the edge of one -sheet and the carboxyl ends of the adjacent 2 and 3 strands (Branden, 1980). Not only are these sites in both CSQs predicted to be binding sites from the topology diagrams, these sites also form hydrophobic grooves that are bound by exposed hydrophilic side chains, with considerable structural similarity to binding sites in other open /-sheet structures.
One of the phenothiazine derivatives, trifluoperazine (TFP), which is a potent inhibitor of calmodulin (CaM), has been shown to bind to a particular sequence on CaM [TFP-binding site or serine acidic hydrophobic site (SAH)], thereby inhibiting interactions between the CaM site and a dibasic hydrophobic (DBH) site on the target enzyme (Levin and Weiss, 1977; Cachia et al., 1986; He et al., 1993; Wang et al., 1998). It is now known that TFP blocks the SAH-DBH interaction by binding to a hydrophobic cleft on CaM, but not with the SAH peptide, as originally thought (Cook et al., 1994). Because the primary sequences of both cCSQ and sCSQ contain sequences similar to the SAH and DBH sites, it was suggested that there is a potential interaction between the SAH and DBH sites with unknown biological significance (Mitchell et al., 1988; He et al., 1993). TFP binding might interfere with an interaction between these two sites and thereby alter the function of CSQ (Fliegel et al., 1987). In fact, later work indicated that TFP interacts with sCSQ by an unknown mechanism, and such ligand binding evidently changes aggregation and Ca2+ binding properties of sCSQ (He et al., 1993).
In agreement with the prediction, the crystal structure reveals an unambiguous interaction between the putative SAH and DBH sites. This SAH-DBH interaction plays a significant role in the dimerization of CSQ molecules by forming much of the helix-helix contact in their dimer interface, rather than intramolecularly as originally proposed (He et al., 1993; Wang et al., 1998).
The unique high-capacity Ca2+-binding by CSQ was also previously reported for its close association with the formation of aggregates (Saito et al., 1984; Maurer et al., 1985; Tanaka et al., 1986; Franzini-Armstrong et al., 1987; Wang et al., 1998; Park et al., 2004). Likewise, electron microscopy and cross-linking studies demonstrate that most CSQs in the SR microsomes are involved in CSQ/CSQ complexes (Saito et al., 1984; Maurer et al., 1985; Franzini-Armstrong et al., 1987; Maguire et al., 1997; MacLennan and Reithmeier, 1998) and that unique high-capacity Ca2+ binding of CSQ is closely associated with the formation of aggregates.
In the crystal structure of CSQ, two different types of dimers, front-to-front and back-to-back, form a continuous, linear polymer. Not only does the polymer formed by these two contacts have the linear morphology inferred for the physiologically relevant aggregation, but also the two contacts that stabilize this polymer have structural details that lend themselves to control by Ca2+ binding. The two dimer interfaces involve burying large surface areas and very intricate interactions, both of which are unlike typical crystal contacts. Furthermore, our studies on sequence conservation across both cCSQ and sCSQ molecules indicate that the residues involved in the back-to-back and front-to-front interfaces are the most highly conserved residues in the entire structure (Park et al., 2003). This situation is reminiscent of the higher conservation of active-site residues of other proteins and thus strongly supports our hypothesis that these dimer interfaces are the functional contacts involved in the coupled CSQ polymerization and low-affinity Ca2+ binding. Through chemical cross-linking, mass spectroscopy and multiangle laser light scattering, we found that at a fixed K+ concentration of 300 mM, increasing Ca2+ concentration results in transitions from a single subunit to dimer and to tetramer. If Ca2+ is fixed, increasing K+ concentration results in a shift toward monomers, and decreasing K+ concentration generates polymers. These data are the first demonstrating the interplay of monovalent and divalent cations on the polymeric behavior of purified CSQ, strongly suggesting that CSQ polymerization is under dynamic regulation by mono- and divalent cations (Park et al., 2003). We have also reported a coupling mechanism between high-capacity Ca2+ binding and the polymerization of CSQ (Park et al., 2004).
As our knowledge of the molecular behavior of CSQs increases, we are attempting to understand the TFP binding in more detail and to identify additional compounds that may interfere with the calcium binding property of CSQ. Because CSQ is one of the major components of both cardiac and skeletal muscle and because of its key role in Ca2+ regulation, drug interactions with CSQ would be expected to produce adverse physiological consequences with a high frequency. Such consequences have not yet been well defined. Although undesirable drug effects or adverse drug reactions are often tolerated because of a high benefit/risk ratio, adverse drug reactions are still a major concern, and understanding the cause of these effects is important. In this regard, cardiotoxicity is a serious side effect of many drugs whose clinical usefulness is often limited by both dose- and time-dependent body accumulation and subsequent toxicity. For example, TFP is currently used as an anti-psychotic drug, but it has notorious side effects of tachycardia and muscle palpitation that often occur and persist even after its therapy is discontinued. The pathophysiologic and biochemical reasons for these side effects are not well understood, but the known interactions between TFP and CSQ suggest a likely source for adverse drug reactions.
In this article, we report investigations of the interactions between cCSQ and several pharmaceutical compounds with reported cardiotoxicity, including TFP, and the consequences of the interactions on the calcium binding capacity of cCSQ. The existence of the thioredoxin-like fold in CSQ molecules, a basic motif that often provides the platform for small molecule binding, is a reasonable target for binding by molecules such as TFP, thus causing various muscle-related side-effects. Even if the interaction between these drugs and the physiologically important protein CSQ is around the micromolar range, the accumulative effect of such interactions may produce a non-negligible problem during the course of long-term treatments. If this hypothesis is correct, then the discovery of alternative molecules with reduced affinity for CSQ provides a strategy for developing drug molecules with reduced skeletal and cardiac-related side-effects.
Materials and Methods
Plasmid Construction, Expression and Purification. Expression plasmid, pET24b-cCSQ was constructed using a polymerase chain reaction strategy. A specific region of canine cardiac calsequestrin was amplified by polymerase chain reaction using the cDNA clone as a template. A sense primer CAL5 (5'CTGTCAACATATGGAAGAGGGGCTCAACTTCCCCA3') that contains an NdeI site and a start codon, and an antisense primer CAL3 (5'CTGATGTCGACTATTACTCATCATCATCATCACTGTCGTC3') that contains a SalI site followed by two stop codons were used for wild-type calsequestrin construct.
Escherichia coli strain BL21-DE3 containing plasmids with the canine cCSQ gene was grown with shaking (250 rpm) at 37°C in 1 liter of Luria-Bertani medium supplemented with kanamycin (30 e/ml), until the optical density at 600 nm reached 0.6eC0.8, at which time 0.4 mM (final concentration) isopropyl -D-thiogalactoside was added. The culture was then shaken for an additional 3 h at 37°C and then harvested by centrifugation for 20 min at 6000g. The cell pellet was resuspended in 50 ml of 50 mM Tris-HCl buffer at pH 7.4. The overexpressed proteins were released from the cells by sonication (550 Sonic Dismembrator; Fisher Scientific, Pittsburgh, PA). The lysate was cleared by centrifugation at 27,000g for 30 min. Crude CSQ protein was separated from E. coli proteins by a phenyl Sepharose (Sigma) column (buffer A, 500 mM NaCl and 20 mM Tris-HCl at pH 7.5; buffer B, 20 mM CaCl2, 500 mM NaCl, in 20 mM Tris-HCI at pH 7.5). The crude cCSQ was desalted and concentrated by ultrafiltration in a stirred Amicon cell with a 10,000 molecular weight cutoff membrane.
Further purification of cCSQ was carried out by anion exchange chromatography on a Mono-Q HR10/10 (Amersham Biosciences) column connected to a fast-performance liquid chromatography system. After purification, CSQ protein was desalted by ultrafiltration with a YM10 membrane with distilled water, concentrated in YM10 Centricon tubes (Millipore Corporation, Bedford, MA), and stored at 4°C for crystallization or -80°C for long-term storage. Purity of the CSQ was characterized by UV-visible absorption spectra and electrospray ionization mass spectrometry.
Isothermal Titration Calorimetry (ITC). The interactions between canine cardiac calsequestrin and different compounds were carried out on a VP-ITC instrument (MicroCal, Northampton, MA) at 25°C following standard procedures. The protein was dialyzed for 3 days at 4°C in 300 mM KCl, 0.1 mM EGTA, and 10 mM MOPS, pH 7.5. The protein was added to the calorimetric reaction cell at a concentration of 0.05 mM with 400 rpm stirring, and a syringe was used to titrate a small molecular ligand dissolved in the same buffer at a concentration of 0.5 mM into a cell containing a cCSQ solution. Compounds that did not dissolve well in dialysis buffer were first dissolved in 100% DMSO (v/v) and went through serial dilution to 2%. The protein solution was prepared to make the final concentration of DMSO in the protein the same as in the ligand solution. cCSQ protein and drug samples were degassed before use. Each titration experiment was performed with 29 injections of 10 e at 300-s equilibration intervals. Heats of dilution for an individual ligand were determined by titrating ligand into the same buffer without protein.
Data were fit to an n-equivalent binding sites model at first by nonlinear least-squares regression with the Origin software package (OriginLab Corp, Northampton, MA). Because the number of binding sites (n) was converged to values between 0.8 and 1.2 by the initial regressions, it was fixed to 1.0 (one binding site model) in the final regressions. The fit of data yields the binding affinity, enthalpy change, entropy change, and binding stoichiometry for the titration. Compounds with positive interactions with calsequestrin were subjected to repeat experiments, and affinities for the interactions were determined by averaging results across experiments. Other compounds that did not show significant heats of reaction were not subjected to repeat experiments.
The compounds that were screened for interaction with calsequestrin were trifluoperazine, promethazine, chlorpromazine, daunorubicin, doxorubicin, 1-anilino napthalene-8-sulfonate (ANS), NADP+, FAD, ATP, caffeine, guanosine, hydralazine, tetracycline, riboflavin, quinine, amitriptyline, nortriptyline, and imipramine. All compounds were purchased from Sigma (st. Louis, MO) except daunorubicin, which was purchased from Fluka Chemical Corp. (Ronkonkoma, NY).
Equilibrium Dialysis. Canine cCSQ at a protein concentration of 1eC3 mg/ml was dialyzed first against distilled water followed by buffer containing a target drug (1 mM). The buffer condition for measuring Ca2+ binding was 10 mM Tris, pH 7.5, 300 mM KCl, and 2 mM NaN3, which was chosen based on our CD, fluorescence spectroscopy, and light-scattering experiments (Park et al., 2003). At this condition, acquisition of secondary and tertiary structure reaches a plateau, and the molecular mass estimate from the light scattering data are 44 ± 1 kDa, corresponding closely to the molecular mass of a single CSQ molecule. Dialysis was performed for 1 week in the cold room with 10 reservoir changes. Upon completion of the buffer exchange, the dialysis bag was removed, and the protein solution was transferred to the half-cell of a modified horizontal-diffusion chamber. cCSQ solution (1.5 ml) was equilibrated against the same volume of various concentrations of CaCl2 solution (0.1eC10 mM) across the 12-kDa molecular cutoff dialysis membrane. Equilibrium was achieved by gentle tilt shaking for 36 h at room temperature.
Atomic Absorption Spectrophotometry. Both the protein and calcium compartments were analyzed for calcium concentration using an Atomic Absorption Spectrophotometer (Shimadzu AA-6200) at the absorption wavelength of 422.7 nm. The protein concentration of each dialysis cell was measured again by Bradford protein assay after equilibrium dialysis. Fractional occupancy (Y = [bound Ca]/[total CSQ]) was calculated as the difference in Ca2+ concentration between two compartments. Even though the ligand in this experiment was an electrolyte, the Donnan effect was not considered in treating the dialysis data, because the concentration of another electrolyte (KCl, 300 mM) was already high enough in both compartments.
ANS Fluorescence. The binding constant of ANS with CSQ was determined by a modified Scatchard analysis as described previously (Semisotnov et al., 1991). In brief, ANS fluorescence intensity (F) was collected for samples with constant CSQ protein (1 mg/ml) and increasing ANS concentration, using an excitation wavelength of 370 nm and emission value of 480 nm. A plot of 1/F versus 1/ gave an estimate of the F value as saturating ANS. At intermediate F values, the fraction of saturation and the amount of free protein were estimated. A Scatchard plot of (bound ANS)/(free ANS) versus (bound ANS) could then be constructed, and the minus value of the slope gave an estimate of the binding constant.
Results
Isothermal Titration Calorimetry. We performed a thermodynamic characterization of various pharmaceutical drugs that have short- and/or long-term cardiotoxicity using ITC. The major muscle-related (both skeletal and cardiac muscle) side effects of the tested drugs range from involuntary movements of muscles (dyskinesias) to arrhythmias and sudden death from heart failure. Measurement of the heat that is generated or absorbed upon ligand-binding by ITC allows accurate determination of the binding constant (Ka), enthalpy and entropy changes, thus providing a complete thermodynamic profile of the molecular interaction.
Thermodynamics values for the binding reaction are summarized in Table 1 and Fig. 1. In every case, heat was released when cCSQ associated with these compounds, indicating that the binding interactions had significant enthalpic contributions. In addition, detailed data analysis revealed slightly unfavorable entropic contributions for five of the eight molecules, possibly indicating that the CSQ structure was slightly stabilized upon binding to these five molecules.
First, we tested TFP and other phenothiazine derivatives that are being used as antipsychotic drugs: promethazine and chlorpromazine (Thorazine). TFP showed the highest affinity (14 e), followed by promethazine (15 e) and chlorpromazine (43 e). Our ITC data, therefore, confirmed the previous suggestion of a CSQ-TFP interaction (He et al., 1993). These three compounds have their branched side chains on the same side of the phenothiazine ring, which was proposed to approximate the conformation of dopamine, but dopamine did not bind to cCSQ in our experiment. Neither the phenothiazine ring nor the phenazine ring alone bound cCSQ.
The phenothiazine derivatives that bound to cCSQ have high hydrophobicity, considering their reported and calculated log S (aqueous solubility) and log P (partition coefficient) values, and most of them approach the hydrophobic extreme of pharmaceutical drugs (Jorgensen, 2004). Therefore, they may well be able to penetrate into the sarcoplasmic reticulum and bind cCSQ or sCSQ.
Second, we tested tricyclic antidepressants (TCAs), such as nortriptyline (secondary amine tricyclics), amitriptyline, and imipramine (tertiary amine tricyclics). The TCAs that we tested generally showed affinity for cCSQ similar to that of the phenothiazine derivatives. Nortriptyline, amitriptyline, and imipramine had dissociation constants of 11, 15, and 77 e, respectively. The TCAs are known to be cardiotoxic, and overdoses are frequently fatal. Electrocardiographic changes evoked by the tricyclics include QRS complex distortion, ventricular ectopic beats, nodal rhythm, bundle branch block, and ventricular tachycardia (Waterfall et al., 1979). The side effects of TCAs have been attributed to their nonspecific interaction with various unidentified proteins (Andrews and Nemeroff, 1994).
Third, we tested anthracycline derivative, daunorubicin, which is a widely used and highly effective anticancer drug. This popular anthracycline drug showed a dissociation constant of 26 e, which is comparable with TCAs and phenothiazine derivatives. Doxorubicin also showed a considerable affinity, with a dissociation constant of 40 e. The tetracycline ring alone showed no significant binding to cCSQ. The anthracycline derivatives are known to interfere with calcium handling mechanisms in the heart (Pessah et al., 1990; Mushlin et al., 1993; Olson et al., 2000; Burke et al., 2003), producing severe cardiotoxicity, often irreversibly damaging the heart, which somewhat limits their therapeutic potential (Olson et al., 1988). Although details of the molecular basis of these side effects are not clearly understood, especially the long-term effects after rapid and serious accumulation of the administered drug in the cardiac myocyte, evidence has been reported of possible interactions between these anthracycline molecules and the unknown components of the SR Ca2+ regulation complex (Zorzato et al., 1985; Pessah et al., 1990; Mushlin et al., 1993; Bowling et al., 1994).
On the other hand, other drugs that are known to have major or minor muscle-related side effects, such as cyclophosphamide, DL-isoproterenol, ephedrine, quinine, flunarizine, hydralazine, mitoxantrone, troglitazone, lonidamine, and etoposide, showed no significant binding to cCSQ. Various biochemical compounds that have a similar flat, multiring structure and similar molecular mass, such as NADP+, FAD, ATP, GTP, tryptophan, and several smaller molecules, such as caffeine, adenine, guanosine, riboflavin, and quinine, also failed to show any significant binding, judging from the ITC signal. Some of typical cases of nonbinding examples are shown in Fig. 1B.
ANS Binding. ANS also binds to cCSQ. To further characterize the binding site in cCSQ, we carried out florescence spectroscopic studies with ANS (Fig. 2). Binding by this probe is relatively easy to detect because of large fluorescence changes, and this probe binds promiscuously to nucleotide, flavin, and other binding sites. ANS binding to hydrophobic patches on protein surfaces is accompanied by increases in fluorescence (typically less than 5-fold), presumably because much of the molecule remains exposed to solvent. In contrast, burying ANS in functional pockets or in the interiors of molten globules leads to 50- to 200-fold increases in fluorescence (Semisotnov et al., 1991). By this criterion, ANS seems largely buried upon binding to CSQ because an increase in fluorescence of >50-fold is observed (Fig. 2). The binding constant of 56 e (Fig. 2, inset) is 2-fold tighter than the values of 100 to 300 e observed for molten globules but within the range of 1eC60 e observed for ANS binding to various proteins having different functional pockets (Bailey et al., 2001).
Calcium Dependence of Drug Binding. To characterize the interdependency between drug binding and Ca2+ binding more thoroughly, cCSQ protein was pre-equilibrated with 0, 3, or 5 mM CaCl2 before conducting the drug-binding calorimetric experiments. The ITC results (Fig. 3) show that the binding affinity of chlorpromazine in the presence of 3 mM CaCl2 was reduced to approximately one quarter (from Kd of 33 e to 140 e) and was abolished almost completely in the presence of 5 mM CaCl2. All of the high-affinity drugs mentioned above show a similar pattern (data not shown).
Reduction in Calcium Binding Capacity of cCSQ upon Drug Binding. Of special interest is whether the interaction of the ligand blocks high-capacity calcium binding by sterically interfering with the formation of CSQ polymer, as we suggested (Wang et al., 1998), involves substantial conformational change, or both. We performed equilibrium dialysis and atomic absorption spectroscopy to check the interference of calcium binding in the presence of drugs that show a high affinity for cCSQ.
Although the calcium binding curve of CSQ has several changes of curvature caused by the formation of various types of oligomer as the concentration of calcium increases (Park et al., 2004), the data were fitted to a simple binding model to check the reduction in calcium binding capacity caused by the interaction with the given ligand. The amount of bound calcium to CSQ was significantly reduced (Fig. 4) as the overall dissociation constant between calcium and cCSQ of 0.9 ± 0.2 mM was changed to 2.1 ± 0.3, 2.5 ± 0.4, and 2.8 ± 0.5 by promethazine, amitriptyline, and trifluoperazine, respectively, in their 1 mM concentrations. The corresponding value by tetracycline that showed no significant binding was within the error range, 1.1 ± 0.2.
Overall, the degree of inhibition for calcium binding tested by equilibrium dialysis was roughly matched with the affinity of the drug measured by ITC experiment. Trifluoperazine is the strongest inhibitor (about 60% inhibition at 1 mM concentration) of calcium binding among the compounds, and promethazine, amitriptyline, nortriptyline, and imipramine showed about 30 to 40% inhibition (at 1 mM, Fig. 5). Our experimental trials with the micromolar range of drugs were not successful, because the differences in the amount of bound calcium came close to the error range of our analysis method. We reported previously that TFP-sCSQ interaction prevented Ca2+-induced aggregation of CSQ and decreased the Ca2+ binding by CSQ (He et al., 1993). Our result here is consistent with those earlier findings. At that time, we suggested the observed interference was caused by intramolecular SAH-DBH interactions, which we later refined by suggesting that the SAH-DBH interactions are between two different CSQ molecules rather than within one molecule and that TFP binding interferes with dimerization (Wang et al., 1998).
Discussion
The tight regulation of Ca2+ release to and clearance from the cytosol is essential for normal cardiac function. Catecholamine-induced polymorphic ventricular tachycardia (CPVT) is a rare arrhythmogenic disorder characterized by syncopal events and sudden cardiac death at a young age during physical or emotional stress, in the absence of structural heart disease (Viskin and Belhassen, 1998; Postma et al., 2002). Missense mutation (D307H) and nonsense mutations were found as the cause of the autosomal recessive form (Lahat et al., 2001; Eldar et al., 2003) and several additional CSQ-related CPVT families were also discovered, suggesting that mutations in cCSQ are more common than previously thought and produce a severe form of CPVT (Postma et al., 2002; Laitinen et al., 2003; Lahat et al., 2004).
In this study, we demonstrated the existence of a putative ligand-binding site in CSQ and severe interference in the Ca2+ binding capacity of CSQ caused by the occupancy of that site by certain kinds of small molecules, which is in accord with other known characteristics of CSQ [e.g., the individual domains of CSQ have a thioredoxin-like fold (Wang et al., 1998) and CSQ can be easily purified with an affinity column containing phenyl moiety (Mitchell et al., 1988)].
The local structure at the putative binding site in each thioredoxin-like domain of both cCSQ and sCSQ seems to be a rather deep, hydrophobic cleft that would seem to be appropriate for certain types of ligands. Many essential endoplasmic reticulum-resident proteins, such as protein disulfide isomerase and calreticulin (CRT), also contain the thioredoxin fold, judging from their amino acid sequences, and these proteins bind Ca2+ with low affinity and high capacity just as CSQ does. The ligand binding activities of protein disulfide isomerase and CRT have been reported previously (Nigam et al., 1994; Michalak et al., 1998; Ferrari and Soling, 1999; Corbett et al., 2000). For example, CRT binds ATP but does not exhibit detectable ATPase activity (Nigam et al., 1994; Corbett et al., 2000). The presence of three thioredoxin folds suggests that CSQ might have a previously unsuspected ligand-binding capacity and that ligand binding might play an important regulatory or adverse role in calcium regulation by CSQ.
Small hydrophobic molecules, such as phenothiazines and tricyclic antidepressants, can certainly diffuse into the sarcoplasmic reticulum (predicted log P ranges from 3.08 to 5.88 in their nonionized form) (Walter and Gutknecht, 1986) and may well accumulate there. CSQ is one of many proteins abundant (100 mg/ml) in both skeletal and cardiac muscles, and our results show that it contains a binding site that modulates Ca2+ binding levels. The interactions between cCSQ and phenothiazines, TCAs, anthracyclines, and many other pharmaceutical compounds will lead to the reduction of the Ca2+ binding function of CSQ and consequently interfere with the normal function of the SR. It is therefore reasonable to suggest that some of the muscle-related symptoms of these drugs, such as tachycardia, may be consequences of the interaction between these drugs and CSQ.
We have proposed previously that polymerization of CSQ is directly linked to its high-capacity, low-affinity Ca2+ binding (Park et al., 2004). Ca2+ largely fills the electronegative pockets formed in the interfaces between CSQ molecules, cross-bridging the subunit and eventually forming a polymer. The propensity of Ca2+-bound CSQ to form linear structures that can be formed and destroyed dynamically after the flux of K+ and Ca2+ makes Ca2+ dissociation and diffusion a rapid event, allowing the two-dimensional diffusion of Ca2+ ions (Fig. 6, left and center). Small amounts of compounds that have significant CSQ binding affinity can disrupt the dynamic polymerization of CSQ by binding to the site near the SAH-DBH interaction in the dimer interface, thus hindering the high-capacity Ca2+ binding of CSQ and reducing the Ca2+-buffering capacity of the SR (Fig. 6, right).
A significant uncertainty still exists as to whether our observation of interfering (or regulating) Ca2+ by binding small, hydrophobic ligands is relevant to CSQ function in vivo, because in our test, recombinant cCSQ has been used, which could disregard the importance of the post-translational modification, including a phosphorylation (Szegedi et al., 1999). In addition, the concentration of the drugs is higher than the usual tissue concentration (around the micromolar range) under the usual treatment with such drugs. However, the very high local concentrations of CSQ within the lumen of the SR means that even low concentrations of a given drug molecule (Kd in the micromolar range) could be sufficient to lead to binding and thereby to important physiological consequences. In addition, a slight alteration in the normal physiological function of the cardiac SR might be enough to cause serious problems in certain persons, and long-term exposure to such drugs could easily generate a cumulative effect (long-term toxicity). The average value of the corresponding dissociation constants listed in Table 1 (32.4 e) gives the fractional value of cCSQ molecules occupied by the drug in tissue reaching 324% of the total cCSQ for the practical tissue concentration of such drugs, 110 e.
The side effects of cardiac or skeletal muscle related symptoms could have several different origins such as nerve and various cellular components. In our experiments, not all the drugs having muscle-related side effects showed an interaction with cCSQ. However, given our evidence that CSQ contains a binding site for certain small molecules (Fig. 7) that alter the Ca2+ binding capacity, it is tempting to speculate that the devastating side effects of some pharmaceuticals, including the drugs we are reporting here, are the result, at least in part, of drug-dependent disruption of the calcium-regulating property of CSQ function in the SR. Reduced cCSQ level was associated with the accelerated Ca2+ discharge, and mutation of cCSQ has been suggested as a plausible explanation for ventricular arrhythmia (Terentyev et al., 2003). Considering the CSQ concentration in muscle tissue, reduced Ca2+ binding capacity caused by drug binding, and the tendency for long-term administration of these types of drugs, the physiological consequences of long-term treatment with such drugs that impair CSQ function are expected to be similar to those of genetic defects of this protein, such as catecholaminergic polymorphic ventricular tachycardia.
The original work described in this article was supported by National Science Foundation grant MCB-0117192), the American Heart Association, and the Murdock Charitable Trust.
References
Andrews JM and Nemeroff CB (1994) Contemporary management of depression. Am J Med 97: 24S-32S.
Bailey RW, Dunker AK, Brown CJ, Garner EC, and Griswold MD (2001) Clusterin, a binding protein with a molten globule-like region. Biochemistry 40: 11828-11840.
Bowling N, Mais DE, Gerzon K, and Watanabe AM (1994) Ryanodine and an iodinated analog: doxorubicin effects on binding and Ca2+ accumulation in cardiac sarcoplasmic reticulum. Eur J Pharmacol 268: 365-373.
Branden CI (1980) Relation between structure and function of a/b proteins. Q Rev Biophys 13: 317-338.
Burke BE, Olson RD, Cusack BJ, Gambliel HA, and Dillmann WH (2003) Anthracycline cardiotoxicity in transgenic mice overexpressing SR Ca2+-ATPase. Biochem Biophys Res Commun 303: 504-507.
Cachia PJ, Van Eyk J, Ingraham RH, McCubbin WD, Kay CM, and Hodges RS (1986) Calmodulin and troponin C: a comparative study of the interaction of mastoparan and troponin I inhibitory peptide [104eC115]. Biochemistry 25: 3553-3562.
Collins JH, Tarcsafalvi A, and Ikemoto N (1990) Identification of a region of calsequestrin that binds to the junctional face membrane of sarcoplasmic reticulum. Biochem Biophys Res Commun 167: 189-193.
Cook WJ, Walter LJ, and Walter MR (1994) Drug binding by calmodulin: crystal structure of a calmodulin-trifluoperazine complex. Biochemistry 33: 15259-15265.
Corbett EF, Michalak KM, Oikawa K, Johnson S, Campbell ID, Eggleton P, Kay C, and Michalak M (2000) The conformation of calreticulin is influenced by the endoplasmic reticulum luminal environment. J Biol Chem 275: 27177-27185.
Damiani E and Margreth A (1990) Specific protein-protein interactions of calsequestrin with junctional sarcoplasmic reticulum of skeletal muscle. Biochem Biophys Res Commun 172: 1253-1259.
Eldar M, Pras E, and Lahat H (2003) A missense mutation in the CASQ2 gene is associated with autosomal-recessive catecholamine-induced polymorphic ventricular tachycardia. Trends Cardiovasc Med 13: 148-151.
Ferrari DM and Soling HD (1999) The protein disulphide-isomerase family: unravelling a string of folds. Biochem J 339: 1-10.
Fliegel L, Ohnishi M, Carpenter M, Khanna V, Reithmeier R, and MacLennan D (1987) Amino acid sequence of rabbit fast-twitch skeletal muscle calsequestrin deduced from cDNA and peptide sequencing. Proc Natl Acad Sci USA 84: 1167-1171.
Franzini-Armstrong C, Kenney LJ, and Varriano-Marston E (1987) The structure of calsequestrin in triads of vertebrate skeletal muscle: a deep-etch study. J Cell Biol 105: 49-56.
He Z, Dunker AK, Wesson CR, and Trumble WR (1993) Ca2+-induced folding and aggregation of skeletal muscle sarcoplasmic reticulum calsequestrin. The involvement of the trifluoperazine-binding site. J Biol Chem 268: 24635-24641.
Jones LR, Suzuki YJ, Wang W, Kobayashi YM, Ramesh V, Franzini-Armstrong C, Cleemann L, and Morad M (1998) Regulation of Ca2+ signaling in transgenic mouse cardiac myocytes overexpressing calsequestrin. J Clin Investig 101: 1385-1393.
Jones LR, Zhang L, Sanborn K, Jorgensen AO, and Kelley J (1995) Purification, primary structure and immunological characterization of the 26-kDa calsequestrin binding protein (junctin) from cardiac junctional sarcoplasmic reticulum. J Biol Chem 270: 30787-30796.
Jorgensen WL (2004) The Many Roles of Computation in Drug Discovery. Science (Wash DC) 303: 1813-1818.
Katti SK, LeMaster DM, and Eklund H (1990) Crystal structure of thioredoxin from Escherichia coli at 1.68 A resolution. J Mol Biol 212: 167-184.
Kobayashi YM, Alseikhan BA, and Jones LR (2000) Localization and characterization of the calsequestrin-binding domain of triadin 1. Evidence for a charged -strand in mediating the protein-protein interaction. J Biol Chem 275: 17639-17646.
Lahat H, Pras E, and Eldar M (2004) A missense mutation in CASQ2 is associated with autosomal recessive catecholamine-induced polymorphic ventricular tachycardia in Bedouin families from Israel. Ann Med 36: 87-91.
Lahat H, Pras E, Olender T, Avidan N, Ben-Asher E, Man O, Levy-Nissenbaum E, Khoury A, Lorber A, Goldman B, et al. (2001) A missense mutation in a highly conserved region of CASQ2 is associated with autosomal recessive catecholamine-induced polymorphic ventricular tachycardia in Bedouin families from Israel. Am J Hum Genet 69: 1378-1384.
Laitinen P, Swan H, and Kontula K (2003) Molecular genetics of exercise-induced polymorphic ventricular tachycardia: identification of three novel cardiac ryanodine receptor mutations and two common calsequestrin 2 amino-acid polymorphisms. Eur J Hum Genet 11: 888-891.
Levin RM and Weiss B (1977) Binding of trifluoperazine to the calcium-dependent activator of cyclic nucleotide phosphodiesterase. Mol Pharmacol 13: 690-697.
MacLennan DH, Abu-Abed N, and Kang C (2002) Structure-Function relationships in Ca2+ cycling proteins. J Mol Cell Cardiol 34: 897-918.
MacLennan DH and Reithmeier RA (1998) Ion tamers. Nat Struct Biol 5: 409-411.
MacLennan DH and Wong PT (1971) Isolation of a calcium-sequestering protein from sarcoplasmic reticulum. Proc Natl Acad Sci USA 68: 1231-1235.
Maguire PB, Briggs FN, Lennon NJ, and Ohlendieck K (1997) Oligomerization is an intrinsic property of calsequestrin in normal and transformed skeletal muscle. Biochem Biophys Res Commun 240: 721-727.
Maurer A, Tanaka M, Ozawa T, and Fleischer S (1985) Purification and crystallization of the calcium binding protein of sarcoplasmic reticulum from skeletal muscle. Proc Natl Acad Sci USA 82: 4036-4040.
Michalak M, Mariani P, and Opas M (1998) Calreticulin, a multifunctional Ca2+ binding chaperone of the endoplasmic reticulum. Biochem Cell Biol 76: 779-785.
Mitchell RD, Simmerman HK, and Jones LR (1988) Ca2+ binding effects on protein conformation and protein interactions of canine cardiac calsequestrin. J Biol Chem 263: 1376-1381.
Mushlin PS, Cusack BJ, Boucek RJ Jr, Andrejuk T, Li X, and RDO (1993) Time-related increases in cardiac concentrations of doxorubicinol could interact with doxorubicin to depress myocardial contractile function. Br J Pharmacol 110: 975-982.
Nigam SK, Goldberg AL, Ho S, Rohde MF, Bush KT, and Sherman MY (1994) A set of endoplasmic reticulum proteins possessing properties of molecular chaperones includes Ca(2+)-binding proteins and members of the thioredoxin superfamily. J Biol Chem 269: 1744-1749.
Ohkura M, Furukawa K, Fujimori H, Kuruma A, Kawano S, Hiraoka M, Kuniyasu A, Nakayama H, and Ohizumi Y (1998) Dual regulation of the skeletal muscle ryanodine receptor by triadin and calsequestrin. Biochemistry 37: 12987-12993.
Olson R, Mushlin PS, Brenner DE, Fleischer S, Cusack BJ, Chang BKand Jr BR (1988) Doxorubicin cardiotoxicity may be caused by its metabolite, doxorubicinol. Proc Natl Acad Sci USA 85: 3585-3589.
Olson RD, Li X, Palade P, Shadle SE, Mushlin PS, Gambliel HA, Fill M, Boucek RJJ, and Cusack BJ (2000) Sarcoplasmic reticulum calcium release is stimulated and inhibited by daunorubicin and daunorubicinol. Toxicol Appl Pharmacol 169: 168-176.
Park H, Park I-Y, Kim EJ, Youn B, Fields K, Dunker AK, and Kang C (2004) Comparing skeletal and cardiac calsequestrin structures and their calcium binding: a proposed mechanism for coupled calcium binding and protein polymerization. J Biol Chem 279: 18026-18033.
Park H, Wu S, Dunker AK, and Kang C (2003) Polymerization of calsequestrin. Implications for Ca2+ regulation. J Biol Chem 278: 16176-16182.
Pessah IN, Durie EL, Schiedt MJ, and Zimanyi I (1990) Anthraquinone-sensitized Ca2+ release channel from rat cardiac sarcoplasmic reticulum: possible receptor-mediated mechanism of doxorubicin cardiomyopathy. Mol Pharmacol 37: 503-514.
Postma AV, Denjoy I, Hoorntje TM, Lupoglazoff JM, Da Costa A, Sebillon P, Mannens MM, Wilde AA, and Guicheney P (2002) Absence of calsequestrin 2 causes severe forms of catecholaminergic polymorphic ventricular tachycardia. Circ Res 91: 21-26.
Saito A, Seiler S, Chu A, and Fleischer S (1984) Preparation and morphology of sarcoplasmic reticulum terminal cisternae from rabbit skeletal muscle. J Cell Biol 99: 875-885.
Semisotnov GV, Rodionova NA, Razgulyaev OI, Uversky VN, Gripas' AF, and Gilmanshin RI (1991) Study of the "molten globule" intermediate state in protein folding by a hydrophobic fluorescent probe. Biopolymers 31: 119-128.
Szegedi C, Sarkozi S, Herzog A, Jona I, and Varsanyi M (1999) Calsequestrin: more than `only' a luminal Ca2+ buffer inside the sarcoplasmic reticulum. Biochem J 337: 19-22.
Tanaka M, Ozawa T, Maurer A, Cortese JD, and Fleischer S (1986) Apparent cooperativity of Ca2+ binding associated with crystallization of Ca2+-binding protein from sarcoplasmic reticulum. Arch Biochem Biophys 251: 369-378.
Terentyev D, Viatchenko-Karpinski S, Gyorke I, Volpe P, Williams SC, and Gyorke S (2003) Calsequestrin determines the functional size and stability of cardiac intracellular calcium stores: mechanism for hereditary arrhythmia. Proc Natl Acad Sci USA 100: 11759-11764.
Tijskens P, Jones LR, and Franzini-Armstrong C (2003) Junctin and calsequestrin overexpression in cardiac muscle: the role of junctin and the synthetic and delivery pathways for the two proteins. J Mol Cell Cardiol 35: 961-974.
Viskin S and Belhassen B (1998) Polymorphic ventricular tachyarrhythmias in the absence of organic heart disease: classification, differential diagnosis and implications for therapy. Prog Cardiovasc Dis 41: 17-34.
Walter A and Gutknecht J (1986) Permeability of small nonelectrolytes through lipid bilayer membranes. J Membr Biol 90: 207-217.
Wang S, Trumble WR, Liao H, Wesson CR, Dunker AK, and Kang C (1998) Crystal structure of calsequestrin from rabbit skeletal muscle sarcoplasmic reticulum. Nat Struct Biol 5: 476-483.
Waterfall JF, Smith MA, Gaston WH, Maher J, and Warburton G (1979) Cardiovascular and autonomic actions of ciclazindol and tricyclic antidepressants. Arch Int Pharmacodyn Ther 240: 116-136.
Yano K and Zarain-Herzberg A (1994) Sarcoplasmic reticulum calsequestrins: structural and functional properties. Mol Cell Biochem 135: 61-70.
Zhang L, Kelley J, Schmeisser G, Kobayashi YM, and Jones LR (1997) Complex formation between junctin, triadin, calsequestrin and the ryanodine receptor. Proteins of the cardiac junctional sarcoplasmic reticulum membrane. J Biol Chem 272: 23389-23397.
Zorzato F, Salviati G, Facchinetti T, and Volpe P (1985) Doxorubicin induces calcium release from terminal cisternae of skeletal muscle. A study on isolated sarcoplasmic reticulum and chemically skinned fibers. J Biol Chem 260: 7349-7355.