Portal de Periódicos da CAPES

Comparing Skeletal and Cardiac Calsequestrin Structures and Their Calcium Binding

2004; Elsevier BV; Volume: 279; Issue: 17 Linguagem: Inglês

10.1074/jbc.m311553200

1083-351X

HaJeung Park, Il Yeong Park, Eunjung Kim, BuHyun Youn, Kelly Fields, A. Keith Dunker, ChulHee Kang,

Force Microscopy Techniques and Applications

Calsequestrin, the major calcium storage protein of both cardiac and skeletal muscle, binds and releases large numbers of Ca2+ ions for each contraction and relaxation cycle. Here we show that two crystal structures for skeletal and cardiac calsequestrin are nearly superimposable not only for their subunits but also their front-to-front-type dimers. Ca2+ binding curves were measured using atomic absorption spectroscopy. This method enables highly accurate measurements even for Ca2+ bound to polymerized protein. The binding curves for both skeletal and cardiac calsequestrin were complex, with binding increases that correlated with protein dimerization, tetramerization, and oligomerization. The Ca2+ binding capacities of skeletal and cardiac calsequestrin are directly compared for the first time, with ∼80 Ca2+ ions bound per skeletal calsequestrin and ∼60 Ca2+ ions per cardiac calsequestrin, as compared with net charges for these molecules of -80 and -69, respectively. Deleting the negatively charged and disordered C-terminal 27 amino acids of cardiac calsequestrin results in a 50% reduction of its calcium binding capacity and a loss of Ca2+-dependent tetramer formation. Based on the crystal structures of rabbit skeletal muscle calsequestrin and canine cardiac calsequestrin, Ca2+ binding capacity data, and previous light-scattering data, a mechanism of Ca2+ binding coupled with polymerization is proposed. Calsequestrin, the major calcium storage protein of both cardiac and skeletal muscle, binds and releases large numbers of Ca2+ ions for each contraction and relaxation cycle. Here we show that two crystal structures for skeletal and cardiac calsequestrin are nearly superimposable not only for their subunits but also their front-to-front-type dimers. Ca2+ binding curves were measured using atomic absorption spectroscopy. This method enables highly accurate measurements even for Ca2+ bound to polymerized protein. The binding curves for both skeletal and cardiac calsequestrin were complex, with binding increases that correlated with protein dimerization, tetramerization, and oligomerization. The Ca2+ binding capacities of skeletal and cardiac calsequestrin are directly compared for the first time, with ∼80 Ca2+ ions bound per skeletal calsequestrin and ∼60 Ca2+ ions per cardiac calsequestrin, as compared with net charges for these molecules of -80 and -69, respectively. Deleting the negatively charged and disordered C-terminal 27 amino acids of cardiac calsequestrin results in a 50% reduction of its calcium binding capacity and a loss of Ca2+-dependent tetramer formation. Based on the crystal structures of rabbit skeletal muscle calsequestrin and canine cardiac calsequestrin, Ca2+ binding capacity data, and previous light-scattering data, a mechanism of Ca2+ binding coupled with polymerization is proposed. Calsequestrin (CSQ) 1The abbreviations used are: CSQ, calsequestrin; cCSQ, cardiac CSQ; sCSQ, skeletal CSQ; SR, sarcoplasmic reticulum; CPVT, catecholamine-induced polymorphic ventricular tachycardia. binds and releases large quantities of Ca2+ through its high capacity (40-50 mol of Ca2+ ion per molecule) and relatively low affinity interactions with Ca2+ (Kd = 1mm) (1Mitchell R. Simmerman H. Jones L. J. Biol. Chem. 1988; 263: 1376-1381Abstract Full Text PDF PubMed Google Scholar). Because of this Ca2+-buffering capacity of CSQ in the lumenal space, the concentration of free Ca2+ in the sarcoplasmic reticulum (SR) can be maintained below the inhibitory level of the Ca2+ pump (1 mm), and simultaneously, the SR can maintain the ability to rapidly deliver a high capacity Ca2+ signal to the cytoplasm. Even though the lumenal space is minuscule compared with the extracellular space, the high concentrations (∼100 mg/ml) of CSQ make the SR an efficient storage compartment for Ca2+ (2MacLennan D. Wong P. Proc. Natl. Acad. Sci. U. S. A. 1971; 68: 1231-1235Crossref PubMed Scopus (426) Google Scholar). CSQ is associated physically with the RyR protein by a nucleation event that involves CSQ binding to the basic lumenal domains of triadin (3Guo W. Campbell K. J. Biol. Chem. 1995; 270: 9027-9030Abstract Full Text Full Text PDF PubMed Scopus (203) Google Scholar) or junctin (4Zhang L. Kelley J. Schmeisser G. Kobayashi Y. Jones L. J. Biol. Chem. 1997; 272: 23389-23397Abstract Full Text Full Text PDF PubMed Scopus (460) Google Scholar). These two proteins interact with RyR in the junctional face region of the SR, and this network of interacting proteins assures that high concentrations of Ca2+ are stored very near to the site of Ca2+ release. Ca2+ release from CSQ through the Ca2+ release channel is regulatory but not limiting. The Ca2+ binding and dissociation mechanisms of CSQ are not yet clearly understood. Ca2+ binding sites in CSQ are supposed to be very different from those in the Ca2+ pump (sarco(endo)plasmic reticulum calcium ATPase (SERCA)), calmodulin, and troponin C. CSQ sites need to be made and broken but not over the low cytosolic Ca2+ concentration range or with the same stoichiometry and precision as those formed and subsequently disrupted in the Ca2+ pump or those which are intrinsic to the EF-hand structure (5MacLennan D. Abu-Abed N. Kang C. J. Mol. Cell. Cardiol. 2002; 34: 897-918Abstract Full Text PDF PubMed Scopus (63) Google Scholar). Therefore, the high capacity and low affinity Ca2+ binding by CSQ is likely nonspecific, although the first few ions that bind at low Ca2+ concentrations may have some specificity. Instead of a distinct Ca2+ binding site such as the EF-hand motif (6Strynadka N. James M. Annu. Rev. Biochem. 1989; 58: 951-998Crossref PubMed Google Scholar), pairs of acid residues have been proposed to bind Ca2+, with Ca2+ binding being driven to a significant degree by the entropy gain from liberation of many water molecules from the hydrated cations (7Wright D. Holloway J. Reilley C. Anal. Chem. 1965; 37: 884-892Crossref Scopus (65) Google Scholar, 8Krause K. Milos M. Luan-Rilliet Y. Lew D. Cox J. J. Biol. Chem. 1991; 266: 9453-9459Abstract Full Text PDF PubMed Google Scholar). The crystal structure of rabbit skeletal CSQ (sCSQ) shows that this protein is made up of three domains, each with a thioredoxin fold, which is a five β-strand sandwiched by four α-helices (9Wang S. Trumble W. Liao H. Wesson C. Dunker A. Kang C. Nat. Struct. Biol. 1998; 5: 476-483Crossref PubMed Scopus (206) Google Scholar). The individual domains of CSQ show very low sequence similarity with thioredoxins, but the locations of hydrophobic and hydrophilic residues can be overlapped after superimposing the secondary structural components of these two proteins. Each domain of CSQ has a hydrophobic core with acidic residues on the exterior, generating electronegative potential surfaces. Individual domains are connected by short sequences located interior to the domains themselves. These connecting loops and the secondary structural elements that fill the inter-domain space contain mostly acidic residues, making the overall center of the protein hydrophilic rather than hydrophobic. Therefore, cations are required to stabilize the acidic center of CSQ. Divalent cations, which can provide cross-bridging, would be expected to be more effective in this regard than monovalent cations. CSQ forms regular, highly elongated structures that appear as crystalline arrays in the lumen of the SR (10Saito A. Seiler S. Chu A. Fleischer S. J. Cell Biol. 1984; 99: 875-885Crossref PubMed Scopus (420) Google Scholar, 11Franzini-Armstrong C. Kenney L. Varriano-Marston E. J. Cell Biol. 1987; 105: 49-56Crossref PubMed Scopus (169) Google Scholar, 12Somlyo A. Gonzalez-Serratos H. Shuman H. McClellan G. Somlyo A. J. Cell Biol. 1981; 90: 577-594Crossref PubMed Scopus (273) Google Scholar). We previously showed that the crystal lattice of the rabbit sCSQ contains a linear polymer joined by two distinct dimerization contacts, the front-to-front and back-to-back interfaces (9Wang S. Trumble W. Liao H. Wesson C. Dunker A. Kang C. Nat. Struct. Biol. 1998; 5: 476-483Crossref PubMed Scopus (206) Google Scholar). These linear polymers provide a reasonable basis for both the crystalline arrays seen in vivo and the needle-shaped crystals often grown in vitro in the presence of Ca2+. We also reported that oligomerization is Ca2+/K+-dependent and proposed several ways this dynamic polymerization may have biological significance (13Park H. Wu A. Dunker A. Kang C. J. Biol. Chem. 2003; 278: 16176-16182Abstract Full Text Full Text PDF PubMed Scopus (103) Google Scholar). According to our proposal, the front-to-front and back-to-back contacts between the CSQ subunits permit formation of a Ca2+-dependent linear polymer that is inhibited as the concentration of monovalent ion, K+, increases. Ca2+ largely fills the electronegative pockets formed in these two contacts cross-bridging the subunit, which monovalent cations cannot do. This dynamic formation of CSQ polymer can provide a highly charged surface onto which Ca2+ ion is adsorbed. The attractive forces exerted by such an extended surface would have a longer range than those from an isolated molecule, and a sparingly soluble ion such as Ca2+ would tend to spread over the surface of the CSQ polymer, forming a readily exchangeable film (14MacLennan D. Reithmeier R. Nat. Struct. Biol. 1998; 5: 409-411Crossref PubMed Scopus (57) Google Scholar). Thus, Ca2+ diffusion from CSQ to the Ca2+ release channel is likely to involve surface diffusion, a more rapid process than diffusion through liquid (14MacLennan D. Reithmeier R. Nat. Struct. Biol. 1998; 5: 409-411Crossref PubMed Scopus (57) Google Scholar, 15Williams R. Biochim. Biophys. Acta. 1978; 505: 1-44Crossref PubMed Scopus (149) Google Scholar). In mammals two different CSQ genes, cardiac and skeletal CSQ (cCSQ and sCSQ), have been identified that are localized in two different cellular environments that obviously require different types of calcium regulation. Apparently, the rabbit cCSQ shows a higher sequence identity with canine cCSQ (92.1%) than with its own sCSQ (66.5%), clearly reflecting its functional divergence. Among the ∼360 residues in either isoform of CSQs from various species, more than 110 residues are either Asp or Glu, making CSQ one of the most acidic self-folding proteins in existence. For example, rabbit sCSQ has a net charge of -80, whereas canine cCSQ has a net charge of -69 at neutral pH. Despite this small overall difference in net charge, sCSQ is known to bind about twice the amount of Ca2+ compared with the cardiac molecule (16Slupsky J. Ohnishi M. Carpenter M. Reithmeier R. Biochemistry. 1987; 26: 6539-6544Crossref PubMed Scopus (79) Google Scholar). In general, cCSQs have more negatively charged amino acids in their C terminus than sCSQs, except for those from frog. For example, the C-terminal tail of the canine cCSQ has about twice as many negatively charged amino acids as does the rabbit sCSQ. Therefore, comparing the atomic structures of the two isoforms of CSQ is likely to be very informative. CSQ has medical importance as well. Overexpression of CSQ impairs Ca2+ signaling, leading to severe cardiac hypertrophy (17Knollmann B. Knollmann-Ritschel BE Weissman N. Jones L. Morad M. J. Physiol. 2000; 525: 483-498Crossref PubMed Scopus (102) Google Scholar, 18Linck B. Boknik P. Huke S. Kirchhefer U. Knapp J. Luss H. Muller F. Neumann J. Tanriseven Z. Vahlensieck U. Baba H. Jones L. Philipson K. Schmitz W. J. Pharmacol. Exp. Ther. 2000; 294: 648-657PubMed Google Scholar), sporadic Ca2+ sparks (19Wang W. Cleemann L. Jones L. Morad M. J. Physiol. 2000; 524: 399-414Crossref PubMed Scopus (34) Google Scholar), depressed contractility in the heart coupled with an induction of a fetal gene expression program (20Sato Y. Ferguson D. Sako H. Dorn G.n. Kadambi V. Yatani A. Hoit B. Walsh R. Kranias E. J. Biol. Chem. 1998; 273: 28470-28477Abstract Full Text Full Text PDF PubMed Scopus (155) Google Scholar), and premature death (21Langdown M. Holness M. Sugden M. Biochem. J. 2003; 371: 61-69Crossref PubMed Scopus (27) Google Scholar, 22Cho M. Rapacciuolo A. Koch W. Kobayashi Y. Jones L. Rockman H. J. Biol. Chem. 1999; 274: 22251-22256Abstract Full Text Full Text PDF PubMed Scopus (105) Google Scholar, 23Jones L. Suzuki Y. Wang W. Kobayashi Y. Ramesh V. Franzini-Armstrong C. Cleemann L. Morad M. J. Clin. Invest. 1998; 101: 1385-1393Crossref PubMed Scopus (247) Google Scholar). Recently, a mis-sense mutation in a highly conserved region of cCSQ, D306H, was found to be the cause of the autosomal recessive form of catecholamine-induced polymorphic ventricular tachycardia (CPVT) (24Lahat H. Pras E. Olender T. Avidan N. Ben-Asher E. Man O. Levy-Nissenbaum E. Khoury A. Lorber A. Goldman B. Lancet D. Eldar M. Am. J. Hum. Genet. 2001; 69: 1378-1384Abstract Full Text Full Text PDF PubMed Scopus (542) Google Scholar, 25Eldar M. Pras E. Lahat H. Trends Cardiovas. Med. 2003; : 148-151Crossref PubMed Scopus (66) Google Scholar). Additionally, three CSQ-related CPVT families were discovered; evidently, CSQ mutations are more common than previously thought and can produce a severe form of CPVT (26Postma A. Denjoy I. Hoorntje T. Lupoglazoff J. Da Costa A. Sebillon P. Mannens M. Wilde A. Guicheney P. Circ. Res. 2002; 91: 21-26Crossref PubMed Google Scholar). The atomic resolution structure of cCSQ can help us to understand the functional perturbations of these hereditary mutations. In this contribution we have completed the x-ray crystallographic analysis of cCSQ in order to perform comparative studies on the cardiac and skeletal isoforms of CSQs. As described below, the three-dimensional x-ray crystal structure of canine cCSQ was refined to 2.6 Å of resolution, the structures of which now allow incisive comparative studies to be made between the cardiac and skeletal isoforms of the CSQs. We also have established accurate Ca2+ binding curves of both CSQs using atomic absorption spectroscopy and correlated these with their Ca2+-dependent oligomerization patterns. These structural and Ca2+ binding comparisons yield a mechanism for Ca2+-associated CSQ polymerization that provides important new insights for understanding the structure-function relationships for this important protein. Plasmid Constructs—Based on our previous experiences crystallizing sCSQ and cCSQ, the last 27 residues (ΔC27), which are all negatively charged, were removed to make the back-to-back contact less sensitive to divalent ions. The expression plasmid pTYB1-cCSQ ΔC27 was constructed using a PCR strategy. Specific regions of canine cardiac CSQ were amplified by PCR using the cDNA clone as a template. A sense primer, CAL5 (5′-CTGTCAACATATGGAAGAGGGGCTCAACTTCCCCA-3′), which contains an NdeI site and a start codon, and an antisense primer, CAL3-27Im (5′-CTAATGGCTCTTCCGCACCCGTCATCCCCCTCTT-3′), which contains a SapI site, were used for ΔC27 cCSQ (amino acids 1-364). PCR-amplified products were digested with NdeI/SapI and ligated into the expression vector pTYB1 (New England Biolabs). The sequence of the construct was confirmed by DNA sequencing. Expression and Purification—The Escherichia coli strain ER2566 transformed with pTYB1-cCSQs ΔC27 was grown at 37 °C in LB containing 100 μg/ml ampicillin to an A600nm of 0.6, then induced with 1 mm isopropyl-1-thio-β-d-galactopyranoside for 6 h at room temperature. The ΔC27 cCSQ mutant was purified through a chitin affinity column (New England Biolabs) followed by an anion exchange column. Briefly, cells were pelleted and suspended in sonication buffer (10 mm Tris-HCl (pH 9.0), 500 mm NaCl, 1 mm EDTA, 1% Triton X-100), sonicated on ice, and centrifuged for 15 min at 20,000 × g. Supernatant was applied to a chitin column and washed with a 10-column volume of sonication buffer without Triton X-100. On-column cleavage was performed by incubating the chitin resin with cleavage buffer containing 10 mm Tris-HCl (pH 9.0), 1 m NaCl, 1 mm EDTA, and 90 mm dithiothreitol. After overnight incubation, the cleaved CSQ protein was eluted with elution buffer (cleavage buffer without dithiothreitol). The elution was loaded onto an Uno-Q12 (Bio-Rad) column after concentrating and substituting the buffer to 20 mm Tris-HCl (pH 7.5), 20 mm NaCl with an ultrafiltration unit (Amicon). The column was then washed with a linear gradient from 100 mm to 1 m NaCl. The mutant CSQs eluted at ∼400 to 600 mm NaCl were concentrated with Centriplus-10 (Amicon). Purified protein was dialyzed against 300 mm KCl with 10 mm Tris-HCl (pH 7.5) and 5 mm dithiothreitol. Crystallization and Structure Determination of cCSQ—Crystals of ΔC27 cCSQ were grown at room temperature by the vapor diffusion method in 5 mg/ml protein, 15% polyethylene glycol 400, 50 mm sodium citrate, 0.25% n-dodecyl-β-d-maltoside, and 0.1 m Tris-HCl at pH 8.5. Drops of ∼4 μl in size were equilibrated against a 0.5-ml reservoir solution containing 30% polyethylene glycol 400, 100 mm sodium citrate, and 0.2 m Tris-HCl at pH 8.5. Typically, diffraction quality crystals were fully grown 3 days after setup and stored in a 4 °C cold room until they were used. Crystals were briefly transferred to reservoir solution before freezing in a cryo-stream at a temperature of -170 °C. The ΔC27 cCSQ crystal belongs to the tetragonal space group I4 with two molecules in an asymmetric unit (a = b = 145.188 Å, c = 99.82 Å, α = β = γ = 90°). A data set of 2.6 Å was collected on an ADSC Q315 CCD detector at the Advanced Light Source beam line 5.0.2. The diffraction data were processed with HKL2000. The structure of canine cCSQ was solved by molecular replacement methods using our previous rabbit sCSQ coordinates (Protein Data Bank code 1A8Y) and the software package AMoRe (27Navaza J. Acta Crystallogr. Sect. D Biol. Crystallogr. 1994; 50: 157-163Crossref Scopus (5029) Google Scholar). The rigid-body refinement of the initial position was carried out using 15.0 to 3.0 Å resolution data and produced an R-value of 29%. After several cycles of positional refinement, temperature factor refinement, and simulated annealing omit map, we were able to fit most of the residues to the electron density. The backbones of the canine cCSQ did not change significantly from the starting coordinates of rabbit sCSQ. As in the sCSQ structure, the electron density corresponding to the C-terminal residues starting from residue 351 was not visible from the early stage of refinement. The R-factor for the final models containing 5731 non-hydrogen atoms for two cCSQ and solvents is 19.3% (Rfree = 24.2%). The reflection numbers above the 2σ level were 25,014 (80% completeness) between 10.0 and 2.6 Å of resolution. The root mean square deviations (from standard geometry) of the cCSQ are 0.018 Å for bonds and 3.54° for angles. The coordinate of the canine cCSQ structure has been deposited in the Protein Data Bank (1SJI). Equilibrium Dialysis—Wild-type sCSQ, cCSQ, and a mutant cCSQ (ΔC27) at a protein concentration of 1-3 mg/ml were dialyzed first against distilled water followed by buffer containing 10 mm Tris (pH 7.5) with 300 mm KCl and 2 mm NaN3. 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. 1.5 ml of protein solution was equilibrated against the same volume of various concentrations of calcium chloride solution (0.1 to 40 mm) across the 12-kDa molecular weight cutoff dialysis membrane. Equilibrium was achieved by gentle tilt shaking for 36 h at room temperature. Both the protein and ligand compartments were analyzed for Ca2+concentration using atomic absorption spectrophotometer (Shimadzu AA-6200) at the absorption wavelength of 422.7 nm. Unlike alterative approaches for measuring Ca2+ ion concentrations, the complete vaporization of the sample by this method means that Ca2+ binding to protein, even to polymerized protein, should present no significant experimental difficulties. Protein concentrations of each dialysis cell were measured again by the Bradford protein assay after equilibrium dialysis. Fractional occupancy (Y = [bound Ca2+]/[total CSQ]) was calculated by a difference in Ca2+ concentration of both compartments. Even though the ligand in this experiment is an electrolyte, the Donnan effect was not considered in treating the dialysis data, since the concentration of another electrolyte (KCl, 300 mm) was already high enough in both compartments. Overall Structures—The asymmetric unit of crystalline canine ΔC27 cCSQ has two independent molecules, which are virtually superimposable, with a root mean square deviation of 0.87 Å between backbone atoms of two molecules. Consistent with this crystal lattice packing, our previous static light-scattering experiments with ΔC27 cCSQ solutions gave an apparent molecular weight corresponding to a dimer (13Park H. Wu A. Dunker A. Kang C. J. Biol. Chem. 2003; 278: 16176-16182Abstract Full Text Full Text PDF PubMed Scopus (103) Google Scholar). The overall three-dimensional structure of canine cCSQ is very similar to that of rabbit sCSQ, with a root mean square deviation of 1.4 Å, reflecting the high level of sequence identity (66%) between two CSQ isomers from two species (Fig. 1). As observed in the structure of sCSQ, the overall structure of cCSQ is composed of three thioredoxin-like domains (Fig. 1B). Individual thioredoxin domains have a five-stranded β-sheet sandwiched by four α-helices composed of ∼100 residues (residues 12-124, 125-228, 229-352). As shown in Fig. 2, the surface electric potential of cCSQ is less negative than that of sCSQ due to the lower net charge of every domain. The net charges of the domains I, II, and III of the sCSQ are -21, -13, and -32 respectively, and -14 at its disordered C-terminal tail. On the other hand, for cCSQ these net charges are reduced to -8, -11, and -22 in its three domains, but there are twice as many negatively charged residues at its C-terminal tail (Fig. 3). All three domains of both cCSQ and sCSQ show stable hydrophobic cores with high aromatic amino acid composition, perhaps to balance the destabilizing effects of the very high negative net charge of this acidic protein. Statistical studies on peptide sequences suggest that charge imbalance is an important factor favoring the unfolded state and that high aromatic content is an important factor favoring the folded state (28Xie Q. Arnold G. Romero P. Obradovic Z. Garner E. Dunker A. Genome Inform Ser. Workshop Genome Inform. 1998; 9: 193-200PubMed Google Scholar). Most of the aromatic residues, especially in domains II and III, are highly conserved between cCSQ and sCSQ (Fig. 3).Fig. 2The molecular surface of canine cardiac calsequestrin (A) and rabbit skeletal calsequestrin (B) (1A8Y). The molecular surface shows the electrostatic potential from -45.24 kBT to 35.44 kBT (where kB is the Boltzman constant, and T is the absolute temperature). Red is negative, blue is positive, and white is uncharged or hydrophobic. Overall, both isomers show extreme negative values of their surface electropotential. The electrostatic potential surface was calculated by the program GRASP (35Nicholls A. Sharp K.A. Honig B. Proteins. 1991; 11: 281-296Crossref PubMed Scopus (5316) Google Scholar).View Large Image Figure ViewerDownload Hi-res image Download (PPT)Fig. 3Comparison of amino acids sequences for rabbit skeletal and canine cardiac calsequestrins. The individual domains, I, II, and III, are marked with different colors. Non-identical residues are highlighted by bold characters.View Large Image Figure ViewerDownload Hi-res image Download (PPT) Two N-terminal residues and 12 C-terminal residues starting from Asn-351 are disordered, and the corresponding electron densities for these regions are not visible, which is similar to the crystal structure of sCSQ. Therefore, it is very likely that the C-terminal 38 amino acids of the wild-type cCSQ including the 27 truncated residues, which are composed of large numbers of aspartic acids and glutamic acids, are all disordered. The reported phosphorylation site of Thr-353 is located at the beginning of the stretch of structural disorder as in most cases of the known phosphorylation sites of other proteins (29Johnson L. O'Reilly M. Curr. Opin. Struct. Biol. 1996; 6: 762-769Crossref PubMed Scopus (90) Google Scholar). In general, the loop areas connecting the helices and strands show elevated temperature factors and in many cases have different conformations in the two molecules of the asymmetric unit and also in cCSQ and sCSQ. In particular, the loop-containing residues 328-333, which is disordered in the case of sCSQ, becomes ordered in cCSQ. The other 4 areas of partially high temperature factors are residues 38-47, 153-158, 201-204,and 257-261. The residue Asp-306, which is known to be the cause of the autosomal recessive form of CPVT, is located near the front-to-front dimer interface in a solvent-exposed loop (Fig. 4A). Dimer Interface—The asymmetric unit of ΔC27 cCSQ crystal lattice is a front-to-front dimer (Fig. 4A). This cCSQ dimer has a root mean square deviation value of 1.4 Å with the same type of front-to-front dimers observed in sCSQ. As seen in the front-to-front interface of sCSQ, the arm exchange of the N-terminal 13 amino acids between the subunits is important in this electronegative interface of the cCSQ dimer (Fig. 4B). Previously, we showed that deletion of the first 13 residues (ΔN13 cCSQ) leads to a high molecular weight aggregate, probably due to random interactions among CSQ molecules (13Park H. Wu A. Dunker A. Kang C. J. Biol. Chem. 2003; 278: 16176-16182Abstract Full Text Full Text PDF PubMed Scopus (103) Google Scholar). All but 3 of these 13 residues are identical between canine cCSQ and skeletal sCSQ (Fig. 3). The back-to-back type molecular contact and consequent linear polymer observed in the crystal lattice of sCSQ are not observed in this C-terminal-truncated cCSQ. This enforces the importance of the two kinds of dimer interfaces, back-to-back and front-to-front, in the polymeric behavior of this protein. Previously, we determined the cation dependence of the apparent molecular weight of ΔC27 cCSQ using size exclusion chromatography equipped with static light-scattering instruments (13Park H. Wu A. Dunker A. Kang C. J. Biol. Chem. 2003; 278: 16176-16182Abstract Full Text Full Text PDF PubMed Scopus (103) Google Scholar). This truncation mutant shows the same Ca2+-dependent dimerization pattern after an increase in calcium ions despite its 27 fewer acidic residues in its tail but stays in the dimer state and does not show any further Ca2+-dependent polymerization as observed in the case of wild-type cCSQ. Because this C-terminal truncation mutant has an intact N terminus, the dimers seen on these mutants were predicted to be a front-to-front type of dimer, which is now confirmed by the crystal structure of ΔC27 cCSQ. Ca2+ Binding Property—To study the potential for coupling between Ca2+ binding and the polymerization of CSQ, Ca2+ binding capacities of wild-type cCSQ, ΔC27 cCSQ, and wild-type sCSQ were analyzed by atomic absorption spectroscopy. Previously, the maximum capacity of canine cCSQ for Ca2+ was reported as ∼18 mol of Ca2+/molecule, which is about half the capacity of rabbit sCSQ (2MacLennan D. Wong P. Proc. Natl. Acad. Sci. U. S. A. 1971; 68: 1231-1235Crossref PubMed Scopus (426) Google Scholar, 16Slupsky J. Ohnishi M. Carpenter M. Reithmeier R. Biochemistry. 1987; 26: 6539-6544Crossref PubMed Scopus (79) Google Scholar, 30Campbell K. MacLennan D. Jorgensen A. Mintzer M. J. Biol. Chem. 1983; 258: 1197-1204Abstract Full Text PDF PubMed Google Scholar, 31Ikemoto N. Nagy B. Bhatnagar G. Gergely J. J. Biol. Chem. 1974; 249: 2357-2365Abstract Full Text PDF PubMed Google Scholar, 32MacLennan D. J. Biol. Chem. 1974; 249: 980-984Abstract Full Text PDF PubMed Google Scholar, 33Ostwald T. MacLennan D. Dorrington K. J. Biol. Chem. 1974; 249: 5867-5871Abstract Full Text PDF PubMed Google Scholar, 34Cozens B. Reithmeier R. J. Biol. Chem. 1984; 259: 6248-6252Abstract Full Text PDF PubMed Google Scholar), but there has never been any report measuring Ca2+ binding capacities of both cCSQ and sCSQ using the same method and the same binding conditions. In our study the buffer condition for measuring Ca2+ binding was chosen based on circular dichroism (CD), fluorescence spectroscopies, 2H. J. Park, I. Y. Park, E. J. Kim, B. Youn, K. Fields, A. K. Dunker, and C. H. Kang, unpublished data. and light-scattering experiments (13Park H. Wu A. Dunker A. Kang C. J. Biol. Chem. 2003; 278: 16176-16182Abstract Full Text Full Text PDF PubMed Scopus (103) Google Scholar). At 300 mm KCl and 10 mm Tris-HCl (pH 7.5), both CD and fluorescence signals, which reflect acquisition of secondary and tertiary structure respectively, reach a plateau. The molecular mass estimate from the light-scattering data for CSQ at this condition is 44 (±1) kDa, corresponding closely to the molecular mass of one CSQ molecule. As shown in Fig. 5A, sCSQ shows a higher Ca2+ binding capacity than cCSQ along the entire range of Ca2+ concentration we have tested, and the ΔC27 cCSQ shows a significantly reduced binding capacity, ∼50% of its wild type. In both wild-type sCSQ and cCSQ there are several changes of curvature that do not fit into a simple binding model, which is clearer after converting the binding data to a Scatchard-type plot (Fig. 5B). Therefore, the binding characters of both CSQs are not even throughout the Ca2+ concentration; instead, they vary or make transitions after a change of Ca2+ concentration. Both cCSQ and sCSQ show sharp increases in their Ca2+ binding capacities between 0 and 1 mm Ca2+ concentration (Fig. 5A) and reach approximate values of 20 and 34 bound Ca2+, respectively, at 1 mm.At 5 mm CaCl2 concentration, Ca2+ binding capacities of cCSQ and sCSQ reached the level of 34 and 45 Ca2+. In the equilibrium dialysis set of ΔC27 cCSQ beyond the Ca2+ concentration of 5 mm, precipitation of this mutant protein was clearly observed, but this insoluble Ca2+-protein complex was readily dissolved by dilution with 300 mm KCl buffer. This precipitation was not apparent in either wild-type sCSQ or cCSQ at the same high Ca2+ concentration. After 5 mm, both cCSQ and sCSQ plateau at about 36 and 50 mol of Ca2+/molecule, respectively, but contrary to a previous report (16Slupsky J. Ohnishi M. Carpenter M. Reithmeier R. Biochemistry. 1987; 26: 6539-6544Crossref PubMed Scopus (79) Google Scholar) both show a sudden increase in the range of 7-10 mm Ca2+ and then rise slowly and steadily after that. In the same range of Ca2+ concentration, the ΔC27 cCSQ rather slowly increases its Ca2+ binding capacity, forming a substantial amount of precipitates. We showed that neither a mutant with reduced C-terminal charges (ΔC27) nor a mutant with its C-terminal tail completely removed (ΔC38) is able to form Ca2+-dependent tetramer (13Park H. Wu A. Dunker A. Kang C. J. Biol. Chem. 2003; 278: 16176-16182Abstract Full Text Full Text PDF PubMed Scopus (103) Google Scholar). Therefore, we attribute this precipitation to ΔC27 aggregating randomly instead of forming a linear polymer via tetramer formation in response to high Ca2+ concentration above ∼5 mm. Canine cCSQ and rabbit sCSQ have a very similar structure, and the surface of cCSQ is less electronegative than that of sCSQ. We also found that the C-terminal tail of cCSQ, which is longer than sCSQ in general and has more negatively charged amino acids, is disordered, as observed in the structure of sCSQ. Truncation of the C-terminal 27 amino acids (ΔC27) still maintains the identical front-to-front interaction as observed in sCSQ through N-terminal arm exchange, but the back-to-back interface is not present. On the other hand, complete removal of the C-terminal tail, ΔC38, dimerizes cCSQ at lower ionic strength in a Ca2+-independent manner. That is, at the KCl concentration of 300-500 mm (without Ca2+), ΔC38 cCSQ already has a substantial dimer population (13Park H. Wu A. Dunker A. Kang C. J. Biol. Chem. 2003; 278: 16176-16182Abstract Full Text Full Text PDF PubMed Scopus (103) Google Scholar). Based on our crystal structure of ΔC27 cCSQ and our previous light-scattering data we conclude, therefore, that the front-to-front dimer forms before the back-to-back dimer and that the dimer observed in the solution of ΔC38 cCSQ might be a nonspecific back-to-back dimer. The negatively charged C-terminal tail in wild-type cCSQ, therefore, inhibits or slows down the formation of the nonspecific back-to-back dimerization. By the same logic, 11 disordered amino acids at the C-terminal tail of ΔC27 cCSQ may not be long enough to induce (or stabilize) the back-to-back dimer interface even at high Ca2+ concentrations but is still long enough to provide enough charge repulsion to prevent the formation of the Ca2+-independent (or nonspecific) back-to-back dimer. Fig. 6 explains the possible coupling mechanisms between high capacity Ca2+ binding and the polymerization of CSQs. From our previous light-scattering data (13Park H. Wu A. Dunker A. Kang C. J. Biol. Chem. 2003; 278: 16176-16182Abstract Full Text Full Text PDF PubMed Scopus (103) Google Scholar), the dimerization of the CSQ molecule occurs in the Ca2+ concentration range of 0-1 mm, and at 3 mm CaCl2 there is already a detectable tetramer population of CSQ. The part of the sharp increase in Ca2+ binding observed at the 0-1 mm concentration, therefore, is due to the formation of the front-to-front dimer. The next increase, observed at 3-5 mm Ca2+, is probably due to the dimer-tetramer transition capturing substantial amounts of Ca2+ ion in the back-to-back interface in addition to the tetramer surface (Fig. 6). It appears that slightly higher concentrations of Ca2+ ion are required for sCSQ to approach each other to form a dimer compared with the cCSQ (inset in Fig. 6). Consequently dimers of sCSQ can sequester more Ca2+ ions than dimers of cCSQ even though the single molecule state (Ca2+ concentration less than ∼0.4 mm) of sCSQ can bind only slightly more Ca2+ ions than cCSQ (inset in Fig. 6). At 0-1 mm Ca2+ concentration ΔC27 cCSQ already shows reduced Ca2+ binding even though it undergoes the same transition to the front-to-front dimer as wild-type cCSQ (Fig. 5A). This dimer of ΔC27 cCSQ reaches the maximum binding capacity of ∼25 Ca2+ ions per molecule, which matches well with the net charge of ΔC27 cCSQ, -49. The sCSQ dimer can bind ∼40 ions per molecule (first plateau of the slope), which is also consistent with the total net charge of the sCSQ molecules, -80. On the contrary, the maximum number of bound Ca2+ per wild-type cCSQ dimer is ∼30, which is less than its charge of -69. Both cCSQ and sCSQ bind many more Ca2+ ions than predicted by their net negative charges as they form their polymeric states. At the highest Ca2+concentrations tested, each sCSQ binds an excess of ∼40 Ca2+ ions, and each cCSQ binds an excess of ∼35 ions. The principle of electroneutrality means that polymer formation involves the binding of about 80 negative charges for sCSQ and about 69 negative charges for cCSQ. Thus, Ca2+regulation within the SR by CSQ polymerization and depolymerization must involve substantial anion binding. One possibility is that the transmembrane flux of a particular anion plays an additional, currently unrecognized regulatory role in binding and releasing Ca2+ via CSQ polymerization and depolymerization. If so, studies of the anion dependence of Ca2+-associated CSQ polymerization could reveal important new insights regarding Ca2+ regulation within the sarcoplasmic and endoplasmic reticula. In summary, the Ca2+-dependent and sequential formation of two different types of dimer interface, front-to-front and back-to-back, through its N-terminal arm and its acidic C-terminal tail is a key feature in the polymerization of both sCSQ and cCSQ. This dynamic polymerization can be directly linked to the high capacity, low affinity Ca2+ binding of CSQ. Our studies on sequence conservation across both cCSQ and sCSQ molecules indicate that the residues involved in those two interfaces are the most highly conserved residues in the entire structure. 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. This polymerization of CSQ is promoted by Ca2+ and inhibited by K+, which is explained well in terms of trafficking K+ and Ca2+ in SR because of the high concentration of CSQ (100 mg/ml) and physiologically varying concentrations (opposite direction) of those two ions. The likely added importance of anion binding in this process provides a new insight that requires further investigation. The use of Ca2+ as a cross-linker rather than as a tightly bound form free of H2O also speeds dissociation (14MacLennan D. Reithmeier R. Nat. Struct. Biol. 1998; 5: 409-411Crossref PubMed Scopus (57) Google Scholar). In our crystal structures, sCSQ shows lower electric potential on its surface than cCSQ. Therefore, growing sCSQ polymers provide more charged surface than cCSQ onto which more Ca2+ ions can be adsorbed. Our data strongly imply the Ca2+-bound state forms linear structures that can be assembled and destroyed dynamically after the flux of K+ and Ca2+. The strong coupling between Ca2+ binding and protein oligomerization provide a distinctive mechanism for facilitated Ca dissociation and for making diffusion in CSQ even more rapid than would be the case for the free three-dimensional diffusion of Ca2+ ions.