Calmodulin (CaM), a conserved eucaryotic protein, can bind specifically to a large number of intracellular proteins and modulate their activity in response to the Ca2+ concentration.1–3 This small 17-kDa acidic protein belongs to a family of homologous calcium-binding proteins4 that bind Ca2+ through the EF-hand motif (e.g., parvalbumin5 or troponin C6). A compact, calcium-free, apo form of CaM is converted to an extended dumbbell-shaped form on binding Ca2+.7
The extended conformation of CaM has been by far the most thoroughly studied, especially by X-ray crystallography.8–13 It consists of two structurally similar domains separated by a flexible 28-residue helix. Each domain has two EF-hand motifs with bound Ca2+. The calcium-induced extension of CaM exposes two hydrophobic pockets, one per domain, which represent the binding sites for target proteins. In some protein targets, the CaM-binding region was located to a sequence of 18 amino acids predicted to form an α-helix.14 On binding to the protein target, the central CaM helix unwinds, and the two hydrophobic pockets wrap around the α-helix of the protein target.15 Structural plasticity of the hydrophobic pockets and flexibility of the central helix are thought to account for the ability of CaM to interact with a variety of different targets in a sequence-independent fashion.14, 16
We have determined the crystal structure of calcium-bound CaM from Caenorhabditis elegans (ceCaM) as a part of the Structural Genomics of C. elegans project.17 Besides the conserved features typical for all CaM's, the ceCaM structure has the straightest central helix so far observed in CaM's. This relatively straight helix may be induced by different crystallization conditions and/or by the crystal symmetry.
The standard expression and purification procedures for the C. elegans genome are described at the website http://sgce.cbse.uab.edu. For CaM, however, only the protein with a precise native sequence was useful for crystallographic experiments. The gene was amplified by polymerase chain reaction (PCR) and cloned into the vector pET28b between the restriction sites NcoI and XhoI. The protein was expressed in the Escherichia coli strain BL21(DE3) at 291 K. The bacterial pellet was resuspended in buffer A (50 mM Tris, pH 7.5, 0.5 M NaCl, 1 mM ethylenediaminetetraacetic acid, 0.5 mM phenylmethanesulfonyl fluoride, 5 mM 2-mercaptoethanol), and lysed by sonication. The supernatant was heated to 373 K for 5 min and cooled rapidly on ice. The clarified supernatant was dialyzed against buffer B (50 mM sodium phosphate, pH 7, 1.5 M ammonium sulfate, 5 mM CaCl2) loaded on a phenylsepharose column and washed by the buffer B, with the ammonium sulfate gradient from 1.5 M to 0 M. The CaM was eluted by 20 mM isopropanol, dialyzed against 5 mM CaCl2 with 5 mM sodium cacodylate, pH 6, and concentrated to 5 mg/mL. Crystals were grown by vapor diffusion in hanging drops made of 5 μL of the protein solution and 5 μL of the reservoir solution, consisting of 50% (v/v) 2-methyl-2,4-pentanediol (MPD), 45% (w/v) polyethylene glycol (PEG) 3350, 12% (v/v) dioxane, 5 mM CaCl2, and 50 mM sodium citrate, pH 5.7. The crystals are monoclinic, space group C2, a = 103.39 Å, b = 24.00 Å, c = 60.77 Å, β = 112.14°, and there is one CaM molecule per asymmetric unit. The diffraction data to 2.11 Å were collected at the Advanced Photon Source (APS) beamline 14-BMD. A crystal frozen in liquid nitrogen was mounted in a cryostream at 100 K, and diffraction images were collected in oscillation mode with a step of 2°. The wavelength of the incident radiation was 1.107 Å. Data were processed in HKL200018 (Table I) and phased by molecular replacement in AMoRe.19 A drosophila CaM molecule, with just one conservative amino acid difference, was used as a search model [Protein Data Bank (PDB) code: 4CLN].
Table I. Data Collection and Refinement Statistics
R=free was calculated with a subset of 5% of randomly selected reflections.
The new atomic coordinates were refined in the Crystallography & NMR System (CNS)20 and sessions of interactive model building were performed in O.21 Terminal residues 1–5 and 147–149 were severely disordered and not included in the final model. The crystallographic R-factor is 21.2% and R-free is 27.1% for F > 4σF. Refinement statistics are summarized in Table I. The final model includes 141 amino acid residues, 4 calcium ions, and 91 water sites. The structure was validated in Whatcheck22 and deposited in the PDB (code: 1OOJ).
Results and discussion.
The anisotropic shape and inherent dynamics of the CaM molecule present a challenge for crystallization and model refinement. As in previous isotropically refined CaM structures, the R-factor of ceCaM tends to be higher, despite low-resolution cutoffs and filtering F/σF. In addition to disordered termini, there are more residues with weak electron density or poorly defined side chains, especially in the central helix and in the hydrophobic binding pockets. The only atomic-resolution crystal structure refined anisotropically has shown discretely disordered residues (multiple conformations) in these regions and suggested that, as a whole, the CaM molecule occupies a large number of conformational substates.13 Thus, the isotropic refinement model, commonly used for medium- and high-resolution structures, cannot properly account for the CaM dynamics in the crystal lattice.
The ceCaM molecule has all the typical features of an extended CaM structure (Fig. 1). Four fully occupied calcium-binding sites are formed by each EF-hand motif, the motifs are paired across a short, two-strand antiparallel β-sheet to form domains, and the domains are connected by the long central helix. A superposition of the 141 equivalent Cα atoms between ceCaM (residues 6–146) and the 1-Å structure of the fungal CaM (PDB code: 1EXR) gives a root-mean-square deviation (RMSD) of 1.91 Å. However, the fit is much better if individual domains are superimposed. The 63-residue N-terminal (9–71) and C-terminal (82–144) domains from ceCaM and 1EXR have RMSDs of 0.53 Å and 0.66 Å, respectively. Similar trends are observed between ceCaM and the other CaM's. The N- and C-terminal domains of a given CaM molecule superimpose to within an RMSD of 1 Å.
Most differences between CaM's come from the central helix, which is either significantly bent8 or curved9, 10, 12, 13 in the previously determined structures. The radius of curvature of the central helix (residues 65–92) was determined by fitting a circle to the centers of ideal helices for each of the five consecutive residues. This yields results similar to the previously proposed method.11 The previous CaM structures have radii of curvature falling in the range of 48–75 Å. The central helix of ceCaM, with radius of 116 Å, is the straightest of all compared structures (Fig. 1).
The calcium-induced CaM from various species has been crystallized essentially from MPD, and the crystals have always grown in the space group P1, with very similar unit cell parameters.8–13 In this study, addition of polar dioxane and PEG (see Experimental section) yielded crystals with different packing in the space group C2, and with approximately 8% lower solvent content. In the previous structures, 14 molecules surrounded 1 CaM molecule within 5 Å, whereas only 13 surrounding molecules in the present structure are within the same distance. Despite the tighter packing, the long central helix in ceCaM has almost no contacts with the symmetry-related molecules, similar to the structures in P1.
Although there is an identical sequence in the central helix region of C. elegans, drosophila, rat, and human CaM, a significant difference is observed in the radius of curvature in ceCaM (Fig. 1). The fungal CaM has seven generally conservative sequence changes in the region, and the radius of curvature is relatively comparable with drosophila, rat, and human CaM. Thus, the observed curvature of the central helix is not a function of the sequence. Neither are there any direct symmetry interactions that could alter the helix geometry. The possible solvent-induced curvature and distortion of an α-helix23 have been proposed in the fungal CaM structures.11, 12 However, the recent atomic-resolution stucture,13 from the same species, and crystallized under the same conditions, did not show any ordered solvent molecules interacting with the central helix that could elicit the significant curvature. Similarly, there are very few water molecules in the hydrogen-bonding distance from the central helix in the ceCaM structure.
The relatively straight central helix in ceCaM may be induced by the packing restraints imposed by the crystallographic, two-fold axis of the space group C2. The molecules pack along the unique axis, which is approximately perpendicular to the helical axis, and the central helices are arranged in an antiparallel mode, whereas in space group P1, the central helices can be arranged only in a parallel mode, and a significant helix curvature is always observed.