Portal de Periódicos da CAPES

Conformation of the Drosophila Motor Protein Non-claret Disjunctional in Solution from X-ray and Neutron Scattering

2001; Elsevier BV; Volume: 276; Issue: 27 Linguagem: Inglês

10.1074/jbc.m103618200

1083-351X

Dmitri I. Svergun, Giuseppe Zaccaı̈, Marc Malfois, Richard H. Wade, Michel H. J. Koch, Frank Kozielski,

Biochemical and Molecular Research

The quaternary structures of monomeric and dimeric Drosophila non-claret disjunctional (ncd) constructs were investigated using synchrotron x-ray and neutron solution scattering, and their low resolution shapes were restoredab initio from the scattering data. The experimental curves were further compared with those computed from crystallographic models of one monomeric and three available dimeric ncd structures in the microtubule-independent ADP-bound state. These comparisons indicate that accounting for the missing parts in the crystal structures for all these constructs is indispensable to obtain reasonable fits to the scattering patterns. A ncd construct (MC6) lacking the coiled-coil region is monomeric in solution, but the calculated scattering from the crystallographic monomer yields a poor fit to the data. A tentative configuration of the missing C-terminal residues in the form of an antiparallel β-sheet was found that significantly improves the fit. The atomic model of a short dimeric ncd construct (MC5) without 2-fold symmetry is found to fit the data better than the symmetric models. Addition of the C-terminal residues to both head domains gives an excellent fit to the x-ray and neutron experimental data, although the orientation of the β-sheet differs from that of the monomer. The solution structure of the long ncd construct (MC1) including complete N-terminal coiled-coil and motor domains is modeled by adding a straight coiled-coil section to the model of MC5. The quaternary structures of monomeric and dimeric Drosophila non-claret disjunctional (ncd) constructs were investigated using synchrotron x-ray and neutron solution scattering, and their low resolution shapes were restoredab initio from the scattering data. The experimental curves were further compared with those computed from crystallographic models of one monomeric and three available dimeric ncd structures in the microtubule-independent ADP-bound state. These comparisons indicate that accounting for the missing parts in the crystal structures for all these constructs is indispensable to obtain reasonable fits to the scattering patterns. A ncd construct (MC6) lacking the coiled-coil region is monomeric in solution, but the calculated scattering from the crystallographic monomer yields a poor fit to the data. A tentative configuration of the missing C-terminal residues in the form of an antiparallel β-sheet was found that significantly improves the fit. The atomic model of a short dimeric ncd construct (MC5) without 2-fold symmetry is found to fit the data better than the symmetric models. Addition of the C-terminal residues to both head domains gives an excellent fit to the x-ray and neutron experimental data, although the orientation of the β-sheet differs from that of the monomer. The solution structure of the long ncd construct (MC1) including complete N-terminal coiled-coil and motor domains is modeled by adding a straight coiled-coil section to the model of MC5. non-claret disjunctional piperazine-N,N′-bis(2-ethanesulfonic acid), e/nm3, electrons/nm3 The superfamily of kinesin motor proteins has about 265 members thus far and can be subdivided into at least 10 different subfamilies. The two kinesin motors that have been investigated in the most detail are conventional kinesin, a member of the kinesin heavy chain subfamily, and Drosophilancd1 (Fig.1), which belongs to the C-terminal subfamily, which is the largest and most divergent subfamily (1Kim A.J. Endow S.A. J. Cell Sci. 2000; 113: 3861-3862PubMed Google Scholar). Both proteins are dimers and have a three-domain structure. Their motor domains, which are some 340 residues long, contain the ATP- and microtubule-binding sites. Dimerization is mediated by a stalk domain predicted to form an α-helical coiled-coil. In conventional kinesin, the coiled-coil domain is predicted to be interrupted by hinge regions (2Bloom G.S. Endow S.A. Protein Profile. 1995; 2: 1105-1171PubMed Google Scholar), which increase the flexibility of the entire molecule. In contrast, ncd has a continuous, uninterrupted coiled-coil. The third domain is a small globular tail responsible for cargo binding. Conventional kinesin has its motor domain at the N terminus of its polypeptide chain and moves processively to the plus end of microtubules, making several hundred steps before detaching (3Howard J. Hudspeth A.J. Vale R.D. Nature. 1989; 342: 154-158Crossref PubMed Scopus (749) Google Scholar, 4Block S.M. Goldstein L.S.B. Schnapp B.J. Nature. 1990; 348: 348-352Crossref PubMed Scopus (856) Google Scholar, 5Hackney D.D. Nature. 1995; 377: 448-450Crossref PubMed Scopus (163) Google Scholar). In contrast, ncd has its motor domain at the C terminus of the polypeptide chain (6Chandra R. Salmon E.D. Erickson H.P. Lockhart A. Endow S.A. J. Biol. Chem. 1993; 268: 9005-9013Abstract Full Text PDF PubMed Google Scholar) and moves to the minus end of microtubules (7Walker R.A. Salmon E.D. Endow S.A. Nature. 1990; 347: 780-782Crossref PubMed Scopus (294) Google Scholar). There is growing evidence that ncd is not processive (8deCastro M.J. Ho C.H. Stewart R.J. Biochemistry. 1999; 38: 5076-5081Crossref PubMed Scopus (60) Google Scholar, 9Foster K.A. Gilbert S.P. Biochemistry. 2000; 39: 1784-1791Crossref PubMed Scopus (47) Google Scholar). Although conventional kinesins and C-terminal motors move to opposite ends of microtubules, the three-dimensional structures of their motor domains are remarkably similar (10Kull F.J. Sablin E.P. Lau R. Fletterick R.J. Vale R. Nature. 1996; 380: 550-555Crossref PubMed Scopus (581) Google Scholar, 11Sablin E.P. Kull J.F. Cooke R. Vale R.D. Fletterick R.J. Nature. 1996; 380: 555-559Crossref PubMed Scopus (324) Google Scholar, 12Sack S. Müller J. Marx A. Thormählen M. Mandelkow E.-M. Brady S.T. Mandelkow E. Biochemistry. 1997; 36: 16155-16165Crossref PubMed Scopus (181) Google Scholar, 13Gulick A.M. Song H. Endow S.A. Rayment I. Biochemistry. 1998; 37: 1769-1776Crossref PubMed Scopus (89) Google Scholar). Therefore, the opposite directionality cannot be explained on the basis of the crystal structures of their motor domains. Studies of the three-dimensional structure of dimeric kinesin and ncd motor domains on microtubules in the presence of 5′-adenylylimidodiphosphate provided a valuable hint (14Arnal I. Metoz F. DeBonis S. Wade R.H. Curr. Biol. 1996; 6: 1265-1270Abstract Full Text Full Text PDF PubMed Scopus (86) Google Scholar, 15Hirose K. Lockhart A. Cross R. Amos L.A. Proc. Natl. Acad. Sci. U. S. A. 1996; 93: 9539-9544Crossref PubMed Scopus (130) Google Scholar). In both studies, it was clearly shown that the motor-microtubule complexes have one attached and one unattached head per tubulin dimer. The unattached heads of kinesin and ncd have distinctly different conformations and tilt toward the microtubule plus end for kinesin and toward the microtubule minus end for ncd. This suggested at once that directionality could be determined by differences in the dimer conformations. The crystal structures of dimeric kinesin (16Kozielski F. Sack S. Marx A. Thormählen M. Schönbrunn E. Biou V. Thompson A. Mandelkow E.-M. Mandelkow E. Cell. 1997; 91: 985-994Abstract Full Text Full Text PDF PubMed Scopus (361) Google Scholar) and dimeric ncd (17Sablin E.P. Case R.B. Dai S.C. Hart C.L. Ruby A. Vale R.D. Fletterick R.J. Nature. 1998; 395: 813-816Crossref PubMed Scopus (185) Google Scholar, 18Kozielski F. De Bonis S. Burmeister W.P. Cohen-Addad C. Wade R.H. Structure Fold. Des. 1999; 7: 1407-1416Abstract Full Text Full Text PDF Scopus (48) Google Scholar), all in the microtubule-independent ADP form, do indeed have different overall conformations. Only one crystal structure is currently available for dimeric kinesin, and this displays an asymmetric conformation, with the two heads being related to each other by a rotation of about 120°. For ncd, three different dimeric crystal structures from two independent studies are available. The first dimeric ncd structure has a perfect 2-fold symmetry, with two identical heads (17Sablin E.P. Case R.B. Dai S.C. Hart C.L. Ruby A. Vale R.D. Fletterick R.J. Nature. 1998; 395: 813-816Crossref PubMed Scopus (185) Google Scholar). The two ncd dimers in the asymmetric unit of the crystal form of the second study are different (18Kozielski F. De Bonis S. Burmeister W.P. Cohen-Addad C. Wade R.H. Structure Fold. Des. 1999; 7: 1407-1416Abstract Full Text Full Text PDF Scopus (48) Google Scholar). In one dimer, the two heads are related by a 179° rotation about an axis defined by the coiled-coil, whereas in the other one, the second head is rotated about 10° away from the 2-fold symmetry position. Comparison between experimental solution scattering curves and those evaluated from crystallographic structures is an established method to verify the structural similarity between macromolecules in crystals and in solution (19Svergun D.I. Barberato C. Koch M.H.J. Fetler L. Vachette P. Proteins. 1997; 27: 110-117Crossref PubMed Scopus (72) Google Scholar, 20Trewhella J. Curr. Opin. Struct. Biol. 1997; 7: 702-708Crossref PubMed Scopus (44) Google Scholar, 21Svergun D.I. Petoukhov M.V. Koch M.H.J. König S. J. Biol. Chem. 2000; 275: 297-303Abstract Full Text Full Text PDF PubMed Scopus (44) Google Scholar). The present x-ray and neutron scattering study aims to check the three available dimeric ncd crystal structures against the solution scattering data. The structures of several monomeric and dimeric ncd constructs are analyzed, and an attempt is made to model the conformation of the C-terminal region, which has not been resolved in any of the known ncd crystal structures thus far. The plasmids pET/MC1, pET/MC5, and pET/MC6 were a kind gift of Sharon Endow (Duke Medical Center, Durham, NC). All protein purification steps were performed at 4 °C. The purification of monomeric MC6 (corresponding to residues Met333-Lys700 of ncd and 14 additional amino acids), dimeric MC5 (corresponding to residues Ala295-Lys700 and 14 additional amino acids), and dimeric MC1 (corresponding to residues Leu209-Lys700 of ncd and 13 additional amino acids) was performed as described previously (6Chandra R. Salmon E.D. Erickson H.P. Lockhart A. Endow S.A. J. Biol. Chem. 1993; 268: 9005-9013Abstract Full Text PDF PubMed Google Scholar), with some modifications. Cells were grown overnight at 37 °C in 2 liters of Luria-Bertani medium supplemented with 150 mg/ml ampicillin and induced with 0.05 mm/ml isopropyl-β-d-thiogalactopyranoside. The temperature was lowered to 20 °C, and the cells were further induced for 12 h. The cells were centrifuged and passed twice through a French press and centrifuged at 15,000 × g for 1 h. Further purification steps were performed as described previously (6Chandra R. Salmon E.D. Erickson H.P. Lockhart A. Endow S.A. J. Biol. Chem. 1993; 268: 9005-9013Abstract Full Text PDF PubMed Google Scholar). As an additional step, the proteins were concentrated using a CENTRICON YM-30 (Amicon) and loaded onto a self-packed gel filtration column (Sephacryl S-300; Amersham Pharmacia Biotech) previously equilibrated in buffer A (20 mm PIPES, pH 7.4, 200 mm NaCl, 1 mm MgCl2, 1 mm dithiothreitol, and 1 mm EGTA). The eluted peak fractions were concentrated, and fractions of 150 μl/tube in the concentration range of 1–20 mg/ml were frozen in liquid nitrogen and stored at −80 °C until use. Albumin (67 kDa), ovalbumin (43 kDa), chymotrypsinogen (25 kDa), and ribonuclease A (13.7 kDa) were used as standards. For data collection, the freshly thawed proteins were either measured directly in 100-μl quartz cuvettes, centrifuged at 15,000 ×g for 30 min at 4 °C, or subjected to an additional gel filtration to obtain the protein solutions under the best possible monodisperse conditions. Equilibrium binding studies performed at 24.7 °C showed that in the presence of different nucleotides, MC1 has a tendency to aggregate (22Foster K.A. Correia J.J. Gilbert S.P. J. Biol. Chem. 1998; 273: 35307-35318Abstract Full Text Full Text PDF PubMed Scopus (48) Google Scholar). Additional experiments indicated that the protein is more stable at lower temperatures. Therefore, the x-ray scattering measurements were performed at 12 °C, and the neutron scattering experiments were performed at 6 °C. The synchrotron radiation x-ray scattering data were collected on the X33 camera (23Koch M.H.J. Bordas J. Nucl. Instrum. Methods. 1983; 208: 461-469Crossref Scopus (287) Google Scholar, 24Boulin C. Kempf R. Koch M.H.J. McLaughlin S.M. Nucl. Instrum. Methods. 1986; A249: 399-407Crossref Scopus (296) Google Scholar, 25Boulin C.J. Kempf R. Gabriel A. Koch M.H.J. Nucl. Instrum. Methods. 1988; A269: 312-320Crossref Scopus (232) Google Scholar) of the European Molecular Biology Laboratory on the storage ring DORIS III of the Deutsches Elektronen Synchrotron (DESY) with multiwire proportional chambers with delay line readout (26Gabriel A. Dauvergne F. Nucl. Instrum. Methods. 1982; 201: 223-224Crossref Scopus (139) Google Scholar). The scattering patterns were recorded at a sample-detector distance of 2.0 m to cover the range of momentum transfer 0.23 nm−1 < s < 3.5 nm−1 (s = 4πsinθ/λ, 2θ is the scattering angle, and λ = 0.15 nm is the wavelength). The data were normalized to the intensity of the incident beam and corrected for the detector response, the scattering of the buffer was subtracted, and the difference curves were scaled for protein concentration. All procedures involved statistical error propagation using the SAPOKO program. 2D. I. Svergun and M. Koch, unpublished observations. The data obtained on samples with protein concentrations between 1 and 20 mg/ml were extrapolated to zero concentration following standard procedures (27Feigin L.A. Svergun D.I. Structure Analysis by X-ray and Neutron Scattering. Plenum Press, New York1987: 70Google Scholar). For MC5 and MC1, the data were also collected at a sample-detector distance of 3.5 m covering the range 0.13 nm−1 < s < 2.0 nm−1 and merged with the higher angle data to yield the final composite scattering curves. The molar masses of the solutes were calculated by comparison with the forward scattering from reference solutions of bovine serum albumin (molar mass = 66 kDa). The maximum dimensions of the particles in solution (D max) were estimated using the orthogonal expansion program ORTOGNOM (28Svergun D.I. J. Appl. Crystallogr. 1993; 26: 258-267Crossref Scopus (106) Google Scholar). The forward scattering I(0) and the radii of gyration (R g) were evaluated using the Guinier approximation (27Feigin L.A. Svergun D.I. Structure Analysis by X-ray and Neutron Scattering. Plenum Press, New York1987: 70Google Scholar) and the indirect transform package GNOM (29Svergun D.I. Semenyuk A.V. Feigin L.A. Acta Crystallogr. 1988; A44: 244-250Crossref Scopus (277) Google Scholar,30Svergun D.I. J. Appl. Crystallogr. 1992; 25: 495-503Crossref Scopus (2987) Google Scholar), the latter of which also provided the distance distribution function p(r) of the particles. Neutron scattering experiments on MC5 were performed on the D22 small-angle scattering instrument at the Institut Laue-Langevin in Grenoble, France (31Büttner H.G. Lelièvre-Berna E. Pinet F. The Yellow Book, Guide to Neutron Research Facilities at the ILL. Institut Laue-Langevin, Grenoble, France1996: 76-77Google Scholar). Samples with protein concentrations between 1 and 10 mg/ml were contained in quartz cells (Helma) of 1.0- and 2.0-mm optical path length for H2O and D2O solutions, respectively. The data were collected using neutrons with a wavelength of λ = 0.6 nm and a spectral width (Δλ/λ) of 8% with a sample detector distance of 4.0 m, to cover the range of momentum transfer 0.1 nm−1 < s < 2.3 nm−1. Data were corrected for buffer scattering and normalized to the scattering of 1.0 mm of H2O, in the standard manner. The molar mass of the solute was computed from the normalized data extrapolated to infinite dilution and to zero scattering angle (32Zaccai G. Jacrot B. Annu. Rev. Biophys. Bioeng. 1983; 12: 139-157Crossref PubMed Scopus (114) Google Scholar, 33Jacrot B. Zaccai G. Biopolymers. 1981; 20: 2414-2426Crossref Scopus (290) Google Scholar). The atomic model of the monomeric ncd was taken from the kinesin home page, and the models of dimeric ncd were taken from the Protein Data Bank (34Bernstein F.C. Koetzle T.F. Williams G.J.B. Meyer Jr., E.F. Brice M.D. Rodgers J.R. Kennard O. Shimanouchi T. Tasumi M. J. Mol. Biol. 1977; 112: 535-542Crossref PubMed Scopus (8183) Google Scholar), entries 2NCD and 1CZ7, respectively. The scattering curves from the atomic models were calculated using the CRYDAM program, 3M. Malfois and D. I. Svergun, manuscript in preparation. an enhanced version of the CRYSOL (35Svergun D.I. Barberato C. Koch M.H.J. J. Appl. Crystallogr. 1995; 28: 768-773Crossref Scopus (2790) Google Scholar) and CRYSON programs (36Svergun D.I. Richards S. Koch M.H.J. Sayers Z. Kuprin S. Zaccai G. Proc. Natl. Acad. Sci. U. S. A. 1998; 95: 2267-2272Crossref PubMed Scopus (805) Google Scholar) that takes the scattering from the solvation shell into account. The macromolecule is surrounded by a 0.3-nm-thick hydration layer with an adjustable density (ρ b ) that may differ from that of the bulk solvent (ρ s ). The scattering from the particle in solution is I(s)=‖Aa(s)−ρsAs(s)+δρbAb(s)‖2ΩEquation 1 where A a (s) is the scattering amplitude from the particle in vacuo,A s (s) andA b (s) are, respectively, the scattering amplitudes from the excluded volume and the hydration layer, both with unitary density δρ b =ρ b − ρ s , and stands for the average over all particle orientations (Ω is the solid angle in reciprocal space, s =(s, Ω)). The program uses the multipole expansion of the scattering amplitudes to facilitate the spherical average in Eq. 1. Given the atomic coordinates, the program fits the experimental scattering curve by adjusting the excluded volume of the particle V and the contrast of the hydration layerδρ b to minimize the discrepancy χ2=1N−1∑j=1NI(sj)−Iexp(sj)ς(sj)2Equation 2 where N is the number of experimental points, andI(s),I exp(s), andς(s) denote the calculated intensity, the experimental intensity, and its standard deviation, respectively. The low resolution shapes of the ncd constructs were determinedab initio from the x-ray scattering curves. To remove the scattering contribution due to the internal structure of the particles, appropriate constants were subtracted from the experimental data to ensure that the intensity decays as s−4, following Porod's law (37Porod G. Glatter O. Kratky O. Small-angle X-ray Scattering. Academic Press, London1982: 17-51Google Scholar) for homogeneous particles. This procedure yields an approximation of the shape scattering curve (i.e.scattering due to the excluded volume of the particle filled by constant density), and these data were used in the ab initioanalysis (38Svergun D.I. Biophys. J. 1999; 76: 2879-2886Abstract Full Text Full Text PDF PubMed Scopus (1756) Google Scholar). A sphere of diameter D max is filled by densely packed small spheres (dummy atoms) of radiusr 0 ≪ D max. The structure of this dummy atoms model is defined by configuration vector X, assigning an index to each atom corresponding to solvent (0) or solute particle (1Kim A.J. Endow S.A. J. Cell Sci. 2000; 113: 3861-3862PubMed Google Scholar). The scattering intensity from the dummy atoms model is computed as I(s)=2π2∑l=0∞∑m=−ll‖Alm(s)‖2Equation 3 where the partial amplitudes are (38Svergun D.I. Biophys. J. 1999; 76: 2879-2886Abstract Full Text Full Text PDF PubMed Scopus (1756) Google Scholar, 39Stuhrmann H.B. Acta Crystallogr. 1970; A26: 297-306Crossref Scopus (153) Google Scholar) as shown below. Alm(s)=il2/πva∑jjl(srj)Ylm*(ωj)Equation 4 Here, v a = (4πr 03/3)/0.74 is the displaced volume per dummy atom, the sum runs over the atoms withX j = 1 (particle atoms),r j and ω j are their polar coordinates, and j l (x) denotes the spherical Bessel function. In keeping with the low resolution of the solution scattering data, the method searches for a configurationX minimizing f(X) = χ2+αP(X), where α > 0 is a positive parameter, and the penalty termP(X) ensures that the dummy atoms model has low resolution with respect to the packing radiusr0. The minimization is performed starting from a random configuration using the simulated annealing method (40Kirkpatrick S. Gelatt Jr., C.D. Vecci M.P. Science. 1983; 220: 671-680Crossref PubMed Scopus (31663) Google Scholar); the details of the shape determination procedure are presented elsewhere (38Svergun D.I. Biophys. J. 1999; 76: 2879-2886Abstract Full Text Full Text PDF PubMed Scopus (1756) Google Scholar, 41Svergun D.I. Malfois M. Koch M.H.J. Wigneshweraraj S.R. Buck M. J. Biol. Chem. 2000; 275: 4210-4214Abstract Full Text Full Text PDF PubMed Scopus (30) Google Scholar). This construct includes residues Met333-Lys700 of ncd (6Chandra R. Salmon E.D. Erickson H.P. Lockhart A. Endow S.A. J. Biol. Chem. 1993; 268: 9005-9013Abstract Full Text PDF PubMed Google Scholar) and corresponds to the complete motor domain and part of α7 forming the core-linker interface. This well-characterized ncd motor fragment is known to be monomeric in solution at low concentrations. At concentrations of up to 10 μm, there is no indication that MC6 forms aggregates (22Foster K.A. Correia J.J. Gilbert S.P. J. Biol. Chem. 1998; 273: 35307-35318Abstract Full Text Full Text PDF PubMed Scopus (48) Google Scholar). A very similar construct (Arg335-Lys700) has been crystallized, and its structure has been solved in the ADP form to 2.5 Å resolution (11Sablin E.P. Kull J.F. Cooke R. Vale R.D. Fletterick R.J. Nature. 1996; 380: 555-559Crossref PubMed Scopus (324) Google Scholar). Compared with MC6, the crystallographic model includes 321 amino acids, lacking the first 14 N-terminal and the last 33 C-terminal residues. The coordinates of a head domain from the dimeric ncd structure (18Kozielski F. De Bonis S. Burmeister W.P. Cohen-Addad C. Wade R.H. Structure Fold. Des. 1999; 7: 1407-1416Abstract Full Text Full Text PDF Scopus (48) Google Scholar) that includes the N-terminal α7 allowed us to complement the model at the N terminus of the protein starting from residue Met333. Additionally, loop L11, which is present in the monomeric structure (11Sablin E.P. Kull J.F. Cooke R. Vale R.D. Fletterick R.J. Nature. 1996; 380: 555-559Crossref PubMed Scopus (324) Google Scholar) but absent in the two dimeric crystal structures (17Sablin E.P. Case R.B. Dai S.C. Hart C.L. Ruby A. Vale R.D. Fletterick R.J. Nature. 1998; 395: 813-816Crossref PubMed Scopus (185) Google Scholar, 18Kozielski F. De Bonis S. Burmeister W.P. Cohen-Addad C. Wade R.H. Structure Fold. Des. 1999; 7: 1407-1416Abstract Full Text Full Text PDF Scopus (48) Google Scholar), was added. The completeness of the crystallographic model was thus increased from 87% to 91%, and the modified atomic model of monomeric MC6 is presented in Fig. 2, top row. The processed scattering curve from MC6 in Fig.3 yields a molecular mass of 41 ± 3 kDa, in good agreement with the value estimated from the primary sequence of the monomeric protein (44 kDa). In contrast, the scattering curve computed by the CRYDAM program from the model in Fig. 2(top row) gives a poor fit to the experimental data (Fig. 3, ■; χ = 2.1). The maximum dimension and the radius of gyration calculated from the crystallographic model (7.5 and 2.29 nm, respectively) are significantly smaller than their experimental counterparts (9.0 ± 0.5 and 2.58 ± 0.04 nm, respectively). An attempt to improve the fit by assuming partial dimerization of MC6 in solution yielded an only marginally lower discrepancy (χ = 1.9 at 4% of dimers). It can thus be concluded that MC6 is monomeric in solution, even at relatively high (up to 20 mg/ml) concentrations, but it has a significantly more extended conformation than the modified crystallographic model in Fig. 2 (top row). To obtain independent information about the overall structure of MC6 in solution, its shape was restored ab initio from the experimental data as described under “Materials and Methods.” Several independent restorations starting from different random approximations yielded reproducible results, and a typical solution is presented in Fig. 2 (middle row) in the orientation that best matches the crystallographic model. The calculated scattering curve from the ab initio model (Fig. 2, ○) deviates significantly from the raw experimental data at higher angles, as it should, but neatly fits the shape scattering curve with χ = 0.73. The ab initio restoration thus confirms that the solution structure of MC6 is more anisometric than the crystallographic model. This discrepancy is not surprising given that about 30 C-terminal residues are still missing in the model in Fig. 2 (top row). These residues are disordered in all ncd crystal structures solved thus far (11Sablin E.P. Kull J.F. Cooke R. Vale R.D. Fletterick R.J. Nature. 1996; 380: 555-559Crossref PubMed Scopus (324) Google Scholar, 17Sablin E.P. Case R.B. Dai S.C. Hart C.L. Ruby A. Vale R.D. Fletterick R.J. Nature. 1998; 395: 813-816Crossref PubMed Scopus (185) Google Scholar, 18Kozielski F. De Bonis S. Burmeister W.P. Cohen-Addad C. Wade R.H. Structure Fold. Des. 1999; 7: 1407-1416Abstract Full Text Full Text PDF Scopus (48) Google Scholar). The missing region represents about 9% of the molar mass of the protein and should make a significant contribution to the solution scattering pattern. To find a plausible configuration of this region in solution, the C-terminal residues were added to the atomic model of MC6 in several possible conformations (a two- or three-stranded antiparallel β-sheet or a helix-turn-helix motif) and in different orientations with respect to the atomic model. A dozen models were generated, and for each model, the agreement with the experimental data was computed using the CRYDAM program. The solution scattering patterns were rather sensitive to the configuration and orientation of this region, and the discrepancy computed for these models ranged from χ = 0.96 to χ = 2.7. The models in which the last C-terminal residues form an antiparallel two-stranded β-sheet yielded systematically better fits than those forming a three-stranded antiparallel β-sheet or a helix-turn-helix motif. The best model in Fig. 2 (shown in the bottom row) provides an excellent fit (Fig. 3, ▵) to the experimental data with χ = 0.96 and represents a probable configuration of the missing C-terminal region in solution. The ncd construct MC5 (6Chandra R. Salmon E.D. Erickson H.P. Lockhart A. Endow S.A. J. Biol. Chem. 1993; 268: 9005-9013Abstract Full Text PDF PubMed Google Scholar) contains residues Ala295 -Lys700 of ncd and 14 additional residues at the N terminus of the protein. The protein contains part of the N-terminal coiled-coil region, linker, and the complete motor domain. MC5 has previously been shown to be a dimer in solution (6Chandra R. Salmon E.D. Erickson H.P. Lockhart A. Endow S.A. J. Biol. Chem. 1993; 268: 9005-9013Abstract Full Text PDF PubMed Google Scholar). The molecular mass computed from the composite x-ray scattering curve of MC5 in Fig. 4 was 75 ± 8 kDa, confirming that the protein forms dimers in solution (the value estimated from the primary sequence of the dimer is 80 kDa). The D max and R g values are 13 ± 1 and 4.05 ± 0.04 nm, respectively. Three available crystallographic models were tested against the scattering data. The first published crystal structure (17Sablin E.P. Case R.B. Dai S.C. Hart C.L. Ruby A. Vale R.D. Fletterick R.J. Nature. 1998; 395: 813-816Crossref PubMed Scopus (185) Google Scholar) has been solved using a similar ncd construct encoding residues Glu281-Lys700. The final model contains residues Leu303-Met672, thus missing the N-terminal residues 281–302, the loop L11 region (residues Lys588-Thr596), and the C-terminal residues Thr673-Lys700. The crystal structure contains one head domain per asymmetric unit. This dimeric ncd structure (dimer 1) has a perfect 2-fold symmetry about the coiled-coil axis. The second study (18Kozielski F. De Bonis S. Burmeister W.P. Cohen-Addad C. Wade R.H. Structure Fold. Des. 1999; 7: 1407-1416Abstract Full Text Full Text PDF Scopus (48) Google Scholar) is based on a different crystal form containing two ncd dimers per asymmetric unit. The overall conformation of the two heads in one of the dimers (dimer 2) is similar to that of dimer 1. In the second dimer (dimer 3), the head domains are not related by crystallographic symmetry. The second head domain is rotated about 10° away from the 2-fold symmetry position. The scattering curves computed from dimers 1 and 2 yielded the fits to the experimental data with χ = 1.53 (the models and the fits are not shown). Dimer 3 (Fig. 5, top row) provided the fit to the experimental data (Fig. 4, ■) with χ = 1.28 atV = 107 nm3 and δρb = 46 electrons/nm3. Because this model, like that of MC6 in the previous section, lacked the L11 loop and C-terminal region, an attempt was made to improve the fit by adding these regions to each of the monomers. Using the same orientation of the C-terminal loop as in Fig. 3, however, worsened the fit (χ = 1.41; curve not shown). To further understand this discrepancy, a low resolution model of the particle shape was restored ab initio, assuming P2 symmetry (this assumption is justified at low resolution). The ab initio model is presented in Fig. 5 (middle row), and the fit is presented in Fig. 4 (○). This model suggests that the orientation of the C-terminal loop makes the particle more anisometric in the plane of the 2-fold axis. Assuming that the orientation of the loop may change upon dimerization, several models differing with regard to orientation of the β-sheet were constructed and validated against the scattering data. The best model presented in Fig. 5,bottom row, yielded a fit to the experimental data with χ = 0.98 at V = 111 nm3 and δρb = 44 electrons/nm3 (Fig. 4, ▵).Figure 5Models of the MC5 dimer. Top row, crystallographic model; middle row, ab initio low resolution model; bottom row,crystallographic model with the L11 loop added and the C-terminal region in the best configuration. The right views are rotated as described in the Fig. 2 legend.View Large Image Figure ViewerDownload (PPT) The molecular mass of the protein computed from the neutron scattering of MC5 in H2O was 78 ± 5 kDa, in good agreement with the x-ray data and with the estimated molecular mass of the dimeric protein. The maximum dimension of the particle was found to be 12 ± 1 nm in both H2O and D2O. The radii of gyration were 3.96 ± 0.12 and 3.88 ± 0.05 nm in H2O and D2O, respectively. The difference between the x-ray and neutron scattering curves should be attributed to the influence of the hydration shell. As demonstrated previously (34Bernstein F.C. Koetzle T.F. Williams G.J.B. Meyer Jr., E.F. Brice M.D. Rodgers J.R. Kennard O. Shimanouchi T. Tasumi M. J. Mol. Biol. 1977; 112: 535-542Crossref PubMed Scopus (8183) Google Scholar), bound water around the protein surface is denser than bulk water, and this makes the particles appear larger for x-rays, smaller for neutrons in D2O, and nearly unchanged for neutrons in H2O (the scattering length density of the solvent is close to zero in the last case). A direct comparison of the three scattering patterns in Fig. 6 displays the same trend as observed earlier for other proteins and clearly illustrates the influence of the hydration shell (36Svergun D.I. Richards S. Koch M.H.J. Sayers Z. Kuprin S. Zaccai G. Proc. Natl. Acad. Sci. U. S. A. 1998; 95: 2267-2272Crossref PubMed Scopus (805) Google Scholar). The neutron scattering patterns in H2O and D2O computed from the atomic model of MC5 in Fig. 5, top row, fitted the experimental data with χ = 1.40 and 2.47, respectively (Fig.7). Addition of the missing C-terminal residues in the conformation determined from the x-ray data (Fig. 5,bottom row) improves the agreement for both neutron data sets, yielding χ = 1.06 and χ = 1.78 for the curves recorded in H2O and D2O, respectively. This improvement of the fit to the neutron data lends further credit to the model in Fig. 5, bottom row, deduced solely from x-ray measurements.Figure 7Neutron scattering curves from MC5 in H2O and in D2O (●) and calculated scattering patterns from the atomic model of the MC5 dimer without the missing C-terminal region (■) and with the modeled region added as shown in Fig. 5, bottom row (▵).View Large Image Figure ViewerDownload (PPT) MC1 contains residues Leu209-Lys700 (including the complete predicted coiled-coil region and motor domain) of full-length ncd and 13 additional residues at the N terminus of the protein (6Chandra R. Salmon E.D. Erickson H.P. Lockhart A. Endow S.A. J. Biol. Chem. 1993; 268: 9005-9013Abstract Full Text PDF PubMed Google Scholar). This construct has been shown to be dimeric in solution. The composite x-ray scattering curve from MC1 in Fig. 8yields a molecular mass of 123 ± 12 kDa, which agrees well with the expected value of 114 kDa. The maximum dimension of the construct (25 ± 3 nm) and its radius of gyration (6.7 ± 0.3 nm) suggest that the particle is extremely elongated. Compared with MC5 (D max = 13 ± 1 nm), MC1 has 86 additional residues in the N-terminal region predicted to form an α-helical coiled-coil. Because the pitch of the helix is 0.54 nm, the expected length of the additional 86 residues would be 12.9 nm, yieldingD max = 25.9 nm for the complete MC1 protein, in good agreement with the experimental value. An ab initio low resolution model of MC1 in Fig. 9(middle row) also suggests a linearly extended form of the additional coil-coil region. A tentative model of the MC1 was therefore built by extending the coiled-coil portion of the MC5 construct in Fig.5 by 86 residues, as illustrated in Fig. 9, top row. The scattering pattern computed from this model yields a reasonable fit to the experimental data (Fig. 8, ■) with χ = 1.47 atV = 125 nm3 and δρb = 42 electrons/nm3. To establish the configuration of the missing C-terminal region, several plausible models were tested as described above for the MC5 construct. Addition of the C-terminal region in the conformation found for the monomeric MC6 (Fig. 3) worsened the fit to the experimental data, but the configuration identical to that for MC5 (Fig. 5, right panel) yielded the best fit (Fig. 8, ▵) with χ = 0.95 at V= 135 nm3 and δρb = 36 electrons/nm3. The final model of MC1 shown in Fig.9, bottom row, suggests that the relative orientation of the two heads is independent of the length of the coiled-coil.Figure 9Models of the MC1 dimer. Top row, crystallographic model; middle row, ab initio low resolution model; bottom row,crystallographic model with the L11 loop added and the C-terminal region in the best configuration. The right views are rotated as described in the Fig. 2 legend.View Large Image Figure ViewerDownload (PPT) The present analysis of monomeric and dimeric constructs of the microtubule-dependent molecular motor ncd indicates that parts missing in the crystal structures must be taken into account for successful fitting of the solution scattering data. The modeling of the missing C-terminal region up to Lys700, which has not been resolved in any of the ncd crystal structures thus far, was crucial in this respect. It was found that the most probable configuration of this region is an antiparallel β-sheet. Of course, the orientations of this region presented in Figs. 2, 5, and 9 should be considered as tentative models because they were based on fitting solution scattering data and not on crystallographic information. With this caveat in mind, it is interesting to note that the probable orientation of the C-terminal fragment changes upon dimerization (i.e. from MC6 to MC5) but remains stable in the dimeric protein independent of the coiled-coil lengths in the ncd constructions (MC5 to MC1). One of the most important findings of the present study is that the published crystallographic structures of the dimeric ncd in the ADP-bound form (17Sablin E.P. Case R.B. Dai S.C. Hart C.L. Ruby A. Vale R.D. Fletterick R.J. Nature. 1998; 395: 813-816Crossref PubMed Scopus (185) Google Scholar, 18Kozielski F. De Bonis S. Burmeister W.P. Cohen-Addad C. Wade R.H. Structure Fold. Des. 1999; 7: 1407-1416Abstract Full Text Full Text PDF Scopus (48) Google Scholar) are compatible with the solution scattering data from the MC5 construct. This agrees with the results of Stoneet al. (42Stone D.B. Hjelm Jr., R.P. Mendelson R.A. Biochemistry. 1999; 38: 4938-4947Crossref PubMed Scopus (13) Google Scholar) obtained on three other dimericDrosophila ncd constructs. Although the differences between the three available crystal structures are rather small, the structure lacking the 2-fold symmetry axis along the coiled-coil domain yields a somewhat better fit. It is interesting that a nonsymmetric crystallographic model of dimeric rat kinesin has recently been found to provide better agreement with the solution scattering data ofDrosophila kinesin (43Kozielski F. Svergun D. Zaccai J. Wade D. Koch M. J. Biol. Chem. 2001; 276: 1267-1275Abstract Full Text Full Text PDF PubMed Scopus (15) Google Scholar). The longest dimeric ncd construct analyzed in the present study, MC1, covers the entire N-terminal coiled-coil domain and catalytic motor core. The model obtained by addition of a straight coiled-coil region to the above-mentioned model of MC5 yields a remarkably good fit to the solution scattering data. This suggests that MC1 has a straight coiled-coil domain, whereas the relative orientation of the two heads is independent of the length of the coiled-coil domain. This creates some confidence that working with truncated motor protein fragments instead of full-length proteins does not alter the overall shape of these molecules. It has been demonstrated in the present study that the overall conformation of the dimeric ncd in solution agrees with the crystallographic models (17Sablin E.P. Case R.B. Dai S.C. Hart C.L. Ruby A. Vale R.D. Fletterick R.J. Nature. 1998; 395: 813-816Crossref PubMed Scopus (185) Google Scholar, 18Kozielski F. De Bonis S. Burmeister W.P. Cohen-Addad C. Wade R.H. Structure Fold. Des. 1999; 7: 1407-1416Abstract Full Text Full Text PDF Scopus (48) Google Scholar). A tentative configuration of the C-terminal loop in monomeric and dimeric ncd is proposed, based on the scattering data. Future x-ray and neutron scattering experiments will include the investigation of nucleotide-dependent conformational changes of dimeric ncd in solution. In addition, different members of the highly divergent C-terminal subfamily should be compared to determine whether they have the same overall conformation. We thank Sharyn Endow for plasmids pET/MC1, pET/MC5, and pET/MC6.