Nature Structural Biology
9, 745 - 749 (2002)
Published online: 16 September 2002; | doi:10.1038/nsb842
Structure-based design of a potent purine-based cyclin-dependent kinase inhibitorThomas G. Davies1, Johanne Bentley2, Christine E. Arris2, F. Thomas Boyle3, Nicola J. Curtin2, Jane A. Endicott1, Ashleigh E. Gibson2, Bernard T. Golding2, Roger J. Griffin2, Ian R. Hardcastle2, Philip Jewsbury3, Louise N. Johnson1, Veronique Mesguiche2, David R. Newell2, Martin E.M. Noble1, Julie A. Tucker1, Lan Wang2
& Hayley J. Whitfield2, 41 Laboratory of Molecular Biophysics and Department of Biochemistry, University of Oxford, Oxford, OX1 3QU, UK. 2 Cancer Research Unit and Department of Chemistry, University of Newcastle, Newcastle upon Tyne, NE2 4HH, UK. 3 AstraZeneca Pharmaceuticals, Alderley Park, Cheshire, SK10 4TG, UK. 4 This work is dedicated to the memory of Hayley J. Whitfield.
Correspondence should be addressed to David R. Newell herbie.newell@ncl.ac.ukAberrant control of cyclin-dependent kinases (CDKs) is a central feature of the molecular pathology of cancer. Iterative structure-based design was used to optimize the ATP- competitive inhibition of CDK1 and CDK2 by O6-cyclohexylmethylguanines, resulting in O6-cyclohexylmethyl-2-(4'- sulfamoylanilino)purine. The new inhibitor is 1,000-fold more potent than the parent compound (Ki values for CDK1 = 9 nM and CDK2 = 6 nM versus 5,000 nM and 12,000 nM, respectively, for O6-cyclohexylmethylguanine). The increased potency arises primarily from the formation of two additional hydrogen bonds between the inhibitor and Asp 86 of CDK2, which facilitate optimum hydrophobic packing of the anilino group with the specificity surface of CDK2. Cellular studies with O6-cyclohexylmethyl-2-(4'- sulfamoylanilino) purine demonstrated inhibition of MCF-7 cell growth and target protein phosphorylation, consistent with CDK1 and CDK2 inhibition. The work represents the first successful iterative synthesis of a potent CDK inhibitor based on the structure of fully activated CDK2−cyclin A. Furthermore, the potency of O6-cyclohexylmethyl-2-(4'- sulfamoylanilino)purine was both predicted and fully rationalized on the basis of protein−ligand interactions.
The cyclin-dependent kinases (CDKs) are increasingly recognized as important targets for therapeutic intervention in various proliferative disease states, including cancer1. Inhibitors of CDK have entered clinical trials to treat cancer2,
3. We recently reported the identification of a newly discovered ATP-competitive purine-based inhibitor, O6-cyclohexylmethylguanine (NU2058, Fig. 1a), which has a Ki of 5  M for CDK1 and 12 M against CDK2 but essentially no activity against CDK4 (ref. 4). An important feature of NU2058 is that it has a cellular pharmacological profile that is distinct from that of the benchmark inhibitors olomoucine and flavopiridol4. Furthermore, the crystal structure of NU2058 in complex with monomeric CDK2 revealed protein−ligand interactions that were distinct from ATP and the 6-aminopurine inhibitors olomoucine5, roscovitine6 and purvalanol7. Comparison of the binding modes of NU2058, the 6-aminopurine CDK inhibitors and the natural product indirubin8 suggest that substitution at the N2 position of NU2058 would be tolerated and might lead to inhibitors with increased potency.
 | | |
Here we demonstrate that structure-based development and iterative biological evaluation can be used to rapidly optimize CDK1 and CDK2 inhibition and identify inhibitors with nanomolar potency. Notably, this work is the first successful use of the fully active CDK2−cyclin A complex in a prospective drug design program.
NU2058 in complex with phosphoCDK2−cyclin A
As a starting point for the optimization of NU2058, we determined the structure of the NU2058 inhibitor bound to the fully active Thr 160-phosphorylated CDK2−cyclin A spacing complex (T160pCDK2−cyclinA) (Fig. 1b). Structural rearrangements associated with CDK2 activation suggest that the use of the fully active complex is important for detailed atom-targeted, structure-based design9.
NU2058 in complex with T160pCDK2−cyclinA, as with monomeric CDK2, forms a triplet of hydrogen bonds between its purine ring and the hinge region of CDK2 (residues 81−84) (Fig. 1c). In addition, several van der Waals and hydrophobic contacts are made between the purine ring and the top and bottom of the ATP-binding cleft. An edge−face aromatic−aromatic contact is formed between the purine ring and Phe 80. The O6-cyclohexylmethyl substituent is accommodated in the binding site for the ribose moiety of ATP and forms highly complementary packing and hydrophobic interactions with an apolar pocket in the Gly-rich loop (residues 9−19).
Structure-activity relationships for the O6-position have shown that a cyclic hydrophobic substituent such as cyclohexylmethyl seems to be optimal10, and we have maintained this group in subsequent compounds. To increase the potency of NU2058, we considered addition and elaboration of functional groups at the N2 position of the purine. Assuming a similar binding mode to NU2058, any groups added to N2 would project out of the ATP-binding cleft toward the solvent and contact the 'specificity surface' of CDK2 (ref. 11), with consequences for both potency and selectivity.
Studies with NU2058 derivatives
Several compounds that pack aromatic moieties with the specificity surface of CDK2 have been developed, such as the purvalanols7 and the indirubins8,
12. Taking this into account, we initially synthesized 2-anilino-6-cyclohexylmethoxypurine (NU6094; Fig. 1a), which contains an anilino group at the purine C2 position of NU2058. This compound has an IC50 of 1.00  0.03 M for CDK2 and is similarly active against CDK1 (Table 1), representing over a 10-fold increase in affinity for CDK2 relative to the parent compound NU2058.
| | Table 1. Inhibition of CDK activity and human MCF-7 breast carcinoma cell growth by O6-cyclohexylmethylguanine CDK inhibitors | | |  |
 Full Table |
| |
To confirm the binding mode of NU6094 and rationalize the observed increase in affinity, the structure of NU6094 in complex with T160pCDK2−cyclinA was determined (Fig. 2a). Within the ATP-binding cleft, NU6094 adopts the same binding mode as NU2058. The anilino group projects out of the adenine site though a largely hydrophobic tunnel constituted by the side chains of Phe 82 and Ile 10 (Fig. 2b), and packs against the kinase surface, forming a  − stacking interaction with the peptide backbone between Asn 85 and Asp 86. The increase in potency associated with this additional group likely arises from desolvation of the hydrophobic plane of the anilino group on binding. Assuming the measured IC50 values are proportional to the Kd for the CDK2−inhibitor interaction, the relative free energy difference between NU2058 and NU6094 ( G°2058 6094) is  -7 kJ mol-1. The upper limit for the free energy associated with desolvation of hydrophobic groups has been estimated13 to be -136 J mol-1 Å-2. The binding of NU6094 buries 50 Å2 of apolar surface, leading to a calculated free energy change of -6.8 kJ mol-1, which is in good agreement with the experimentally measured free energy difference.
| | Figure 2. Binding of NU6094, NU6086 and NU6102 to T160pCDK2−cyclinA. | | |  | a, T160pCDK2−cyclinA−NU6094 structure. Selected CDK2 residues are drawn in ball-and-stick representation, with carbon atoms colored green (inhibitor) and yellow (CDK2). The final 2Fo-Fc electron density contoured at 0.24 e- Å-3 for NU6094 is included. Hydrogen bonds in all panels except (b) are denoted by dashed lines. b, NU6094 bound to the CDK2 active site. NU6094 is depicted as yellow CPK spheres. The CDK2 molecular surface is colored by atom type: carbon, oxygen and nitrogen atoms are green, red and blue, respectively. The figure illustrates the complementarity of the NU6094 anilino ring to the shape of the hydrophobic tunnel leading to the specificity surface. c, T160pCDK2−cyclinA−NU6086 structure. NU6086 and selected CDK2 residues are rendered in ball-and stick-representation, with carbon atoms colored as in (a). Both conformers of the anilino ring (denoted I and II) are included. The final 2Fo-Fc electron density for NU6086 is contoured at 0.24 e- Å-3. d, T160pCDK2−cyclinA−NU6102 structure. NU6102 and selected CDK2 residues are rendered in ball-and-stick representation, with carbon atoms colored as in (a). The final 2Fo-Fc electron density for NU6102 is contoured at 0.24 e- Å-3. e, NU6102 bound to the CDK2 active site. The CDK2 molecular surface is rendered in transparent gray so that interactions between the NU6102 sulfonamide group and the backbone nitrogen and side chain oxygen of Asp 86 are visible. Hydrogen bonds are depicted by dotted lines. NU6102 is rendered in ball and stick, with carbon atoms colored green.
Full Figure and legend (100K) |
| |
The planes of the purine and anilino rings are tilted relative to each other with an interplanar angle of 50°. A search of the Cambridge Structural Database (CSD) for small molecule crystal structures14 that contain two aromatic rings coupled by a single nitrogen showed the modal angle to be 48°, close to that observed for NU6094 bound to CDK2. The binding of NU6094 to CDK2 may, therefore, have the additional advantage that the conformation of the complex will be relatively strain-free and entropically favored because of pre-organization of the free ligand.
The observation for the purvalanols that CDK2 has a preference for chlorine atoms on the specificity surface7 led to the synthesis and testing of 6-cyclohexylmethoxy-2-(3'-chloroanilino)purine (NU6086, Fig. 1a). However, the affinity of NU6086 for both CDK2 and CDK1 (Table 1) was not improved over NU6094. The crystal structure of NU6086 in complex with T160pCDK2−cyclinA (Fig. 2c) revealed a mode of inhibitor binding identical to NU6094. Initial difference density for the chloro-substituted anilino group indicated the presence of two partially occupied conformations. In conformer I, the meta chloro atom forms van der Waals contacts with Phe 82 and His 84, whereas in conformer II it forms a hydrogen bond with the protonated form of Asp 86 (rCl−O = 3.0 Å). The packing and interplanar anilino-purine angle is different in each case, with the interaction between Asp 86 and the inhibitor causing a 13° rotation of the anilino ring for conformer II relative to conformer I and NU6094.
The development of NU6102
The NU6094 and NU6086 results demonstrate that increased affinity could be achieved through the addition of an anilino group at C2 of NU2058. To further increase the potency of such compounds, a sulfonamide group was introduced at the anilino para position of NU6094 in an attempt to form a hydrogen bond to Lys 89.
O6-cyclohexylmethoxy-2-(4'-sulfamoylanilino)purine (NU6102; Fig. 1a) was synthesized and found to be a highly potent CDK inhibitor, with Ki values of 9 1 nM and 6 0.5 nM for CDK1 and 2, respectively (Table 1). The crystal structure of T160pCDK2−cyclinA in complex with NU6102 reveals the interactions formed and provides an explanation for its tight binding. NU6102 adopts an identical binding mode to the other N2-substituted purines discussed above. The O6-cyclohexyl group superimposes closely with both NU6086 and NU6094 in the inhibitor-bound complexes, and the usual triplet of hydrogen bonds are formed between the purine core and the backbone of residues Leu 83 and Glu 81 (Fig. 2d). Although the anilino group packs closely with the specificity surface, the sulfonamide does not form the designed hydrogen bond with Lys 89. Instead, the NH2 group of the sulfonamide donates a hydrogen bond to a side chain oxygen of Asp 86 (rNH2−O = 2.9 Å), and one sulfonamide oxygen accepts a hydrogen bond from the backbone nitrogen of Asp 86 (rNH−O = 3.1 Å) (Fig. 2e). The formation of these interactions with good geometry leads to rotation of the anilino group by 13°, as observed for conformer II of NU6086. When this occurs, the packing between the surface of the protein and the plane of the anilino ring is enhanced and stabilized. With NU6102 in essentially its lowest energy conformation, the sulfonamide group is positioned to form protein−ligand hydrogen bonds with good geometry. Furthermore, this 'fine-tunes' the orientation of the anilino group for optimum packing with the kinase, as well as for stabilizing hydrophobic contacts. The result is a 1,000-fold increase in affinity relative to NU2058 and maintenance of selectivity for CDK1 and CDK2 over CDK4. To confirm the selectivity of NU6102, the compound was independently tested as an inhibitor of 28 kinases, including CDK2−cyclin A in the same assay system, at an ATP concentration of 100 M. The only kinases to be inhibited by NU6102 under these conditions with IC50 values <1 M were CDK2 (30 nM), Rho-dependent protein kinase II (ROCKII at 600 nM), phospholipid dependent kinase 1 (PDK1 at 800 nM) and dual-specificity tyrosine phosphorylated and regulated kinase 1A (DYRK1A at 900 nM). Thus, NU6102 has at least 10-fold selectivity for the target kinases CDK1 and 2.
Cellular effects of CDK inhibitors
All compounds inhibited growth of MCF-7 cells in a concentration-dependent manner, with NU6102 being the most potent (Table 1). The activities of NU2058 and NU6102 on G1/S-associated CDKs were further examined by investigating their ability to inhibit phosphorylation of downstream CDK target proteins, the retinoblastoma protein (pRb) and CDK1. pRb contains 16 consensus sites for CDK phosphorylation that have been characterized as either specific for CDK4−cyclin D1, CDK2−cyclin E or CDK2−cyclin A, or able to be phosphorylated by combinations of these kinases15. Both NU2058 and NU6102 induced a concentration-dependent decrease in pRb hyperphosphorylation (Fig. 3a). Furthermore, phosphorylation of pRb at Thr 821 (Fig. 3b), a site that is specific for CDK2-mediated phosphorylation events, was markedly reduced after treatment with NU2058 and NU6102 (50% inhibition at 31 7 and 5 2 M, respectively). Phosphorylation of the CDK1 target protein was measured using the TG3 antibody, which binds to CDK1−cyclin B phosphorylated nucleolin16 (Fig. 3d). Cells arrested in M-phase by treatment with 0.5 g ml-1 nocodazole for 24 h were found to contain high levels of phosphorylated nucleolin, and the presence of NU2058 or NU6102 inhibited nucleolin phosphorylation (50% inhibition at 39 7 and 3 1 M, respectively) in a concentration-dependent manner.
| | |
Discussion
This paper describes the development of a CDK inhibitor with nanomolar potency using structural information initially derived from a complex of the inhibitor NU2058 with fully activated CDK2−cyclin A. An important feature of NU2058 was that it binds to monomeric CDK2 in a conformation distinct from ATP and other adenine-based CDK inhibitors4. Furthermore, NU2058 has a pattern of activity against a panel of tumor cell lines that is distinct from that of the known CDK inhibitors flavopiridol and olomoucine4.
In the current study, molecules were designed to probe interactions with the 'specificity surface' on CDK2. Structural analyses had indicated that an aromatic ring at the N2 position of NU2058 would improve inhibitory activity against CDK1 and CDK2. This was found to be the case, with the resulting compound maintaining specificity for the inhibition of CDK1/2 over CDK4. Additions to the anilino ring at the meta position, including chlorine (NU6086), bromine, fluorine or methoxy (data not shown), did not improve or reduced CDK-inhibitory activity and tumor cell growth inhibition. In contrast, substitutions at the para position, exemplified by addition of a sulfonamide group, substantially improved activity against CDK1 and CDK2. Furthermore, excellent specificity over CDK4 (1,000-fold) and 27 other kinases was maintained, and inhibition of tumor cell growth was enhanced. The structure of a complex of NU6102 with T160pCDK2−cyclinA revealed that the triplet of hydrogen bonds observed in the NU2058−T160pCDK2−cyclinA structure was maintained, and additional hydrogen bonds involving Asp 86 were formed. These interactions facilitate optimal hydrophobic packing of the anilino group with the specificity surface of CDK2. Consistent with inhibition of CDK1 and CDK2, both CDK2-specific pRb phosphorylation and CDK1-specific nucleolin phosphorylation was inhibited by NU6102. Further modification of the NU6102 structure is necessary to improve the pharmacological properties of these compounds. In particular, despite in vitro Ki values <10 nM against CDK1 and CDK2, micromolar concentrations are required to inhibit cell growth and CDK activity in whole cells (Table 1; Fig. 3); poor cellular penetration may explain this discrepancy.
In summary, this paper describes the prospective application of structure-based drug design to the development of inhibitors of CDK2, where a fully active CDK−cyclin complex has been used. The resulting compound, NU6102, is selective and one of the most active CDK2 inhibitors described so far. The potency of the compound has been predicted and is fully rationalized on the basis of protein−ligand interactions.
Methods
Chemistry.
The synthesis of the CDK inhibitors used in these studies has been described (NU2058)4 or will be described elsewhere (NU6094, NU6086 and NU6102).
Expression and purification T160pCDK2−cyclinA, and crystallization of T160pCDK2−cyclinA−inhibitor complexes.
Thr 160-phosphorylated CDK2−cyclin A was purified as described17. Crystals were obtained by the hanging drop method, using protein at a concentration of 10 mg ml-1, 5% (v/v) dimethylsulfoxide (DMSO) and 5 mM inhibitor. The reservoir solution contained 0.8 M KCl and 1.2 M (NH4)2SO4 in 40 mM HEPES, pH 7.0, and 5 mM dithiothreitol (DTT). The protein was incubated with inhibitor on ice for 1 h and spun through a 0.22 M filter before setting up crystallization trials.
X-ray crystallography data collection and processing.
Before data collection, crystals were briefly immersed in cryo-protectant, which consisted of mother liquor and 24% (v/v) glycerol (NU2058 and NU6094 structures) or 8 M sodium formate (NU6086 and NU6102 structures), before flash-freezing using an Oxford Cryostream at 100 K. Data were collected using a rotating anode source and Mar image plate (NU6094) or a MarCCD at ID14-EH1 at the ESRF (NU2058, NU6086 and NU6102). Data processing was carried out using MOSFLM18 and SCALA19.
Structure solution and refinement.
The structures were solved by molecular replacement (MR) using MOLREP18. A high-resolution, well-refined structure of the T160pCDK2−cyclinA complex with another inhibitor (J.A.E, L.N.J & M.E.M.N, unpub. data) was used as the starting model for MR, and a clear solution was obtained with two CDK2−cyclin A heterodimers in the asymmetric unit. This model was subjected to rigid-body refinement followed by individual atomic refinement using REFMAC20. Strict noncrystallographic restraints were applied initially and gradually relaxed during refinement. Inhibitor models were built using SYBYL (Tripos), and model building and fitting were carried out in O21. Water molecules were added using ARP22. Data collection, processing and refinement statistics are given in Table 2.
| | Table 2. X-ray data collection and refinement statistics for the four structures | | | |
Full Table |
| |
Kinase assays.
Inhibition of CDK1−cyclin B1 and CDK2−cyclin A was assayed as described4. Inhibition of CDK4−cyclin D1 was determined in a similar assay using an assay buffer composed of 50 mM HEPES, 10 mM MnCl2, 1 mM DTT, 100 M sodium fluoride, 100 M sodium vanadate, 10 mM sodium glycerophosphate, 5 g ml-1 aprotinin, 2.5 g ml-1 leupeptin, 100 M phenylmethylsulfonyl flouride (PMSF) and a recombinant retinoblastoma peptide (encoding residues 792−928) as substrate. CDK4−cyclin D1 (gift from AstraZeneca Pharmaceuticals) was prepared as a GST-tagged complex from baculoviral-infected insect cell lysate. The final ATP concentration in each assay was 12.5 M. The IC50 concentration for each inhibitor is the concentration required to inhibit enzyme activity by 50% under the assay conditions used. The activity of NU6102 against 28 kinases, including CDK2−cyclin A, was studied as described23 using an ATP concentration of 100 M (data obtained in collaboration with P. Cohen, University of Dundee, UK).
MCF-7 cell growth inhibition.
The effects of compounds on the growth of MCF-7 human breast carcinoma cells was investigated. Cells were plated at a density of 1  103 per well in 96-well plates containing RPMI medium (supplemented with 10% (v/v) fetal calf serum) (Gibco), allowed to grow for 72 h and then treated with a range of inhibitor concentrations (1−100 m) for 48 h. Cell growth, relative to control cells treated with 1% (v/v) DMSO, was measured using the cell proliferation assay kit (Boehringer).
Western blotting.
Cells were lysed (20% (v/v) glycerol, 4% (w/v) SDS and 100 mM Tris-HCl, pH 6.8) and heated to 85 °C for 5 min. Following addition of equal volume of sample buffer (0.001% (w/v) bromophenol blue and 100 mM DTT), 20 g protein was resolved on 3−8% polyacrylamide Tris-acetate (pRb) and 4−20% (nucleolin) NUPAGE gels (Invitrogen) and blotted onto Hybond ECL nitrocellulose membrane (Pharmacia) using NUPAGE transfer buffer. Blots were blocked in TTBS (20 mM Tris, 140 mM NaCl and 0.1 % (v/v) Tween 20, pH 7.6) containing 5 % (w/v) dried milk for 1 h and incubated overnight in primary antibody (either retinoblastoma-specific (1:250, sc-50 Santa Cruz), retinoblastoma-phosphospecific (1:1000, pT821 Biosource International) or nucleolin-specific (1:500, TG3 a generous gift from P. Davies, Albert Einstein College of Medicine, New York)). Labeled proteins were detected using Supersignal West Dura extended duration ECL substrate (Pierce).
Coordinates.
Coordinates and structure factors for the four structures (NU2058, NU6094, NU6086 and NU6102) have been deposited with the Protein Data Bank (accession codes 1H1P, 1H1Q, 1H1R and 1H1S, respectively).
Received 25 April 2002; Accepted 5 August 2002; Published online: 16 September 2002.
REFERENCES
- Sielecki, T.M., Boylan, J.F., Benfield, P.A. & Trainor, G.L. J. Med. Chem. 43, 1−18 (2000) | Article | PubMed | ISI | ChemPort |
- Senderowicz, A.M. & Sausville, E.A. J. Natl. Cancer Inst. 92, 376−387 (2000). | Article | PubMed | ChemPort |
- Meijer, L., Leclerc, S. & Leost, M. Pharmacol. Ther. 82, 279−284 (1999). | Article | PubMed | ISI | ChemPort |
- Arris, C.E. et al. J. Med. Chem. 43, 2797−2804 (2000). | Article | PubMed | ISI | ChemPort |
- Schulze-Gahmen, U. et al. Proteins 22, 378−391 (1995). | PubMed | ChemPort |
- De Azevedo, W.F. Jr, LeClerc, S., Meijer, L., Havlicek, L., Strnad, M. & Kim, S.-H. Eur. J. Biochem. 243, 518−526 (1997). | Article | PubMed | ChemPort |
- Gray, N.S. et al. Science 281, 533−538 (1998). | Article | PubMed | ISI | ChemPort |
- Hoessel, R. et al. Nature Cell Biol. 1, 60−67 (1999). | Article | PubMed | ISI | ChemPort |
- Russo, A.A., Jeffrey, P.D. & Pavletich, N.P. Nature Struct. Biol. 3, 696−700 (1996). | Article | PubMed | ISI | ChemPort |
- Gibson, A.E. et al. J. Med. Chem. 45, 3381−3393 (2002). | Article | PubMed | ISI | ChemPort |
- Davies, T.G., Pratt, D.J., Endicott, J.A., Johnson, L.N. & Noble, M.E.M. Pharmacol. Ther. 95, 125−133 (2002). | Article |
- Davies, T.G. et al. Structure 9, 389−397 (2001). | Article | PubMed | ISI | ChemPort |
- Juffer, A.H., Eisenhaber, F., Hubbard, S.J., Walther, D. & Argos, P. Protein Sc. 4, 2499−2509 (1995). | ISI | ChemPort |
- Fletcher, D.A., McMeeking, R.F. & Parkin, D. J. Chem. Inf. Comput. Sci. 36, 746−749 (1996). | Article | ISI | ChemPort |
- Zarkowska, T. & Mittnacht, S. J. Biol. Chem. 272, 12738−12746 (1997). | Article | PubMed | ISI | ChemPort |
- Peter, M., Nakagawa, J., Doree, M., Labbe, J.C. & Nigg, E.A. Cell 60, 791−801 (1990). | Article | PubMed | ISI | ChemPort |
- Brown, N.R., Noble, M.E.M., Endicott, J.A. & Johnson, L.N. Nature Cell Biol. 1, 438−443 (1999). | Article | PubMed | ISI | ChemPort |
- Leslie, A.G.W. Joint CCP4 and ESF-EAMCB Newsletter on Protein Crystallography 26 (Science and Engineering Research Council UK, Daresbury Laboratory, Warrington; 1992).
- Evans, P.R. in Proceedings of CCP4 Study Weekend (eds Sawyer, L., Isaacs, N. & Bailey, S.) 114−122 (Science and Engineering Research Council UK, Daresbury Laboratory, Warrington; 1993).
- Murshudov, G.N., Vagen, A.A. & Dodson, E.J. Acta Crystallogr. D 53, 240−255 (1997). | Article | PubMed | ISI | ChemPort |
- Jones, T.A., Zou, J.Y., Cowan, S.W. & Kjeldgaard, M. Acta Crystallogr. A 47, 110−119 (1991). | Article | PubMed | ISI |
- Lamzin, V.S. & Wilson, K.S. Acta Crystallogr. D 49, 129−147 (1993). | Article | PubMed | ISI | ChemPort |
- Davies, S.P., Reddy, H., Caivano, M. & Cohen, P. Biochem J. 351, 95−105 (2000). | Article | PubMed | ISI | ChemPort |
Acknowledgments
We would like to thank T. Hunt for his gift of the human CDK2 and cyclin A3 cDNAs and C. Man for the Saccharomyces cerevisiae CIV1 sequence. We thank N. Hanlon, N. Brown and D. Barford for the development of the CDK2-CIV1 co-expression strategy. We also thank the beamline staff at ESRF, Grenoble, France, who provided excellent facilities and advice during data collection. We wish to acknowledge the use of the EPSRC's Database Service at Daresbury. At the LMB, the authors would like to thank I. Taylor, E. Garman, A. Lawrie, R. Bryan, Y. Huang, K. Measures and S. Lee, and L. Meijer at CNRS Station Biologique, Roscoff, for advice on the nucleolin western blotting procedure. This research was supported by grants from Cancer Research UK, The Royal Society UK, Medical Research Council UK, BBSRC and AstraZeneca PLC UK.
Competing interests statement:
The authors declare that they have no competing financial interests. |
