Crystallization of a complex between human CDK6 and a virus-encoded cyclin is critically dependent on the addition of small charged organic molecules.

Human CDK6 plays an important role in controlling entry into the eukaryotic cell cycle. An activated complex of human CDK6 with a viral cyclin from herpesvirus saimiri was purified to homogeneity and crystallized using polyethylene glycol 3350 as precipitant. Crystallization was critically dependent on a narrow range of calcium acetate concentration and the presence of sulfo-betaine 201 as additive. Crystals belong to the hexagonal space group P6(1)22 or P6(5)22, with unit-cell parameters a = b = 70.14, c = 448.77 A, gamma = 120 degrees, and diffract X-rays to at least 3.1 A resolution.

Crystal structure of a viral cyclin, a positive regulator of cyclin-dependent kinase 6.

BACKGROUND: Cyclin-dependent kinases (CDKs) have a central role in cell-cycle control and are activated by complex formation with positive regulatory proteins called cyclins and by phosphorylation. The overexpression and mutation of cyclins and CDKs has been associated with tumorigenesis and oncogenesis. A virus-encoded cyclin (v-cyclin) from herpesvirus saimiri has been shown to exhibit highest sequence homology to type D cyclins and specifically activates CDK6 of host cells to a very high degree. RESULTS: We have determined the first X-ray structure of a v-cyclin to 3.0 A resolution. The structure of the core domains is very similar to those of cyclin A and cyclin H from human cells. To understand the structural basis for the v-cyclin specificity for CDK6 and the insensitivity of the complex to inhibitors of the p21 and INK4 families, a v-cyclin-CDK2 model was built on the basis of the known structures of human cyclin A in complex with CDK2 and the CDK inhibitor p27(Kip1). CONCLUSIONS: Although many critical interactions between cyclin A and CDK2 would be conserved in a v-cyclin-CDK2 complex, some appear sterically or electrostatically unfavorable due to shifts in the backbone conformation or sidechain differences and may contribute to v-cyclin selectivity for CDK6. The insensitivity of v-cyclin-CDK6 complexes to inhibitors of the p21 family is probably due to structural changes in v-cyclin that lead to a flatter surface area offering fewer potential contacts with the protein inhibitor. In addition, sequence changes in v-cyclin eliminate hydrogen-bonding partners for atoms of the p27(Kip1) inhibitor. This structure provides the first model for interactions between v-cyclins and host cell-cycle proteins; these interactions may be important for virus survival as well as oncogenic transformation of host cells.

High-resolution crystal structures of human cyclin-dependent kinase 2 with and without ATP: bound waters and natural ligand as guides for inhibitor design.

Inhibition of the cell cycle is widely considered as a new approach toward treatment for diseases caused by unregulated cell proliferation, including cancer. Since cyclin-dependent kinases (CDKs) are key enzymes of cell cycle control, they are promissing targets for the design and discovery of drugs with antiproliferative activity. The detailed structural analysis of CDK2 can provide valuable information for the design of new ligands that can bind in the ATP binding pocket and inhibit CDK2 activity. For this objective, the crystal structures of human CDK2 apoenzyme and its ATP complex were refined to 1.8 and 1.9 A, respectively. The high-resolution refinement reveals 12 ordered water molecules in the ATP binding pocket of the apoenzyme and five ordered waters in that of the ATP complex. Despite a large number of hydrogen bonds between ATP-phosphates and CDK2, binding studies of cyclic AMP-dependent protein kinase with ATP analogues show that the triphosphate moiety contributes little and the adenine ring is most important for binding affinity. Our analysis of CDK2 structural data, hydration of residues in the binding pocket of the apoenzyme, flexibility of the ligand, and structural differences between the apoenzyme and CDK2-ATP complex provide an explanation for the results of earlier binding studies with ATP analogues and a basis for future inhibitor design.

Structural basis for specificity and potency of a flavonoid inhibitor of human CDK2, a cell cycle kinase.

The central role of cyclin-dependent kinases (CDKs) in cell cycle regulation makes them a promising target for studying inhibitory molecules that can modify the degree of cell proliferation. The discovery of specific inhibitors of CDKs such as polyhydroxylated flavones has opened the way to investigation and design of antimitotic compounds. A novel flavone, (-)-cis-5,7-dihydroxyphenyl-8-[4-(3-hydroxy-1-methyl)piperidinyl] -4H-1-benzopyran-4-one hydrochloride hemihydrate (L868276), is a potent inhibitor of CDKs. A chlorinated form, flavopiridol, is currently in phase I clinical trials as a drug against breast tumors. We determined the crystal structure of a complex between CDK2 and L868276 at 2.33 angstroms resolution and refined to an Rfactor 20.3%. The aromatic portion of the inhibitor binds to the adenine-binding pocket of CDK2, and the position of the phenyl group of the inhibitor enables the inhibitor to make contacts with the enzyme not observed in the ATP complex structure. The analysis of the position of this phenyl ring not only explains the great differences of kinase inhibition among the flavonoid inhibitors but also explains the specificity of L868276 to inhibit CDK2 and CDC2.

Structural basis for chemical inhibition of CDK2.

The central role of cyclin-dependent kinases (CDKs) in cell cycle regulation makes them a promising target for discovering small inhibitory molecules that can modify the degree of cell proliferation. The three-dimensional structure of CDK2 provides a structural foundation for understanding the mechanisms of activation and inhibition of CDK2 and for the discovery of inhibitors. In this article five structures of human CDK2 are summarised: apoprotein, ATP complex, olomoucine complex, isopentenyladenine complex, and des-chloro-flavopiridol complex.

Multiple modes of ligand recognition: crystal structures of cyclin-dependent protein kinase 2 in complex with ATP and two inhibitors, olomoucine and isopentenyladenine.

Cyclin-dependent kinases (CDKs) are conserved regulators of the eukaryotic cell cycle with different isoforms controlling specific phases of the cell cycle. Mitogenic or growth inhibitory signals are mediated, respectively, by activation or inhibition of CDKs which phosphorylate proteins associated with the cell cycle. The central role of CDKs in cell cycle regulation makes them a potential new target for inhibitory molecules with anti-proliferative and/or anti-neoplastic effects. We describe the crystal structures of the complexes of CDK2 with a weakly specific CDK inhibitor, N6-(delta 2-isopentenyl)adenine, and a strongly specific inhibitor, olomoucine. Both inhibitors are adenine derivatives and bind in the adenine binding pocket of CDK2, but in an unexpected and different orientation from the adenine of the authentic ligand ATP. The N6-benzyl substituent in olomoucine binds outside the conserved binding pocket and is most likely responsible for its specificity. The structural information from the CDK2-olomoucine complex will be useful in directing the search for the next generation inhibitors with improved properties.

Crystal structure of a peptide complex of anti-influenza peptide antibody Fab 26/9. Comparison of two different antibodies bound to the same peptide antigen.

The three-dimensional structure of the complex of a second anti-peptide antibody (Fab 26/9) that recognizes the same six-residue epitope of an immunogenic peptide from influenza virus hemagglutinin (HA1; 75-110) as Fab 17/9 with the peptide has been determined at 2.8 A resolution. The amino acid sequence of the variable region of the 26/9 antibody differs in 24 positions from that of 17/9, the first antibody in this series for which several ligand-bound and free structures have been determined and refined. Comparison of the 26/9-peptide with the 17/9-peptide complex structures shows that the two Fabs are very similar (r.m.s.d. 0.5 to 0.8 A) and that the peptide antigen (101-107) has virtually the same conformation (r.m.s.d. 0.3 to 0.8 A) when bound to both antibodies. A sequence difference in the 26/9 binding pocket (L94; His in 26/9, Asn in 17/9) results in an interaction with a bound water molecule that is not seen in the 17/9 structures. Epitope mapping shows that the relative specificity of 26/9 and 17/9 antibodies for individual positions of the peptide antigen are slightly different. Amino acid substitutions in the peptide, particularly at position SerP107, are tolerated to different extents by 17/9 and 26/9. Structural and sequence analysis suggests that amino acid differences near the peptide-binding site are responsible for altering slightly the specificity of 26/9 for three peptide residues and illustrates how amino acid substitutions can modify antibody-antigen interactions and thereby modulate antibody specificity.

Detailed analysis of the free and bound conformations of an antibody. X-ray structures of Fab 17/9 and three different Fab-peptide complexes.

A new orthorhombic crystal form of Fab 17/9 has been determined in complex with a 7-mer peptide from influenza virus hemagglutinin (HA1 101-107, acetylated and amidated). The three-dimensional structure was resolved to 2.8 A with an improved refinement and better geometry than two previously determined Fab 17/9-peptide (HA1 100-108) complexes, facilitating a detailed description of the Fab-peptide interactions. The binding pockets and the peptide antigen are structurally similar in all three peptide complexes of Fab 17/9. The peptide adopts an extended conformation (residues 100 to 103) and a type I reverse turn (residues 104 to 107). Additionally, the antigenic determinant described here correlates well with previous epitope mapping studies. The structures of the free and antigen bound Fab illustrate the role of induced fit as a mechanism for antibody-antigen recognition. Fab 17/9 undergoes a large conformational change, mainly in the H3 loop, upon peptide binding. As a result, the shape of the binding pocket changes substantially in the liganded Fab. However, the backbone conformations of the other hypervariable loops (L2, L3, H1 and H2) show no significant difference between free and bound structures. The conformation of the L1 loop is also maintained in all structures, but its position relative to the framework varies in different crystal environments. The availability of three X-ray structures of an Fab-peptide complex in three different space groups makes it possible to clearly distinguish between crystal packing and antigen binding as the cause of structural differences. Two distinct H3-loop conformations, free and bound, are observed with no evidence otherwise for multiple conformations of the hypervariable loops (CDRs) or increased flexibility in either the free or bound forms.

Crystal structure of glycinamide ribonucleotide transformylase from Escherichia coli at 3.0 A resolution. A target enzyme for chemotherapy.

The atomic structure of glycinamide ribonucleotide transformylase, an essential enzyme in purine biosynthesis, has been determined at 3.0 A resolution. The last three C-terminal residues and a sequence stretch of 18 residues (residues 113 to 130) are not visible in the electron density map. The enzyme forms a dimer in the crystal structure. Each monomer is divided into two domains, which are connected by a central mainly parallel seven-stranded beta-sheet. The N-terminal domain contains a Rossmann type mononucleotide fold with a phosphate ion bound to the C-terminal end of the first beta-strand. A long narrow cleft stretches from the phosphate to a conserved aspartic acid, Asp144, which has been suggested as an active-site residue. The cleft is lined by a cluster of residues, which are conserved between bacterial, yeast, avian and human enzymes, and likely represents the binding pocket and active site of the enzyme. GAR Tfase binds a reduced folate cofactor and glycinamide ribonucleotide for the catalysis of one of the initial steps in purine biosynthesis. Folate analogs and multi-substrate inhibitors of the enzyme have antineoplastic effects and the structure determination of the unliganded enzyme and enzyme-inhibitor complexes will aid the development of anti-cancer drugs.

Structural evidence for induced fit as a mechanism for antibody-antigen recognition.

The three-dimensional structure of a specific antibody (Fab 17/9) to a peptide immunogen from influenza virus hemagglutinin [HA1(75-110)] and two independent crystal complexes of this antibody with bound peptide (TyrP100-LeuP108) have been determined by x-ray crystallographic techniques at 2.0 A, 2.9 A, and 3.1 A resolution, respectively. The nonapeptide antigen assumes a type I beta turn in the antibody combining site and interacts primarily with the Fab hypervariable loops L3, H2, and H3. Comparison of the bound and unbound Fab structures shows that a major rearrangement in the H3 loop accompanies antigen binding. This conformational change results in the creation of a binding pocket for the beta turn of the peptide, allowing TyrP105 to be accommodated. The conformation of the peptide bound to the antibody shows similarity to its cognate sequence in the HA1, suggesting a possible mechanism for the cross-reactivity of this Fab with monomeric hemagglutinin. The structures of the free and antigen bound antibodies demonstrate the flexibility of the antibody combining site and provide an example of induced fit as a mechanism for antibody-antigen recognition.

Structural aspects of antibodies and antibody-antigen complexes.

The structures of several Fab fragments and Fab-antigen complexes have now been solved at high resolution. These structures of antibodies in complex with proteins, peptides and various other haptens have enabled us to gain insights into the structural basis of immune recognition. Early structures of Fab fragments with and without bound haptens showed the antibody combining sites to be pockets or grooves. More recent Fab-protein complex structures have shown the antibody-antigen interactions to be more extensive with flatter, more undulating binding surfaces. We have solved the structures of three Fab fragments in their native form and as complexes with their respective antigens. Two of these are anti-peptide Fab fragments, the other an anti-progesterone Fab. Comparison of the free and bound structures indicates small but significant changes in the antibody on ligand binding. An analysis of the Fab complexes solved so far indicates that the antibodies can have very differently shaped binding sites, depending on the antigen.

Monoclonal antibodies against an identical short peptide sequence shared by two unrelated proteins.

Antipeptide antibodies provide the opportunity to explore the molecular basis for antigen-antibody recognition and to test theories of immune recognition. We investigated the possibility of raising monoclonal antipeptide antibodies against a specific epitope consisting of six amino acid residues, which is common to two unrelated proteins. The goal of this investigation was to analyze the reactivity of these epitope specific antibodies towards the same sequence in these two different proteins. A correlation between antibody reactivity and secondary structures of the same peptide sequence in different proteins could help to understand the ability of antipeptide antibodies to react with their cognate sequence in intact folded proteins. Monoclonal antibodies were raised against one hexamer sequence, PGTAPK, that is present in both thioredoxin and Fab New lambda-light chain. The antipeptide antibodies reacted only with thioredoxin but not with Fab New in ELISA's, immune precipitation and Western blots. Determination of the antibody specificity through binding tests with peptide analogs revealed the influence of the residue N-terminal from the hexamer epitope on antibody binding. Because of the observed influence of the N-1 adjacent residue in peptide analogs, the discrimination between the protein antigens could not be interpreted clearly as the result of the different hexamer conformations present in the native structures of the two proteins. However, analysis of the antibody reactivity with peptide analogs with varying "frame residues" surrounding the hexamer epitope indicates the possible discrimination of different peptide conformations by the antibody.

Preliminary crystallographic data, primary sequence, and binding data for an anti-peptide Fab and its complex with a synthetic peptide from influenza virus hemagglutinin.

X-ray quality crystals which diffract to high resolution (less than or equal to 1.9-2.1 A) have been grown of an anti-peptide Fab and its complex with a 9-residue peptide antigen. Both crystals are monoclinic P2(1), with unit cell dimensions a = 90.3 A, b = 82.9 A, c = 73.4 A, beta = 122.5 degrees for the native Fab and a = 63.9 A, b = 73.0 A, c = 49.1 A, beta = 120.6 degrees for the complex. The peptide sequence corresponds to residues 100-108 of all influenza virus hemagglutinins (HA1) of the H3 subtype (1968-1987). The peptide antigen has been well characterized immunologically (Wilson, I.A., Niman, H.L., Houghton, R.A., Cherenson, A.R., Connolly, M.L., and Lerner, R.A. (1984) Cell 37, 767-778; Wilson, I.A., Bergmann, K.F., and Stura, E.A. (1986) in Vaccines '86 (Channock, R.M., Lerner, R.A., and Brown, F., eds) pp. 33-37, Cold Spring Harbor Laboratory, Cold Spring Harbor, NY), structurally, as a free peptide by NMR (Dyson, J.H., Cross, K.J., Houghton, R.A., Wilson, I.A., Wright, P.E., and Lerner, R.A. (1985) Nature 318, 480-483; Dyson, J.H., Lerner, R.A., and Wright, P.E., (1988) Annu. Rev. Biophys. Chem. 17, 305-324), as part of the intact antigen by x-ray crystallography (Wilson, I.A., Skehel, J.J., and Wiley, D. C. (1981) Nature 289, 366-373) and by binding studies to the HA molecule (White, J.M., and Wilson, I.A. (1987) J. Cell Biol. 105, 2887-2896). Knowledge of the three-dimensional structure of the complex will elucidate the details of how anti-peptide antibodies recognize a small peptide antigen and provide insights into the recognition of the same sequence in the intact protein antigen. As both native Fab and the peptide-Fab complex have been crystallized, we can also determine in addition whether changes in the structure of the antibody accompany antigen binding. The nucleotide sequence of the mRNA coding region of the anti-peptide Fab has been determined to provide the amino acid sequence ultimately required for the high resolution three-dimensional structure determination.

Immunogenicity of loop-structured short synthetic peptides mimicking the antigenic site A of influenza virus hemagglutinin.

In order to assess the relevance of conformation for the antigenic site A of the hemagglutinin of influenza virus we synthesized two peptides, comprising two variant sequences of the central part of site A (amino acids 140 - 146 of subunit HA1) inserted into an artificial peptide skeleton, which imposes a loop-like structure on the respective sequence stretch. Assuming that the loop structure in the synthetic peptides would roughly approximate to the structure of the cognate protein sequence we tried to raise protein-reactive anti-peptide antibodies. The antibodies obtained indeed showed reactivity against influenza virus, although the discriminating specificity with regard to a mutation at position 144 was lost for virus binding in contrast to the highly specific peptide binding. Considering the failures in raising anti-hemagglutinin antibodies against the site A by immunization with short flexible peptide our results support the hypothesis that conformation makes a major contribution to the immunogenic and antigenic characteristics of site A in influenza hemagglutinin.

Towards assignment of secondary structures by anti-peptide antibodies. Specificity of the immune response to a beta-turn.

In an attempt to assign secondary structure elements to protein primary structures with antibodies, we synthesized a model peptide (beta-peptide: TVTVTDPGQTVTY) with a putative beta-turn structure and analysed the anti-peptide antibodies for their specificity towards the turn sequence. At least 50% of the peptide fraction adopts the intended conformation of a beta-turn (DPGQ) inserted between the two segments of an antiparallel beta-sheet structure. The specific anti-beta-peptide antibodies of the hyperimmune response bind the beta-turn containing epitope of the immunogenic beta-peptide with a three orders of magnitude higher affinity than the synthetic control peptide (Gly-peptide: GGGGGDPGQGGGG). The affinity of the antibodies with specificity for the beta-turn region increases from the primary to the hyperimmune response. Therefore, probing of secondary structure elements, i.e., of individual beta-turn regions, by anti-peptide antibodies now seems feasible for proteins of known sequence and may result in sequence assignments of secondary structures.