FGF Receptor

Investigators:

Luca Pellegrini, Nic Harmer, K H Kim, Tom L Blundell

Collaborators:

Barbara Mulloy, National Bureau of Standards, London

John Heath and Jeremy Thurnbull,, Biochemistry, Birmingham

Sue Malcolm, Institute of Child Health, London

Abstract:

Our recent structure of fibroblast growth factor receptor 2 (FGFR2) in complex with its ligand (FGF1) and heparin at 2.3 Šresolution (Pellegrini et al., 2000) demonstrates how heparin, an analogue of the low affinity receptor heparan sulphate, can bring together two FGFR2: FGF1 complexes and dimerise the receptor. FGF binding to the receptor involves extensive hydrophobic and polar contacts with D2, the linker region, and D3, creating a continuous interface that buries 1700 Ų. The protein-heparin interface is large at 2240 Ų (divided in 617 Ų with FGF1 and 631 Ų with domain D2 in subunit A, and 992 Ų with FGF1 in subunit B). Protein heparin interactions involve ion pairs between basic side chains and heparin sulphate groups as well as H-bonds and non-polar interactions.

Future objectives include the in vitro and in crystal characterisation of the composition and structure of complexes with heparins of specific sequence with:

• other human FGFs and FGFRs and their splice forms,
• mutants found in craniosynostosis,
• fly and worm FGF and FGFR

References:

Pellegrini L., Burke D.F., von Delft F., Mulloy B., Blundell TL (2000) Crystal Structure of fibroblast growth factor receptor ectodomain bound to ligand and heparin. Nature 407, 1029-1034

Nagendra, H.G., Harrington, A.E., Harmer, N.J., Pellegrini, L., Blundell, T.L. and Burke, D.F. (2001). Sequence analyses and comparative modeling of fly and worm fibroblast growth factor receptors indicate that the determinants for FGF and heparin binding are retained in evolution. FEBS Letters 501, 51-58.

HGF/SF and Met receptor

Investigators:

Dima Chirgadze, Avi Bandapaudhy, Tom Blundell

Collaborators:

Ermanno Gherardi, Daniel Leitha, MRC Centre Cambridge.

Abstract:

Hepatocyte growth factor/scatter factor, HGF/SF, is essential for the development of the liver and other cells in the mammalian embryo. The domain organisation resembles that of plasminogen, including an N-terminal domain (N), four kringle domains (K) and a catalytically inactive serine proteinase domain.

A crystal structure at 2.3 Å resolution of the NK1 fragment, a naturally occurring partial agonist, has been defined [Chirgadze et al. 1999, Figure A],. NK1 is a splice variant of the polypeptide growth factor HGF/SF that consists of the N-terminal and first kringle domain acts as a partial agonist and requires heparan sulphate or soluble heparin for activity. In the crystals the two subunits in the dimer interact through an extensive interface (2010 Ų⁾, although the fragment is a monomer in solution. A surface region, defined by mutagenesis experiments, that appears to interact with the receptor has been defined [Chirgadze et al. 1999].

The crystal structures of NK1-heparin complexes show that heparin does not bridge sub-units (Figure B), but appears to act electrostatically [Leitha et al., 2001]. Mutagenesis experiments show that heparin binding to the site on the N domain is important for dimerization in solution and biological activity of NK1. Heparan sulphates play a complex role in biological activity of HGF/SF in which binding to the primary site in the N domain is essential for activity whereas binding to the K domain is inhibitory.

Thus, unlike NGF the HGF/SF NK1 molecules are monomeric in solution, but form dimers when heparin fragments, and therefore heparin sulphate bind. Interestingly, the same dimers are formed by NK1 on crystallisation in several different crystal forms. The dimers are presumably stabilised at the receptor by heparan sulphate or heparin, so leading to dimerisation of the receptor.

Future studies include determination of the structure of:

• NK2 and larger fragments of HGF/SF in complex with heparins
• Met receptor ectodomain
• Met tyrosyl kinase

References:

Chirgadze, D.Y., Hepple, J.P., Zhou, H., Byrd, R.A., Blundell, T.L. & Gherardi, E. (1999). Crystal structure of the NK1 fragment of HGF/SF suggests a novel mode for growth factor dimerization and receptor binding . Nature Structural Biology 6, 72-79.

Leitha D, Chirgadze DY, Mulloy B, Blundell TL, Gherardi E (2001) Crystal structures of NK1 heparin complexes reveal the basis for NK1 activity and enable engineering of potent agonists and the MET receptor. EMBO J. 20, 5543-5555

Nerve Growth Factor (NGF)

Investigators:

Tom Blundell

Collaborator:

Marja Makarow , Institute of Biotechnology, University of Helsinki, Finland

Abstract:

Nerve growth factor (NGF) is a neurotrophic factor that promotes the differentiation and survival of certain populations of neurons in the central and peripheral nervous systems. It is the prototype for a family of such factors, each with its own subset of target neurons. An increasing range of non-neuronal tissues have been found to be sensitive to NGF induced responses, most notably immune-related haematopoietic cells.

The active neurotrophin, beta-NGF is biologically active as a dimer [16]. The 2.3 Å crystal structure of this dimer [McDonald et al. 1991] showed that beta-NGF adopts the cystine-knot fold, which was later shown to be common to several other growth factors (e.g. transforming growth factor-beta and platelet derived growth factors). The dimer interface is principally hydrophobic, and buries 3032 Å2 of surface area per dimer. This large buried surface area leads to the formation of the extremely stable dimer (Kd = 10-13 M).

7S NGF is an alpha2beta2gamma2 complex in which the beta-NGF dimer is associated with four serine proteinases of the glandular kallikrein family (two alpha-NGF and two gamma-NGF subunits). gamma-NGF is an active serine proteinase capable of processing the precursor form of beta-NGF while alpha-NGF is an inactive serine proteinase. The 3.1 Å crystal structure of 7S NGF [Bax et al., 1998] shows that the two gamma-NGF subunits bind close to the termini of the beta-NGF dimer (3033 Å2 surface buried) and make extensive interactions (2440 Å2) with the beta-NGF subunits). The two alpha-NGF subunits are locked in the zymogen conformation, they are bound on opposite sides of the beta-NGF dimer, and they interact with the dimer via a short region of antiparallel beta-sheet (2130 Å2 surface buried in total). Two zinc ions, which stabilize the complex, are bound at the relatively small interfaces between the alpha-NGF and gamma-NGF subunits. The regions of the beta-NGF dimer that contact the alpha-NGF subunits overlap with those known to bind p75 neurotrophin receptor. The structure of 7SNGF shows how the two-fold axis of the central beta-NGF dimer organizes the symmetry of this multiprotein growth factor complex.

The structures of the ?-NGF dimer, the 7S-NGF complex and the TrkA receptor domain complex show that a stable symmetrical dimer is very efficient organiser of the symmetry of multiprotein complexes in which it is involved, and this brings together receptor dimers in a way that would facilitate transphosphorylation of receptor tyrosyl kinases and activation of the downstream signalling pathway. The structures also provide us with insights into how specific signalling can arise in multiprotein complexes, allowing reliable interactions in crucial developmental pathways and immune cell development.

Future objectives: In 1992 we obtained soluble p75 receptor from Parke Davis, but this failed to crystallise, although it formed a well defined 2:2 complex in solution with ß-NGF. We now have an opportunity to revisit this project in collaboration with Marja Makarow at the Institute of Biotechnology, University of Helsinki, Finland. They are using Saccharomyces cerevisiae, as heterologous hosts to produce the secreted extracellular part of p75 in a functional form. We wish to compare different binding sites on NGF dimers and to study mechanisms by which this growth factor can interact with several unrelated proteins.

References

McDonald N, Lapatto R, Murray-Rust J, Gunning J, Wlodawer A and Blundell TL (1991) New Protein fold revealed by a 2.3Å resolution crystal structure of nerve growth factor. Nature 345: 411-414

Bax, B., Blundell, TL., Murray-Rust, J. and McDonald N.Q. (1997) Structure of mouse 7S NGF: a complex of nerve growth factor with four binding proteins. Structure. 5: 1275-1285.

Insulin, IGF and their receptors

Investigators:

Tom L Blundell

Collaborator:

Ken Siddle, University of Cambridge.

Abstract

The determination of crystal structure of insulin in the Dorothy Hodgkin laboratory in Oxford in 1969 provided the first insights into the architecture of a polypeptide hormone/growth factor and correlation with the results available on sequences, chemical modification, receptor binding and biological activity indicated that the general topology of the insulin molecule was required for activity, and that a large region of the surface of the monomer was interacting with the receptor [Blundell et al., 1972].

The insulin receptor and the homologous type 1 IGF receptor are Mr 350,000 glycoproteins, composed of two disulphide-linked subunits, each containing a single transmembrane helix and an intracellular tyrosine kinase region. We have contributed to showing that the extracellular region of the receptors comprises two homologous globular domains (L1 and L2) flanking a cystine-rich domain, followed by three fibronectin III repeats. The L1 and L2 domains of IGFR have ?-helical structures and are connected to the cystine-rich domain by flexible linkers, indicating that the relative positions of these domains in the crystal structure may not be retained in complex with the ligand. Studies with chimaeric insulin/IGF-I receptors and alanine mutagenesis suggest that the C-terminus of cystine-rich domain, the LI domain and a C-terminal peptide (residues 692-702) form the ligand-binding site.

Future work will include studies of the structure of the insulin receptor ectodomain.

References:

Marino-Buslje, C., Mizuguchi, K., Siddle, K., Blundell, T. (1998). A third fibronectin type III domain in the extracellular region of the insulin receptor family. FEBS Letters 441, 331-336.

Marino-Buslje C, Martin-Martinez M, Mizuguchi K, Siddle K and Blundell TL (1999) The insulin receptor: from protein sequence to structure. Biochemical Society Transactions 27. 715-726.

Intracellular signalling and cell cycle

Cdk6 complex with cell cycle inhibitor p19INK4

Investigator:

Tom Blundell

Collaborator:

Ernest Laue , Cambridge.

Abstract:

In memory of Venogopal Dhanaraj who solved the structure but tragically died of a heart attack in 2001.

The structure of the cyclin D-dependent kinase Cdk6 complexed with cell cycle inhibitor p19INK4d at 1.9 Šresolution, defined in collaboration with Ernest Laue (Brotherton et al., 1998), (Figure 1e) provided the first structural information for a cyclin D-dependent protein kinase and showed that the INK4 family of Cdk inhibitors do not interfere directly with ATP or substrate binding but rather stabilise a conformation that does not bind ATP. The conformation is likely to be low energy as the surface is mixed hydrophobic and polar, and buries an area of 1700 Ų.

References:

Brotherton, DH., Dhanaraj, V., Wick, S., Brizuela, L., Domaille, PJ., Voyanik, E., Xu, X., Parisini, E., Smith, OB., Archer, SJ., Serrano, M., Brenner, SL., Blundell, TL. and Laue, ED. (1998) Crystal structure of the complex of the cyclin D-dependent kinase Cdk6 bound to the cell cycle inhibitor p19INK4d.

Nature 395, 244-250

CK2 protein kinase

Investigators:

Muhammed Sayed, Victor Bolanos Garcia, Tom Blundell

Collaborators:

Jorge and Catherine Allende, Santiago, Chile.

Abstarct:

Protein kinases CK2 (casein kinase 2) are a ubiquitous family of eukaryotic enzymes, usually composed of protein kinase catalytic subunits (a) and (a'), and smaller regulatory subunits (ß) arranged as heterotetramers. They may be involved in the regulation of cell division and indeed the a-subunits are most closely related in sequence to the cyclin-dependent kinases. However, the regulatory subunits, CK2 (, show no significant sequence similarity to those of various cyclins, except for the "cyclin destruction box". Together with our collaborators Catherine Allende and Jorge Allende (Santiago) we have designed and made site-directed mutants of CK2a, and constructed a three-dimensional model, to investigate the basis for the dual specificity for ATP or GTP. The structural and functional effects of the regulatory subunit are not well understood, although the structures of the two separate subunits and a complex of Zea mays CK2a and a peptide of CK2( subunit have been defined elsewhere. CK2ß binds tightly to several substrates of CK2a such as p53, thus contributing to an increase in the effective substrate concentration in the vicinity of the catalytic active site.

Future objectives:

We have produced crystals of the CK2 tetrameric complex (a(2ß2), but they exhibit very weak diffraction; we will attempt to improve these. The next objective is to study the structural features of the interaction of CK2( peptides with several substrates of CK2 such as p53 and p21WAF1/CIP1. CK2( 1-44, 49-63, 201-215 interact with p21 46-65 and CK2( 71-149 interacts with p53 320-339. Expression in baculovirus should improve yields of folded subunits and fragments.

References:

Korn I, Gutkind S, Srinivasan N, Blundell TL, Allende CC, Allende JE (1999) Interactions of protein kinase CK2 subunits. Molecular and Cellular Biochemistry. 191, 75-83.

Srinivasan N, Antonelli M, Jacob G, Korn I, Romero F, Jedlicki A, Dhanaraj V, Sayed M F-R, Blundell TL, Allende CC, Allende JE (1999) Structural interpretation of site-directed mutagenesis and specificity of the catalytic subunit of protein kinase CK2 using comparative modelling. Protein Engineering. 12, 119-127.

DYRK Protein kinase

Investigators:

Muhammed Sayed, Tom Blundell

Collaborators:

Walter Becker, Aachen, Germany

Len Packman, Cambridge.

Abstarct:

DYRK kinases play essential roles in the control of growth and development in eukaryotes and have a conserved Tyr-Xxx-Tyr motif in the "activation loop". Human DYRK1A auto-phosphorylates on Tyr, but phosphorylates exogenous protein only on Ser or Thr. We have shown by mutagenesis and direct mass-spectrometric analysis that Tyr321, but not Tyr319 in the Tyr-Xxx- Tyr motif is auto-phosphorylated both in vitro and in vivo and mutation of Tyr321 (but not Tyr319) impairs kinase activity (Himpel et al., 2001). DYRK kinases contain non-catalytic N-terminal and C-terminal segments (699 and 424 amino acids) that are involved in the regulation of their activity. The conserved DH-Box (DDDNxDY), immediately preceding the catalytic domain, probably represents a site of interaction with substrates. Other regions represent putative nuclear localization and rapid turnover signals. Understanding these regulatory "domains" is the longer-term objective of this project.

Initially we seek to define the structure of the catalytic domain of DYRK1A. Our collaborator Walter Becker has provided a C-terminal His-tagged construct encompassing the catalytic domain of DYRK1A, which was expressed in E.coli and purified using metal affinity and gel filtration to give very thin needle-like crystals, whichstill need to be improved for X-ray analysis. In the longer term we seek to define globular regions in the regulatory regions that can be crystallised or studied by NMR. We will also synthesise shorter peptides for cocrystallisation with other interacting proteins. We need to move to baculovirus to improve expression.

References:

Himpel S, Panzer P, Eirmbter K, Czajkowska H, Sayed M, Packman LC, Blundell TL, Kentrup H, Grotzinger J, Joost H-G, Becker W (2001) Identification of the autophosphorylation sites and characterisation of their effects in the protein kinase DYRK1A Biochem. J. 359, 497-505.

TRABID M-J L

The Structure, Specificity & Mechanism of Li⁺-sensitive/Mg²⁺-dependent Phosphatases

Investigators:

Sahil Joe Patel and Tom L Blundell

Collaborators:

Armando Albert, Instituto de Quimica-Fisica Rocasolano, Madrid

Ramon Serrano, Valencia, Spain.

Abstract:

The Li+-sensitive/Mg2+-dependent phosphatases have been identified in most species on the basis of their two highly conserved sequence motifs. We have earlier defined the crystal structures of inositol phosphatases from yeast Hal2p from Saccharomyces cerevisiae.

The crystal structures of six Hal2 metal ion complexes were solved at 1.3-1.75Å resolution in order to understand ion selectivity and catalytic mechanism. The discrimination of heavier cations from water and magnesium was based on the presence or absence of anomalous signal. These crystal structures highlighted the difference in cation accessibility of the three metal ion-binding sites. The crystal structure of a protein-substrate complex, solved to high resolution, provides insight into the side chain arrangement required for catalysis and identifies the nucleophilic water molecule responsible for phosphoester bond hydrolysis. The proposed mechanism involves a deprotonation relay between this water, Thr147 and Asp49.

The crystal structure of RnPIP from Rattus norvegicus in a product complex was solved at 1.7Å resolution by a selenomethionine MAD experiment at the European Synchrotron Radiation Facility (Patel et al., 2001). The structure was built in an automated fashion using ARP/wARP with minimal user input and produced a near complete model of the protein. This novel dual specificity enzyme displays a core fold similar to other Li+-sensitive/Mg2+-dependent phosphatases with greatest similarity to that of IPPase whose activity it also shares. Modelling shows that the active site can accommodate both PAP and I1,4P2. A mechanism similar to that of Hal2 is proposed.

A homology model of HsPIP based on the crystal structure of RnPIP shows the enzymes to be identical in overall fold. Sequence differences, other than those localised in loop regions, are conservative in nature. The failure to crystallise is thought to be due, in part, to the Gln158?Glu amino acid change that is responsible for important hydrophobic crystal contacts in the RnPIP structure.

References:

Albert A, Yenush L, Gil-Mascarell MR, Rodriguez PL, Patel S, Martinez-Ripoll M, Blundell TL, Serrano R (2000) X-ray structure of yeast Hal2p, a major target of lithium and sodium toxicity and identification of framework interactions determining cation sensitivity. Journal of Molecular Biology 295, 927-938.

Patel S, Blundell TL (2002) Crystal Structure of an Enzyme Displaying both Inositol-Polyphosphate 1-Phosphatase and 3'-Phosphoadenosine-5'-Phosphate Phosphatase Activities: A Novel Target for Lithium Therapy. J. Mol. Biol. 315, 677-686.

Farnesylated transducin-beta gamma from bovine retina complexed with phosducin

Investigators:

Tom Blundell

Collaborators:

Ben Bax, Birkbeck College

Andreas Loew and Y-K Ho, Chicago

Abstract:

The structure of farnesylated transducin-beta gamma from bovine retina complexed with phosducin at 2.8Šresolution (Loew et al., 1998; Figure 1f) led to the unexpected discovery of a conformational change in transducin-beta gamma leading to a binding pocket for farnesyl and stabilisation of the Gbeta gamma - phosducin as a soluble complex. This showed how the complex is sequestered away from the membrane. The surface areas involved are large, with Gbeta to Ggamma of 5,100 Ų and Gbeta with phosducin of 4,400 Ų, characteristic of stable, adaptor, and inhibitory functions.

References

Loew, A., Ho Y-K., Blundell, TL. and Bax (1998) Phosducin induces a structural change in transducin betagamma. Structure 6, 1007-1019

DNA Damage Signalling Signalling networks link the molecular machinery responsible for DNA repair to the cell cycle. Failure of this molecular circuitry leads to genome instability. We are pursuing projects on enzymes central to the two major pathways of DNA repair in eukaryotic organisms: homologous recombination and non-homologous end joining (NHEJ).

XRCC4 - DNA ligase IV complex

Investigators:

Lynn Sibanda, Luca Pellegrini, Andy Dore, Tom Blundell.

Collaborators:

Steve Jackson, Wellcome/CR-UK Institute, Cambridge

Abstarct:

The structure of the XRCC4 - DNA ligase IV complex determined in our laboratory gives the first definitive indications as to how the signalling from double strand DNA breaks is mediated by multicomponent complexes (Sibanda et al., 2001). It shows an unusual XRCC4 dimer with an N-terminal globular head that probably binds DNA and a coiled-coil C-terminus that binds the linker region between the BRCT domains of ligase IV. The interactions involve both ion pair and non-polar interactions. The structure differs radically from the previously defined XRCC4 tetramer where the C-terminal helices are flayed open. In the region of binding, the XRCC4 sequence is conserved and the coiled-coil is distorted. Repair of chromosomal double strand breaks requires intervention of a group of proteins comprising the protein kinase, DNA-PK, the Ku hetero-dimer, XRCC4, and DNA Ligase IV in the NHEJ pathway. Ku recruits a complex of XRCC4-DNA Ligase IV to the DNA ends, while DNA PK provides a signalling link with the cell cycle proteins.

Future work will be directed first at crystallisation of the complete C-terminal region of DNA ligase IV, which contains the BRCT repeats, in complex the XRCC4. We will then map the regions of the XRCC4-Ligase IV complex that interact with other components of the NHEJ DNA repair complex. The XRCC4 functional core 3-D structure has a suggestive helix-turn-helix region in the globular head that likely binds DNA, and the long coiled-coil helical region is a strong candidate for assembly of Ku. We will test these hypotheses by site-directed mutagenesis and by defining structures of the multicomponent complexes.

We have also cloned and expressed most of the components in yeast, which differ greatly in sequence from those in human and where DNA PK appears to be absent. Expression in baculovirus should improve yields of the more complex multidomain proteins.

References:

Sibanda BL, Critchlow S, Begun J, Pei XY, Jackson SP, Blundell TL, Pellegrini L (2001) Insight into the mechanism of DNA end joining from the structure of an Xrcc4 dimer in complex with DNA ligase IV. Nature Structural Biology 8, 1015-1019

Interactions of tumor suppressor protein BRCA2 with RAD51

Investigators:

Thomas Lo, Tom Blundell, Luca Pellegrini.

Collaborators:

Ashok Venkitaraman, Cambridge

Abstract:

Mutations in the BRCA1 and BRCA2 gene are responsible for the majority of familial cases of breast cancer. BRCA2, a protein of over 3000 amino acids which has essential roles in embryonic proliferation and tumor suppression in adult cells. It binds RAD51, which performs homologous DNA reconstitution and has an evolutionary conserved core resembling bacterial RecA protein. The binding through eight short BRC repeats, coded within the BRCA2 exon 11, is essential for maintaining chromosomal stability. In collaboration with Ashok Venkitaraman (Cambridge) we have produced and set up crystallisation trials of the RecA-like RAD51 core (RCT) in complex with both BRC3 and BRC1. We have defined the minimal size of BRC repeats necessary for RAD51 binding by limited proteolysis of the complex, and established that the stoichiometry of the RAD51-BRC3 complex is 1:1, surprising as free RAD51 is oligomeric.

Our future work will focus on structures of RAD51-BRCA2 BRC complexes. We will determine the RAD51 structure on its own, to ascertain how its oligomeric nature is affected by BRC binding. We will investigate complexes with both a single repeat, and multiple repeats. We are also interested in regions near the C-terminus of BRCA2 that appear to be structurally organised. A large number of proteins beside RAD51 are involved in homologous recombination: RAD52, XRCC2, XRCC3 etc. As their role in the process becomes clearer, we plan to subject them to structural analysis.

Structural and biochemical studies of biosynthesis enzymes

Pantothenate biosynthesis

Scheme 1: The Pantothenate Pathway in E. coli

Aspartate decarboxylase (ADC)

Abstract:

We have defined the structure of ADC at 1.9Å resolution (Albert et al. 1998) and studied the in vitro processing of ADC involving an active site pyruvoyl, and isolated ADC tetramers in different states of processing. We have prepared a series of mutants to determine the contributions made by various amino acids to protein processing and catalysis of the decarboxylation reaction [specifically S25C, G24S, A24' (introducing an extra amino acid), A26', Ser25T, Y58F, K9Q or K9A, H11A]. All the constructs have been transferred to vectors such that the wild-type and mutant forms of ADC are expressed with N-terminal His6 tags, to facilitate purification. We have characterised the processing and catalytic activity of the mutant proteins, and are carrying out structural studies on several of them. The S25A mutant, in which the serine destined to become the pyruvoyl group is absent, prevents any processing, and has no activity. Converting Ser25 to Cys slows processing and reduces activity, while the S25T mutant does not process at all (nor has activity). The G24S mutant behaves like wild type in processing, but has much reduced catalytic activity. Adding an alanine immediately before Gly24 or after Ser25 slows processing considerably; the former is inactive, whereas the latter has some residual catalysis. These results are particularly relevant to the idea that loop strain drives the processing. Looking at presumed catalytic residues: K9A is like wild type, K9Q processes more slowly and H11A more quickly. Y58F processes normally, but has no activity. We have high-resolution crystals for S25A, and lower resolution crystals for S25T.

We have solved the following liganded structures of ADC (all to approximately 1.8 Å):

1. bound (as an imine at the pyruvoyl group) to aspartate (substrate),
2. bound to methyl aspartate,
3. bound to beta-isopropyl-beta-alanine (where the carboxyl group is replaced with an isopropyl group),
4. bound to beta-alanine (product), and
5. with beta-alanine bound and then the imine reduced by borohydride.

On the basis of these structures we are proposing a detailed mechanistic scheme for catalysis that involves a previously unrecognised active site flap (manuscript in preparation)

References:

Albert, A., Dhanaraj, V.,Genschel, U., Khan, G., Ramjee, MK., Pulido, R., Sibanda, BL., von Delft, F., Witty, M., Blundell, TL., Smith, AG and Abel, C. (1998) Crystal structure of aspartate decarboxylase at 2.2Å resolution provides evidence for an ester in protein self-processing. Nature Structural Biology 5(4): 289-293.

von Delft, F., Dhanaraj, V., Witty, M., Pulido, R., Blundell, T.L., Smith, A.G. and Abell, C. New insights into the catalytic mechanism of L-aspartate-gamma-decarboxylase from high resolution structures of enzyme ligand complexes. Manuscript in preparation

Ketopantoate-hydroxymethyl-transferase (KPHMT)

Abstract:

We have demonstrated that the structure of E. coli KPHMT is decameric by ultracentrifugation. We solved the crystal structure of the enzyme to 1.8 Å resolution using a combination of native and selenomethionine proteins. Solving this structure was a significant technical achievement, as there were 160 selenium atoms in the unit cell! The protomer is a closed beta-barrel. The decamer is toroidal comprising two closed superposed circles, each of five protomers. Rather serendipitously, the product, ketopantoate is bound in the active site. We have been able to model in the substrate alpha-ketoisovalerate, and to propose the position of binding of the cofactor methylenetetrahydrofolate (manuscript in preparation). We have constructed several site-specific mutants of the panB gene, and tested them for their ability to complement a panB auxotroph. By this means E181 (the active site base) has been shown to be an essential residue.

Ketopantoate Reductase (KPR)

Abstract:

At the start of this grant we did not know the identity of the gene encoding KPR. Work in S. typhimurium established that the apbA gene, initially identified as a gene involved in the alternative pyrimidine biosynthesis pathway, encoded an enzyme with ketopantoate reductase activity, and was allelic to panE. Using this information, we were able to amplify the E. coli panE gene by PCR, and clone it into an expression vector. The 31 kDa enzyme protein is expressed to very high levels (13 mg/L of culture), and was easily purified in three chromatographic steps.

Using the selenomethionine MAD method, we solved the structure of the apoenzyme to 2.4 Å. The enzyme is monomeric and has two domains, the N-terminal domain has an alpha-beta fold of Rossman type, while the C-terminal domain is all alpha helical. The structure is topographically similar to N-(1-D-carboxylethyl)-L-norvaline dehydrogenase, which catalyses the NADH-dependent reaction between hydrophobic L-amino acids and ?-ketoacids. This work has been published in Biochemistry (Matak-Vinkovic et al., 2001).

Pantothenate synthetase (PtS)

Abstract:

We have solved the unliganded structure of E. coli PtS to 1.7 Å. This revealed that pantothenate synthetase is a member of the aminoacyl-tRNA synthetase superfamily, and enabled us to identify the ATP binding site and a putative pantoate binding site. We have proposed a catalytic mechanism that involves initial binding of pantoate, then binding of ATP, with an associated hinge bending conformational change. This work has been published in Structure (von Delft et al., 2001).

References:

Matak-Vinkovic, D., Vinkovic, M., Saldanha, S.A., Ashurst, J.L., von Delft, F., Inoue, T., Miguel, R.N., Smith, A.G., Blundell, T.L. and Abell, C. (2001) Crystal structure of Escherichia coli ketopantoate reductase at 1.7 angstrom resolution and insight into the enzyme mechanism. Biochemistry 40: 14493-14500.

von Delft, F., Lewendon, A., Dhanaraj, V., Blundell, T.L., Abell, C. and Smith, A.G. (2001) The crystal structure of E. coli pantothenate synthetase confirms it as a member of the cytidylyltransferase superfamily. Structure 9: 439-450.

von Delft, F., Inoue, T., Saldanha, S.A., Dhanaraj, V., Witty, M., Abell, C., Smith, A.G. and Blundell, T.L. The crystal structure of E. coli ketopantoate hydroxymethyltransferase with product bound, solved by locating 160 selenomethionine sites using the multiwavelength anomalous dispersion (MAD) method. Manuscipt in preparation

Biosynthesis of the Antibiotic Chloroeremomycin

Investigators:

Ricardo Núñez Miguel, Bojana Popovic, Tom L Blundell

Collaborators:

Dudley H. Williams and Jonathan B. Spencer , Department of Chemistry, Cambridge.

Abstarct:

Over the past two or three decades an increasing number of microbial pathogens have become antibiotic-resistant. Chloroeremomycin from Amycolatopsis orientalis is a member of the vancomycin group of antibiotics that are of great importance in combating infections due to methicillin-resistant Staphylococcus aureus. Vancomycin is produced by Amycolatopsis orientalis and was first used in the clinic in 1959. The acquisition of resistance by Staphylococcus aureus to virtually all antibiotics in clinical use has propelled the vancomycin group to the forefront of the fight against these bacteria. The prospect of engineering new antibiotics in this family to fight further resistance that will inevitably arise will depend critically on understanding the mechanism of synthesis of these complex molecules.

The glycoheptapeptide backbone of chloroeremomycin contains two 4-hydroxyphenylglycine (HPG) and one 3,5-dihydroxyphenylglycine (DHPG) residues and two 4-epi-vancosamine (4eV) and one glucose sugars. The glycoheptapeptide backbone has the sequence: R-Leu-R-Tyr-S-Arn-R-HPG-R-HPG-S-Tyr-S-DHPG. DHPG moiety is derived from acetate, probably via a polyketide mechanism, whereas the HPG residues are derived from tyrosine. Our research is directed towards a better understanding of the structure and catalytic mechanism of the enzymes involved in chloroeremomycin biosynthesis, with a view to using that information to produce new glycopeptide antibiotics.

Scheme of the open reading frames of enzymes involved in the biosynthetic pathway of chloroeremomycin:

A.M.A. van Wageningen, P.N. Kirkpatrick, D.H. Williams, B.R. Harris, J.K. Kershaw, N.J. Lennard, M. Jones, S.J.M. Jones and P.J. Solenberg. Chem. Biol. 5 (1998) 155-162.

3D structural protein-cofactor-substrate complex models have been obtained by comparative modelling, from which several residues have been identified as mutation targets that could allow the production of new antibiotics.

The biosynthesis of HPG from tyrosine is achieved in several steps, involving 4-hydroxymandelate synthase, 4-hydroxymandelate oxidase and 4-hydroxyphenyl-glyoxylate transaminase. The model of 4-hydroxymandelate synthase has allowed the identification of three important residues in the active site as targets for mutagenesis. By studying the model of NDP-4-keto-6-deoxyglucose 3,5-epimerase, involved in the production of 4eV, it has been concluded that Tyr 139 is hydrogen bonded to the hydroxyl group in C2 of the sugar, while Thr 141 is the closer residue to the hydroxyl group in C3. Probably Thr 141 is involved in the isomerization of C3. If mutations of T141Y and Y139T are made, it is possible that the situation could be reversed and a 2,5-epimerization instead of a 3.5-epimerization be produced.

The model of orf22 the 4-hydroxymandelate oxidase (left) and orf17 the 4-hydroxyphenyl-glyoxylate transaminase (right).

The biosynthesis of DHPG is achieved by utilising four malonate molecules in a polyketide mechanism to produce 3,5-dihydroxyphenylacetate. The enzyme coded for by orf27 is chalcone synthase-like enzyme and an unusual type III polyketide synthase. The highly conserved catalytic triad characteristic for orf27 and a variety of plant polyketide synthases consists of Cys 160, His 296, Asn 329. The chalcone synthase acts as a homodimer of 42kDa protomers. Each protomer consists of the N-terminal thioesterase domain (alpha-beta-alpha-beta-alpha pseudo-symmetric motif) and a C-terminal substrate-binding pocket. The fold consists of two similar domains, each with a five-stranded sheet. The three downstream genes coded for by orf28, orf29 and orf30 have homologies to enoyl-CoA hydratase/dehydratase (orf28 and orf30 ) and dehalogenase (orf29). They are involved in synthesis of 3,5-dihydroxymendalate and 3,5-dihydroxyphenyl glycolate respectively.

There are three cross-links in the biosynthesis of chloroeremomycin. All are produced by p450 proteins, which have also been modelled.

In parallel to the modelling studies we have now expressed and puified several of the enzymes, including the chalcone synthase-like enzyme (orf27) as well as hydrogenase/dehydrogenase enzyme (orf29), ready for crystallization and X-ray analysis.

References:

R. Nunez Miguel, T.L. Blundell, J. Spencer and D.H. Williams. In preparation.

Tsung-Lin Li, Oliver W. Choroba, Hui Hong, Dudley H. Williams, Jonathan B. Spencer (2001) Biosynthesis of the vancomycin group of antibiotics: characterisation of a type III polyketide synthase in the pathway to (S)-3,5-dihydroxyphenylglycine. Chem. Commun. 2156-2157

Williams, D.H. and Bardsley, B. (1999) The vancomycin group of antibiotics and the fight against resistant bacteria. Angew. Chem. Int. Ed. 38, 1172-1193.