G. J. Hol, Ph.D.
Professor of Biochemistry and Biological Structure
Head, Biomolecular Structure Center
University of Washington, Box 357742
Health Sciences Bldg., Room K-428
Seattle, WA 98195
PH: (206) 685-7044
FX: (206) 685-7002
Email address: email@example.com
(updated 29 November 2012)
The major goals of our laboratory are to unravel the three-dimensional structures of protein molecules; to explore the relationships among protein structure, function, and dynamics; and to exploit, where possible, this insight for the design of new medically relevant molecules, in particular in the field of infectious diseases.
Specifically, the group focuses on providing a solid structural basis for the design of therapeutic agents to treat the following global diseases:
A. RESEARCH SUMMARY - Wim G.J. Hol
Overview of the results from mid 1994 until mid 2004
The studies in the last ten years were mainly focused on unraveling the crystal structures of key proteins from major tropical pathogens in parallel with structure-based development of inhibitors of these proteins. Some examples are given below.
In addition, the structure of human topoisomerase I has been elucidated which is of relevance for the development of new anticancer agents. The mode of action of the DNA repair enzyme Tdp1 has been unraveled. Studies of the pyruvate dehydrogenase and related very large multi-enzyme complexes, have provided detailed insight into the architecture of these molecular machines, and also resulted in a profound understanding of maple syrup urine disease.
Several contributions have also been made to the development of methods in protein crystallography. Most recently these have been made largely within the framework of the Structural Genomics of Pathogenic Protozoa (SGPP) Consortium which has expressed thousands of genes from trypanosomatid and Plasmodium species, purified hundreds of proteins from these pathogens, and solved dozens of crystal structures.
All these results could only be obtained because of:
RESEARCH SUMMARY PART I mid 1999 to early 2004
1. Structure determinations and Structure-based inhibitor Design
1.1 The cholera toxin family: Action and Inhibition
Cholera toxin (CT) and E. coli heat labile enterotoxin (LT) are two very similar heterohexameric AB5 protein toxins, secreted by V. cholerae and enterotoxigenic E. coli, respectively. The B-pentamer is responsible for GM1 receptor recognition on the surface of host epithelial cells and the A subunit for ADP-ribosylating a specific arginine of GSα (1) . Once covalently modified, GSα continuously stimulates adenylyl cyclase leading to excess levels of cAMP, and export of salts and fluids (2).
Based on our activation hypothesis (3), our collaborator Randall Holmes (Colorado) discovered a constitutively active Y30S A-subunit CT mutant whose structure not only led to the first view of an active member of the CT family (4) but also provided insight into the binding mode of NAD+.
Our extensive structural information on LT and CT, (5-15) has been used for the design of inhibitors of receptor recognition in collaboration with the groups of Verlinde (modeling) and Fan (synthesis) at the University of Washington. Successive structure-guided combinatorial chemistry approaches have resulted in monovalent galactose variants with a 104 -fold improved affinity over galactose (16-19) . Recent results with bivalent inhibitors have revealed intriguing gains in affinity (20) . Incorporation of galactose derivatives into pentavalent and decavalent ·macro-inhibitors· with molecular weights over 14 kDa, have increased affinity by more than seven orders of magnitude compared to our starting monovalent ligand galactose (21-23). It appeared that the effective linker length is the key in gaining affinity (24).
Figures I 1.1
1.2 Glycosomes in trypanosomatids
Glycosomes, unique and essential organelles in trypanosomatids, contain a large number of glycolytic and other enzymes (25). Import of these nuclear-encoded proteins into the glycosome is a fascinating process involving about 25 proteins, called peroxins, whose function is only partially understood (26-29). We are studying trypanosomatid glycosomal enzymes and several glycosomal import proteins in collaboration with the groups of Michels in Brussels (parasitology, expression systems), and Gelb (chemistry), Fan (chemistry), Verlinde (modeling), Van Voorhis and Buckner (in vivo testing) at the University of Washington.
Our structure determination of L. mexicana glycerol-3-phosphate dehydrogenase (GPDH) (30) has been followed up by a study of complexes both with inhibitors (31) and with a bisubstrate adduct. The latter showed a major conformational change of a crucial active site loop and domain closure of this glycosomal enzyme (32).
Glycosomal matrix proteins carry either a PTS1 or a PTS2 target signal which is recognized by the receptor peroxins 5 or 7, respectively (33) . Signal recognition is followed by a very complex series of events, involving numerous peroxins. Our studies of the peroxins have led to the structure of a triple Tetratrico Peptide Repeat (TPR) fragment of T. brucei peroxin 5 (34) where one pair of helices adopted a surprisingly elongated conformation, suggested the possibility of a "jack-knife" motion during the functioning of this protein. Detailed insight into the interactions of T. brucei peroxin 5 with peroxin 14 was obtained by mutagenesis studies (35,36).
Figure I 1.2
1.3 Eukaryotic Topoisomerases I and Tyrosyl DNA Phosphodiesterases (Tdp1)
Topoisomerases are involved in critical processes in the cell such as DNA relaxation, DNA recombination and chromatin remodeling (37-40). Several Topoisomerase I (Topo I) poisons have proven to diminish growth of Plasmodium falciparum and of Leishmania (41-43). Soluble expression of the malaria enzyme, and numerous fragments thereof, has been challenging, but major progress has recently been made in the expression of the two subunits of trypanosomatid Topo I. These pathogens contain heterodimeric Topo I's, of which only parts are related to the monomeric human enzyme studied previously in our laboratory (44-47). By combining the proper variants of the two trypanosomatid Topo I subunits active enzyme has recently been obtained.
Tdp1 is a DNA repair enzyme that prevents cell death by cleaving the bond between the crucial catalytic Topo I tyrosine and DNA to remove the large Topo I polypeptide chain from the 3' end of the broken DNA strand when Topo I religation has stalled (48-50). Our studies on human Tdp1 have not only fully unraveled its structure (51,52) but also its catalytic mechanism, by trapping the enzyme, peptide, DNA, and vanadate in a transition state-mimicking quaternary complex (53) . This result has made it possible to explore the characteristics of the peptide and DNA binding sites of Tdp1 by unnatural compounds (54). Expression of the trypanosomatid Tdp1 is making good progress. Our collaborators Heidrun Interthal and Jim Champoux (Microbiology, UW) have recently carried out knock-out experiments revealing that Tdp1 is essential in T. brucei.
Figures I 1.3
1.4 Mycobacterial proteins
M. tuberculosis is able to sense oxygen starvation and go into a persistent stage which causes the long duration of tuberculosis therapies (55-57). Three mycobacterial proteins, of potential relevance for the persistent stage, are being investigated.
1.4.1 The iron-dependent regulator (IdeR) of M. tuberculosis
IdeR, an essential gene in M. tuberculosis, is involved in iron-metabolism and possibly in the signaling cascade leading to persistence in macrophages (58). Our structure determination of fully metal-activated IdeR (59), a collaboration with the Holmes group (U. Colorado), was followed recently by that of activated IdeR, in complex with mbtA/mbtB operator DNA controlling siderophore synthesis (60). As we found previously with the C. diptheriae homolog DtxR in complex with DNA (61), the SH3-like third domain is engaged in metal-binding. A cyclic peptide, designed by the Verlinde group and synthesized and tested by the Beeson group (South Carolina), appears to modulate the affinity of IdeR for DNA, providing a platform for the next generation of the regulator.
1.4.2 The dormancy-survival regulator DosR
DosR (62) is a key protein in the oxygen sensing mechanism and the point of convergence of two kinases as shown in a recent paper (63) with our collaborator David Sherman (Pathobiology,UW). We have therefore started structural studies of DosR which have so far led to crystals of the C-terminal domain (1.7 Å) and of the C-terminal domain in complex with consensus hypoxic promoter DNA (3.2Å).
1.4.3 The special thymidylate synthase - ThyX.
M. tuberculosis employs an unusual F420 flavin cofactor-containing thymidylate synthase, ThyX (64), which is possibly crucial for survival of the bacterium in low oxygen conditions (65). In collaboration with Carol Sibley (Genome Sciences, UW), who provided an expression system, we have obtained recently crystals which diffracted to 2.8 Å resolution. Structure determination is in progress.
Figures I 1.4
1.5 Malaria proteins
P. falciparum has been reported (66,67) to be sensitive to inhibitors of the bacterial peptide deformylase (PDF) enzyme. Therefore P. falciparum PDF is considered a promising malaria drug target. In collaboration with Dehua Pei (Ohio State University), we solved the structure of the apo-enzyme, with crystals displaying complicated non-crystallographic symmetry (68). An engineered P. falciparum PDF variant gave improved and less complex crystals (69). Recently, structures of several complexes of engineered PDF with cyclic inhibitors, designed and synthesized in collaboration with the Verlinde and Pei groups (70), have revealed the ligand binding mode (71).
Entry into host cells by the P. falciparum sporozoites is an extremely sophisticated multi-step process of great interest for drug design (72). Recently, we assisted the group of Victor Nussenzweig (New York) in demonstrating that in P. falciparum the glycolytic enzyme aldolase connects the TRAP tail with actin in the inner-membrane complex which is responsible for moving parasite all into the host cell.(73). Building on our previously solved structure of P. falciparum aldolase (74), we obtained new crystal forms of malaria aldolase in complex with TRAP tail-derived peptides. This has led to most intriguing, peptide-like densities bound to aldolase, which are, however, still too ill-defined so far to draw definite conclusions.
Figure I 1.5
2. Molecular machines
2.1 Type II secretion system of Vibrio cholerae and enterotoxigenic E. coli.
Cholera toxin (CT) and Heat-labile enterotoxin (LT) are assembled in the periplasm and subsequently translocated across the outer membrane of these pathogens with the Type II secretion machinery (75-78). This machinery spans two membranes and is composed of about 14 proteins forming an assembly of over 1.5 MDa. We have solved the structure of V. cholerae EpsE, the first Type II secretion NTPase of known structure (79). This resulted in a model for the presumed hexameric arrangement of EpsE near the surface of the inner membrane. In addition, the crystal structure of the periplasmic domain of EpsM, the first structure of any periplasmic domain in Type II secretion apparati, showed a remarkably simple abb-abb; topology which had not yet been seen (80) . Recently, we solved the structure of the cytoplasmic domain of EpsL, which interacts with EpsE. Expression attempts of all components C to M of the type II secretion pathway proteins are under way. Several initial genes were provided by our collaborators Bagdasarian (Michigan State) and Sandkvist (Red Cross, DC).
Figures I 2.1
2.2 The RNA editing machinery of Trypanosoma brucei
Trypanosomatids have an unbelievably complex mitochondrial DNA, consisting of ~50 maxicircles and hundreds of concatenated minicircles (81). About 12 mitochondrially encoded proteins require extensive editing of their premature mRNA by a complex of about 20 nuclear encoded, but mitochondrially located, proteins, called the "editosome" (82-86). This machinery is crucial for the survival of the blood stream form of the parasite (87). In collaboration with Ken Stuart (SBRI), we have very recently succeeded in solving structures of the catalytic domain of the key RNA ligase of T. brucei (TbRel1) at 1.2 Å resolution in complex with ATP. This structure shows a deep adenosine binding pocket which is very promising for the design of inhibitors of this critical ligase (88).
Figures I 2.2
3. Structural Genomics of Pathogenic Protozoa (SGPP)
The Structural Genomics of Pathogenic Protozoa (SGPP) consortium was initiated, in order to explore protozoan genome sequences of which are just becoming available. The SGPP consortium (www.sgpp.org) consists of 14 investigators in six institutions engaged in solving crystal structures of proteins from Plasmodium falciparum, P. vivax, Trypanosoma brucei, T. cruzi, and several Leishmania species. Target selection is aimed to optimize the medical relevance of the proteins under investigation.
The initial years of SGPP have been characterized by developing and testing new methods, including:
In one of the SGPP malaria protein structures solved very recently, the asymmetric unit appeared to contain 12 subunits / 5 dimers plus two half dimers.
4. Other structural studies
4.1. Cephalosporin acylase
Cephalosporin acylase deacylates substrates like glutaryl-7 amino cephalosporanic acid (89). The structure determination of the enzyme and its precursor (90-92) assisted in obtaining a profound understanding of this class of enzymes of great biotechnological interest.
4.2 GAF domain of Phosphodiesterase.
Phosphodiesterases play a major role in cell signaling keeping the concentrations of the internal messenger camp at the appropriate levels. These complex enzymes are regulated by cGMP-binding GAF domains. In collaboration with Joe Beavo (Pharmacology, UW) the structures of the GAF-A and GAF-B domains were elucidated. The GAF-B domain:cGMP complex was the first view of cGMP bound to a signaling protein (93,94). The structure corrected the predicted cGMP binding mode based on the structure of a distant homolog.
Figures I 4.2
4.3 Heat-shock protein Hsp31 from E. coli.
Heat-shock protein Hsp31 is a member of a large protein family (95,96) and a distant relative of human DJ-1, a protein involved in Parkinson's disease (97-101). Our structure determination of E. coli Hsp31 led to a proposal of its mode of action wherein two flexible loops expose hydrophobic patches in a termperature-dependent manner (102,103). Support for this proposal has been obtained by mutational and functional studies by our collaborator Francois Banyex (Bioengineering, UW) (104).
Figures I 4.3
See References for Research Summary Part I
RESEARCH SUMMARY PART II mid 1994 to mid 1999
1. Glycosomes in Trypanosomatids: Functioning and Inhibition
Several Trypanosomatid species are the cause of major tropical diseases such as (i) African trypanosomiasis, or sleeping sickness, due to Trypanosoma brucei infections, (ii) American trypanosomiasis or Chagas' disease, caused by T. cruzi, and (iii) many forms of leishmaniasis, caused by eleven different Leishmania species throughout the tropics (20-23). Opperdoes and Borst (24) discovered in trypanosomatids a unique organelle, the glycosome, which contains a large number of glycolytic enzymes. Most glycosomal matrix proteins contain glycosomal target signals (called PTS-1 and PTS-2) which ensures proper import into the organelle.
The bloodstream form of T. brucei is entirely dependent on glycolysis for its energy supply (25) and hence glycolytic enzymes are an interesting target for drug design (25,26). Recent metabolic flux control simulations of glycolysis in T. brucei (27) have shown that the best enzymes to target are glyceraldehyde-3-phosphate dehydrogenase (GAPDH), phosphoglycerate kinase (PGK), glycerol-3-phosphate dehydrogenase (GPDH) and fructose 1,6-bisphosphate aldolase. We are studying these four enzymes in close collaboration with the groups of Drs. Paul Michels and Fred Opperdoes in Brussels (parasitology, expression systems), and Drs. Michael Gelb (chemistry), Christophe Verlinde (modeling), Wes van Voorhis and Fred Buckner (in vivo testing) at the University of Washington. In several instances we are also investigating these enzymes from Leishmania mexicana.
The de novo structure of L. mexicana glycerol-3-phosphate dehydrogenase (GPDH) has been solved recently in our group by MAD phasing methods. This 78 kDa dimeric protein (28) appears to have an unexpected evolutionary relationship with plant acetohydroxy acid isomeroreductase (29). Not only the N-terminal NAD-binding domain, but also parts of the catalytic domain appear to be structurally equivalent. Another surprise was the presence of a completely buried fatty acid-like density approaching a Cys residue at one end but without making a covalent link with this residue. The physiological relevance of this discovery still needs to be established. The NAD binding site of GPDH is very different from that in the homologous human enzyme, in accordance with the limited 30% sequence identity of the two enzymes. This bodes well for the synthesis of selective GPDH inhibitors which has recently been initiated by the Gelb group.
The crystal structures of both L. mexicana and T. brucei aldolase, 160 kDa tetrameric proteins (30), have recently been determined in my group (Chudzik et al., in preparation). Both aldolases show a completely unexpected dimeric "handshake" structure for the PTS-2 N-terminal glycosomal import signal (31-33) with potentially important implications for the import mechanism. This is the first time that the conformation of a PTS-2 signal, a nonapeptide, has been visible in a protein structure.
The crystal structure of a ternary complex of T. brucei PGK with two substrates (34) was solved in a fully closed state, the first time any PGK has been observed in this conformation (35). The ternary complex revealed many aspects of the phosphoryl transfer mechanism (34,36). Subsequent structures with a bisubstrate analogue (37) revealed a 2 Å shift of one of the helices in the "constant" C-terminal domain upon bisubstrate analogue binding. The structure of PGK in complex with 2-amino-6-(p-hydroxy phenethylamino) adenosine showed that the adenine ring had "flipped" 180° in the adenosine derivative compared to ADP. In particular the latter discovery opens avenues for arriving at new inhibitors with high affinity and selectivity for the adenosine binding pocket of T. brucei PGK.
The structure of L. mexicana GAPDH was solved to 2.8 Å resolution (38,39) and this formed the basis for dramatic improvements in the affinity of adenosine derivatives for this parasite enzyme. In close collaboration with the groups of Drs. Verlinde and Gelb, a N6-(1-naphthalenemethyl)-2’-(3-methoxybenzamido) adenosine derivative was obtained with an IC50 for Leishmania GAPDH of 280 nM, an improvement of over five orders of magnitude compared to the starting compound adenosine (40,41). These compounds appear to have an ED50 of 10-30 µM for T. brucei and T. cruzi while leaving fibroblasts essentially unaffected at these concentrations (41).
2. The cholera toxin family: Action and Inhibition
Cholera toxin (CT) and E. coli heat labile enterotoxin (LT) are two very similar heterohexameric AB5 molecules with the B pentamer responsible for GM1 receptor recognition on the surface of host epithelial cells and the A subunit for ADP-ribosylating a catalytically critical arginine of GSa (42-45). The covalently modified GSa is no longer able to hydrolyze bound GTP and continuously stimulates adenylyl cyclase leading to excess levels of cAMP, and export of salts and fluids. The WHO estimates that approximately 400,000 children die annually from infections of enterotoxigenic E. coli (46,47).
The structure determination of an inactive LTA (R7K):LT B5 mutant (provided by Dr. Rino Rappuoli, Siena) showed that the loop comprising residues 47-56 in the active site had become completely flexible due, most likely, to loss of interactions between this loop and the newly introduced lysine residue (48). This observation led to a proposal (48)for the mechanism of A subunit activation by the reduction of the Cys 187-Cys 199 disulfide bridge and proteolytic cleavage of the 192-195 loop (49,50). The reduction and cleavage events, which take place ~ 20 Å from the active site, are proposed to result in a displacement of the 47-56 loop in wild type toxin similar to that seen in the R7K mutant in order to provide space for substrates to bind to the active site (48).
The extensive structural information on LT and CT (51-57) has been used for the discovery and design of inhibitors of the assembly and receptor recognition processes, in collaboration with the groups of Drs. Verlinde (modeling) and Fan (synthesis) at the University of Washington. For the development of assembly antagonists we focused on a ring of hydrophobic residues observed (58) at the point of entry of the A subunit into the pore of the B pentamer. This ring has preserved its hydrophobic nature in B pentamers of several different toxins. A pharmacophore model was constructed, in collaboration with the Verlinde group, on the basis of two phenyl rings from the A-subunit interacting with the hydrophobic ring of the B pentamer pore. A search through the Available Chemical Database led to several possible pore binding candidates, of which 3-methylthio-1,4-diphenyl-1H-1,3,4-triazolium (MDT) was co-crystallized with the B pentamer of LT. MDT was indeed observed to be present in the targeted area (59).
The crystal structure of the ~60 kDa complex of the CT B pentamer with five copies of the pentasaccharide head group of GM1 was determined at a very high resolution of 1.25 Å (60). This provided profound insight into saccharide conformation, water molecule positions, and anisotropic displacement parameters, and revealed an unusually strongly bent peptide unit. This peptide is located quite close to the saccharide binding site and, consequently, might be the result of conformational changes due to receptor binding.
The receptor binding site has been subject of a series of investigations where evaluation of a first generation of commercially available galactose derivatives resulted in m-nitrophenyl galactoside (MNPG) which bound ~100 times better than the starting compound galactose (61). Interestingly, the nitro group of MNPG displaced a water molecule (named W2) which was observed to be unperturbed in all other B pentamer-galactose derivative complexes elucidated (56,62,63). This observation was followed up by computational studies (64), and in the Fan group, by the synthesis of MCPG, the monocarboxylate homologue of MNPG. Interestingly, MCPG binds about 10 times less well to the receptor site than MNPG and is also unable to displace water W2 from its binding site according to a very recent 1.3 Å crystal structure (Minke et al., in preparation). These studies provide deep insight into charge and solvation aspects of protein ligand interactions.
Most recently, structure-guided combinatorial chemistry approaches have resulted in galactose variants with 103 fold improved affinity compared to galactose (65). Moreover, incorporation of galactose into molecules with five fold symmetry increased affinity by more than four orders of magnitude (Fan et al., in preparation).
3. Eukaryotic Topoisomerases I
Topoisomerases are involved in critical processes in the cell such as DNA relaxation, DNA recombination and chromatin remodeling (66-69). Among several classes of topoisomerases, the eukaryotic, or type IB, topoisomerases I are unique in that they are monomeric, require no ATP or Mg2+ for DNA relaxation, and can relax both overwound and underwound DNA. In collaboration with Dr. James Champoux (University of Washington) we have elucidated three crystal structures of human topoisomerase I in complex with DNA. Two of the topo:DNA complexes contained intact duplex DNA due to the fact that the critical Tyr 723, which forms a temporary bond with the 3' OH of the scissile strand, had been mutated into a Phe. Using a reconstituted, active wild type enzyme consisting of a "core" and the C-terminal domain, along with a phosphorothioate linkage at the cleavage site, a covalent topoisomerase:DNA complex structure was also obtained. On the basis of these first topoisomerase:DNA complexes a catalytic mechanism as well as an isomerization mechanism, called "controlled rotation", has been proposed (70-72).
Human topoisomerase I is the sole target of the "topo poison" camptothecin, and derivatives thereof. Some of the latter, such as topotecan and irinotecan (73-75), have been recently approved as anticancer drugs. These compounds act by blocking the religation step of topoisomerase. We have proposed a model for the binding mode of camptothecin (71) while a quite different mode of action has been put forward by Pommier and colleagues (76). Attempts to obtain crystallographic evidence for the binding mode of this inhibitor are currently underway.
4. Mycobacterial proteins
We are interested in potential drug targets from M. tuberculosis since this organism is the single most important infectious agent today. Approximately one third of the world population is currently infected, 6.7 million new cases appear annually, about 2.4 million deaths occur per year from tuberculosis (77), and the prospects for controlling the disease are grim (78-80). The crystal structures of the following four mycobacterial proteins have been solved in my group in the last five years:
·Chaperonin-10 of Mycobacterium leprae, a close relative of M. tuberculosis and the causative agent of leprosy, with protein kindly supplied by Dr. Barry Bloom, Albert Einstein, NY. The seven 10kDa subunits of this protein form a dome with at the top an oculus with a diameter of ~10 Å (81). The M. leprae chaperonin 10 heptamer shares many features with the structure of E. coli GroES solved by Hunt et al. (82). The inner surface of the dome appeared to be highly hydrophylic and loop 17 to 35 is very flexible. The role of GroES/chaperonin 10 has been the subject of elegant studies of many investigators, including the crystal structure of the GroES:GroEL complex by the Sigler-Horwich groups (83).
· The iron-dependent regulator (IdeR) of M. tuberculosis (84) in collaboration with the group of Dr. R.K. Holmes, Denver, who provided protein and an expression system (85). The structure of a close relative of IdeR, DtxR from Corynebacterium diphtheriae, was also solved in collaboration with the Holmes group, in complex with a variety of divalent transition state metals (86-88) as well as in the apo form (89). These structures revealed a hinge motion of the DNA-binding domain with respect to the dimerization domain, and a SH3-like third domain. Structures of DtxR have also been reported by other groups (90-92) The information obtained by the structure of a DNA:mutant DtxR complex by White et al. (93) was confirmed and extended by our recent DNA:wt DtxR structure which revealed in addition a metal-binding property of the SH3-like third domain (94).
· Dihydrofolate reductase (DHFR) of M. tuberculosis, with protein provided by Dr. W. Sirawaraporn, Bangkok. DHFRs from human and pathogens are the targets of antineoplastic, antimicrobial and antiprotozoan drugs (95-97). This new DHFR structure, solved in binary complex with NADPH and in three ternary complexes with NADPH and inhibitors, suggests several directions by which these inhibitors might be modified such that they gain affinity for the M. tuberculosis enzyme while losing affinity for the enzyme from humans (98).
· Dihydropteroate synthase (DHPS) from M. tuberculosis, also with protein provided by Dr. Sirawaraporn. DHPS, the target of sulfa drugs (99-103), is absent in the human host, and the M. tuberculosis DHPS structure is likely to provide clues for the development of high affinity inhibitors once the structure is further refined.
5. Architecture of the Pyruvate Dehydrogenase Multienzyme Complex (PDC)
The PDC is the prototype of a number of extraordinarily large multienzyme complexes including the PDC itself, the a-ketoglutarate dehydrogenase complex and the branched-chain a-ketoacid dehydrogenase complex. The total molecular weight of the complexes ranges from 4 to 10 million daltons (1-5). In mammalian cells these three complexes occur in the mitochondrion. The complexes all contain multiple copies of three enzymes: (i) E1: the a-keto acid dehydrogenase, a thiamine diphosphate dependent enzyme; (ii) E2: the dihydrolipoyl acyl transferase, a dynamic multidomain multifunctional enzyme forming the highly symmetric (cubic or dodecahedral) core of the complexes, providing binding domains for E1 and E3, and containing lipoyl domains which swirl around visiting three active sites in the complex; (iii) E3: dihydrolipoamide dehydrogenase, a dimeric flavoenzyme. Prior to joining HHMI, my group made major contributions to the understanding of the architecture of these complexes by solving several E3 structures (6-9) and the hollow cubic E2 core of the pyruvate dehydrogenase complex (10-12).
More recently, the complex of dihydrolipoamide dehydrogenase (E3) from Bacillus stearothermophilus with the binding domain of E2 was unraveled (13), explaining why only one small 40-residue binding domain interacts with the 100 kDa E3 dimer (14). Subsequently, the structure of the 1.5 million Dalton 60-meric dodecahedral E2 core from B. stearothermophilus and Enterococcus faecalis was determined at medium resolution (15). It appeared possible (15) to arrive at a detailed understanding of the relationship between the cubic and dodecahedral cores, based on Euclidean geometry (16) and Caspar & Klug's principle of quasi-equivalence (17).
Very recently, the 170 kDa heterotetrameric E1b structures of Pseudomonas putida (18) and of Homo sapiens (19) were elucidated. These structures reveal an intimate association of two a and two b subunits, a long channel at a subunit interface leading to the thiamine diphosphate cofactor, the position of two crucial K+ ions, and the mode of inactivation by phosphorylation of the primary phosphorylation site. These structures also led to global proposals for the binding modes of the E2 lipoyl domain and the E2 binding domain. Moreover, the severity of the symptoms of maple syrup urine disease caused by some 20 amino acid mutations could be correlated with their effects on function and stability of the E1b structure.
As a result of these studies my group has produced the first views at the atomic level of E1, E3, and the major part of E2 from the PDC family. The following groups provided protein for these investigations: Drs. Aart de Kok, Univ. of Wageningen, The Netherlands; Richard Perham, Cambridge, UK; John Sokatch, Univ. of Oklahoma; David Chuang, Southwestern Medical Center at Dallas.
6. Contributions to Methodological Advances in Protein Crystallography
Methodologically oriented results include:
1. A computer program Difference Density Quality (DDQ) which evaluates the quality of protein structures by analyzing the physical chemical environment of water molecules included in the coordinate set, as well as the errors in difference electron density maps (104).
2. A flash annealing protocol for cryo-crystallography (105).
3. Monte Carlo and Data Base methods for rapid positioning of large fragments of known protein structures into electron density maps of new structures (106,107).
4. The development of a so-called "RIG plasmid" encoding tRNAs for three rarely used E. coli codons which, in contrast, are frequently used in the AT-rich genome of Plasmodium species (Baca & Hol, in press).
5. A computer program which simulates the hanging drop crystallization experiment by numerical analysis, leading to excellent agreement with experimental results (108).
6. The demonstration that two-fold density averaging can result in full recovery of a significant portion of a protein not included in the initial molecular replacement search model (109).
7. The demonstration that fragments containing as little as 8.6% of the scattering mass of the asymmetric unit can be oriented and positioned successfully by molecular replacement methods (110).
University of Washington Academic Departments:
Department of Biochemistry
Department of Biological Structure
Department of Pharmacology
University of Washington Interdisciplinary Graduate Programs
Biological Physics, Structure & Design
Biomolecular Structure Center