Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Crystallographic studies of bacterial sialyltransferases Chiu, Cecilia P. C. 2007

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Notice for Google Chrome users:
If you are having trouble viewing or searching the PDF with Google Chrome, please download it here instead.

Item Metadata


831-ubc_2007-317518.pdf [ 19.12MB ]
JSON: 831-1.0100622.json
JSON-LD: 831-1.0100622-ld.json
RDF/XML (Pretty): 831-1.0100622-rdf.xml
RDF/JSON: 831-1.0100622-rdf.json
Turtle: 831-1.0100622-turtle.txt
N-Triples: 831-1.0100622-rdf-ntriples.txt
Original Record: 831-1.0100622-source.json
Full Text

Full Text

C R Y S T A L L O G R A P H I C STUDIES O F B A C T E R I A L S I A L Y L T R A N S F E R A S E S by C E C I L I A P. C. C H I U B . S c , The University of British Columbia, 2001 A THESIS S U B M I T T E D I N P A R T I A L F U L F I L L M E N T O F T H E R E Q U I R E M E N T S F O R T H E D E G R E E OF D O C T O R OF P H I L O S O P H Y in T H E F A C U L T Y OF G R A D U A T E S T U D I E S (Biochemistry and Molecular Biology) T H E U N I V E R S I T Y O F B R I T I S H C O L U M B I A M a y 2007 © Cecil ia P. C. Chiu, 2007 11 A B S T R A C T Sialic acids terminate oligosaccharide chains on the surfaces of mammalian cells and many microbial species, often playing critical biological roles in recognition and adherence. The enzymes which transfer the sialic acid moiety to these key terminal positions are known as sialyltransferases. Despite their important biological roles very little is understood about the mechanism of action or molecular structure of these enzymes. Campylobacter jejuni, a highly prevalent food-borne pathogen that causes acute gastroenteritis in humans, contains two versions of a sialyltransferase: a monofunctional a-2,3-sialyltransferase Cst-I and a bifunctional a-2,3/8-sialyltransferase Cst-II. Both of these enzymes are responsible for lipooligosaccharides (LOS) sialylation to camouflage the bacterial surface from the host, and thus evade the immune system. In addition, sialylated-glycoconjugates on C. jejuni often mimic human gangliosides, contributing to the molecular basis of Guillain-Barre syndrome, an autoimmune disease of the peripheral nervous system that often develops post-infection. The sialyltransferase reaction is believed to proceed through an inversion mechanism, catalyzing the transfer of sialic acid from CMP-A^-acetylneuraminic acid onto different acceptors. This thesis is to understand through high-resolution structural characterization, site specific mutagenesis and kinetic analysis, the mechanism of the glycosyl transfer(s) in both monofunctional and bifunctional Csts. Crystals of Cst-II were obtained and the complex structures with bound CMP, inert donor sugar analogue CMP-3-fluoro-7V-acetylneuraminic acid (CMP-3FNeu5Ac) and . inhibitor CDP were solved using M A D phasing from incorporated selenomethionines. Work within this study represents the first known structure of a sialyltransferase. Based on the position of I l l the substrates, the active site of Cst-II has been elucidated. Site-directed mutagenesis of conserved residues in the active site was performed and mutants were characterized using enzyme kinetics. A reaction mechanism was proposed based on the kinetic assay. A directed evolution methodology was designed for glycosyltransferases using Cst-II as the model system. A single mutation, F91Y was found to substantially increase the reaction rate of the enzyme with a fluorescent-coupled acceptor, bodipy-lactose. The crystal structure of this Cst-II F91Y mutant was solved and it revealed an unexpected flip of the tyrosine side chain of Y91 from the core of the enzyme into the solvent region, exposing a hydrophobic pocket which seems to be capable of accommodating the bodipy ring structure. Together with kinetic analyses, the crystallographic study was able to explain the observed increase in the catalytic rate for this novel sugar acceptor. A monofunctional variant of Cst, Cst-I, also isolated from Campylobacter jejuni, was characterized crystallographically and kinetically. The conservation of active site residues supports the proposed mechanism for GT-42 sialyltransferases. The complex structure of the Cst-I enzyme with the donor analogue CMP-3FNeu5Ac provides a platform for molecular modeling of various acceptors into the active sites of Cst-I and Cst-II. The modeling shed lights upon the understanding of differences in substrate specificity. The structures of these complexes will be used as templates to design therapeutic inhibitors against this common human pathogen. IV T A B L E OF C O N T E N T S ABSTRACT , ii TABLE OF CONTENTS iv LIST OF TABLES vii LIST OF FIGURES viii ABBREVIATIONS x ACKNOWLEDGEMENTS xiii CO-AUTHORSHIP STATEMENT . xv CHAPTER 1 - INTRODUCTION 1 1.1 C A R B O H Y D R A T E S A N D G L Y C O B I O L O G Y 1 1.2 S L A L Y L A T E D G L Y C O C O N J U G A T E S : 3 1.2.1 Sialylated glycoconjugates and bacterial pathogenicity 7 1.3 G L Y C O S Y L T R A N S F E R A S E S 11 1.3.1 Structure of glycosyltransferases 12 1.3.2 Reaction mechanism 14 1.4 SlALYLTRANSFERASES 1 7 1.4.1 Sialyltransferase from Campylobacter jejuni 2 0 1.5 O B J E C T I V E S OF THESIS 2 4 CHAPTER 2 - STRUCTURAL AND BIOCHEMICAL CHARACTERIZATION OF BIFUNCTIONAL SIALYLTRANSFERASE CST-II FROM CAMPYLOBACTER JEJUNI. 25 2.1 INTRODUCTION 2 5 2.2 METHODS . . . . . 2 7 2.2.1 Cloning, protein expression, and purification 2 7 . 2.2.6 Static light scattering 28 2.2.2 Crystallization and data collection 2 8 2.2.4 Site-directed mutagenesis via PCR 3 0 2.2.5 Kinetic assays... 3 0 V 2.2.6 Overexpression and purification of 13C-histidine labeled samples 31 2.2.7 pKa titration and 1 3 C - N M R data collection ., 32 2.3 R E S U L T S , 33 2.3.1 Oligomerization and membrane attachment 33 2.3.2 Overall architecture 37 2.3.3 C M P binding 39 2.3.4 CMP-3Fneu5Ac binding 41 2.3.5 CDP binding 47 2.3.6 Kinetic analysis of active site mutants 49 2.3.7 Histidine titration ; 51 2.4 D I S C U S S I O N '. 54 CHAPTER 3 - STRUCTURAL INVESTIGATION OF CST-II MUTANT FROM DIRECTED EVOLUTION.... 62 3.1 I N T R O D U C T I O N 62 3.2 M E T H O D S 65 3.2.1 D N A manipulations 65 3.2.2 Screening Cst-II libraries by FACS 65 3.2.3 Screening, purification and kinetic analysis of Cst-II F91Y mutant 67 3.2.4 Crystallization, Data Collection and Structure Determination 67 3.3 R E S U L T S 69 3.3.1 Detection and sorting of ST activity in intact E. coli cells 69 3.3.2 Selection of Cst-II library for increase in sialyltransferase activity 74 3.3.3 Kinetic analysis of Cst-II F91Y mutant 76 3.3.4 Crystallographic analysis of Cst-II F91Y mutant 78 3.4 D I S C U S S I O N 84 CHAPTER 4 - STRUCTURAL AND BIOCHEMICAL CHARACTERIZATION OF MONOFUNCTIONAL CST-I FROM CAMPYLOBACTER JEJUNI „ 88 4.1 I N T R O D U C T I O N :.. 88 4.2 M E T H O D S .' ; 90 4.2.1 Cloning and protein expression 90 4.2.2 Purification of the Cst-I 1" 2 8 5 product and kinetic assay 90 vi 4.2.3 Crystallization and data collection 91 4.2.4 Structure determination, refinement and modeling 92 4.3PvESULTS 94 4.3.1 Sequence and Kinetic analysis of Cst-I .• 94 4.3.2 Determination of the Cst-I 1" 2 8 5 structure 95 4.3.3 Comparison of structures of the apo- and substrate-bound monofunctional Cst-I 1" 2 8 5 . 99 4.3.4 Catalytic residues in Cst-I 102 4.3.5 Docking of acceptor sugars to the active site of Cst-I 1" 2 8 5 103 4.3.6 Docking of acceptors to bifunctional Cst-II and comparison to Cst-I 106 4.4 D I S C U S S I O N 112 CHAPTER 5 - CONCLUSIONS AND FUTURE DIRECTIONS 115 5.1 S U M M A R Y A N D S I G N I F I C A N C E O F R E S U L T S 115 5.2 F U T U R E D I R E C T I O N S 119 5.2.1 Monofunctional Cst-II from C. jejuni 123 5.2.2 Nst from /V. meningitidis 124 5.2.3 ST3Gal I and ST3Gal HI from Homo sapiens 125 REFERENCES 127 APPENDIX - PUBLICATIONS ARISING FROM GRADUATE WORK 145 vii LIST OF T A B L E S Table 1.1 Bacterial species and their molecular mimicking molecules found on their capsule or LOS 9 Table 1.2 Sialyltransferase families 18 Table 2.1 Data collection and structure refinement statistics '. 35 Table 2.2 Michaelis-Menten parameters for CMP-Neu5Ac with various Cst-II 1" 2 5 9 mutants :-. :'. 51 Table 2.3 Michaelis-Menten parameters for acceptors with wildtype Cst-II 1" 2 5 9 54 Table 3.1 The evolved Cst-II F91Y mutant: catalytic efficiency for the transfer of CMP -Neu5Ac to different acceptors 78 Table 3.2 Data collection and structure refinement statistics 80 Table 4.1 Data collection and structure refinement statistics 97 Table 5.1 Attempts for obtaining ternary complex structure for Csts 119 Table 5.2 Summary for work in progress for characterization of various members of CAZy sialyltransferase families 126 LIST OF FIGURES Figure 1.1 7Y-acetylneuraminic acid, sialic acid 3 Figure 1.2 Schematic of cell surface sialylated glycoconjugates 4 Figure 1.3 Main gangliosides of human brain 6 Figure 1.4 Schematic for reaction catalyzed by glycosyltransferases 11 Figure 1.5 Ribbon diagrams of representative GT folds 14 Figure 1.6 Glycosyltransferase mechanisms 16 Figure 1.7 a and P linkage at the anomeric center 16 Figure 1.8 Universal donor CMP-Af-acetylneuraminic acid for sialyltransferases 17 Figure 1.9 Sequence alignment of representative CAZy GT-42 sialyltransferases 22 Figure 1.10 Reaction scheme of Cst 23 Figure 2.1 SDS-PAGE and static light scattering analysis of Cst-II 1" 2 5 9 33 Figure 2.2 Arrangement of the Cst-II 1" 2 5 9 tetramer 36 Figure 2.3 Schematic membrane attachment of Cst-II 1" 2 5 9 37 Figure 2.4 View of Cst-II 1" 2 5 9 monomer showing the N-terminal domain and the lid domain with bound donor sugar analogue '. 38 Figure 2.5 Interactions of CMP and active site residues : 41 Figure 2.6 The donor substrate analogue CMP-3FNeu5Ac 43 Figure 2.7 Interactions of CDP and active site residues '. 48 Figure 2.8 Coupled kinetic assay system developed for Cst-II 50 Figure 2.9 2D HSQC spectra of [ring 2-13C]histidine-labeled Cst-II 1" 2 5 9 53 Figure 2.10 Proposed reaction mechanism for Cst-II 58 Figure 2.11 Mapping of the amino acids that differ between mono- and bifunctional Cst-II in the active site of bifunctional Cst-II 61 Figure 3.1 Cell based assay for sialyltransferase activity 70 Figure 3.2 Fluorescent acceptor sugars used in this study 71 Figure 3.3 Cell based sialyltransferase assay for cells expressing wildtype Cst-II 1" 2 5 9 and containing the pUC18 plasmid 73 Figure 3.4 Library selection through three iterative rounds of sorting by FACS 75 Figure 3.5 Location of Phe91 residue in the structure of wildtype Cst-II 79 Figure 3.6 Structure of the F91Y Cst-II mutant 81 ix Figure 3.7 Surface representation of the active site cleft and the Tyr91 region in the F91Y Cst-II mutant 83 Figure 4.1 Sequence alignment of bifunctional Cst-II1"2 5 9, monofunctional Cst-II 1" 2 5 9, and monofunctional Cst-I1"285'.' , 98 Figure 4.2 Structural alignment of the three Campylobacter sialyltransferases 99 Figure 4.3 Interactions of CMP-3FNeu5 Ac with active site residues of Cst-I" 101 Figure 4.4 Molecular modeling of Gal-6-l,3-GalNAc into the active site of Cst-I 105 Figure 4.5 Molecular modeling of various acceptors into the active site of Cst-II 109 Figure 4.6 Overlap of sialyl lactose from molecular modeling for Cst-II into the active site of Cst-I I l l Figure 5.1 Structural comparison of GT-42 and GT-80 123 A B B R E V I A T I O N S 2D two-dimensional 3D three-dimensional A B C ATP-binding cassette A 3 4 0 . absorbance at 340 nm ADP adenosine 5'-diphosphate ADSC Area Detection Systems Corporation A L S Advanced Light Source A M P adenosine 5'-monophosphate ATP adenosine 5'-triphosphate B S A bovine serum albumin CAZy carbohydrate active enzymes C C D charge-coupled device CDP cytidine 5'-diphosphate C M P . cytidine 5'-monophosphate CNS Crystallography and N M R System CTP cytidine 5'-triphosphate D A N A 2,3-dehydro-2-deoxy-A^-acetylneuraminic acid D M density modification D N A deoxyribonucleic acid DTT dithiothreitol E D T A ethylenediaminetetraacetic acid FACS fluorescence-activated cell sorting FITC fluorescein isothiocyanate Gal galactose GalNAc A-acetyl-galactosamine GBS Guillain-Barre syndrome GDP guanosine 5'-diphosphate G H glycosidase Glc glucose GlcNAc Af-acetyl-glucosamine GT glycosyltransferase HSQC heteronuclear single quantum coherence HEPES 4-(2-hydroxyethyl)-l-piperazineethane sulfonic acid HTS high throughput screening IPTG isopropyl-p-D-thiogalactopyranoside K D N 2-keto-3-deoxy-D-glyero-D-galacto-nononic acid KIE kinetic isotope effect Lac lactose LacNAc Af-acetyl-lactosamine L B Luria Bertoni media L D H lactose dehydrogenase LOS lipooligosaccharides MFS Miller-Fisher syndrome M M E monomethyl ether MPD 2-methyl-2,4-pentanediol mRNA messenger ribonucleic acid N A D + nicotinamide adenine dinucleotide (oxidized form) N A D H nicotinamide adenine dinucleotide (reduced form) N C A M neural cell adhesion molecule NDP nucleoside diphosphate N M P K myokinase N M R nuclear magnetic resonance NSLS National Synchrotron Light Source NTP nucleotide triphosphate O D 6 0 0 optical density at 600 nm PBS phosphate buffered saline PCR polymerase chain reaction PDB Protein Databank PEG polyethylene glycol PEP phosphoenolpyruvate P K pyruvate kinase P S A poly-a-2,8-sialic acid r.m.s. root mean squared R T reverse transcriptase S A D single anomalous diffraction SeMet selenomethionine S D S - P A G E sodium dodecyl sulfate polyacrylamide gel electrophoresis T E A Triethanolamine T L C thin layer chromatography Tris 2-amino-2-(hydroxymethyl)-13-propanediol U D P Uridine 5'-diphosphate U V ultraviolet vdW van der Waals W T wildtype Y T yeast extract tryptone X l l l A C K N O W L E D G E M E N T S First of all, I would like to thank my parents, my brother, and my fiance for their unconditional support over the years. I am grateful that they always believe in me and give me the freedom to pursue my interests. I consider myself extremely fortunate to have Dr. Natalie Strynadka as my supervisor. She has given me tremendous amount of support over the years. I would like to thank her for giving me this interesting and challenging project, her invaluable direction in my research and manuscript writing, and her faith in my abilities. I would also like to thank my supervisory committee members, Drs. Lawrence Mcintosh, Michael Murphy, and Stephen Withers, who provided me valuable insights and guidance for my work. This work is greatly enhanced by excellent collaborations with Dr. Stephen Withers in the Chemistry department of U B C , and Dr. Warren Wakarchuk at National Research Council in Ottawa. Specifically, I would like to thank Luke Lairson, Drs. Amir Aharoni and Michel Gilbert for their remarkable collaborations on different aspect of this project. In addition, I would like to thank Dr. Mark Okon from Dr. Lawrence Mcintosh laboratory for his assistance in N M R data collection. I would like to extend my appreciation for the assistance I obtained from all members of the Strynadka laboratory, both past and present. Special thanks to Liza deCastro, for maintaining an enjoyable working environment, for her ongoing encouragement and stress-releasing conversations. I want to thank Drs. Daniel Lim, Y u Luo, Michela Bertero, Andrew Lovering, and Igor d'Angelo for teaching me protein crystallography, Richard Pfuetzner, Tanya Hills, Drs. Lodovica Loschi, Haizhong Zhu, Calvin Yip and Trevor Moraes for discussions on protein purification and molecular xiv biology techniques. Thanks to all the coffee buddies who went for coffee breaks with me, rain or shine. I am extremely grateful for the financial support from the Canadian Institutes of Health Research, and the Michael Smith Foundation for Health Research throughout my graduate studies. I would like to extend my appreciation to Drs. Michael Murphy and Stephen Withers for their time and effort to supplement my award application with their reference letters. X V C O - A U T H O R S H I P S T A T E M E N T Chapters 2 and 4 of this thesis describe work that was previously published. Manuscripts for these chapters were written by myself and revised by my supervisor Dr. Natalie Strynadka. Chapter 2 contains portions from a manuscript published in Nature Structural and Molecular Biology [Chiu, C.P.C., Watts, A .G . , Lairson, L .L . , Gilbert, M . , Lim, D., Wakarchuk, W.W., Withers, S.G:, and Strynadka, N . C J . Structural analysis of the sialyltransferase Cst-II from Campylobacter jejuni in complex with a substrate analogue. Nature Structural and Molecular Biology. 2004 11:163-70.] Dr. Warren Wakarchuk provided the initial full length construct of Cst-II. I designed a truncation construct and performed molecular cloning into the expression vector. I developed the protein purification protocol. Crystallization, data collection, structure determination, and refinement were all performed by myself. Site directed mutagenesis for active site mutants were performed by Dr. Warren Wakarchuk and the protein samples were prepared by myself. Kinetic analysis of active site mutants were performed by Luke Lairson and rescue assay of H188A mutant was performed together by Luke Lairson and 1 ^ I. Preparation of C-histidine labeled samples was performed by myself, and data collection was assisted by Dr. Mark Okon in Dr. Lawrence Mcintosh laboratory. A l l text and figures in this article were prepared by myself. Chapter 3 contains portions from a manuscript published in Nature Methods [Aharoni, A. , Thieme, K. , Chiu, C.P.C., Buchini, S., Lairson, L .L . , Chen, H. , Strynadka, N.C.J. , Wakarchuk, W.W., and Withers, S.G. High throughput screening of glycosyltransferase libraries: Evolution of a sialyltransferase with a novel substrate xvi specificity. Nature Methods, 2006 3:609-14.] The directed evolution, high-throughput screening method and identification of F91Y mutant were performed by Dr. Amir Aharoni. Subcloning of Cst-II F91Y mutant into the appropriate expression plasmid, purification and crystallization of the enzyme, data collection of X-ray diffraction pattern, structure solution and refinement, as well as structural analysis/interpretation were all performed by myself. This purified enzyme sample was used by Luke Lairson to obtain kinetic parameters. Most of the text and figures were prepared by Dr. Amir Aharoni, whilst I prepared the text and figures for the crystallographic portion of the manuscript. Chapter 4 contains portions from a manuscript submitted to Biochemistry [Chiu, C.P.C., Lairson, L .L . , Gilbert, M . , Wakarchuk, W.W., Withers, S.G., and Strynadka, N.C.J. Structural analysis of the a-2,3-sialyltransferase Cst-I from Campylobacter jejuni in both apo and substrate-analogue bound forms. Biochemistry. 2007.] Dr. Warren Wakarchuk provided the plasmid of the truncated Cst-I with MalE-tagged in expression vector and the purification protocol. Protein crystallization, data collection, structure solution and refinement, and molecular modeling were all performed by myself. Kinetic analyses were performed by Luke Lairson. A l l text and figures in this article were prepared by myself. 1 C H A P T E R 1 - Introduction 1.1 Carbohydrates and glycobiology Carbohydrates are fundamental to life and are the most abundant biological molecules on Earth. In addition to serving as an energy source in simple sugar forms, they are also present as linear and highly branched polysaccharides on the cell surface (as glycoproteins, glycolipids and polysaccharides) in virtually all species. Their existence in the extracellular matrix is associated with their numerous biological roles in processes relevant to health and disease. They are a crucial part in the transmission of important biological signals into and between cells. Specifically, they are involved in cell-cell recognition, cell adhesion, cell and tissue development, functioning of the immune system and regulation of blood coagulation, protein folding, interaction with microbials and toxins, are implicated in cancer (e.g. perturbation of specific glycan-selectin, glycan-integrin interaction in cancerous cells may cause metastasis), as well as structural components (e.g. cellulose in plants, chitin in animals). Oligosaccharides also serve as unique cell surface markers on mammalian cells. Linear carbohydrate chains that are attached to protein are primarily glycosaminoglycans, which are long polymers of sulphated disaccharide repeats 0-linked to a core protein (Raman, al. 2005). Branched glycans exist as TV-linked and 0-linked glycosylation on glycoproteins or on glycolipids (Lowe and Marth 2003). A^-glycans can be subdivided into three distinct groups (high mannose type, hybrid type, and complex type) and are linked through JV-acetylglucosamine to the amino acid asparagine within the consensus sequence Asn-X-Ser/Thr (where X can be any amino 2 acid except proline). O-glycans are divided into core 1 to 4 subgroups, and are linked to proteins through ^-acetylgalactosamine to the amino acids serine or threonine. Sugar molecules are considered as building blocks, analogous to amino acids and nucleotides. In contrast to amino acid and nucleotides, the remarkable diversity of glycosidic linkages and the lack of template (protein sequence being determined by the genetic code) make studying functional glycomics. a non-trivial task. Different combinations of chemical modifications along with distinctions in. the stereochemical arrangement of bonds allow enormous structural diversity in glycan chains. In addition, the biosynthesis of these glycans is a multi-enzyme process in which the sequence of the end product is dependent on both substrate concentration and the availability of these enzymes in different stages of development. Methods such as chemical and enzymatic cleavage of specific glycosidic bonds, capillary electrophoresis, gel electrophoresis, mass spectrometry, lectin affinity liquid chromatography and N M R have been used to study and annotate the composition and structure of complex carbohydrates. Alternatively, the problem can be tackled from the opposite end by studying the enzymes that are responsible for glycan biosynthesis. Understanding the mechanism and stereospecificity of the enzymes responsible for the biosynthesis of the glycan chains, as well as their regulation and interacting partners will shed light on the structure and function of the glycoconjugate glycan chains they produce. Ultimately this knowledge may lead to a new generation of effective carbohydrate therapeutics. 3 1.2 Sialylated Glycoconjugates Many biologically active glycans contain an essential 9-carbon sugar that is known as sialic acid/A/-acetylneuraminic acid (Neu5Ac, Figure 1.1), which is abundantly expressed on secreted and cell-bound glycoproteins and glycolipids (Figure 1.2). A specific example of this class is the Sialyl-Lewis X glycan found on cell surface glycolipids and glycoproteins. This glycan was shown to be the specific ligand responsible for selectin-mediated cell adhesion between leukocytes and endothelial cells during an inflammatory response. The interaction allows the leukocytes in the bloodstream to slow down, traverse the endothelial cell barrier and then enter the tissue (Fukuda, Hiraoka et al. 1999). This sialo-glycan molecule provides a good example of how a critical biological function, the inflammatory response, is regulated by the specific interaction of a glycoconjugate with an adhesion molecule. Sialyl-Lewis X is also involved in the immune system in the recognition of the sialo-ligand by Siglecs (sialic acid binding immunoglobulin Ig-like lectins) receptors (von Gunten, Yousefi et al. 2005; von Gunten and Simon 2006). 6'-sulfo-sialyl-Lewis X has been identified as the ligand for Siglec-8, which has been implicated in the regulation of cell death under inflammatory conditions (Bochner, Alvarez et al. 2005). OH O Figure 1.1 /V-acetylneuraminic acid, sialic acid 4 Poly-a-2,8-sialic acid (PSA) is a unique glycan found mainly on the neural cell adhesion molecule ( N C A M ; Figure 1.2). The N C A M molecule has been implicated in numerous normal and pathological processes, including cell adhesion, signaling, migration, plasticity in the central nervous system, and tumor cell metastasis (Kiss and Muller 2001). The negatively charged P S A molecule on N C A M attenuates cell-cell interactions by exerting intermembrane repulsion during early stages of development (Bruses and Rutishauser 2001). This post-translational modification of N C A M creates an excluded volume around the glycoprotein, enhances cell motility, prevents homophilic attachment to other N C A M molecules on opposing cell surfaces, and increases neural plasticity (Rutishauser 1996). Recently the P S A itself on N C A M was shown to have a very important role in brain development. P S A deficiency in N C A M results in brain defects, progressive hydrocephalus, postnatal growth retardation, and precocious death. (Weinhold, Seidenfaden et al. 2005) • 5 Gangliosides represent another group of cell surface glycoconjugates containing Neu5Ac (Figure 1.2, 1.3). They comprise a large group of sialylated glycosphingolipids found in all vertebrate cell types and constitute the major glycans of the nerve cells of both the peripheral and central nervous systems, as well as being the major sialic acid-bearing glycoconjugates in the brain (Vyas and Schnaar 2001). They can amount to 6% of the weight of lipids from the brain, and are typically found on the neuronal membranes. They are also present in low levels in the plasma membrane of animal tissue cells. Gangliosides are believed to be the functional ligands for a specific myelin-associated glycoprotein to regulate myelin stability and to control nerve regeneration (Vyas and Schnaar 2001). In addition, gangliosides act as receptors for microorganisms and bacterial toxins, regulate cell growth and differentiation and contribute to cell-cell and cell-matrix interactions (Lloyd and Furukawa 1998). For example, the Vibrio cholerae B subunit (cholera toxin) binds with high specificity to the G M I ganglioside (Masserini, Freire et al. 1992). The sialylation state of gangliosides can also provide a significant positive or negative effect on the roles of gangliosides in the signal transduction events across the membrane (Ladisch, Becker et al. 1992). 6 GMI | Galp l -3Ga lNA C p i^Ga lp l^Glcp l -Ceramide Neu5Aca2-3 GM21 GalNAcpl-4Galpl^Glcpi-Ceramide Neu5Aca2-3 GM31 Neu5Aca2-3Galpl-4Glcpl-Ceramide Galp l -3GalNAcpi^Gaip i^Glcp i -Ceramide Neu5Aca2-3 Neu5Aca2-3-Gpibl . Gaipi-3GalNA C pi^Gaipi-4Glcpl-Cerarr i ide Neu5Aca2-3 Neu5Aca2-8 Gxil Galp l-3GalNAcp l-4Gaip l-4Glcp 1-Ceramide Neu5Aca2-3 Neu5Aca2-3 Neu5Aca2-8 Figure 1.3 Main gangliosides of human brain Nomenclature of gangliosides follows Svennerholm shorthand (Svenrierholm 1994), where G is for ganglioside, M for monosialo-, D for disialo- and T for trisialo-ganglioside. The numbers 1, 2, and 3 refer to the order of migration of the gangliosides on TLC . ? Another function for sialic acid-containing glycoconjugates is in their association with circulating proteins, where the presence or absence of terminal Neu5 Ac is correlated with the biological half-life of the protein. Certain circulating glycoproteins lacking the terminal Neu5Ac are cleared. rapidly by either the mannose receptor or the asialo-glycoprotein receptor (Raju, Briggs et al. 2001). Surface sialic acids on circulating proteins shield them from asialoglycoprotein receptor present in the liver, which binds to glycoproteins containing oligosaccharide structures with terminal galactose residues. (Stockert 1995). A specific example is that of sialic acid-containing carbohydrates, which have been shown to affect the activity, receptor-binding, and clearance rate of erythropoietin (Elliott, Egrie et al. 2004). A n understanding of these interactions can be employed with the sialylation of therapeutic glycoprotein to increase its serum half-life (Smith, Bousfield et al. 1993). 1.2.1 Sialylated glycoconjugates and bacterial pathogenicity In prokaryotes, sialic acid also plays an essential role in pathogenesis. Many of the bacterial pathogens that invade mammalian cells have exploited the presence of sialo-glycoconjugates on their host to their advantage (Moran, Prendergast et al. 1996; Kahler and Stephens 1998). These bacteria, which include Escherichia coli, Neisseria meningitidis, N. gonorrhoeae, Haemophilus influenza, H. ducreyi, Streptococcus agalctiae, Campylobacter jejuni, and Pasteurella multocidai, display some of the same sialylated structures present on their lipooligosaccharides (LOS) or as a part of their capsule. Table 1.1 summarizes these pathogens and their molecular mimicking structures. A l l of these carbohydrates contain sialic acid as the terminal sugar. Specifically, sialic 8 acid present in the bacterial capsule (E. coli, N. meningitidis, P. haemolytica, and S. agalctiae) is found to imitate PSA structures in the host, and the sialylated LOS structures all mimic human cell surface glycolipids. It is thought that the presence of these carbohydrate mimics camouflages the cell surface of these pathogens, and allows them to escape detection by the host immune system since these molecules are not considered foreign (Moran, Prendergast et al. 1996; Kahler and Stephens 1998). The presence of these sugar chains also impedes the lytic effects of serum complement (Vogel, Claus et al. 1999). In addition, sialylation of LOS may decrease or prevent adherence to human neutrophils and the subsequent oxidative burst or phagocytosis (Rest and Frangipane 1992; Wetzler, Barry et al. 1992). Furthermore, it has been proposed that some pathogens might use cell surface receptors on the host that recognize these sialylated glycoconjugates as a means for transmission (Preston, Mandrell et al. 1996; Harvey, Porat et al. 2000). 9 Table 1.1 Bacterial species and their molecular mimicking molecules found on their capsule or LOS Bacterial Species Molecular Mimic Capsule/LOS E. coli K l PSA oc-2,8-Neu5Ac Capsule K92 PSA mixed a-2-8/2-9 Capsule P. haemolytica PSA a-2,8-Neu5Ac Capsule N. meningitidis B PSA a-2,8-Neu5Ac Capsule a-2,3-sialyl-lacto-N-neotetraose LOS N. meningitidis C PSA a-2,9-Neu5Ac Capsule a-2,6-sialyl-Pk LOS TV, gonorrhoeae a-2,3-sialylo-lacto-N-neotetraose LOS a-2,6-sialyl-Pk (?) LOS H. ducreyi a-2,3-sialylo-lacto-N-neotetraose LOS a-2,3-sialyl-lactose LOS H. influenzae a-2,3-sialyl-lactose LOS a-2,8/a-2,3-disialyl-lactose LOS C. jejuni Numerous ganglioside mimics LOS S. agalactiae a-2,3-sialyl-lacto-N-neotetraose Capsule A specific example in which bacterial molecular mimicry leads to the development of disease is illustrated in Campylobacter jejuni. C. jejuni is the most prevalent food borne pathogen and the leading cause of diarrhoeal disease in developed countries (Aspinall, Fujimoto et al. 1994; Salloway, Mermel et al. 1996; Blaser 1997; Ketley 1997; Saida, Kuroki et al. 1997; Hao, Saida et al. 1998; Sheikh, Nachamkin et al. 1998; Moran and Prendergast 2001; Hye and Nachamkin 2007). Infection with C. jejuni is one of the most recognized antecedents to the development of Guillain-Barre syndrome (GBS) (Mishu and Blaser 1993; Vriesendorp, Mishu et al. 1993; Alios, Lippy et al. 1998; 10 Sheikh, Ho et al. 1998; Moran and Prendergast 2001). GBS is an acute inflammatory and demyelinating paralytic disease of the peripheral nervous system and is currently the most common cause of rapidly acquired paralysis in the United States, affecting one to two people in.every 100,000 ( Terminal oligosaccharides identical to those of G M I 3 , G M 2 , G M 3 , Goia, G D I C , GD3 , and G n a gangliosides have all been found in various C. jejuni strains (Penner and Aspinall 1997). In a few cases, infection with C. jejuni serotypes 0:2 and 0:23 leads to the development of Miller-Fisher syndrome (MFS), an infrequent variant of GBS (Salloway, Mermel et al. 1996; Shin, Ackloo et al. 1997; Endtz, Ang et al. 2000). MFS, like its more common variation GBS, is also a neurophagy characterized by abnormal muscle coordination, paralysis of eye muscles, and absence of the reflexes. Previous studies have demonstrated that molecular mimicry of C. jejuni sialylated LOS to human gangliosides is responsible for the onset of immunopathogenesis of GBS (Aspinall, Fujimoto et al. 1994; Aspinall, McDonald et al. 1994; Yuki, Taki et al. 1994; Oomes, Jacobs et al. 1995; Moran and Prendergast 2001). The association of infection with C. jejuni to development of GBS is further supported by the association of certain C. jejuni serotypes to the development of either GBS or MFS, and the isolation of autoreactive anti-GMi, anti-Goia, (Figure 1.3) and anti-Gojb antibodies to gangliosides from the sera of patients suffered from GBS and MFS after C. jejuni infection (Saida, Kuroki et al. 1997; Hao, Saida et al. 1998). 11 1.3 Glycosyltransferases The biosynthesis of glycan structures is catalyzed mainly by glycosyltransferases (GTs), while at times combinations of glycosidases (also known as glycoside hydrolases, GHs) and GTs coordinate together to produce the complex sequence of the glycan chains. Glycosidases hydrolyze the glycosidic bond between two or more carbohydrates or between a carbohydrate and a non-carbohydrate moiety, while glycosyltransferases transfer a sugar moiety from an activated (typically nucleotide or lipidated) sugar donor onto specific acceptors, generating new glycosidic bonds (Figure 1.4). The seven major nucleotide sugar donors are GDP-fucose, GDP-mannose, UDP-glucose, UDP-galactose, UTJP-TV-acetylglucosamine, UDP-Af-acetylgalactosamine, and CMP-7Y-acetylneuraminic acid (Paulson and Colley 1989). In addition, a few families of glycosyltransferases utilize dolichyl-phospho-sugars, sugar-1-phosphates, and lipid diphospho-sugars as the donors. A l l known and putative G T sequences have been classified into families on the basis of amino acid sequence similarity ( (Coutinho, Deleury et al. 2003)). There are currently 87 families of glycosyltransferases (second largest family of C A Z y database, while G H family is the biggest), which together comprise over 12,000 characterized or putative glycosyltransferase sequences. The number of G T families and sequences is constantly increasing with the discovery of new G T genes. + Figure 1.4 Schematic for reaction catalyzed by glycosyltransferases 12 While some CAZy families members (e.g. GT-3, glycogen synthase) share high specificity for the type of glycosidic bond created, diversity exists within other GT families. For example, GT-2 houses a variety of enzymes (over 6,500 entries) including cellulose synthase, chitin synthase, dolichyl-phosphate P-D-mannosyltransferase, dolichyl-phosphate P-glucosyltransferase, A^acetylglucosaminyltransferase, N-acetylgalactosaminyltransferase, hyaluronan synthase, chitin oligosaccharide synthase, P-1,3-glucan synthase, P-l,4-mannan synthase, P-manriosylphosphodecaprenol-mamiooligosaccharide a-l,6-mannosyltransferase, and a-l,3-L-rhamnosyltransferase. In families such as GT-2 that contain enzymes of different functions, their sequence similarities are restricted to the catalytic domain only. It is therefore challenging to predict the function of a putative GT, and in particular, the precise stereochemistry of the resulting glycosidic bond formed, based only on sequence similarity. A specific example is given by blood group A and B transferases, which differ by only four amino acids. They use different glycosyllic donor: UDP-GalNAc for A transferase and UDP-Gal for B transferase but the same H-antigen receptor (Yamamoto, Clausen et al. 1990), and thus generating different epitopes on red blood cells. Two versions of the sialyltransferase Cst-II from Campylobacter jejuni which differ by eight amino acids, also display different acceptor specificities (discussed below). 1.3.1 Structure of glycosyltransferases In eukaryotic cells, the majority of glycosyltransferases are type II membrane proteins that reside in the Golgi apparatus. Their topology is characterized by an N -terminal cytoplasmic domain, followed by a transmembrane helix, a stem region and a 13 large catalytic domain that projects into the lumen of the rough endoplasmic reticulum or the Golgi apparatus (sites of glycosylation). Bacterial glycosyltransferases are typically monotopic membrane proteins that contain a membrane association domain to anchor the enzymes on the bacterial cell surface. The first nucleotide-sugar dependent glycosyltransferase crystal structure was that of bacteriophage T4-glucosyltransferase, solved in 1994 (Vrielink, Ruger et al. 1994). Since then, 43 glycosyltransferase protein structures from prokaryotes and eukaryotes (with more than 100 PDB entries) corresponding to 23 distinct GT families have been reported. A 3D-glycosyltransferase database has been created to provide up-to-date structural information concerning this class of enzymes ( (Breton, Snajdrova et al. 2006). From the available 3D structures of the nucleotide activated glycosyltransferases, this class of enzyme is characterized by the conservation of their architecture (in.contrast to glycosidases where structures are more diverse). Collectively, the structures of GTs can be categorized into two superfamilies, GT-A and GT-B (Figure 1.5). In the GT-A fold, the enzyme contains two distinct domains (Figure 1.5a): an N-terminal Rossmann nucleotide binding domain (Rossmann and Argos 1978), which is a characteristic mixed a/p/a sandwich with a seven-stranded P-sheet (3214567 topology) and recognizes the activated donor sugar, and a second smaller C-terminal domain that contains the acceptor binding site. In contrast, the GT-B fold contains two similar Rossmann nucleotide binding domains, which are connected by a linker region and sandwich the active site inbetween (the N-terminal domain is responsible for acceptor binding whereas the C-terminal domain binds the donor sugar) 14 (Figure 1.5b). In both G T - A and G T - B , a metal binding site is observed within the cleft of the enzymes. Figure 1.5 Ribbon diagrams of representative GT folds. (a) GT-A fold, Bos taurus a1,3-galactosyltransferase a3GalT complexed with UDP, M n 2 + and LacNAc (PDB code 1o7q) (Zhang, Swaminathan et al. 2003) (b) GT-B fold, Bacteriophage T4 DNA b-glucosyltransferase BGT, in complex with UDP-GIc (PDB code 1nzd) (Lariviere, Gueguen-Chaignon et al. 2003). Bound nucleotide sugars are represented in stick form. Helices and loops are shown in blue and p-strands are shown in green. 1.3.2 Reaction mechanism B y analogy to the well-studied glycosidases, glycosyltransferases can be categorized mechanistically as either retaining or inverting, based on the configuration of 15 the anomeric center before and after the reaction (Figure 1.6). In the retaining mechanism, the enzymatic reaction was first proposed to proceed through a SN*-like mechanism in which leaving group departure and nucleophilic attack occur in a collaborative but asynchronous manner on the same face of the glycoside (Sinnott and Jencks 1980; Gibson, Turkenburg et al. 2002; Ly, Lougheed et al. 2002). Alternatively, a different mechanism via a double displacement involving the formation and subsequent breakdown of a covalent glycosyl-enzyme intermediate has also been proposed. In the first step of the double displacement mechanism, a general base in the enzyme active site acts as a nucleophile and attacks the donor substrate to displace the aglycone portion of the molecule. The cleavage of the glycosidic bond is assisted by a general acid (another residue or metal ion). In the second step, this general acid then acts as a base to activate the acceptor molecule to nucleophilically attack the anomeric center of the donor sugar, forming the new glycosidic bond. After the retaining reaction, the anomeric center retains the same configuration. For the inverting mechanism, the catalysis involves a direct displacement in which the acceptor is activated by a general base to nucleophilically attack the anomeric center of the donor sugar, with the possibility of a general acid simultaneously aiding in the cleavage of the carbon-oxygen bond. The resulting glycosidic bond has an inverted configuration as compared to the starting molecule. The exact catalytic mechanism (inverting or retaining) of a particular glycosyltransferase cannot be reliably deduced from sequence comparison or the predicted fold (GT-A or GT-B) alone. 16 Retaining acceptor Inverting A = general acid B = general base donor ^ c 5 acceptor Figure 1.6 Glycosyltransferase mechanisms In addition to an inverting or retaining mechanism, glycosyltransferases can also be classified as either "a" or "P" according to the anomeric configuration of the glycosidic bond formed. (Figure 1.7) Alpha-linkage OH Beta-linkage OH HO Figure 1.7 a and (3 linkage at the anomeric center 17 1.4 Sialyltransferases Within the CAZy classification sialyltransferases are found in families 29, 38, 42, 52 and 80 (Table 1.2). They are responsible for the biosynthesis of sialylated glycoconjugates. The universal donor for this class of GT is CMP-P-7Y-acetylneuraminic acid, which is activated by the CMP molecule (Figure 1.8) (Kean 1991). On the other hand, the specific acceptor varies for each enzyme. Sialyltransferases are classified as inverting glycosyltransferases due to the inversion of the anomeric configuration from the P-linkage of the donor molecule to an a-linkage in the new glycosidic bond. Sequences from different families are very diverse, yet they catalyze a highly similar activity. N H , H O H O O H O H H O O Figure 1.8 Universal donor CMP-iV-acetylneuraminic acid for sialyltransferases 18 Table 1.2 Sialyltransferase families according to CAZy database C A Z y Source 3 Known Activities Statistics GT29 Eukaryotes (323), Viruses (5) a-2,3/a-2,6/o-2,8-sialyltransferase CAZy entries: 328 GenBank/GenPept: 895 Swissprot: 161 GT38 Neisseria meningitidis (13), Escherichia coli (5) a-2-8/a-2,9-polysialyltransfearse CAZy entries: 18 GenBank/GenPept: 22 Swissprot: 4 GT42 Campylobacter jejuni (56), Haemophilus influenzae (3), Helicobacter acinonychis (2), Pasteurella multocida (1), unknown (23) a-2,3/a-2,8-sialyltransferase CAZy. entries: 85 GenBank/GenPept: 127 Swissprot: 9 3D: 1 PDB: 3 GT52 N. gonorrhoeae (15), N. meningitidis (5), Salmonella enterica (21), H. influenzae (3), Aeromonas punctata (1), P. multocida (1), Psychrobacter arcticum (1), Streptococcus agalactiae (1) a-2,3-sialyltransferase; a-glucosyltransferase CAZy entries: 48 GenBank/GenPept: 56 Swissprot: 20 GT80 P. multocida (2), /-/. ducreyi (1), Photobacterium damselae (1) a-2,3/a-2,6-sialyltransferase CAZy entries: 4 GenBank/GenPept: 6 Swissprot: 0 3D: 1 PDB: 8 The number inside the bracket corresponds to the specific number of CAZy sequences identified for that species Eukaryotic sialyltransferases are all found in the GT-29 family (307 sequences). Recent analysis of the human genome has identified 20 distinct sialyltransferase genes, 19 which can be subdivided into four subfamilies: ST3Gal, ST6Gal, ST6GalNAc, and ST8Sia (Harduin-Lepers, Vallejo-Ruiz et al. 2001; Takashima, Tsuji et al. 2002; Harduin-Lepers, Mollicone et al. 2005). Whilst work on these mammalian enzymes has been ongoing for the past 25 years, only a few of the approximately 20 enzymes from the human genome have been significantly characterized (Paulson, Rearick et al. 1977; Rearick, Sadler et al. 1979), despite the importance of these enzymes in making a large number of bio-active glycoconjugates. This is mainly due to the fact that the recombinant expression of these enzymes is mostly performed in mammalian cell lines, or other eukaryotic expression systems that do not yield large quantities of soluble protein, thereby precluding structural analyses. In addition, all vertebrate sialyltransferases are type II transmembrane glycoproteins which lead to heterogeneous protein preparations and highly mobile glycan chains, both of which are problematic for protein crystallization. Structure-function studies performed so far have been limited to site directed mutagenesis on the four conserved sialyl motifs: motif L(large), motif S (small), motif III, and motif VS (very small) (Datta and Paulson 1995; Geremia, Harduin-Lepers et al. 1997; Datta, Sinha et al. 1998; Jeanneau, Chazalet et al. 2004) with some investigations into kinetic mechanisms (Rearick, Sadler et al. 1979; Horenstein and Bruner 2002; Jeanneau, Chazalet et al. 2004). Recently, with the availability of the C. jejuni sialyltransferase crystal structure (Chiu, Watts et al. 2004), attempts for the molecular modeling of human sialyltransferases using Cst as template have been reported (Sujatha and Balaji 2006). Detailed examination of the bacterial sialyltransferases is a much more recent undertaking than the analysis of their mammalian counterparts. Earlier attempts to clone these prokaryotic sialyltransferases using D N A probes based on the mammalian sialyl 20 motifs were unsuccessful (Rest and Mandrell 1995). The molecular cloning and characterization of the GT-42 and GT-52 families of enzymes was started through the application of expression cloning and the use of synthetic acceptor molecules (Gilbert, Watson et al. 1996; Gilbert, Brisson et al. 2000). These bacterial enzymes might present a better model system to study sialyltransferase structure and function, due to the higher yield of protein in bacterial expression systems. In addition, purified bacterial sialyltransferases would be less prone to heterogeneity as they are not glycosylated. 1.4.1 Sialyltransferase from Campylobacter jejuni The mucosal pathogen C. jejuni has been recognized as an important cause of acute gastroenteritis in humans and is believed to be the antecedent to the onset of the autoimmune disease GBS (Salloway, Mermel et al. 1996; Blaser .1997; Ketley 1997; Saida, Kuroki et al. 1997; Hao, Saida et al. 1998; Sheikh, Nachamkin et al. 1998; Moran and Prendergast 2001; Hye and Nachamkin 2007). The genes responsible for the biosynthesis of the ganglioside mimics in C. jejuni have been cloned using a combination of approaches (Gilbert, Brisson et al. 2000; Gilbert, Karwaski et al. 2002), thereby isolating a distinct membrane-associated enzyme with cc-2,3-sialyltransferase activity. Sequence analysis and comparison has identified a family of sialyltransferases that occur in C. jejuni, H. influenzae and perhaps Pasteurella multocida (Coutinho, Deleury et al. 2003). These sialyltransferases are grouped into the CAZy family GT-42, which currently contains 70 entries, 47 of these being bacterial proteins, of which 40 are from different Campylobacter strains. A sequence alignment of representative members of this family is shown in Figure 1.9. 21 Sialyltransferases from C. jejuni dominate this GT family and are termed Cst. There are three different isoforms of Csts, namely Cst-I, Cst-II from serostrain 019, and Cst-II from serotype OH4384, a GBS-associated strain (Aspinall, McDonald et al. 1994). The two Cst-II enzymes differ by only eight amino acids, whilst Cst-I and Cst-IIs share about 40 % overall sequence identity. A l l three versions of Cst can catalyze the first transfer step which involves a donor sugar CMP-Neu5Ac and an acceptor sugar galactose derivative (Figure 1.10). The sialic acid is transferred onto the P-galactose acceptor sugar at position 3 to form an a-2,3 linkage with the release of CMP. Interestingly, despite very high sequence identity (97.3 %) between two of these enzymes Cst-IIo:i9 and Cst-IIOH4384, they display different substrate specificities (Gilbert, Brisson et al. 2000). Cst-IIOH4384 carries an additional a-2,8-sialyltransferase activity in which the initially formed a-2,3-linked sialyl-galactose is used as the acceptor substrate in a second transfer step (Figure 1.10). Hence Cst-IIoH4384 is termed bifunctional, while both Cst-IIo:i9 and Cst-I are termed monofunctional. Understanding the rationale for the difference in substrate specificities of Csts may allow us to extend this knowledge to other families of GT. 22 « 2 0 * 40 * 60 « 0_19 t MKKVIIAGNGPSLKEIDYSRLPNDFDVFRCN 0_36 i - MKKVI IAGNGPSLKEIDYSRLPNDFDVFRCN 0_4 : HKKVI IAGNGPSLKEIDYSRLPNDFDVFRCN O_10 : - - MKKVI IAGNGPSLKEIDYSRLPNDFDVFRCN 0_41 : MKKVI IAGNGPSLKEIDYSRLPNDFDVFRCN OH43B4 : MKKVI IAGNGPSLKEIDYSRLPNDFDVFRCN L i c 3 A _ 3 7 5 : M SINQSINQSINQSINQSINQSINQSINQSINQSKSVIIAGNGTSLKSIDYSLLPKDYDVFRCN L i c 3 A _ 4 B 6 ; - MSINQSINQSINQSINQSINQSINQSKSVIIAGNGTSLKSIDYSLLPKDYDVFRCN L i c 3 A _ R M l l : MNGTICQSINQSINQSINQSINQSINQSINQSINQSINQSINQSINQSKSVIIAGNGTSLKSIDYSLLPKDYDVFRCN C s t - I : MTRTRMENEL I VS K N MQ N I 11 A G N G P S L K N I N Y K R L P R E Y D V F R C N 3 1 3 1 3 1 3 1 3 1 3 1 6 4 5 6 7 8 4 e 8 0 10 0 12 0 14 0 1 9 3 6 4 1 0 0_4 1 0H4 3 8 4 L i c 3 A _ 3 7 5 L i C3A_4 8 6 L i C 3A_RM11 C s t - I Q F Y FEDKYYLGKKCKAVFYTPNFFFEQYYTLKHLIQNQE-Q F Y F E D K Y Y L G K K C K T V F Y T P N F F F E Q Y Y T L K H L I Q N Q E -QFYFEDKYYLGKKCKAVFYTPGFFFEQYYTLKHLIQNQE-QFYFEDKYYLGKKFKAVFYNPGLFFEQYYTLKHLIQNQE-QFYFEDKYYLGKKCKAVFYNPSLFFEQYYTLKHLIQNQE-Q F Y F E D K Y Y L G K K C K A V F Y N P I L F F E Q Y Y T L K R L I Q N Q E -Q F Y F E D H Y F L G K K I K K V F F N C S V I F E Q Y Y T F M Q L I K N N E Y K Y E Y A D I I L A S F L N L G D S T L K K I Q H L E K L L P Q I D L G H C Q F Y F E D H Y F L G K K I K K V F F N C S V I F E Q Y Y T F M Q L I K N N E Y K Y K Y A D I I L A S F L N L G D S T L K K I Q H L E K L L P Q I D L G K C Q F Y F E D H Y F L G K K I K K V F F N C S T I F E Q Y Y T F M Q L I K N N E - - Y E Y A D I I L S S F V N L G D S E L K K I K N V Q K L L T Q V D I G H Y QFYFSDKYYLGKKIKAV FFNPGVFLQQYHTAKQLILKNE- - Y E I K N I F C S T F N L P F I E S N D F L H Q F Y N F F P D A K L G Y E YETELIMCSNYNQAHLENENFVKTFYDYFPDAHLGYD YETELIMCSNYNQAHLENENFVKTFYDYFPDAHLGYD YETELIMCSNYNQAHLENENFVKTFYDYFPDAHLGYD YETELIMCSNYNQAHLENENFVKTFYDYFPDAHLGYD YETELIMCSNFNQAHLENQNFVKTFYDYFPDAHLGYD YETELIMCSNYNQAHLENENFVKTFYDYFPDAHLGYD 10 7 10 7 10 7 10 7 10 7 10 7 14 2 13 4 15 4 12 2 160 * 180 * 200 * 0_19 : F F K Q L K E F N A Y F K F H E I Y F N Q R I T S G V Y M C A V A I A L G Y K E I Y L S G I D F Y Q - N G S 0_36 : F F K Q L K E F N A Y F K F H E I Y F N Q R I T S G V Y M C A V A I A L G Y K E I Y L S G I D F Y Q - N G S 0_4 : F F K Q L K E F N A Y F K F H E I Y F N Q R I T S G V Y M C A V A I A L G Y K E I Y L S G I D F Y Q - N G S O_10 : F F K Q L K E F N A Y F K F H E I Y L N Q R I T S G V Y M C A V A I A L G Y K E I Y L S G I D F Y Q - N G S 0_41 : F F K Q L K E F N A Y F K F H E I Y F N Q R I T S G V Y M C T V A I A L G Y K E I Y L S G I D F Y Q - N G S OH4384 : FFKQLKDFNAYFKFHEIYFNQRITSGVYMCAVAIALGYKEIYLSGIDFYQ-NGS L i c 3 A _ 3 7 5 : YLKKLRAFNAHLQYHELYENKRITSGVYMCAVATAMGYKDLYLTGIDFYQEKGN L i c 3 A _ 4 8 6 : YLKKLRAFNAHLQYHELYENKRITSGVYMCAVATAMSYKDLYLTGIDFYQEKEN L i c 3 A _ R M l l : YLNKLPAFDAYLQYNELYENKRITSGVYMCAVATVMGYKDLYLTGIDFYQEKGN C s t - I : V I E N L K E F Y A Y I K Y N E I Y F N K R I T S G V Y M C A I A I A L G Y K T I Y L C G I D F Y E - G D V 2 2 0 * SYAFDTKQENLLKLAPDFKNDRSH SYAFDTKQENLLKLAPDFKNDRSH SYAFDTKQENLLKLAPDFKNDRSH SYAFDTKQENLLKLAPDFKNDRSH SYAFDTKQKNLLKLAPNFKNDNSH SYAFDTKQKNLLKLAPNFKNDNSH PYAFHHQKENIIKLLPSFSQNKSQ PYAFHHQKENIIKLLPSFSQNKSQ PYAFHHQKENIIKLLPSFSQNKSQ I Y P F E A M S T N I K T I F PGIKDFKP-18 4 18 4 18 4 18 4 18 4 18 4 2 2 0 2 12 2 3 2 19 8 1 9 3 6 4 I 0 4 1 OH 4 3 8 4 L i c 3 A _ 3 7 5 L i c 3 A L i c 3 A C s t - I 4 8 6 RM 1 1 240 * 260 • Y I G H S K N T D I K A L E F L E K T Y K I K L Y C L C P N S L L A N F I Y I G H S K N T D I K A L E F L E K T Y K I K L Y C L C P N S L L A N F 1 Y I G H S K N T D I K A L E F L E K T Y K I K L Y C L C P N S L L A N F I Y I G H S K N T D I K A L E F L E K T Y K I K L Y C L C P N S L L A N F I YIGHSKNTDIKA L E F L E K T Y E I K L Y C L C P N S L L A N F I Y I G H S K N T D I K A L E F L E K T Y K I K L Y C L C P N S L L A N F I SDIHSMEYDLNALYFLQKHYGVNIYCISPESPLCNYF SDIHFMEYDLNALYFLQKHYGVNIYCISPESPLCNYF SDIHSMEYDLNALYFLQKHYGVNIYCISPESPLCNYF S N C H S K E Y D I E A L K L L K S I Y K V N I Y A L C D D S I L A N H F 280 * 300 * ELAPNLN - SNFI IQEKNN-YTKDILIPS - -E L A P N L N - S N F I I Q E K N N - Y T K D I L I P S E L A P N L N - S N F I I Q E K N N - Y T K D I L I P S - -E L A P N L N - S N F I I Q E K N N - Y T K D I L I P S E L A P N L N - S N F I I Q E K N N - Y T K D I L I P S E L A P N L N - S N F I I Q E K N N - Y T K D I L I P S - -P L S P L N N P I A F I P E E K K N - Y T Q D I L I P P P L S P L N N P I A F I P E E K K N - Y T Q D I L I P P - -P L S P L N N P I T F I L E E K K N - Y T Q D I L I P P PLSININ-NNFTLENKHNNSINDILLTDNTPGVSFYKNQLK 2 4 7 2 4 7 2 4 7 2 4 7 2 4 7 2 4 7 2 8 4 2 7 6 2 9 6 2 7 5 0_ 1 9 0_3 6 0_4 0_1 0 0_4 1 OH4 3 8 4 L i c 3 A _ 3 7 5 L i c 3A_4 8 6 L i c 3 A _ R M l l C s t - I 3 2 0 * 3 4 0 . SEAYGKFSKN-. SEAYGKFSKN-. SEAYGKFSKN-SEAYGKFSKN-. SEAYGKFTKN-. SEAYGKFSKN-* 3 6 0 I N F K K I K I K E N V Y Y K L -• I N F K K I K I K E N V Y Y K L -• I N F K K I K I K E N V Y Y K L -• I N F K K I K I K E N I Y Y K L -• I N F K K I K I K E N I Y Y K L -* 3 8 0 * IKDLLRLPSDIKHYFKGK IKDLLRLPSDIKHYFKGK IKDLLRLPSDIKHYFKGK IKDLLRLPSDIKHYFKGK IKDLLRLPSDIKHYFKGK I N F K K I K I K E N I Y Y K L IKDLLRLPSDIKHYFKGK KFVYKKIG I Y S K P R I Y Q N L I F R L FWDILRLPNDIKHALKSR KFVYKNIG V Y S K P R I Y Q N L I F R L IWDILRLPNDIKHALKSR - KFVYKKIG I Y S K P R I Y Q N L I F R L IWDILRLPNDIKHALKSR A D N K I M L N F Y N I L H S K D N L I K F L N K E I A V L K K Q T T Q R A K A R I Q N H L S Y K L G Q A L I I N S K S V L G F L S L P F I I L S I V I S H 2 9 1 2 9 1 2 9 1 2 9 1 2 9 1 2 9 1 3 2 5 3 17 3 3 7 3 5 3 1 9 3 6 0_ 0_ 0_4 0_1 0 0_4 1 OH4 3 8 4 L i c 3 A _ 3 7 5 L i e 3 A_4 8 6 L i c 3 A _ R M l 1 C s t - I KWD KWD - - -KWD KQEQKAYKFKVKKNPNLALPPLETYPDYNEALKEKECFTYKLGEEFIKAGKNWYGEGYIKFIFKDVPRLKREFEKGE 3 2 8 3 2 0 3 4 0 4 3 0 Figure 1.9 Sequence alignment of representative CAZy GT-42 sialyltransferases Cyan highlights amino acids that are strictly conserved within the sequences compared while light gray highlights similar residues and dark gray highlights residues conserved in C. jejuni. The first five sequences are sialyltransferases from different serotypes of C. jejuni (serotype name 23 indicated in the sequence identification). The Lic3A sequences are from different serotypes of Haemophilus influenzae. The last Cst-I sequence is the monofunctional variant of Cst-II. Donor NH2 CMP-NeuAc ?f\ n - o -„ .o> ' \ ) „ o / ™\ H 0 ^ ^ _ ^ , O H T HO OH ~H&*" 7 ^ q 7 V 0 i i H HO / C 3 " O Acceptor Galp -1.3-R OH HO B linkage Inversion CMP >a-2,3 linkage ° * ? ? H O H a-2,8 linkage 0 3 ' Cst-II (OH4384) Cst-II (0:19) Cst-I O R O H Cst-II (OH4384) O ^ O ' O H H O Ac&H O H O H OH O OR OH Monofunctional O R Bifunctional Figure 1.10 Reaction scheme of Cst. Sialyltransferase Csts from Campylobacter use CMP-Neu5Ac as donor sugar and a galactose derivative as acceptor sugar. The bonds created in the monofunctional and bifunctional reaction are highlighted by the green circle. Monofunctional Cs t - l l 0 : i 9 and Cst-I can only perform the first transfer step onto galactose while bifunctional Cst-II0H4384 can perform the second transfer onto sialyl galactose from the first transfer. \ . 24 1.5 Objectives of thesis Despite their important role in the biosynthesis of sialylated glycoconjugates, little is known about the structural details and the catalytic mechanism of sialyltransferases. At the start of this thesis investigation, no high-resolution structural information for any sialyltransferase was available. The rationale for the differences in substrate specificities observed in different sialyltransferases was also poorly understood. The objective of this thesis study is to understand the structure and kinetic mechanism of this important class of glycosyltransferases, with particular focus on C. jejuni sialyltransferases Csts. Chapter 2 describes the structural and kinetic characterization of Cst-II in its truncated form from C. jejuni. This is a bifunctional enzyme which can catalyze both an a-2,3 and a-2,8 linkage. Crystallographic and mutagenesis studies on Cst-II in complex with an inert donor sugar analogue, CMP-3FNeu5Ac, are presented. Most of this work has been published in Nature Structural and Molecular Biology (Chiu, Watts et al. 2004) Chapter 3 describes a directed evolution study of Cst-II aimed at producing variants of the sialyltransferase with novel substrate specificities . and products. Crystallographic and kinetic analyses of a generated Cst-II mutant are presented. Most of this work has been published in Nature Methods (Aharoni, Thieme et al. 2006) Chapter 4 describes the structural characterization of Cst-I in its truncated form, a monofunctional sialyltransferase with only a-2,3-transferase activity, also isolated from C. jejuni. Crystallographic study on Cst-I in apo form and in complex with an inert donor sugar analogue, CMP-3FNeu5Ac, is presented. Modeling results of various acceptors into the active sites of Cst-I and Cst-II are also presented. This work has generated a manuscript currently under consideration at Biochemistry. . • 25 CHAPTER 2 - Structural and biochemical characterization of bifunctional sialyltransferase Cst-II from Campylobacter jejuni 2.1 I N T R O D U C T I O N The human mucosal pathogen Campylobacter jejuni has been shown to express variable cell surface carbohydrate mimics of gangliosides that are associated with virulence (Penner and Aspinall 1997; Guerry, Szymanski et al. 2002). Terminal oligosaccharides identical to seven different gangliosides have all been found in various C. jejuni strains (Penner and Aspinall 1997). The molecular mimicry between C. jejuni LOS outer core structures and gangliosides has also been suggested to act as a trigger for autoimmune mechanisms in the development of Guillain-Barre syndrome (Endtz, Ang et al. 2000). The terminal sialic acid residues are the main sources of diversity in these structures. Genes responsible for the biosynthesis of the ganglioside mimics in C. jejuni were isloated by the laboratory of Dr. Warren Wakarchuk, using a combination of approaches (Gilbert, Brisson et al. 2000; Gilbert, Karwaski et al. 2002), thereby identifying a distinct membrane-associated enzyme (Figure 1.10) (Cst-II from C A Z y family GT-42 (Coutinho, Deleury et al. 2003). The essential role of sialylated oligosaccharides in the biology of both pathogen and host suggests that the enzymes involved in their biosynthesis represent potential targets for therapeutic intervention. To date there has been very little work on fundamental structure/function aspects of the sialyltransferases from any species. This is likely a result of difficulties in expressing these typically membrane-associated proteins in either bacterial or eukaryotic systems. Using designed constructs from which predicted C-terminal membrane association domains have been removed, but which 26 importantly retain full activity, the first detailed structure of a sialyltransferase from any source, Cst-II from C. jejuni strain OH4384 was obtained in this study. The structure of this bifunctional enzyme has been determined in the absence and presence of a sialic acid donor sugar analogue. This chapter represents a major advance in our understanding of how sialyltransferases catalyze their essential reactions and provides a structural basis for the design of novel anti-bacterial agents that function through inhibition of their action. 27 2.2 M E T H O D S 2.2.1 Cloning, protein expression, and purification Cloning of Cst-II from C. jejuni strain OH4384 was performed as previously described (Gilbert, Brisson et al. 2000). Primers with sequences 5' GC ATT A C G C A T A T G A A G A A A G T T ATT ATTGC3 ' • and 5' G C A T T A C G T C G A C T T A A T T A A T A T T T T T T G 3 ' were used to remove the C-terminal 32 residues. The PCR product was inserted into pET-28a and pET-41a (Novagen) vectors, transformed into E. coli strain BL21(DE3) and grown at 37 °C. Overexpression of the protein was induced with 0.5 m M IPTG when the OD600 reached ~ 0.60, and cells harvested after overnight incubation at 20 °C. 6-His-tagged Cst-II 1 - 2 5 9 was 94-purified with a N i -chelating sepharose column (Pharmacia). The 6-His tag was cleaved at 4 °C overnight with 1:1,000 dilution of thrombin (Sigma) leaving three extraneous residues at the N-terminus (Gly^Ser-His). This cleaved product was purified by a Mona-co anion-exchanger (Pharmacia) using a linear gradient of NaCl in 20 m M Tris-HCl buffer, pH 8.3. A tagless Cst-II 1 - 2 5 9 construct was expressed as detailed above but was selectively precipitated with 1.0 M ammonium sulphate to aid in purification. The pellet was resuspended in, and dialyzed overnight against, 20 m M Tris-HCl pH 8.3. The dialysate was then applied to a Q-sepharose column, followed by Mono-Q and Superdex200 columns (Pharmacia). Purified proteins were concentrated to -12 mg mL" 1 for crystallization trials. Selenomethionyl Cst-II 1 - 2 5 9 was prepared using previously described protocols (Doublie 1997). 28 2.2.6 Static light scattering Static light scattering experiments were performed at 25 °C on a Shodex Protein KW-803 column using 50 m M H E P E S p H 7.5, 100 m M N a C l . Refractive index and Mini-dawn light scattering detectors (Wyatt Technology) were calibrated using bovine serum albumin (Sigma). 2.2.2 Crystallization and data collection Protein mixtures containing 10 mg m L _ 1 protein, 10 m M MgCL. , and varying substrates at 10 m M (donor: C M P - N e u 5 A c or CMP-3FNeu5Ac ; acceptor: lactose, 3'deoxylactose or sialyl-lactose; inhibitor: CDP) were used for hanging drop vapor diffusion crystallization trials. Tetragonal crystals of 6-His-tagged protein crystallized in 100 m M H E P E S p H 7.5, 10 % P E G 6000 and 5 % M P D and monoclinic crystals of non-tagged protein crystallized in 100 m M bicine p H 9.0, 10 % P E G M M E 2000 and 100 m M N a C l . Data were collected at 100 K with 15 % (v/v) M P D plus mother liquor as cryoprotectant. The C M P complex crystal (P4) has unit cell dimensions of a = b = 115.43, c = 41.06 A. A multiwavelength data set was collected at the N S L S , beamline X 2 5 , using an A D S C Quantum Q315 C C D detector. C M P - 3 F N e u 5 A c was synthesized as described (Watts, Damager et al. 2003). For the C M P - 3 F N e u 5 A c and C D P complexes (P2i), data sets were collected at N S L S beamline X 8 - C using an A D S C Quantum Q4R C C D detector. The unit cell dimensions of these crystals are a = 83.66, b = 66.03, c = 99.17 A, p angle = 94.53°. Data were processed using D E N Z O and S C A L E P A C K (Otwinowski and Minor 1997). Statistics for data collection and processing are summarized in Table 2.1. 29 2.2.3 Structure determination and refinement The Cst-II 1 - 2 5 9 structure was solved by M A D phasing from tetragonal crystals grown with protein containing incorporated SeMet using a 3-wavelength dataset. Four out of the six selenium atom positions were determined using CNS 1.1 (Brunger, Adams et al. 1998) from Patterson map, and initial phases improved by R E S O L V E (Terwilliger and Berendzen 1999) and electron density maps were improved with D M (Cowtan 1994). The initial model was built using X T A L V I E W (McRee 1999) and the final model was obtained after iterations of refinement with CNS 1.1 (Brunger, Adams et al. 1998). The sequence differs from the published sequence at amino acid position 53 (I53S). In the 6-His-tagged construct of Cst-II 1 - 2 5 9 a second random mutation occurred at position 222 (E222G). This latter position was not mutated in the tagless construct and overlap of the two structures show that this surface localized amino acid (on a solvent exposed disordered loop) has little effect on the structure. Sequence differences were confirmed by D N A sequencing. The refined Cst-II 1 - 2 5 9 structure from the P4 crystal (solvent and CMP removed) was used as the starting model for the CMP-3FNeu5Ac complex in the P2i crystal form with AMoRe (Navaza 1994) and a resolution range of 15 - 4 A . The dictionaries for substrates were generated by X P L 0 2 D (Kleywegt and Jones 1997). Both complexes were refined with CNS 1.1 (Brunger, Adams et al. 1998). Quality of the models were analyzed with PROCHECK (Laskowski, MacArthur et al. 1993) (88 % in the most favorable region of the Ramachandran plot) and summarized in Table 2.1. Residues 31 and 158, well-ordered residues in the active site, adopt disallowed main chain parameters. The coordinates of Cst-II 1 - 2 5 9 in complex with CMP and CMP-3FNeu5Ac were deposited into PDB database with code ID 1R08 and 1R07 respectively. •. • ' 30 2.2.4 Site-directed mutagenesis via PCR Both a two stage PCR mutagenesis protocol and QuikChange™ site-directed mutagenesis protocol were used to generate Cst-II 1 - 2 5 9 mutants. For the two stage PCR mutagenesis, two separate PCR reactions were performed to generate two overlapping gene fragments, one of which contained the mutation. The primers used were from the 5' (primer 1) and 3' (primer 2) ends of the gene as well as two internal primers. One internal primer contained the mutation and the other was chosen such that the two PCR products would overlap by 100 base pairs. These two products were gel purified and then used as a template for a third PCR reaction containing primers 1 and 2. This produced the full length version of the gene with the mutation incorporated. Primers 1 and 2 contained Ndel and Sail restriction sites that were used to subclone the final PCR product into pET28a. For the QuikChange™ site-directed mutagenesis, coding and antisense primers containing a single mutagenic site were used to for PCR amplification. Parental double stranded template was subsequently digested with Dpnl enzyme. Constructs were sequenced to verify the presence of only the mutation of interest. 2.2.5 Kinetic assays Kinetic studies were performed at 37 °C in 20 m M HEPES, pH 7.5 containing 0.1 % (w/v) B S A and 50 m M KC1. Transferase activity was monitored using a continuous coupled assay analogous to that previously described (Matyas and Morre 1983). Briefly, release of CMP in the reaction catalyzed by Cst-II was coupled to the oxidation of N A D H (k = 340 nm, e = 6.22 mM" 1 cm - 1) using the enzymes nucleoside monophosphate kinase, pyruvate kinase and lactate dehydrogenase. Rescue assays were performed with 500 uM 31 CMP-Neu5Ac, 3 m M sialyl-lactose, with either 500 pg mL"1 mutant enzyme or 10 pg mL"1 wildtype enzyme. The transferase activity was rescued by addition of 0, 25, 100, 250, 500, and 1000 m M sodium formate. Measurements were undertaken to ensure the consumption of contaminating nucleotide phosphates, and a stable rate of the significant spontaneous hydrolysis of CMP-Neu5Ac was established, before protein sample was added to initiate assays. Absorbance measurements were obtained using a Cary 300 U V -Vis spectrophotometer equipped with a circulating water bath. Grafit 4.0 was used to calculate kinetic parameters by direct fit of initial rates to the respective equations. 2.2.6 Overexpression and purification of 13C-histidine labeled samples The plasmids for the 6-His-tagged version of wildtype Cst-II 1" 2 5 9 and Cst-II1"2 5 9-H188A mutant were transformed into an E. coli BL21/DE3 histidine auxotroph strain hisG:Tnl0 (a gift from Dr. David Waugh in National Cancer Institute, Maryland) with a tetracycline resistant marker. 45 mL of overnight wildtype or HI 88A mutant cultures grew in L B were spun and resuspended in 1.5 L of M9 (Sigma) media prepared in D 2 O (Spectra Stable Isotopes). Cultures were supplemented with 1 m M MgS04, 0.1 m M CaCL., 1 UM FeCL;, lmg mL"1 thiamine, 1 % glucose, 50 m M kanamycin, 12 m M tetracycline, and 100 mg L" 1 ring-2-13C-L-histidine (Cambridge Isotope). Cultures were induced with 0.5 m M IPTG when the OD600 reached -0.6 and were grown further for 18 hours at 20 °C. Purification of 13C-histidine-labeled samples was performed as described above. Final protein samples were concentrated to ~10 mg mL"1 in 10 m M HEPES pH 7.5, 50 m M NaCl in D 2 0 . 32 2.2.7 pKa titration and 1 3 C-NMR data collection One-dimensional 1 3 C - N M R spectra were recorded at 35 °C for samples at pH 7.5 with a Varian INOVA spectrometer equipped with cryoprobe operating at 600 MHz for protons and 150 M H z for carbons. Subsequently, the pH of the samples was adjusted by addition of microliter aliquots of 0.5 N DC1 or NaOD to cover a range of 5.5 to 8 at ~ 0.3 intervals and monitored by a microelectrode (Ingold). Two-dimensional 'H- 1 3 C-HSQC spectra were recorded with 2400 complex points over a spectral width of 12000 Hz for protons, and 16 complex points over a spectral width of 600 Hz for carbons. The spectra were 'H-^C-decoupled and centered at 4.8 ppm and 138 ppm for protons and carbons respectively. 192 transients were collected per spectrum over a two-hour period. The data were processed with NMRPipe (Delaglio, Grzesiek et al. 1995) using a sine bell apodization and 1024-point zero-filling and visualized with Sparky (Goddard and Kneller). 33 2.3 RESULTS 2.3.1 Oligomerization and membrane attachment Bacterial sialyltransferases are typically anchored to the inner membrane via a monotopic membrane-association domain or an anchoring transmembrane helix. In order to prevent Cst-II from aggregating, a portion of the putative C-terminal membrane-association domain was deleted. A total of 32 amino acids were removed including the basic residues (Lys261, Lys262, Lys264, Lys266, Lys272, Arg279, Lys285, Lys289, and Lys291) and hydrophobic and aromatic residues (Phe260, Ile263, Ile265, Ile269, Tyr270, Tyr271, Leu273, Ile274, Leu277, Leu278, Leu280, Ile284, Tyr287 and Phe288) typical of those observed at membrane/protein interfaces. Our construct was also designed to contain an Ile53Ser mutation which had previously been shown to enhance a-2,8-sialyltransferase specificity and stabilize the enzyme (Gilbert, Karwaski et al. 2002). The resulting truncated protein (Cst-II 1" 2 5 9 , 31 kDa) retains complete activity when compared to the full length enzyme and was found to form a tetramer in solution (-124 kDa) as identified by static light scattering analysis (Figure 2.1). 1 2 V o l u m e (m l ) Figure 2.1 SDS-PAGE and static light scattering analysis of Cst-II 1-259 34 Molecular weight of Cst-II 1" 2 5 9 as a denatured state and in-solution state was checked with S D S -P A G E (denatured) and static light scattering (in-solution). For static light scattering, the refractive index (the horizontal line) of the sample was measured as the sample was eluted from a size-exclusion column. The concentration of the sample eluting from the column is indicated by the bell shaped curve. Well-ordered crystals were produced in the presence of different ligands, namely CMP, an inert analogue of the donor sugar CMP-3-fluoro-A7-acetyl-neuraminic acid (CMP-3FNeu5Ac), and CDP. Two different space groups were observed in these crystals, P4 for the Cst-II1 " 2 5 9 - C M P complex, and P2i for both the Cst-II1 " 2 5 9 - C M P -3FNeu5Ac and Cst-II1 " 2 5 9 -CDP complexes. In the latter, there is a tetramer in the. asymmetric unit of the crystal, agreeing with the results from static light scattering in solution. In the P4 space group, there are two molecules in the asymmetric unit, with the 4-fold crystallographic axis creating two similar tetramers to that observed in the asymmetric unit of the monoclinic space group (r.m.s. deviation for the 708 Coc atoms is 0.34 A; Figure 2.2). There is approximately 1400 A 2 of buried surface area between each pair of monomers in the tetramer with a large number of stabilizing hydrophobic residues (from Phe56, Tyr60, Tyr98, Phe99, ProlOO, Pro246, Tyr251, Lys253 and Phe254 of one monomer and His85, A l a l l 7 , Phel21, Ilel24, Tyrl25 and Phel26 of another) and hydrogen bonding (Phe56 O to Tyrl25 OH, Asp97 OD1 to Asn82 ND2, Asp97 OD2 to Tyrl06 OH, Asp97 O to Lysl20 NZ, Lys253 N Z to Glul23 OE1 and O, and Lys253 O to Asnl27 N) interactions. The C-terminii of the four monomers form an independent folding domain and align on the same side of the tetramer, suggesting a role for the oligomer in effective attachment to the bacterial inner membrane (Figure 2.3). 3 5 Table 2.1 Data collection and structure refinement statistics. Data collection Derivative (Se-Met) Complex Peak Inflection Remote C M P - C D P Crystal parameters space group P4 P2i P2i Diffraction statistics Beamline X25 X8-C X8C Resolution 30-2.1 30-2.3 30-2.3 30-1.8 30-2.0 Wavelength ' 0.97912 0.97961 0.97900 1.00000 1.00000 Total reflections 363065 98127 99539 429958 296053 Unique reflections 62730 41979 42343 99748 73988 Redundancy 5.8 2.3 2.4 4.3 4.0 Completeness (%Y 93.7 (68.8) 88.9 (55.3) 89.7 (58.7) 99.6 (98.5) 98.4 (90.1) <l/ol>1 23.9 (10.6) 26.6 (11.2) 28.3 (11.5) 23.6 (2.0) 19.1 (2.0) 5.7 (9.1) 3.1 (6.3) 3.1 (6.4) 5.4 (52.6) 5.8 (45.9) Refinement Statistics and model stereochemistry Resolution (A) 30-2.1 30-1.8 30-2.0 Number of Atoms Protein 3948 8164 7448 Substrate 41 167 75 Water 89 334 160 Rcrys/Rfree (%)"* 21.9/25.6 21.9/25.2 25.2/29.4 R.m.s. deviations Bonds (A) 0.006 0.007 0.007 Angles (°) 1.17 1.39 1.14 Average B-factor (A 2) , Protein 39.4 34.7 44.1 Substrate 44.2 43.3 56.3 Water 35.8 33.0 38.3 36 High resolution shell (2.18 A - 2.10 A for C M P complex, 1.86 A - 1.80 A for C M P -3FNeu5Ac complex and 2.08 A - 2.00 A for C D P complex) statistics are in parentheses Rsym =A(lhki)-<l>V£( ,hki), where Ihki is the integrated intensity of a given reflection. RCryst=(Z\F0-Fc\)/(£F0), where F0 and Fc are observed and calculated structure factors. 5% of total reflections were excluded from the refinement to calculate Rfree. Figure 2.2 Arrangement of the Cst-II1'259 tetramer. Each monomer is colored differently. CMP-3FNeu5Ac is shown as a stick representation in magenta indicating the location of the catalytic centre. 37 Figure 2.3 Schematic m e m b r a n e a t t a c h m e n t of Cst-II1"259. All four terminii of four Cst-II monomers are on the same side of the tetramer. The truncated 32 amino acids at the C-terminii are hypothesized to form a membrane attachment domain and are depicted as solid cylinders 2.3.2 Overall architecture Each monomer of C s t - I I 1 - 2 5 9 consists of 259 residues organized into two domains (Figure 2.4). The first domain (residues 1 to 154, 189 to 259) is composed o f a mixed a/p fold. The central, parallel seven-stranded twisted p-sheet (topology P8, P7, p i , p2, P4, P5, P6) is flanked by five helices on one side (D, E , F, I and J) and four helices (A, B , C and K ) on the other. Hel ix A , the N-terminal region of helix B , helix H , helix J, and the C-terminal region of helix K adopt a 3io-helical conformation. Helices F, I, and part of the P-sheet (p8, P7, p i , P2, and P4), create the nucleotide-binding domain in the form of a Rossmann fold (Rossmann and Argos 1978). The binding site of the nucleotide sugar is located in a coil region at the edge of the beta sheet, and is on the same side of the sheet as the C-terminus. The active site is not at an interface between monomers of 38 the tetramer, suggesting that the oligomerization of the enzyme has no direct role in catalysis. The second, smaller domain (residues 155 to 188) is composed of a coil extending from 07, followed by a-helix G and 3io-helix H , and a coil that links back to helix I. This domain forms a lid-like structure that folds over the active site. We observe from our structures that a region o f this domain (residues 175 to 187) only becomes ordered upon the binding of the intact CMP-s ia l ic acid substrate. Figure 2.4 View of Cst-II1"259 monomer showing the N-terminal domain and the lid domain with bound donor sugar analogue. CMP-3FNeu5Ac is depicted in a red stick representation. The N-terminus, C-terminus, individual strands and individual helices are labeled. 39 From the currently published structures, glycosyltransferases have been categorized into two groups based on overall fold. The GT-A (for GlycosylTransferase A) group is composed of oc/p proteins with a single Rossmann domain, a conserved D X D (Asp-X-Asp) motif and a conical active site cleft formed by two closely associated domains. The GT-B group consists of two Rossmann domains separated by a deep substrate-binding cleft (Unligil and Rini 2000; Breton, Snajdrova et al. 2006). Since Cst-II is composed of a single Rossmann domain it belongs to the GT-A group. However, significant differences of Cst-II from other members of the GT-A group are observed in terms of the connectivity of the secondary structural elements and the absence of the conserved D X D motif. Structural alignment of Cst-II 1 - 2 5 9 against all reported glycosyltransferase structures (both inverting and retaining) was performed with the Secondary Structure Matching server of the European Bioinformatics Institute ( The closest match obtained was that of rabbit muscle glycogenin (Gibbons, Roach et al. 2002) (PDB code 1112) with a r.m.s. deviation of 4.46 A for 136 equivalent C a atoms (primarily the Rossmann domains). A general search of the D A L I database ( dali/index.html) indicates that the closest match in the structural database to Cst-II is the Rossmann-fold containing thiamin pyrophosphokinase (Timm, Liu et al. 2001) (PDB code lig3) with a relatively weak Z-score of 6.4 and a r.m.s. deviation of 3.8 A for 127 equivalent C a atom positions. 2.3.3 C M P binding Cst-II 1 - 2 5 9 was first crystallized in the presence of both the donor sugar, CMP-Neu5Ac, and the acceptor sugar analogue, sialyl lactose. However, in the electron 40 density map only CMP was present, suggesting that the sialic acid moiety of the CMP-Neu5Ac had either been transferred or hydrolyzed by the wild type enzyme or that the sialic acid moiety was disordered in the crystal structure. Indeed, kinetic analysis clearly shows significant hydrolysis of CMP-Neu5Ac in the absence of an acceptor catalyzed by both the truncated and the full length Cst-II. CMP binds to a deep cleft in the nucleotide-binding domain at the C-terminal end of the central (3-sheet. This location of the active site at the protein/membrane interface presumably facilitates transfer of the sugar onto the terminus of the LOS, which is assembled on the cytoplasmic face of the inner membrane and transported onto the cell surface via an A B C transporter (Karlyshev, Linton et al. 2000). CMP is relatively buried within the active site cleft (-630 A 2 or 68 %), with several favorable interactions formed within the active site (Figure 2.5): the cytidine ring is held in position by aromatic stacking with the conserved Tyrl56 and the cytidine carbonyl 02 forms hydrogen bonds with main chain nitrogen atoms of Asp 154 and Phel55. N4 of the base donates a hydrogen bond to a main chain carbonyl of Serl61. The ribose ring adopts a C2'exo conformation with the cytidine base axial to the ring. 02 ' interacts with the OG1 of Thrl31 and the main chain amide of Glyl33 while 0 3 ' interacts with main chain amide of Serl32. The phosphate is stabilized by hydrogen bonds to the hydroxyl groups of Tyrl56 and Tyrl62, ND2 of Asn31, and by a water-mediated hydrogen bond interaction with OD1 of Asn9. Interestingly, this mode of CMP binding is reminiscent of the ring stacking and phosphate coordination observed for Tyr75 within the A M P binding site of glycogen phosphorylase (Stura, Zanotti et al. 1983). An extended loop of Cst-II 1 - 2 5 9 (residues 175 to 187), which presumably lies adjacent to the active site, is highly disordered in the Cst-II 1 _ 2 5 9 -CMP complex. 4 1 Figure 2.5 Interactions of CMP and active site residues. CMP is depicted in CPK coloring with carbon atoms in magenta, nitrogen atoms in blue, oxygen atoms in red, and phosphorus atom in green. Active site residues involved in CMP binding and catalysis are labeled and shown with carbon atoms in beige, nitrogen in blue and oxygen in red. Dotted lines indicate hydrogen bonding. 2.3.4 CMP-3Fneu5Ac binding In order to overcome the problem of hydrolysis o f the donor sugar, and to allow stable complex formation with acceptor sugars, the inert donor substrate analogue C M P -3FNeu5Ac (Figure 2.6a) was synthesized and utilized to form co-crystals with the truncated wi ld type enzyme. Substitution of the hydrogen at the C 3 ' position on the sialic acid with an electronegative fluorine atom prevents turnover by inductively destabilizing the oxocarbenium ion-like transition state for the reaction (Burkart, Vincent et al. 2000). Furthermore, this also stabilizes the otherwise labile sugar donor against spontaneous hydrolysis. Kinetic and mass spectroscopic analysis showed that CMP-3FNeu5Ac is not 42 hydrolyzed by our Cst-II 1 - 2 5 9 construct in solution. The measured K{ value for the binding of this inert sugar to Cst-II 1" 2 5 9 (K\ = 657 uM) is comparable to the KM value for the natural substrate CMP-Neu5Ac (KM = 460 uM). The near equivalence of these numbers is unsurprising given that the fluorine has replaced a hydrogen atom, and that no close interactions to the enzyme are seen at this position (Figure 2.6a). While there is very little difference in overall, structure between the CMP and CMP-3FNeu5Ac complexes (r.m.s. deviation ~ 0.2 A on 246 C a atoms) there is a substantial local change involving the ordering of residues 175 to 187, thereby creating an effective lid which closes oyer and substantially buries the donor sugar. This lid likely plays a number of roles in catalysis; directly liganding the donor sugar, creating the acceptor sugar binding site and shielding the enzyme active site from bulk solvent, thereby minimizing undesirable side reactions such as hydrolysis of substrate. Indeed, a similar ordering of the active site has also been observed in other glycosyltransferase studies (Unligil, Zhou et al. 2000; Persson, Ly et al. 2001; Qasba, Ramakrishnan et al. 2005). 43 Figure 2.6 The donor substrate analogue CMP-3FNeu5Ac . (a) Chemical drawing of CMP-3FNeu5Ac. (b) Observed electron density of CMP-3FNeu5Ac in a refined 2Fo - Fc map contoured at 1.5 a. (c) Interactions of CMP-3FNeu5Ac and key active site residues of Cst-II 1" 2 5 9. CMP-3FNeu5Ac is depicted in C P K coloring with carbon atoms in magenta, nitrogen atoms in blue, oxygen atoms in red, phosphorus atom in green, and fluorine atom in light blue. Active site residues involved in CMP-3FNeu5Ac binding and catalysis are labeled and shown with carbon atoms in beige, nitrogen in blue and oxygen in red. Water molecules are 44 shown in cyan spheres. Helix, F is highlighted as a blue ribbon. Dotted lines indicate hydrogen bonding. Well-ordered electron density defines C M P - 3 F N e u 5 A c in the active site of Cst-I I 1 - 2 5 9 (Figure 2.6b), with the C M P nucleotide moiety binding in a very similar manner to that seen in the C M P complex, despite the presence of the now ordered l id domain. The only significant difference is a C 3 ' endo conformation of the ribose resulting in a more equatorial position of C 5 ' which consequently positions the remaining chain of phosphate and sialic acid in contact with active site residues. The sugar ring of the sialic acid, Neu5Ac, adopts a distorted conformation best described as °Ss skew boat conformation, with C 2 " and the carboxylate group in the .. same plane as C 3 " and the ring oxygen (Figure 2.6b,c). Although this conformation of the sugar is of higher energy than the normal chair or boat conformation, it has also been observed in sialic acid complexes with Trypanosoma rangeli sialidase (Buschiazzo, Tavares et al. 2000) (PDB code In ly ) and Influenza B virus neuraminidase (Burmeister, Ruigrok et al. 1992) (PDB code lnsc). The Neu5Ac carboxylate oxygens in our structure are highly coordinated by neutral amino acid side chains in the enzyme. One oxygen of the carboxylate hydrogen bonds with the conserved side chains of Asn31, Asn51 and Serl32 and the other oxygen hydrogen bonds with the main chain amide of Serl32 as well as with 0 3 ' of the ribose. In addition, the carboxylate resides at the N-terminus o f the extended a-helix F (initiated by Serl32). This structural feature provides extra electrostatic stabilization of the negatively charged carboxylate through the positive charge arising from the helix dipole. Surprisingly, the observed coordination of the sialic acid carboxylate in our sialyltransferase structure is very different from that observed 45 previously in sialidase (Burmeister, Ruigrok et al. 1992; Crennell, Garman et al. 1993; Crennell, Garman et al. 1994; Gaskell, Crennell et al. 1995; Crennell, Garman et al. 1996; Luo, L i et al. 1998; Luo, L i et al. 1999; Crennell, Takimoto et al. 2000) and trans-sialidase (Amaya, Buschiazzo et al. 2003) structures. In these enzymes, a highly conserved arginine triad directly coordinates the carboxylate of sialic acid and seems to play a role in anchoring the substrate during its distortion along the reaction coordinate. Other key interactions observed in our Cst-II 1 _ 2 5 9-CMP-3FNeu5Ac structure include hydrogen bonds between 04" and the main chain amide of Tyrl85, and between the glycerol moiety of the sugar and the enzyme active site: 07" with ND2 of Asn51, 08" and 0 9 " with each of the side chain atoms of Gln32, and 0 9 " with NE2 of Gln58. A hydrophobic interaction of Phel78 with the iV-acetyl group of the sialic acid moiety provides further stabilization and an ordered water molecule mediates the interaction of the nitrogen of the Af-acetyl group with OG of Serl83. The introduced substituent F3", is 2.6 A away from NE2 of Hisl88. Several of the interactions of the sugar moiety with the protein involve residues of the flexible lid domain that is disordered in the Cst-II 1 - 2 5 9 complex with CMP alone. These interactions include electrostatic and hydrophobic contacts (Figure 2.6c), explaining why this loop is disordered in the absence of the sugar moiety. Ordering of the lid domain likely creates the acceptor sugar binding site, a significant cleft adjacent to Neu5Ac in our structure (Figure 2.4, 2.11). Interestingly, the I53S mutation which promotes stabilization of our Cst-II construct lies on the periphery of this cleft where it may stabilize the enzyme via interactions with the acceptor sugar. 46 A sialyl-binding motif L has been previously proposed based on sequence analysis of mammalian sialyltransferases and mutagenesis studies in rat liver alpha-2,6-sialyltransferase (Datta and Paulson 1995). The proposed motif contains 8 residues (Cysl81, Vail84, Leul90, Glyl96, Arg207, Asp219, Val220 and Gly221 in the rat enzyme). However, our structure shows that the majority of these positions would not be predicted to line the sialic acid binding site (Val4, Ala7, Leul3, Serl9, Arg29, Tyr40, Leu41, and Gly42 in Cst-II). Only Ala7 (Vail84 in rat) and Arg29 (Arg207 in rat) are in relatively close proximity to the active site, but are still too far away to bind substrate directly (7 and 9 A from the donor sugar, respectively). Analysis of our structure indicates that at least some of the conserved residues in the sialyl motif L may play an indirect structural role in substrate binding, potentially aiding in correct positioning or stability of adjacent (but less conserved) amino acids in the active site cleft. For example residues 185-487, which follow the conserved V a i l 84 in the rat enzyme, correspond to Gly8, Asn9 and GlylO in our Cst-II 1" 2 5 9 structure, all of which line the bottom of the sialic acid cleft. Similarly, arginines 208 and 209, which follow the invariant Arg207 in the rat enzyme would correspond to Cys30 and Asn31 in Cst-II, both of which line the side of the CMP-binding pocket. Unlike other glycosyltransferases with a GT-A fold, no bound metal was observed in the active site of Cst-II 1" 2 5 9. Furthermore, Cst-II lacks the ' D X D ' sequence motif found in a wide range of glycosyltransferases. This motif is known to coordinate the divalent cation involved in binding of the nucleotide sugar via interaction with the diphosphate moiety (Wiggins and Munro 1998; Hagen, Hazes et al. 1999). The metal is generally considered to act as a Lewis acid catalyst in the reaction mechanism by 47 stabilization of the leaving nucleoside diphosphate. Mutagenesis studies of the conserved Asp residues in the D X D motif of glycosyltransferases from various species have shown that the removal of the Asp residues, and hence the metal ions, completely eliminates transferase activity (Busch, Hofmann et al. 1998; Shibayama, Ohsuka et al. 1998; Wiggins and Munro 1998; Hagen, Hazes et al. 1999). Although kinetic studies with Cst-II 1" 2 5 9 indicate that both magnesium and manganese enhance the activity somewhat (~ 50 % activity increase for both metal ions), they are not essential for catalysis. Such differences are unsurprising given that the donor substrate in this case is a nucleoside monophosphate sugar (CMP-Neu5Ac), thus no diphosphate moiety is present for metal ion binding. Cst-II presumably uses other means to activate/stabilize the leaving nucleotide (discussed below). 2.3.5 C D P binding CDP has previously been shown to be an effective sialyltransferase inhibitor. For the rat a-2,6-sialyltransferase a K\ of 4.7 uM has been determined (Datta and Paulson 1995) (a value lower than the reported KM of 50 uM), indicating its effectiveness as a competitive inhibitor. Unfortunately, since our continuous assay system measures the activity of Cst-II based on the release of CMP which is then converted to CDP by a coupling enzyme, it was not possible to determine the K\ value of CDP with Cst-II1"2 5 9. Cst-II 1" 2 5 9 crystallizes with CDP under the same conditions as the other substrates and was observed to occupy the same CMP binding site. The cytidine moiety of CDP binds to the protein in the same manner as for the nucleoside monophosphate (Figure 2.7). The ribose ring adopts the same C2' exo conformation and the a-phosphate is in the same 48 position as for the C s t - I l ' " 2 5 9 - C M P complex. The P-phosphate occupies a highly similar position to that of the sialic acid carboxylate in our C M P - 3 F N e u 5 A c structure, forming a favourable electrostatic interaction with the N-terminal dipole of a-helix F and forming strong hydrogen-bonded contacts to the side chains of the conserved Asn31, Asn51, and Serl32 residues. These interactions, which mimic that of the donor substrate, presumably provide C D P with its inhibitory activity. Figure 2.7 Interactions of C D P and active site res idues CDP and enzyme are coloured as in Figures 2.5 and 2.6. Dotted lines indicate hydrogen bonding between the inhibitor and the enzyme. 49 2.3.6 Kinetic analysis of active site mutants A continuous coupled assay system was developed (Figure 2.8) to test the activity , of Cst-II. Kinetic analyses of key active site mutants are presented in Table 2.2. Careful attention was required during kinetic analysis to ensure that rates being determined do in fact reflect the transferase activity of Cst-II. A significant rate of spontaneous background hydrolysis of CMP-Neu5 Ac was observed in the absence of enzyme and was subtracted from all enzyme-catalyzed activities to arrive at true enzyme-catalyzed turnover rates. Cst-II 1" 2 5 9 was also found to catalyze the hydrolysis of CMP-Neu5Ac at an appreciable rate, as determined by incubating Cst-II 1" 2 5 9 with CMP-Neu5Ac in the absence of acceptor (Table 2.2). To ensure that the observed hydrolytic activity was not due to a transfer of sialic acid from CMP-Neu5 Ac to an additional molecule of donor to form CMP-Neu5Ac-Neu5Ac, a result that would not be completely unexpected considering the known bifunctional acceptor specificity, mass spectrometric product analysis was performed and indeed confirmed that no such transfer product was formed, with sialic acid and CMP formed instead (data not shown). The quantification of each of these three possible transformations of CMP-Neu5Ac proved particularly important during the analysis of several Cst-II 1" 2 5 9 mutants. For the H188A and R129A variants, no significant transferase activity above that of enzyme-catalyzed hydrolysis was observed, thus kinetic parameters reflect the hydrolytic process, as confirmed by product analysis using T L C and mass spectrometry (data not shown). No hydrolytic activity could be detected for either the Y156F or the Y162F point mutants, therefore the rate constants reported for these variants represents transferase activity solely. Additionally, the approximate 100 fold decrease in k c a t values resulting from each of these mutations 50 appears to be additive as the Y156/162F double mutant displayed neither transferase nor hydrolytic activity above that of spontaneous hydrolysis (Table 2.2). C M P . NADH NADf Figure 2.8 Coupled kinetic assay system developed for Cst-II Cst-II catalyzed reaction (step 1) is the rate-limiting step in this assay. C M P generated is converted to C D P by myokinase (step 2). C D P is then coupled with phosphoenolpyruvate to generate pyruvate by pyruvate kinase (step 3). The pyruvate is then used by lactose dehydrogenase to convert NADH to N A D + (step 4). The rate of NADH generation is measured by monitoring the decrease in A ^ a n d the decrease in absorbance corresponds to the rate of C M P generation. 51 Table 2.2 Michaelis-Menten parameters for CMP-Neu5Ac with various Cst-II1'259 mutants. Cst-I I 1 " 2 5 9 kcat K M 1 ( S . E . ) kCat/ KM (min - 1) (^M) (min - 1 M"1) W T 43 (±5) 4.6 x 10 ' (± 1 x 10') 9.4 x 10 4 W T Hydrolysis' 1.7 (±0.09) 4.7 x 10 ' (±0 .5x 10') 3 . 6 x 1 0 J R129A J <0.20 (±0.01) 2.6 x 10 ' (±0.4 x 10') 7.8 x 10 ' H188A J <0.16(±0.03) 3.0 x 10 ' (±1.1 x 10') 5 . 3 x 1 0 ' Y156F 4 0.58 (±0.07) 2 . 2 x 1 0 ' (±0 .5x10 ' ) 2.6 x10 3 Y162F 4 0.12 (±0.01) 4.3 x 10 ' (±0.9x 10') 2 . 8 x 1 0 ' Y156/162F S <0.040 n.d. — Kinetic parameters were determined at a constant lactose concentration of 160 mM 2 For enzyme catalyzed hydrolysis kinetic parameters were determined in the absence of acceptor 3 No significant activity above that of enzyme catalyzed hydrolysis (measured in the absence of acceptor) was observed, thus kinetic parameters reflect the hydrolytic process, as confirmed by product analysis using T L C and mass spectrometry (data not shown). 4 No hydrolytic activity could be detected, thus the rate represents transferase activity. 5 No activity above that of the significant rate of spontaneous background hydrolysis was detected up to an enzyme concentration of 500 ng m L - 1 . 2.3.7 Histidine titration ' H N M R spectra revealed that the 13C-histidine was successfully incorporated and that the protein samples were properly folded; however, protein samples were found to be constantly precipitating. Un-rresolved peaks were observed in the two-dimensional ' H -13 C-HSQC spectra (Figure 2.9a). The strongest signal in Figures 2.9a,c,d is likely the peak for the flexible histidine at the N4erminus of the protein resulted from the thrombin cleavage site (Gly-Ser-His). The three most distinct peaks were assigned arbitrarily and their migrations in response to pH change are plotted in Figure 2.9b. Of these peaks, no perturbation in normal histidine pKa values is observed. Comparisons between wildtype 52 and mutant H188A 'H- 1 3 C-HSQC spectra did not distinguish a specific peak for Hisl88 (Figure 2.9c, d). 53 (a) A> 137.0 B.35 8.30 8.25 8.20 <62 - 1 3 C (ppm) (b) x -*-RskA -* \ 1 * v . 29930 2553) 23600 Hz 2565) 25700 8.2 8.1 8.0 - 7.9 (c) (d) 8.< 8,0 7.9 7.8 <u2 - , 3 C <ppm) 8.0 7.9 7.8 137.0 139.0 Figure 2.9 2D HSQC spectra of [ring 2-13C]histidine-labeled Cst-II1'259 (a) H S Q C spectrum of wildtype Cst-II 1" 2 5 9 at pH 6.5. Three peaks were assigned A, B, and C. '(b) Movement of these peaks in pH titration. X-axis value is calculated with the formula 5(Hz) =5 1 H*freq( 1H) + 6 1 3 C * f req( 1 3 C) where freq(1H) = 599.738 and freq( 1 3C) = 150.825. (c) H S Q C spectrum of wildtype Cst-II 1" 2 5 9 at pH 7.0. (d) H S Q C spectrum of H188A mutant at pH 7.0 54 2.4 DISCUSSION A l l structural and kinetic evidence for inverting glycosyltransferases thus far supports a direct displacement mechanism through an oxocarbenium ion-like transition state with general base assistance. Thus Cst-II presumably catalyzes two such direct displacements. In the first reaction the 3-hydroxyl of the galactose acts as the nucleophile, directly attacking the anomeric centre (C2) of CMP-Neu5Ac to form NeuAca-2,3-Gaip-1,3-GalNac with the release of CMP. In the second reaction the 8-hydroxyl from the sialyl moiety of this newly formed sugar now acts as the nucleophile, attacking the anomeric centre of a second CMP-Neu5Ac bound in the active site (Figure 1.10, Table 2.3). Key questions about this mechanism concern the identities of the general base catalyst that deprotonates the attacking nucleophile, and that of any acid catalyst that assists CMP departure, as well as how the.anticipated oxocarbenium ion-like transition state is stabilized. Table 2.3 Michaelis-Menten parameters for acceptors with wildtype Cst-II1"' Acceptor kcat KM k c a t / K M (min - 1) (mM) ( m M - 1 min - 1 ) Lactose 38 (±1) 35 (±3) 1.1 3'-Sialyl lactose 55 (±1) 3.5 (±0.3) 16 - Kinetic parameters were determined at a constant CMP-Neu5Ac concentration of 1 mM. The structure of Cst-II 1" 2 5 9 in complex with CMP-3FNeu5Ac provides some answers to these key questions. The distorted skew boat conformation of the sialyl moiety places the leaving group phosphate in a pseudo-axial position suitable, for 55 departure upon attack of the nucleophilic hydroxyl group. Such an axial disposition for the leaving group is a universally observed feature for both transferases and glycosidases at each step in catalysis. Distortion of the sialyl moiety to a skew boat places 06, C2, C I , C3 and C4 in a planar arrangement as required for oxocarbenium ion formation. Substrate distortion upon binding to enzymes has also been observed previously in crystal structures of glycosidases (Vasella, Davies et al. 2002). The leaving phosphate is twisted above the ring plane such that one of the non-bridging oxygens - the pro R - interacts with the ring oxygen at a distance of 2.6 A (reminiscent of the interaction of the carbonyl oxygen of the nucleophile Asp229 with the ring oxygen in the glycosyl-enzyme intermediate formed by cyclodextrin glucanotransferase (Uitdehaag, Mosi et al. 1999). In both cases, the breakage of the glycosidic bond could be facilitated by the negative charge build-up on the leaving group oxygen, which serves to stabilize the developing oxonium ion character at 06. Departure of the phosphate could also be facilitated by the hydrogen bonding interactions of Tyrl56 and Tyrl62 with the non-bridging pro S oxygen. Discrete acid catalysis is not required for transferases since the second ionization of the leaving group monophosphate has a pK a of around neutrality, in contrast to the alcohol leaving group of pKa «16 released by glycosidases. Indeed the situation is more equivalent to the second step of a retaining glycosidase wherein the leaving group is an enzymatic carboxylate. In these latter examples tyrosine residues also often play roles in stabilizing the departing leaving group (Gebler 1995). Mutational analysis shows that kcat values for the Y156F and Y162F mutants of Cst-II are 75-fold and 360-fold lower than those for the wild type enzyme, and their simultaneous mutation results in the complete loss of any activity above that of spontaneous hydrolysis (Table 2.2). 56 The identity of the general base catalyst is unclear. In glycosidases, including sialidases, this role is played by the carboxylate group of a Glu or Asp residue. However, there are no suitably disposed acidic groups in the active site of Cst-II 1" 2 5 9, the closest being Glu57, which is 14 A away. The closest side chains to the anomeric carbon are Asn31 (3.1 A), Asn51 (4.0 A), Serl32 (3.0 A) and Hisl88 (4.8 A). Of these, the only feasible candidate as suitable base is Hisl88 which lies appropriately on the alpha face of the sugar and adjacent to an open cleft in the enzyme, which would presumably be the most obvious site to bind the acceptor sugar (Figure 2.6c). It is thus well-positioned for the imidazole side chain to act as the general base in catalysis, deprotonating the incoming hydroxyl group of the acceptor galactose during nucleophilic attack at the anomeric carbon of the donor sugar (Figure 2.10). Despite the lack of overall structural similarity and sequence identity, a comparison of the active site of Cst-II 1" 2 5 9 to that of the inverting human pl,3-glucuronyltransferase (GlcAT-I; PDB code lfgg), based on the position of the anomeric centres of the donor sugars, shows that Hisl88 is in an equivalent position to Glu281, the proposed catalytic base of GlcAT-I (-5.6 A away from the anomeric center) (Pedersen, Tsuchida et al. 2000). Our kinetic analysis shows the optimal activity of Cst-II occurs at approximately pH 8. At this pH, the imidazole of His 188 would be expected to be deprotonated, as required for a general base catalyst. Upon proton transfer, the positive charge that develops on the imidazole side chain could be stabilized through electrostatic interactions with the negatively charged phosphate and carboxylate groups of the donor sugar substrate (-5 A away) (Figure 2.10). Mutation of His 188 to alanine results in a loss of all detectable transferase activity, indicating its key catalytic role in the transferase mechanism of Cst-II (Table 2.2). In order to confirm the 57 identity of the general base in the Cst-II-catalyzed.reaction, a histidine mutant rescue assay was performed. HI 88A mutant was able to show transferase activity upon addition of formate. The rescue assay supports our hypothesis that the histidine is well positioned to extract a proton from the acceptor. Our structure shows a conserved arginine residue (Argl29) which lies 5.9 A away from Hisl88 and thus could act to provide an electrostatic shield favoring the deprotonated form of the putative general base His 188 (at physiological pH, the imidazole side chain may not be completely deprotonated without such assistance, dependent on its intrinsic pK a ) . Argl29 is also suitably positioned within the active site to play a potential role in orienting the incoming acceptor sugar. Accordingly, our kinetic analysis shows that mutation of Argl29 to alanine decreases the k c a t value by 210-fold and, similarly to His 188, that the residual "activity" results from enzyme catalyzed hydrolysis (Table 2.2). Interestingly, rate constants for hydrolysis by both HI 88A and R129A variants are ten-fold lower than those for the wild type enzyme implying that the normal catalytic machinery is recruited for hydrolysis (Table 2.2). Hisl88 is thus proposed to be the general base in the catalysis with Argl29 as its pKa modulator (Figure 2.10). Attempts to measure the pKa value of His 188 with N M R titration were performed. There are a total of six histidine residues for Cst-II 1" 2 5 9. Based on our hypothesis, Hisl88 would have a lower pKa value than the other five histidine residues due to the presence of Argl29. Unfortunately, due to the large molecular weight of the Cst-II tetramer in solution (124 kDa), broadening of peaks was observed and comparison of wildtype and HI88A mutant spectra was inconclusive. 58 Figure 2.10 Proposed reaction mechanism for Cst-II His188 is proposed to be the general base of the reaction. Its pKa value is modulated by Arg129. The imidazole side chain of His188 is completely deprotonated and can subtract the hydrogen from 03 of the galactose. The resultant positive charge can be stabilized by the substrate carboxylate group as well as the phosphate group. The oxygen on galactose can then attack the C2 anomeric center of the donor CMP-Neu5Ac. The binding of the donor sugar ring in the flatten transition-state like conformation also favors the forward reaction. Once the reaction is completed, the new glycosylic bond is formed (red). The enzyme is regenerated once the product leaves the active site. Another possible candidate for the general base catalyst could be the carboxylate moiety of the substrate itself. Support of this contention might be the complete absence of any basic amino acid side from the immediate vicinity o f the carboxylate in contrast to the three arginines observed in sialidases. However, the in-plane conformation of the carboxylate seen here is not consistent with this notion since a role as general base catalyst would require that the carboxylate be bound perpendicular to the ring plane. 59 While it is still possible that the conformation would change upon formation of a ternary complex with the acceptor sugar, it is interesting to note that this near-planar conformation of the carboxylate is consistent with ab initio calculations and kinetic isotope effect (KIE) experiments, present in the transition state for the formation of a sialyl-oxocarbenium ion during the spontaneous hydrolysis of CMP-Neu5Ac (Horenstein 1997). Additionally, more recent KIE experiments have cast serious doubt on the ability of the carboxylate of CMP-Neu5 Ac to function as an intramolecular base catalyst during spontaneous hydrolysis (Horenstein and Bruner 1998). With the three-dimensional structure in hand it should be possible to gain insight into the structural basis for the different specificities of the mono- and bifunctional Cst-II enzymes. Kinetic analysis (Table 2.3) reveals that k c a t values of Cst-II 1" 2 5 9 for lactose and sialyl lactose are similar, but that KM values differ by a factor of 10. A comparison of the sequences between mono- and bifunctional Cst-II shows only eight amino acid differences: Asn51Thr, Ile53Asn, Leu54Phe, Aspl l4Glu , Lysl69Glu, Asnl77Asp, Asnl82Arg and Ile269Val (the amino acid codes before the numbers represent residues from bifunctional Cst-IIoH4384 and those after the numbers represent residues from monofunctional Cst-IIo:i9). Based on the structure of Cst-II 1" 2 5 9 from strain OH4384, the structural and/or mechanistic roles of each of these eight amino acids can be extrapolated. Previously, site-specific mutations at Asn51 were shown to be critical for the bifunctionality of Cst-II (Gilbert, Karwaski et al. 2002). In the complex structure, this residue forms a direct interaction with the donor sugar carboxylate group and is open to the acceptor sugar binding cleft, potentially playing a role in distinguishing between the two acceptor sugars, namely Gaip-l,3-GalNAc and NeuAca-2,3-Gaip-l,3-GalNAc 60 (Figure 2.11). Ile53 (Ser53 in our construct) also lines the acceptor cleft, thus the hydrophobicity of Ile53 may stabilize the N-acetyl or the methylene group of the sialyl moiety, facilitating its binding. Leu54 is in close proximity to the acceptor sugar binding cleft and is buried inside the core of the enzyme. Substitution of Leu54 by a bulkier phenylalanine could well disrupt the architecture of the binding cleft. Both A s n l 7 7 and A s n l 8 2 lie in the flexible l id domain of the enzyme. The side chain of Asn l77 faces toward the binding cleft while that for A s n l 8 2 faces towards the exterior of the enzyme. Substitution of A s n l 7 7 with aspartate could potentially disfavour the binding of sialyl galactose through charge repulsion with the carboxylate moiety. Asp 114, L y s l 6 9 and Ile269 lie outside the active site and it seems unlikely, based on the structure, that they would play a role in differentiating between the two acceptor sugars. 6 1 Figure 2.11 Mapping of the amino acids that differ between mono- and bifunctional Cst-II in the active site of bifunctional Cst-II Surface representation of the Cst-II1"259 centered on the active site. The white molecule is the donor analogue and the yellow dashed circle highlights the potential acceptor binding cleft. Five of the eight amino acids that differ between mono- and bifunctional Cst-II are in close proximity to the active site and are mapped here. This structure of a sialyltransferase in complex with an inert substrate analogue provides our first insight into the architecture and mechanism of this important class of enzymes and w i l l be useful for both understanding the basis of the regio- and acceptor specificity in the various types o f sialyltransferases, and as a foundation for the design of novel sialyltransferase inhibitors. 62 CHAPTER 3 - Structural investigation of Cst-II mutant from directed evolution 3.1 I N T R O D U C T I O N Complex carbohydrates occur in a wide range of contexts in biology, including polysaccharides, proteoglycans, glycoproteins, glycolipids and antibodies. They play important roles in a number of functions, including: cell growth, cell-cell interactions (Crocker and Feizi 1996), immune defense (Rudd, Elliott et al. 2001), inflammation (Lowe 2003) and both viral and parasitic infections (Sacks and Kamhawi 2001). Assembly of these complex structures is orchestrated by a series of specific glycosyltransferases which sequentially transfer the monosaccharide moieties of their activated sugar donor to the required acceptor, with the correct positional and stereochemical outcome (Qasba, Ramakrishnan et al. 2005). Consequently a large number of GTs exist, with widely different specificities. The prospect of engineering GTs to generate enzymes of desired specificity is therefore very promising. This is important since, by contrast with the situation for peptides and oligonucleotides, the chemical synthesis of complex carbohydrates is an extremely challenging and labour intensive process, and cannot generally be achieved in an automated fashion or on a large scale. New approaches to complex carbohydrate synthesis remain an urgent need in glycobiology in order to further our understanding as well as to facilitate the development of potential therapeutics. In the past few years directed evolution approaches for protein engineering have proven to be highly useful in improving the stability of enzymes (Arnold, Wintrode et al. 2001; Tao and Cornish 2002), and for altering their substrate specificities (Dalby 2003). 63 One of the crucial steps in any directed evolution experiment is the development of an assay to facilitate the screening of large libraries (Goddard and Reymond 2004). . However, assaying for transfer activity, in particular GT activity, is extremely challenging as no obvious change in fluorescence or absorbance is associated with glycosidic bond formation. As the screening for a desired phenotype, in most cases, is a random process it is highly desirable to develop high throughput screening methodologies (HTS) to facilitate the screening of extremely large libraries (Becker, Schmoldt et al. 2004; Aharoni, Griffiths et al. 2005). These methodologies are particularly valuable for the enrichment and isolation of rare mutants with beneficial activity that may arise from large mutant libraries (Griffiths and Tawfik 2003). This chapter describes the development of a novel high throughput methodology for the directed evolution of sialyltransferases with increased catalytic activity, using the bifunctional Cst-II as a model system. A HTS methodology was developed for the detection and sorting of sialyltransferase activity in intact E. coli cells (Figure 3.1) using a fluorescence activated cell sorter (FACS). The development of a cell-based assay for GT activity using FACS is highly advantageous as it alleviates the need to lyse the cells and perform many other manipulations otherwise necessary for screening large mutant libraries. By using a carefully designed, fluorescently labeled acceptor sugar, and • selectively trapping the sialylated fluorescent product in the cell, the formation of transfer product can be directly correlated to the fluorescence of the cell. This study describes the sensitivity and dynamic range of this screening system and its use to isolate a new variant with a 400-fold increase in catalytic efficiency (as assayed with fluorescent bodipy-labeled acceptor sugars). Interestingly, this large increase was associated with a single 64 mutation F91Y, located 15A from the donor sugar binding site. Crystallographic analysis of this mutant allowed us to observe that this mutation resulted in exposure of a hydrophobic pocket (Figure 3.6) that is thought to create a high affinity aromatic aglycone binding site. Correspondingly, appendage of this aromatic aglycone to a range of otherwise incompetent acceptors endowed them with efficient acceptor activity for this specific mutant. This fluorescence-based HTS method should therefore allow for the evolution of enzymes with new catalytic activities, providing new synthetic routes for complex carbohydrates and other conjugates such as phosphates or sulfates. 65 3.2 METHODS 3.2.1 DNA manipulations The gene encoding Cst-II 1" 2 5 9 (Chiu, Watts et al. 2004) was PCR-amplified from the pET28-Cst-II plasmid and sub-cloned into the pUC18 plasmid using the EcoRI and Sail restriction sites. The CMP-synthetase gene was PCR-amplified from pNSY-05 (Karwaski, Wakarchuk et al. 2002) and subcloned to a low copy number plasmid, pACKC18, using the EcoRI and Sail restriction sites. Cst-II libraries were generated by error-prone PCR, as detailed in previously established protocols (Vartanian, Henry et al. 2001). Briefly, 1-10 ng of pUClS-Cst-II 1" 2 5 9 plasmid was amplified in PCR reactions containing NTPs at 1:5 ratio of A C : T G , supplemented with either 0.25 m M or 0.5 m M MnCb., using the pUC18-seq-fo and -21M13 primers. The PCR product was digested and ligated into pUC18. The ligation mixture was electroporated into E. cloni (Lucigen). The cells were then grown overnight at 37°C in L B supplemented with ampicillin (100 pg mL"1) and the library plasmid D N A was extracted. Several individual clones from the native library were sequenced, and contained an average mutation frequency of ~2 mutations and ~ 4 mutations per gene in the 0.25 m M and 0.5 m M M n C h libraries, respectively. 3.2.2 Screening Cst-II libraries by FACS The pACKC18-CMP-syn plasmid encoding for CMP-synthetase was transformed into JM107 NanA' cells (a derivative of E. coli K12, containing a chromosomal deletion of the Neu5Ac aldolase gene) (Antoine, Heyraud et al. 2005). Electrocompetent cells were prepared of a subsequent clone. Plasmid D N A (pUClS-Cst-II 1" 2 5 9) encoding Cst-II • 66 variants and libraries were transformed into these cells. The cells were then grown overnight at 37 °C. The cells were diluted 50 fold in mineral cultured media as described (Antoine, Heyraud et al. 2005) and grown at 37 °C until an OD600 of 0.6 was reached. At this point 0.5 m M IPTG was added and the cells were transferred to 20 °C and grown overnight. Cells were spun down (1 mL) and resuspended in 50 pX of M9 media supplemented with 1 m M sialic acid and 0.2-0.5 m M of the different fluorescently labeled acceptor sugars (Figure 3.2). Following 20 min to 1 hour of incubation, cells were spun down and excess acceptor sugar and Neu5Ac were removed. The cells were resuspended in L B media and transferred to 37 °C for 10 min, centrifuged and washed twice with PBS before resuspension in 1 mL of PBS. The cells were visually analyzed for fluorescence and taken to the FACS for further analysis and sorting. Cells were diluted -50 fold in PBS, and analyzed in the FACS Aria flow cytometer (Becton-Dickinson) using PBS as sheath fluid. The threshold for event detection was set to forward and side scattering. The average sort rate was -4000 events per second, using a 70 pm nozzle, exciting argon ion (488 nm) and 405 nm lasers, and measuring emissions passing the 530 ± 20 nm (FITC) band-pass filter for the bodipy emission, and the 450 nm (violet 1) filter for the coumarin emission. Cells were sorted into eppendorf tubes containing 200 pL L B medium. Pools of sorted positives were plated on L B agar plates and colonies were grown overnight at 37 °C. The cells were removed from the agar plates for subsequent FACS enrichment as described above. FACS data were processed using the Flowjo Software (Tree Star). 67 3.2.3 Screening, purification and kinetic analysis of Cst-II F91Y mutant Plasmid D N A extracted from the third round of FACS enrichment was transformed to fresh JM107 NanA' E. coli cells containing pACKC18-CMP-syn. Colonies were picked and grown overnight in L B as above. Cells were diluted 1/100 in fresh L B media and grown to an ODgoo of 0.6, induced with 0.5 m M IPTG, transferred to 20 °C and grown overnight for purposes of protein expression. Cells were spun down (1 mL), lysed in 30 uL of BugBuster (Novagen). The transfer activity was tested in reaction buffer containing 10 m M M g C l 2 , 0.25 m M DTT, 2.5 m M CTP, 2.5 m M Neu5Ac and 0.5 m M of bodipy-lactose (Figure 3.2). Transfer activity was analyzed at different time points on T L C plates (4:2:1; ethyl acetate:methanol:H20) and compared to cells expressing wildtype Cst-II 1" 2 5 9 under identical conditions. Clones with significantly higher activity than the wildtype activity were further characterized for transfer activity with a variety of fluorescently labeled acceptor sugars and donor sugars under the same conditions (discussed below). Plasmid D N A from clones expressing a higher activity than the wildtype Cst-II was prepared, and further analyzed by D N A sequencing. The Cst-II F91Y mutant was subcloned into the pET28a plasmid, and protein expressed and purified as described previously (Chiu, Watts et al. 2004). Kinetic analysis of the Cst-II F91Y mutant was performed essentially as described (Gosselin, Alhussaini et ai. 1994). 3.2.4 Crystallization, Data Collection and Structure Determination Pure Cst-II F91Y protein was concentrated to approximately 10 mg mL"1 at room temperature and subjected to co-crystallization screens together with the inert donor sugar analogue CMP-3FNeu5Ac using the vapour diffusion method. Crystallization conditions 68 distinct from that required for the previously determined wildtype mbnoclinic and tetragonal Cst-II crystals resulted in novel body-centered tetragonal crystals obtained with 0.1 m M T E A pH 7.5, 10 % isopropanol and 20 % PEG 400. This 14 crystal had unit cell dimensions of a = b = 116.653, c = 44.086 A. A 2.2 A dataset was collected on a local Rigaku RU200 rotating anode equipped with OSMIC mirrors and a MAR345 image plate detector. Data were processed using DENZO and S C A L E P A C K (Otwinowski and Minor 1997). The structure of the mutant was solved by molecular replacement with Molrep (Vagin and Teplyakov 2000) using a monomer of wild type Cst-II as the starting model (PDB code 1R07) and a resolution range of 15 - 4 A. The initial model was built with X T A L VIEW (McRee 1999) using Cst-II as the template and the final model was obtained after iterations of refinement with CNS 1.1 (Brunger, Adams et al. 1998) and Refmac 6.0 (Winn, Isupov et al. 2001). 3D coordinates of bodipy-lactose were generated by ChemDraw and modeling of bodipy-lactose into the active site of the Cst-II1 " 2 5 9 -F91Y mutant was performed manually. 69 3.3 RESULTS 3.3.1 Detection and sorting of ST activity in intact E. coli cells To develop a fluorescence cell based assay for Cst-II activity, an engineered mutant E. coli cell strain was used. This strain (JM107 NanA'), previously used for the production of sialylated oligosaccharides (Antoine, Heyraud et al. 2005), efficiently transports the Neu5Ac donor and lactose acceptor sugars to the cytoplasm through specific transporters. To prevent catabolism of lactose and Neu5Ac, the mutant strain lacks both (3-galactosidase (lacZ gene) and Neu5Ac aldolase activities (NanA gene). To allow cell-based synthesis of sialosides, the cells express both CMP-Neu5 Ac synthetase and Cst-II 1" 2 5 9. CMP-Neu5Ac synthetase activates the Neu5Ac in situ to CMP-Neu5Ac, and the Cst-II 1" 2 5 9 uses the latter sugar as its donor sugar for sialyltransfer to (3-galactoside acceptors (Chiu, Watts et al. 2004). The second component required for this screen was a fluorescently labeled galactose-containing acceptor that is freely transported into and out of the cell. To this end, a series of fluorescently labeled acceptor sugars were synthesized (Figure 3.2). The general scheme for the detection and sorting of sialyltransferase activity in the cells is shown in Figure 3.1. The engineered cells were incubated with fluorescently labeled acceptor sugars and Neu5Ac. Following an incubation time of 30-60 minutes the cells were subjected to three rounds of centrifugation and resuspension to wash out any unreacted fluorescent lactoside (step 4 in Figure 3.1). At this point, fluorescent sialylated lactoside product remains trapped in the cells due to its size and charge, whilst the unreacted fluorescent lactose is washed away. This wash step is extremely important in the reduction of background fluorescence, and facilitates the detection of weak 70 sialyltransferase activity. Finally the cells are subjected to FACS analysis and sorting to assess the amount of fluorescently labeled sialylated product trapped in the cell (step 5 in Figure 3.1). Library of Cstll genes Cloning into E. coll Expression in JM107 Nan A- strain (2) V Call CMP Neu5Ac syii FACS Analysis and Sorting < 5 » f N e u S A c - a c c e p t o r - i y e Extensive wash (4) acceptor - dye Incubation with NeuSAc and fluorescent acceptor (3) Neu5Ac-acceptor-dye — i f ' NeuSAc and derivatives Figure 3.1 Cell based assay for sialyltransferase activity (1) A gene library is transformed and cloned in E. coli. (2) The encoded ST protein is expressed in the cytoplasm of the engineered JM107 NanA strain together with CMP-Neu5Ac-synthetase. (3) Cells are incubated with Neu5Ac donor sugar and fluorescently labeled acceptor sugars (Figure 3.2). (4) Following incubation, cells are extensively washed to remove unreacted fluorescent acceptor sugar. (5) Cells are directly analyzed and sorted by a Fluorescence Activated Cell Sorter (FACS) . OH FliK)rescein-lactose Bodir^-3SH-lactose Figure 3.2 Fluorescent acceptor sugars used in this study 72 To test the feasibility of this approach, cells expressing the target Cst-II enzyme and cells expressing empty pUC18 plasmid were separately incubated with Neu5Ac (1 mM) and either bodipy-lactose acceptor or bodipy-galactose (Figure 3.2). Following extensive washing, the fluorescence intensity of cells expressing Cst-II was significantly higher than that of control cells, as visualized under an ultraviolet light lamp with the bodipy-lactose-containing cells being more fluorescent than those with bodipy-galactose (Figure 3.3a, Table 3.1). To quantify the difference in fluorescence and test the dynamic range of the cell-based sialyltransferase assay, the samples were subjected to FACS analysis. The mean fluorescence intensity of the cells expressing the Cst-II enzyme and incubated with bodipy-lactose was about 80 fold higher than that of the control cells (Figure 3.3b). This demonstrates the high dynamic range and potential to detect even slow transfer reactions. To verify that the fluorescence measured from the cells arises from the accumulation of sialylated fluorescent product, fluorescent cells were lysed and the crude cell lysate analyzed by TLC. Comparison with an in vitro reaction in which Cst-II 1" 2 5 9 was incubated with CMP-Neu5Ac and bodipy-lactose revealed that identical products were formed under these two conditions (data not shown). 73 (a) (b) 100 102 103 104 105 10 Product fluorescence intensity (530 nm) Figure 3.3 Cell based sialyltransferase assay for cells expressing wildtype Cst-II1"259 and containing the pUC18 plasmid The different cell samples were incubated with Neu5Ac and bodipy-lactose or bodipy-galactose. After extensive washing the cells were analyzed either visually or by F A C S . (a) Eppendorfs containing the four different cell samples as visualized under an UV light lamp, (b) F A C S histogram analysis of the four different cell samples. Overlapping light blue and light green, cells expressing empty pUC18 plasmid incubated with bodipy-galactose and bodipy-lactose respectively; beige, cells expressing wildtype Cst- I I 1 " 2 5 9 incubated with bodipy-galactose; purple, cells expressing wildtype Cst- I I 1 " 2 5 9 incubated with bodipy-lactose. The cell based assay was used to simultaneously detect Cst-II transfer activity to two different acceptor sugars. Cells expressing Cst-II 1 " 2 5 9 were incubated with Neu5Ac (1 m M ) together with two acceptor sugars of different transfer efficiency, bodipy-galactose and coumarin-lactose (Figure 3.2), and the transfer activity compared to that of control cells. F A C S detection and analysis of both dyes in the cells was performed through separate excitation and emission channels. The fluorescence intensity of cells expressing the Cst-II was much higher, as observed for both fluorescently labeled acceptor sugars. Finally, to verify that cells containing active Cst-II enzyme can be 74 sorted from a large heterogeneous cell population by using FACS, a model selection was performed (Mastrobattista, Taly et al. 2005). Following sorting, an enrichment factor of 80 fold was calculated for a cell population in which cells expressing wildtype Cst-II were mixed with a large excess (200 fold) of cells expressing a control plasmid. 3.3.2 Selection of Cst-II library for increase in sialyltransferase activity With these controls in place, plus the establishment of an effective screen, attention turned to the use of this screen to probe libraries of Cst-II mutants. A large Cst-II gene library was constructed by inserting random mutations along the Cst-II gene. This library was cloned into a pUC18 vector and propagated in E. coli cells to yield >106 different colonies. The library plasmid D N A was extracted, transformed to JM107 NanA' cells carrying the CMP synthetase expression vector, cells grown and protein expressed. Cells were incubated with Neu5Ac and the bodipy-lactose, washed, and with greater than 107 cells analyzed and sorted by the FACS (Figure 3.1). Three iterative rounds of enrichment were performed and in each round, multiple 'positive' events (3-5xl04) within the top 1-2 % of the green fluorescence intensity were sorted and collected into growth media, and plated on agar for a new round of selection. Following each round of sorting, an increase in ST activity of crude cell lysates was observed as judged by T L C analysis (Figure 3.4a). Accordingly, an increase in the fluorescence intensity of the library following each round of sorting was observed, and the mean fluorescence of the library following three rounds of sorting was found to be significantly higher than that of the wildtype cells (Figure 3.4b). 75 Fluorescence intensity (530nm) Figure 3.4 Library selection through three iterative rounds of sorting by FACS. (a) Activity analysis of the various rounds of F A C S enrichment. Cst-II transfer activity was measured on crude cell lysates prepared from the pool of cells obtained after each round of enrichment and analyzed by TLC. Activity was measured and compared to cells expressing pUC18 plasmid (control), wildtype Cst-II and the evolved Cst-II F91Y mutant. Lane 1, pUC18 plasmid; lane 2, naive Cst-II library; lane 3-5, library following one, two and three rounds of enrichment respectively; lane 6, evolved F91Y Cst-II mutant; and lane 7, wildtype Cst-II. (b) F A C S histogram analysis of cells expressing pUC18 (pink), wildtype Cst-II (green) and library following three rounds of enrichments (blue) following a 1-hour incubation with 1 mM Neu5Ac and 0.5 mM bodipy-lactose. To identify and isolate single clones with improved transfer activity, the plasmid D N A extracted from the third round of sorting was transformed to fresh JM107 NanA' cells. 20 random clones were picked, individually grown and tested for Cst-II activity. Product formation was analysed by T L C at different time points and was compared to the wildtype Cst-II activity. Approximately 20 % of the clones showed much higher activity than the wildtype clone (using bodipy-lactose and Neu5Ac as substrates). Four of the improved clones were sequenced and one mutation F 9 1 Y was repeated in two of the four most active clones. To verify that the improvements in activity were due to an increase in 76 specific activity rather than an increase in protein expression level, the crude cell lysate of the Cst-II variant containing only the F91Y mutation was compared to the crude cell lysate of the wildtype Cst-II by SDS polyacrylamide gel electrophoresis. No difference in expression level was observed between the two proteins (data not shown). The F91Y mutant showing the highest transfer activity by TLC analysis was further subcloned to a pET28 vector, over-expressed and purified for further characterization. 3.3.3 Kinetic analysis of Cst-II F91Y mutant Given the considerable improvements in transfer efficiency observed, it was necessary to first confirm the nature of the glycosidic linkage formed by the F91Y mutant. Incubation of the bodipy-labeled sialyl lactose product with the specific recombinant a-2,3-neuraminidase from Salmonella typhimurium (NEB) clearly resulted in cleavage of the sialic acid and regeneration of bodipy-lactose, as determined by T L C analysis (data not shown). This observation confirmed that an a-2,3 linkage had been formed. This is also consistent with the observation of disialylation in Figure 3.4a (lanes 4-6), as Cst-II is known to form oligo a-2,8-sialyl linkages on a-2,3-sialyl-lactose. Kinetic analysis of the purified F91Y mutant was performed using a spectrophotometric continuous coupled assay (Gosselin, Alhussaini et al. 1994). The F91Y mutant transferase activity was measured with a variety of acceptor sugars including: bodipy-lactose, bodipy-galactose and bodipy-3SH-lactose (Figure 3.2, Table 3.1). The contribution of the dye to the transfer efficiency of the F91Y mutation was assessed using unmodified lactose and galactose as acceptors (Table 3.1). A dramatic difference in catalytic efficiency of 153- and 367-fold was observed for bodipy-lactose 77 and bodipy-galactose respectively, relative to the natural lactose and galactose sugars. Observation of this dramatic rate improvement with dye-tagged acceptor sugars raised the question of whether lactose analogues that do not function with the wild type enzyme could be turned into useful acceptors for F91Y by dye-tagging. Of particular interest in this regard was an ability to form glycosidase-resistance thioglycosidic linkages which would be metabolically stable. Neither 3SH-lactose nor bodipy-3SH-lactose acts as an acceptor for Cst-II. By contrast bodipy-3SH-lactose acted as a good acceptor for the F91Y enzyme, with a k ^ / K ^ value only five fold lower than its parent bodipy-lactose (Table 3.1). Product analysis by mass spectrometry confirmed" the formation of a sialylated-3-SH lactose derivative, as did T L C analysis (data not shown). A n increase in transfer activity was also detected for the alternative donor sugar CMP-2-keto-3-deoxy-D-glycero-D-galacto-nononic acid (CMP-KDN, representing CMP-Neu5 Ac minus the N-acetyl group) in both cell lysates. Finally, the effects of the F91Y mutation on reaction rate with untagged substrate, as well as the enzyme catalyzed hydrolysis of CMP-Neu5 Ac were determined. The mutation decreases these inherent rates by 3- to 5-fold, highlighting the importance of the bodipy dye binding for the acceleration of the transfer reaction. 78 Table 3.1 The evolved Cst-II F91Y mutant: catalytic efficiency for the transfer of CMP-Neu5Ac to different acceptors Substrate kQJKM (min1 M 1) a Foldb increase Bodipy-lactose . 35.1 153 Lactose 0.23 Bodipy-galactose 11 367 Galactose 0.03 Bodipy-3 SH-lactose 6.5 » 4 0 7 c a Catalytic efficiency (k c a t /K M ) was calculated from a linear fit of plots of initial transfer rates from CMP-Neu5Ac (0,5 mM) to the bodipy acceptors at a range of concentrations from 0.05-0.2 mM. Measurements of transfer to lactose and galactose were performed at higher acceptor concentrations (up to 150 mM). b Fold increase in catalytic efficiency of bodipy acceptor relative to the native acceptor. c Minimum estimation of the fold increase of the bodipy-3SH-lactose relative to 3SH-lactose. 3.3.4 Crystallographic analysis of Cst-II F91Y mutant From the wildtype Cst-II structure, residue Phe91 is not in close proximity to the active site. It is approximately 15 A away from the donor sugar (Figure 3.5). Based on the wildtype structure, two hypotheses can be proposed for the observed increase in activity for F91Y mutant toward the bodipy-labeled acceptor. The first hypothesis is that with the substitution from a phenylalanine residue to a tyrosine residue, the extra hydroxyl group of the tyrosine side chain is protruding into the space where the Asn51-Pro52-Ser-53-Leu54 loop is (Figure 3.5). Since this loop is lining the active site and acceptor binding site, this perturbation makes the enzyme more effective toward bodipy-lactose binding. The second hypothesis is that this mutation is now interacting with the bodipy dye and is stabilizing it for the transferase activity. 79 Figure 3.5 Location of Phe91 residue in the structure of wildtype Cst-II. Donor sugar analogue CMP-3FNeu5Ac is shown in magenta. Red dashed line depicts the closest distance between the Phe91 residue and the donor sugar (CZ of Phe91 to 010 of the N-acetyl group of the donor sugar). The loop of Asn51-Leu54 is highlighted in green. Residues important for catalysis are also shown. In order to study the effect o f the F91Y mutation on the structure and catalytic activity o f Cst-II, the crystal structure o f the F 9 1 Y mutant was solved in complex with C M P - 3 F N e u 5 A c and compared to the structure o f the wildtype Cst-II complexed with the same donor sugar analogue. A 4-fold redundant data set was collected at a home source, with 94.3 % completeness (65.4 %, for the highest resolution shell) and Rsym of 6.2 % (17.5 %, for the highest resolution shell). The structure was solved with molecular replacement and refined to a final crystallographic R/Rfree o f 21/25 % (Table 3.2). 80 Table 3.2 Data collection and structure refinement statistics. Data collection Crystal parameters space group 14 Diffraction statistics Resolution ( A ) 25-2.25 Wavelength 1.514 Total reflections 52576 Unique reflections 13458 Redundancy 3.9 Completeness 93.7 (65.4) <//o7>1 16.9(5.1) R 1 , 2 r\sym 6.6(18.5) Refinement Statistics and model stereochemistry Resolution ( A ) 25-2.25 Number of Atoms Protein 1950 . Substrate 41 Isopropanol 8 Water 91 Rcrys/Rfree (%)3 19.4/26.1 R.m.s. deviations Bonds ( A ) 0.021 Angles (°) 1.79 High resolution shell (2.33 A - 2.25 A ) statistics are in parentheses Rsym ^Ohkd-^VZOhki), where lhki is the integrated intensity of a given reflection. Rcryst=($F0-Fc\)/(£F0), where F0 and Fc are observed and calculated structure factors. 5 % of total reflections were excluded from the refinement to calculate Rfree. 81 In the wildtype Cst-II, the phenyl side chain of F91 packs into the enzyme core where it interacts closely with the surrounding hydrophobic residues (Pro52, Phe55, Cys78, aliphatic chain of Glu87, Phe95, Phe99, Leu l04 , and Phe254; Figure 3.6). Substitution of the tyrosine side chain with its inherent hydroxyl group apparently disrupts this tight hydrophobic packing and results in a dramatic flip of the side chain to a completely solvent-exposed orientation (Figure 3.6), as observed by the electron density map for the mutant. This movement consequently results in the exposure of a hydrophobic pocket that is fortuitously complementary to the fused aromatic ring system of the bodipy dye structure (Figure 3.2). Figure 3.6 Structure of the F91Y Cst-II mutant 82 Structures of the wildtype (white) and F91Y (yellow) mutant of Cst-II are superimposed. Corresponding 2 F 0 - F C electron density at 1 a for the F91Y mutant is 'also shown in blue mesh. The Tyr91 residue in the mutant (yellow) is flipped out, exposing a hydrophobic pocket that is complimentary to the bodipy dye ring structure. Based on the geometry of the exposed hydrophobic pocket, bodipy-lactose was modeled into the acceptor binding site of Cst-II (Figure 3.7). This model suggests that the bodipy dye is specifically bound in the newly formed hydrophobic cavity of the F91Y mutant, and that the lactose would be appropriately positioned in the vicinity of the donor sugar to facilitate the formation of the sialyl lactose product. This model further suggests two adjacent (but distinct) binding sites for the sugar and bodipy dye that together result in an overall dramatic increase in catalytic proficiency. The formation of an additional binding site may explain why tagging of the bodipy dye to otherwise incompetent acceptors (for example, 3-SH-lactose) results in a dramatic increase of the transfer activity, and suggests a general way to increase the transfer activity of glycosyltransferases by generating a specific aglycone binding site without compromising the transfer activity and regioselectivity. In accordance with the notion that the improved catalytic efficiency (kc^/Ku) of the F91Y mutant is mostly due to improved binding of bodipy-lactose, we observed a saturation in the transfer product formation, as judged by TLC, between 0.6 m M and 0.8 m M of bodipy-lactose, whereas we observed no sign of saturation with the wildtype enzyme up to a concentration of 5 m M bodipy-lactose (data not shown). Additionally, the specificity of this newly created dye binding site was probed by monitoring reactions with both coumarin lactose and fluorescein lactose. In either case, no significant 83 improvement in transfer rate was observed for the F 9 1 Y mutant relative to wildtype Cst-II, indicating that the binding site formed was indeed specific for bodipy. Figure 3.7 Surface representation of the active site cleft and the Tyr91 region in the F91Y Cst-II mutant Red depicts negative electrostatic potential, and blue positive electrostatic potential. The CMP-3FNeu5Ac donor sugar (white) is depicted in a stick representation. The bodipy-lactose (yellow) is modeled so that the bodipy is positioned in the exposed hydrophobic pocket, with the lactose directly adjacent to the CMP-3FNeu5Ac moiety. 84 3.4 DISCUSSION We have developed a HTS methodology for the detection of sialyltransferase activity directly in E. coli cells. Coupling of the fluorescence cell based assay to FACS analysis allowed us to screen library of >106 variants in less than two hours. Unlike other bulk selection methodologies such as panning on immobilized ligands (Fernandez-Gacio, Uguen et al. 2003), FACS allows fine-tuning of the selection threshold, enrichment and recovery. We have also demonstrated how parallel selection can be performed on two different acceptor sugars by the use of a different dye for each acceptor sugar, and monitoring the transfer reaction simultaneously in the cell. This approach could be possibly used to exert positive and negative selective pressures for the isolation of highly selective enzyme variants (Varadarajan, Gam et al. 2005). Other methodologies that have been developed for screening large enzyme libraries (Becker, Schmoldt et al. 2004; Aharoni, Griffiths et al. 2005) are based on different display technologies (e.g., phage-display (Fernandez-Gacio, Uguen et al. 2003) or bacterial display (Varadarajan, Gam et al. 2005)). Recently a novel HTS method was developed in which the diffusion of substrate and product is restricted by using double water-in-oil-in-water emulsion (Aharoni, Amitai et al. 2005; Mastrobattista, Taly et al. 2005). However no such technologies have been developed previously for glycosyl transfer reactions. Indeed, the only high throughput screening systems in use are those developed for glycosynthases. This included the use of an agar plate-based coupled enzyme assay for the Agrobacterium sp. P-glucosidase (Abg) glycosynthase in which an endo-cellulase was used to release a fluorescent dye only from the reaction product (Kim, Lee et al. 2004). Recently a selection assay for glycosynthase activity was developed for 85 the E197A mutant of Cel7B from Humicola insolens (Lin, Tao et al. 2004). Using the yeast three-hybrid system, product formation was directly coupled to yeast growth, but this approach was only applied to very small libraries. The outcome of the selection of >106 different Cst-II mutants for increase in transfer activity of Neu5Ac to bodipy-lactose was a Cst-II variant containing a single F91Y mutation. This mutation significantly increased the transfer activity with a variety of bodipy-labeled acceptor sugars (Table 3.1). However the activity of this mutant with unlabeled acceptor sugars or acceptor sugars labeled with different dyes (e.g. coumarin, fluorescein) was barely affected. The structural analysis of this F91Y mutant was an important component of this study as it sheds light into understanding the observed significant increase in catalytic transfer proficiency (kcJKM). From the observed electron density, the F91Y mutation generates a highly specific binding site for the bodipy dye (Figure 3.6, 3.7) and the refined mutant structure provides a foundation for our modeling studies. The model suggests that a bivalent interaction is formed between the enzyme and the bodipy acceptor, with the dye functionality being bound in the observed newly formed hydrophobic pocket and the sugar group bound in a suitable orientation to facilitate a productive transfer reaction to yield the normal a-2,3 linkage with donor sugar (Figure 3.7). This model also gives insight to the substantial increase of 400-fold in k^JKu for to the bodipy dye (Table 3.1). The complementary geometry of the hydrophobic pocket to the bodipy rings might explain the lower binding efficiency observed with other dyes (coumarin and fluorescein). Finally, during the modeling process, it was observed that the linker between the bodipy dye and the lactose is of optimal length to allow both the 86 bodipy and the lactose to bind to their respective binding sites. The lower reactivity of the mutant toward bodipy-galactose (Table 3.1) might be due to the variation in length of this linker region. Generation of a novel binding site for the bodipy dye in the F91Y mutant provides a general strategy to increase the transfer efficiency to poor acceptor sugars by temporarily tagging them with a hydrophobic moiety. Indeed, using this approach a sugar that does not usually function as an acceptor for the wild type Cst-II, 3-SH-lactose, was converted into an acceptor for the F91Y mutant, with a catalytic proficiency (kcJKu) for the transfer of Neu5Ac to bodipy-3SH-lactose that is only ~5 times lower than that for the transfer to bodipy-lactose using the F91Y mutant (Table 3.1). This result suggests that the low transfer activity to 3SH lactose is mainly due to inefficient binding of the acceptor rather than any intrinsic difference in the catalytic transfer mechanism between the 3-thio and 3-hydroxy analogues. The thiosialylated product is of particular interest as thiooligosaccharides are metabolically stable mimics of their naturally occurring counterparts (Witczak and Culhane 2005). Whilst a very limited set of glycosyltransferases has been shown to catalyze the synthesis of thiooligosaccharides (Rich, Szpacenko et al. 2004), no thioglycoside product has been previously reported for any sialyltransferase. This chapter describes the first directed evolution experiment for glycosyltransferases based on a genuinely H T S methodology. This work opens up new avenues for directed evolution experiments of glycosyltransferases, for the glycosylation of a variety of acceptors. In the case of glycosyltransferases that form neutral sugar products, Cst-II could serve as coupling enzyme to trap the reaction product in the cell, 87 thus extending the methodology to other glycosyltransferases that transfer galactose to fluorescently labeled acceptor sugars. This methodology, based on selectively trapping the transfer product, can be extended to detect other transfer reactions (for example, phosphorylation or sulfation) in which a charged moiety is transferred to a variety of acceptors. In addition this methodology is also applicable to the detection of transfer reactions by fluorescence resonance energy transfer by using both acceptors and donors that are fluorescently labeled. 88 CHAPTER 4 - Structural and biochemical characterization of monofunctional Cst-I from Campylobacter jejuni 4.1 I N T R O D U C T I O N In addition to the bifunctional Cst-II isolated from Campylobacter jejuni strain OH4384, a second sialyltransferase was also identified. This variant enzyme, Cst-I, exhibits an ct-2,3-sialyltransferase activity only and shows a high specificity for galactoside acceptors (Figure 1.10). It is thus best described as monofunctional. Gene-specific inactivation of cst-I (in C. jejuni HS:36) showed that it was not essential for lipooligosaccharide sialylation, so the exact role of cst-I is still undefined. Full length Cst-I consists of 430 amino acids and appears to be organized into two domains. The N -terminal domain, (residues 1-300) shows homology (47% identity) with the bifunctional Cst-II sialyltransferase from the same organism. The C-terminal domain (residues 301-430) shows homology with a domain that is present at either the N - or the C- terminal ends of five C. jejuni OH4384 ORFs found in either the cst-I or capsule-biosynthesis loci. Deletions constructs indicate that the minimal Cst-I sequence to show activity is 1-272, (at levels somewhat reduced from the full length construct). Although the truncated C-terminal domain is not essential for enzymatic activity it could play a role in oligomer stabilization and/or cell localization. To further our understanding of differences in substrate specificity as well as sialyltransferase function in general, the X-ray crystal structures of the monofunctional Cst-I in both the apo and substrate-bound forms were solved. Together with a kinetic study characterizing its enzyme activity and a proposed model for acceptor sugar binding, 89 this work provides significant new insight into the mechanism of catalysis and substrate specificity of sialyltransferases. 90 4.2 M E T H O D S 4.2.1 Cloning and protein expression Cloning of cst-I (Protein sequence database #AAF13495) from Campylobacter jejuni OH4384 was performed as previously described (Gilbert, Karwaski et al. 2002). A truncated version of the gene, encoding residues 1-285, was amplified by PWO polymerase with primers 5 'CTT A G G A G G T C A T A T G A C A A G G A C T A G A A T G G A A A A T G A A C 3 ' and 5 ' T T T A G A A T G G T C G A C C T A A T A A A A A T T A A G C A T A A T 3 ' . The PCR product was digested with Ndel and Sail restriction enzymes and subcloned into the pCWori+(-lacZ) vector that carries the E. coli maltose-binding protein MalE ((Protein sequence database #P02928), without a signal peptide, followed by a G G G H linker and JJNPRGSH thrombin cleavage site). The plasmid pCWori-malE-Cst-I 1" 2 8 5 was subsequently transformed into E. coli strain AD202, and cells grown at 37 °C in 2xYT broth with 150 pg mL"1 ampicillin and 0.2 % (w/v) glucose. When the OD600 reached ~ 0.5, a final concentration of 1.0 m M IPTG was added and the cells were switched to 25°C and incubated for 22 hours. Cells were then harvested at 6,200 x g for 15 minutes and stored at-80°C. 4.2.2 Purification of the Cst-I1"285 product and kinetic assay The cell pellet (2.4 g wet weight) was resuspended in 24 mL of 20 m M Tris-HCl pH 7.5 buffer containing 200 m M NaCl, 1.0 m M EDTA and 5.0 m M 2-mercaptoethanol. The cells were lysed in three passes using an Emulsiflex C-5 at ~ 19,000 psi. At this step, a tablet of protease inhibitor cocktail (Roche) was added to the sample. The resulting 91 extract was ullxacentrifuged at 75,000 x g for 35 minutes at 4 °C. The supernatant was then dialyzed overnight against 2.0 L of 20 m M Tris-HCl pH 7.5 buffer containing 200 m M NaCl, 1.0 m M E D T A and 5.0 m M 2-mercaptoethanol before loading onto an amylose affinity column. A column (2.6 x 4.5 cm) packed with amylose resin (New England Biolab E8021L) was equilibrated with 125 mL of 20 m M Tris-HCl pH 7.5 buffer containing 200 m M NaCl, 1.0 m M E D T A and 5.0 m M 2-mercaptoethanol. The overnight dialysate was added to the resin and incubated at 4 °C with rocking for one hour. The resin was packed into a Bio-Rad open column and washed with 100 mL of equilibration buffer at a flow rate of 2.0 mL minute"1. The fusion protein was eluted with 70 mL of the same buffer containing 10 m M maltose. For thrombin cleavage, the fusion protein was incubated at 4 °C overnight with 1:1000 dilution of restriction grade thrombin (Sigma) in I X thrombin cleavage buffer. The sample was then desalted via dialysis against 20 m M Tris-HCl pH 7.5 and applied to a MonoQ HR 10/10 column (GE Healthcare) pre-equilibrated with the 1 285 same buffer. Cst-I" was collected from the flow-through fractions. MalE and the uncleaved fusion protein were removed from the column with a 20 mL gradient of 0 to 0.5 M NaCl in the same buffer. Kinetic analysis of the purified enzyme, using lactose as co-substrate was performed using a continuous coupled assay analogous to that previously described (Chiu, Watts et al. 2004). 4.2.3 Crystallization and data collection Cst-I 1" 2 8 5, purified as above, was concentrated using a 15 mL Amicon Ultra-4 centrifugal filter device (10,000 molecular weight cutoff) spun on an Allegra 21R 92 centrifuge (Beckman Coulter). Tetragonal protein crystals appeared when the concentration of the protein reached ~ 3 mg mL"1. For the substrate-bound structure, both the. acceptor sugar lactose and the inert donor sugar analogue CMP-3FNeu5Ac (synthesized as previously described (Watts, Damager et al. 2003)) were soaked into the crystals for 30 minutes before data collection. The crystals were transferred from 0 to 30 % v/v ethylene glycol at 2.5 % intervals for cryoprotection. Both apo- and substrate-bound crystals adopt the space group 14 with unit cell dimensions a = b = 112.4, c = 58.8 A . For the apo-enzyme, a 2.2 A dataset was collected on a local Rigaku RU200 rotating anode equipped with OSMIC mirrors and a MAR345 image plate detector. For the CMP-3FNeu5Ac complex, a 1.7 A dataset was collected at the Advanced Light Source beamline 8-2-1 using an ADSC Quantum Q210 2x2 CCD array detector. Data were processed using DENZO and S C A L E P A C K (Otwinowski and Minor 1997). Statistics for data collection and processing are summarized in Table 5.1. 4.2.4 Structure determination, refinement and modeling The apo-Cst-I1"285 structure was solved by molecular replacement using Molrep(Vagin and Teplyakov 2000) with a monomer of Cst-II 1" 2 5 9 as the starting model (PDB code 1R07). Model phases were improved by R E S O L V E (Terwilliger and Berendzen 1999). The initial model was built with X T A L V I E W (McRee 1999) using Cst-II 1" 2 5 9 as the template and the final model was obtained after iterations of refinement with CNS 1.1 (Brunger, Adams et al. 1998) and Refinac 5.0.1 (Murshudov, Vagin et al. 1999). The CMP-3FNeu5Ac complex structure was solved using apo-Cst-I1"285 structure as the starting model for refinement. The quality of the models was analyzed with 93 P R O C H E C K ( L a s k o w s k i , MacAxthur et al. 1993) and the results are summarized in Table 4.1. Alignment of structures was performed using A l i g n (Cohen 1997). Modeling of acceptor sugars was performed using a combination of AutoDock3.0 (Morris, Goodsell et al. 1998) and manual adjustments with X T A L V I E W . Required parameter files for AutoDock3.0 were prepared using the Dundee P R O D R G server (Schuttelkopf and van Aalten2004). 94 4.3 R E S U L T S 4.3.1 Sequence and Kinetic analysis of Cst-I Full length Cst-I (residues 1 to 430) contains a predicted N-terminal sialyltransferase domain (based on sequence similarities with other members of CAZy family 42), and an additional C-terminal domain that shows localized sequence homology with two ORFs in the cst-I locus and three ORFs in the capsule biosynthesis locus of C. jejuni OH4384, but no discernible homology with any known deposited structures. Full-length Cst-I is not sufficiently soluble at the higher concentrations required for crystallization trials, thus a truncated version was designed based on sequence alignments with the naturally shorter Cst-II, which lacks an analogous C-terminal domain. This truncated version of Cst-I, (residues 1 to 285 or Cst-I1"2 8 5), retains a high level of activity, showing that the C-terminal domain likely does not contribute in a significant way to the active site or observed a-2,3-transferase activity. Although, one can postulate potential roles for the apparently novel C-terminal domain based on its similarity to ORFs in the capsule biosynthesis locus, including that of cell localization and/or protein scaffolding, clearly further investigation will be required to understand its structure and its contributions to Cst-I action in vivo. Using sialyl lactose as acceptor at concentrations of up to 500 mM, transferase activity could not be detected above the significant (k c a t = 0.1 s"1) enzyme catalyzed hydrolysis of the donor substrate for Cst-I. In addition, product formation could not be detected using TLC or capillary electrophoretic analysis with sialyl lactose as acceptor (data not shown). These results clearly demonstrate the reaction specificity of Cst-I, showing exclusively an a-2,3-transferase activity with Gal-B-R acceptor sugar. The KM 95 value of the donor sugar CMP-Neu5Ac for Cst-I 1" 2 8 5 was determined to be 400 uM, which is similar to the previously determined value of 460 uM for Cst-II (Chiu, Watts et al. 2004). By contrast, monofunctional Cst-I 1" 2 8 5 has a KU value of 500 uM for the acceptor lactose, whilst bifunctional Cst-II has a value of 35 mM. This is a substantial difference, but is likely a consequence of the bifunctional nature of Cst-II, which binds sialyl lactose better, with a KM value of 3.5 mM. 4.3.2 Determination of the Cst-I 1" 2 8 5 structure The structure of the catalytic domain of Cst-I 1" 2 8 5 in the apo form was first solved via molecular replacement methodology using Cst-II (PDB code 1R07) as the starting model (there is -53% local sequence identity between Cst-I 1" 2 8 5 and Cst-II1"2 5 9; see Figure 1 o o c 4.1). The resulting Cst-I " model is well refined, with crystallographic R/Rfree values of 0.22/0.25 for the apo-enzyme (to 2.1 A resolution; Table 4.1). The overall structure is a mixed a/p fold, and is composed of an N-terminal Rossmann nucleotide binding domain and a C-terminal " l id" domain which encompasses the active site region of the enzyme (Figure 4.2). Static light scattering analysis reveals that Cst-I 1" 2 8 5 forms a tetramer in solution (data not shown). Correspondingly, we observe that the predicted biological 4-fold axis of the tetramer in our structure coincides with the crystallographic 4-fold axis in the body-centered tetragonal 14 space group with the putative membrane interface of all 4 molecules aligned on the same face of. the tetramer. The majority of the residues involved in oligomerization are conserved within the Cst family, and are primarily aromatic and hydrophobic residues. which create an extensive hydrophobic interface 96 between the 4 monomers that maintains the tetrameric oligomerization state (-1300 A 3 of buried surface). Although the overall architecture of Cst-I 1" 2 8 5 is similar to that of Cst- II 1" 2 5 9, with a r.m.s. deviation of 0.8 A between 235 C a atoms, there are major structural differences in the lid domain that folds over the active site and we believe that these conformational differences dictate the consequent substrate specificity of the enzymes (Figure 4.2; see below). 97 Table 4.1 Data collection and structure refinement statistics. Data collection Apo CMP-3FNeu5Ac Crystal parameters space group 14 14 Diffraction statistics Xray source Rigaku RU200 A L S 8.2.1 Resolution 25-2.2 50-1.7 Wavelength 1.514 1.00 Total reflections 66105 429679 Unique reflections 18129 40593 Redundancy . 3.65 10.59 Completeness 96.5 (66.8) 99.8 (99:0) <l/ol>1 16.3 (3.5) 28.9 (2.6) p 1.2 4.3 (21.4) 6.4 (52.4) Refinement Statistics and model stereochemistry Resolution ( A ) 25-2.2 40-1.7 Number of Atoms Protein 2122 2270 Substrate 0 42 Water 84 90 Cryoprotectant 20 28 Ion 0 1 Rcrys/Rfree (%)"* 21.9/25.3 0.19/0.22 R.m.s. deviations Bonds ( A ) 0.006 0.015 Angles (°) 1.17 1.431 Average B-factor ( A 2 ) Protein 39.8 37.4 Substrate - 45.5 Water 36.3 39.4 Cryoprotectant 39.7 47.6 Ion - 34.1 98 High resolution shell (2.28 A - 2.20 A for apo form and 1.76 A - 1.70 A for C M P -3FNeu5Ac bound form) statistics are in parentheses RSym =2\0hki)^<l>VZ(lhki), where lhM is the integrated intensity of a given reflection. Rcryst=(AF0-Fc\)/(2F0), where F 0 and F c are observed and calculated structure factors. 5% of total reflections were excluded from the refinement to calculate Rfree bi-Cst-II b i - C s t - I I m o n o - C s t - I I roono-Cst—I M T R T R M E N E L I V S K N p2 H2 (33 sum -+-TT 4 O SMJiSl DViFRCNQiFjYFE'DRYYL'GKK DVFRCNQFYFEDKYjYL'GKK DVFRCNQFYFEDKYYLGKK bi-Cst-II b i - C s t - I I m o n o - C s t - I I m o n o - C s t - I a l SLSISLSLSLSLSISIIISISLSL 6 0 7 0 p5 H 0 A 0 Q QN QN LK T E u a • o2 a3 SUISISUL JLQJLS. 9 0 1 0 0 L jjNE L 3 NE I 2 SKI (56 a4 -*• SUL8JL n o a5 SL3JLSLSLBJL8JL5LSLS18. 1 2 0 bi-Cst-II b i - C s t - I I mono—Cst—II mono-Cst—I a7 T|4 SLSLUQJUl 1 7 0 1 8 0 T T SLSLSULSLS. 1 9 0 bi-cst-n b i - C s t - I I m o n o - C s t - I I m o n o - C s t - I a8 JUlSlSiSlSlSUUl 2 0 0 p8 F H E K T BE K T T,5 T T J2J2AJI 2 2 0 IP N = L l g g n EBA|P P9 I J I J N H O E i IQEU. T L E N 0 H G K 21 K g a9 TI6 JISJUULSJULOJI 2 5 0 S3 SE Figure 4.1 Sequence alignment of bifunctional Cst-II1"259, monofunctional Cst-II1"259, and monofunctional Cst-I 1 ' 2 8 5 Residues important for coordinating the donor sugar (Asn51/66, Ser132/147, Tyr156/171, and Tyr162/177) are highlighted by green asterisks. Proposed general base, His188/202, and its pKa modulator, Arg129/144, are highlighted by blue asterisks. Red box with white character shows strict identity between the sequences. Red character indicates similarity in a group and blue frame depicts similarity across groups. Secondary structural elements from Cst-II 1" 2 5 9 are shown above the sequence. 99 Figure 4.2 Structural alignment of the three Campylobacter sialyltransferases. The C a trace of apo-Cst - l 1 " 2 8 5 is shown in white, and substrate-bound form in blue, and that for Cst-II 1" 2 5 9 in yellow. Donor sugar analogue is shown in magenta color. 4.3.3 Comparison of structures of the apo- and substrate-bound monofunctional Cst-I 1 2 8 5 The structure o f Cst-I 1 " 2 8 5 in complex with the inert donor sugar analogue C M P - 3 -fluoro-Neu5Ac was solved using difference Fourier methods in the same crystal form as that of the apo Cs t - I 1 ' 2 8 5 and is well refined with R/Rfree values of 0.19/0.22 (to 1.7 A resolution). Cst - I 1 " 2 8 5 binds the C M P - 3 F N e u 5 A c in the N-terminal Rossmann nucleotide binding domain. The cytidine base of C M P is tucked inside a cavity between Tyr l71 and 100 Gly25, and is stacked over the aromatic side chain of Tyrl71 (Figure 4.3). The base also forms intermolecular hydrogen bonds with the main chain amides of Asp 169, Phel70, Tyrl71, and the main chain carbonyl oxygen of Ilel76. The ribose moiety of the nucleotide base adopts a C3 ' endo conformation with the cytidine base axial to the ring. It is held in place with hydrogen bonds to the main chain amides of Asn24 and Serl47, and side chain OG1 of Thrl46. The two free oxygens of the phosphate group form a complex hydrogen bonding network to the side chain amide of Asn46, 08 of the sialic acid, and the side chain hydroxyl atoms of tyrosines 171 and 177. The sialic acid moiety of the donor nucleotide sugar also forms several interactions with the active site residues. The carboxylic acid group is coordinated by both the main chain amide and side chain OG of Serl47. It also forms hydrogen bonds to ND2 of Asn66, and intramolecularly to 03 of the ribose moiety. 04 interacts with the main chain nitrogen of Serl99, 06 and 07 of the sugar ring are within hydrogen bonding distance of ND2 of Asn66, and 08 and 09 are hydrogen bonded, respectively, to NE2 and 0E1 of the Gln47 side chain. The N-acetyl group forms hydrophobic interactions with the phenyl side chain of Phel96. Although the residues in the lid domain (residues 171 to 199) are not as well ordered as the rest of the enzyme (as indicated by their higher temperature factors), the majority of the residues can be modeled unambiguously in the electron density map. In contrast, in the apo-enzyme structure where the substrate is absent, two loop regions of the lid domain (residues 171 to 181, and residues 194 to 199) are highly disordered and cannot be modeled in the electron density map. These regions include some of the key residues involved in donor sugar binding, such as the two conserved tyrosine residues, Tyrl71 and Tyrl77. These tyrosine residues interact with the phosphate oxygen of the nucleotide and 101 are thought to stabilize the negative charge that develops on the phosphate oxygens during catalysis. Indeed, a double mutation of the analogous tyrosines into alanines in Cst-II was shown to abolish the enzyme transferase activity (Chiu, Watts et al. 2004). Recently, Jeanneau et al have also shown via mutagenesis and kinetic analysis that an invariant tyrosine residue in the conserved sequence motif (sialyl motif III) of the human sialyltransferase ST3Gal I may play a similar catalytic role(Jeanneau, Chazalet et al. 2004). Figure 4.3 Interactions of CMP -3FNeu5Ac with active site residues of Cst-I 1' 2 8 5. 102 Donor analogue CMP-3FNeu5Ac is depicted with carbon atoms in magenta, nitrogen atoms in blue, oxygen atoms in red, and phosphorus atom in green. Active site residues involved in CMP-3FNeu5Ac binding and catalysis are labeled and shown with carbon atoms in cyan, nitrogen in blue and oxygen in red. Dotted lines indicate hydrogen bonding. 4.3.4 Catalytic residues in Cst-I Cst-I catalyzes the reaction by a direct displacement mechanism with overall inversion of configuration at the anomeric center and the requirement of a proximal functional group serving as the general base to activate the acceptor sugar for transfer to occur. His202, which is positioned immediately after the more mobile l id domain region, is highly ordered, with the lowest observed B-factors of the residues that comprise ; the active site. This residue is also conserved in Cst-II (Hisl88) and, based on the orientation of the imidazole side chain relative to the anomeric carbon of the donor sugar, has been proposed as a likely candidate for the general base in the catalytic mechanism. In this role it would abstract a proton from the 3'-hydroxyl group of the incoming acceptor molecule during its nucleophilic attack on the anomeric center of the donor 1 285 C M P - N e u 5 A c . In the Cst-I" structure we also observe a close electrostatic influence from the nearby A r g l 4 4 (5 A). A similar interaction in the Cst-II active site (Argl29) was proposed to play a role in regulating the p K a of the general base histidine. Our observations of these conserved features in Cst-I 1 " 2 8 5 , despite many other point differences in the active site, support the previously proposed mechanism of C A Z y family 42 sialyltransferases (Chapter 2). Several additional active site residues, identified from the bifunctional Cst-II structure and mutagenesis studies, are conserved in Cst-I and occupy roughly an analogous position, namely, Asn31(Asn46), Asn51(Asn66), Serl32(Serl57), Tyr l56(Tyr l71) , and Tyr l62(Tyr l77) . These residues are involved in 103 the binding of CMP-Neu5Ac. Since these catalytic players are conserved between the monofunctional Cst-I and bifunctional Cst-II variants, other critical point mutations in the active site must presumably be responsible for the different substrate specificities of these enzymes, as is discussed further below. 4.3.5 Docking of acceptor sugars to the active site of Cst-I1"285 Despite extensive attempts to obtain a donor-acceptor ternary complex structure via soaking of crystals with a range of acceptor analogues, no such complexes were obtained. At best with donor/acceptor Cst-I 1" 2 8 5 soaks, we observe a small amount of residual density at 1 sigma, possibly representing a very weakly occupied potential acceptor site situated on the alpha face of the donor sugar ring and in close proximity to the proposed general base His202 (Figure 4.4). Our density suggests that of a molecule of the cryoprotectant additive ethylene glycol, which is essential for the crystallographic data collection (an extensive screen of other cryoprotectants failed), and is successfully competing with the acceptor sugar at this site site due to the abundant amount used in cryoprotection (30 % v/v). The refined model of the bound ethylene glycol indeed shows interactions with the enzyme (the hydroxyl groups hydrogen bond with O of Ilel45, 2.7 A, and OD 1 of Asn66, 2.9 A) that may mimic that of the hydroxyl groups of the natural acceptor sugar. Using this information as a guide, the binding mode of the acceptor sugar, Gal-P-l,3-GalNAc, was modeled into the active site of Cst-I 1" 2 8 5 using a combination of AutoDock 3.0 and manual fitting. The final model of the bound Gal-P-1,3-GalNAc is oriented at a -50° angle to the donor sugar ring and is sandwiched between the side chains of Asn66 and Argl44. The distance between 03 of the galactose and the anomeric C2' of the donor sugar is approximately 3.9 A. The imidazole ring of 104 the proposed general base, His202, is ~ 3.5 A away from the same 03 atom. Both of these distances are compatible with a potential deprotonation event by His202 and subsequent nucleophilic attack on the anomeric center of the donor sugar. Galactose is anchored in place by stacking with the amide side chain of Asn66, forming hydrophobic interactions with this key residue that were postulated previously (in our studies on Cst-II). Hence, Asn66 plays a key role in defining the substrate specificity of the enzyme. 02 of the galactose is within hydrogen bonding distance of the 07 ' of the donor sugar, and 04 and 06 of the acceptor are within hydrogen bonding distance of the side chains of Glul02 and Argl44, respectively. The GalNAc portion of the acceptor is situated in a more solvent-exposed environment than the galactose. The only potential hydrogen bond to the enzyme is between 04 of GalNAc and the main chain carbonyl group of IlelOl. There is also a potential set of hydrophobic interactions between the JV-acetyl group and the side chain of Phel90. The smaller number of specific interactions of GalNAc to the enzyme would explain the tolerance of the enzyme for other galactose derivatives as acceptor sugars (for example, the readily available lactose substrate used in our kinetic assays). Modeling analysis of lactose into the active site of Cst-I 1" 2 8 5 shows low energy forms that are able to bind in a similar orientation and with retention of similar interactions to active site residues as predicted for Gal-P-1,3-GalNAc. Even though sialyl lactose is not an acceptor for Cst-I, it was also used in modeling analysis for this enzyme. After several docking attempts, all the resulting models show an overall higher estimated free energy of binding (close to zero or positive in the range of -0.5 kcal/mol to 1.4 kcal/mol) as compared to Gal-P-1,3-GalNAc or lactose (~ -4 kcal/mol). Furthermore, the program failed to generate a model with 08 of 105 the sialic acid moiety close to the anomeric center of donor sugar in a correct orientation (with the end of the sugar chain pointing toward the core of the enzyme). Collectively this suggests the inability of this enzyme to bind sialyl lactose tightly or in the correct orientation for the sialylation to occur. Figure 4.4 Molecular model ing of Gal -R-1,3-GalNAc into the active site of Cst-I Active site residues are shown in cyan, donor analogue is shown in magenta, and the model of Gal-U-1,3-GalNAc is shown in green. The site at which the sialic acid would be transferred onto the acceptor, 0 3 of the galactose moiety, is highlighted by an orange circle. The orange line connects the imidazole ring of the proposed base His202 to the 0 3 of the galactose and onto the C2 anomeric center of the donor sugar. Dashed lines indicate potential hydrogen bonds between the acceptor and donor/active site residues. 106 4.3.6 Docking of acceptors to bifunctional Cst-II and comparison to Cst-I Although the overall structures of Cst-I" and Cst-II~ are similar, the two enzymes have fundamental differences in substrate specificity. Cst-I can perform only a single a-2,3 transfer of sialic acid onto galactose derivatives. In contrast, bifunctional Cst-II can transfer sialic acid first onto position 3 of galactose and perform a second transfer onto position 08 of the sialyl galactose generated from the initial transfer step. Indeed, when comparing the primary sequences of the two versions of Csts, the majority of the differences occur in the lid domain region which orders over the substrate to form part of the active site. This region also exhibits the least structural conservation between crystal forms, implying that although the enzymes utilize the same set of active site residues in catalysis, their lid domain region plays the role of selecting different acceptor molecules for the reaction to proceed. In order to elucidate the detailed differences in the binding of the acceptor sugar in Cst-II, docking of both Gal-P-1,3-GalNAc (a-2,3 transfer; Figure 4.5a) and Neu5Ac-a-2,3-Gal-P-1,4-Glc (sialyl lactose, a-2,8 transfer; Figure 4.5b) to the active site of Cst-II was performed. When the binding of Gal-P-l,3-GalNAc to the two enzymes is compared, galactose is seen to occupy a relatively similar space with respect to the reaction center of the donor sugar and the postulated histidine general base, with the major difference lying in the binding of the GalNAc moiety. In the docked model of Cst-1 285 I" , the GalNAc ring stacks against the main chain atoms of Gly68, which forms part of a P-turn composed of Asn66-Pro67-Gly68 and Val69. In Cst-II, the replacement of Gly68 by an isoleucine (Ser53 in Cst-II) sterically blocks GalNAc from occupying the analogous position due to the protrusion of the serine side chain hydroxyl. As a result, 107 the yV-acetyl group is forced closer to Tyrl85 allowing favourable hydrophobic interactions with the side chain aromatic ring. The carbonyl oxygen of the iV-acetyl group can also form a hydrogen bond to NH1 of Argl29 in this position, and 01 of the GalNAc is within hydrogen bonding distance to Asnl77. These predicted interactions between the enzyme and the acceptor sugar are quite different from those in Cst-I, implying that even though both enzymes utilize Gal-p-l,3-GalNAc as acceptor sugars, the terminal sugar is accommodated in the active site quite differently. Docking of the bulkier sialyl lactose acceptor sugar involved in the cc-2,8 transfer to Cst-II allows for appropriate placement of 08 ' in the glycerol moiety of the sialic acid relative to the general base His 188 (-3.4 A) and the anomeric center of the donor sialic acid (~4.5 A) (Figure 4.5b). In this position, Argl29 from the enzyme could also form a hydrogen bond to 07 of the glycerol moiety. Unlike in Cst-I 1" 2 8 5 where Asn66 stacks with the galactose unit of the acceptor sugar, the equivalent Asn51 of Cst-II does not interact directly with the docked acceptor. Also, the side chain of Ser53 again protrudes into the acceptor binding cleft and hinders close interaction of the acceptor to the |3-turn of Asn51-Pro52-Ser53-Leu54, and stabilizes the acceptor in the altered position through a potential interaction from its side chain hydroxyl to the P carboxyl group of sialic acid. It is interesting to note that the conserved Argl29 on the opposite face of the sugar ring is not suitably positioned to interact with the carboxylate of the acceptor sugar. As docked, the galactose ring of the sugar does not form many direct contacts with the enzyme active site residues, but appears rather to be stabilized by intramolecular interactions such as contacts between 02 of the galactose and 07 of the sialic acid, as well as 05 of the galactose and 03 of the glucose. The glucose ring forms favorable vdW interactions with 108 the amide side chain of Asnl77. In the monofunctional version of Cst-II (isolated from serotype 019 (Gilbert, Karwaski et al. 2002)) which contains only 8 amino acids different from its bifunctional isoform (Figure 4.1), an aspartate residue is located at position 177 (1 of the 8 different amino acids). Preliminary activity data of designed point mutants shows that substitution of this aspartate residue with a neutral asparagine significantly enhances a-2,8 activity of monofunctional Cst-II (data not shown). Clearly a negative charge at this position would disfavor the negatively charged sialic acid moiety of the acceptor sugar. Figure 4.5 Molecular modeling of various acceptors into the active site of Cst-II. 110 (a) Gal-p-1,3-GalNAc (b) sialyl lactose. Active site residues are shown in yellow, donor analogue is shown in magenta, and the model of either Gal-B-1,3-GalNAc or shown in green. The site at which the sialic acid would be transferred onto the acceptor, 0 3 of the galactose moiety or 0 8 of the sialyl moiety, is highlighted by an orange circle. The orange line connects the imidazole ring of the proposed base His188 to the 0 3 of the galactose/08 of sialic acid and onto the C2 anomeric center of the donor sugar. Attempts to dock the same acceptor, sialyl lactose, to the active site of the mono-functional Cst-I 1" 2 8 5 provide clear indications as to the inability of this sialyltransferase variant to effectively transfer the bulkier acceptor sugar (Figure 4.6). Since docking of sialyl lactose specifically for Cst-I failed to yield a reasonable model, the model of sialyl lactose for Cst-II is overlapped into the active site of Cst-I. While the active site of Cst-I 1" 2 8 5 can accommodate the entire substrate, the negatively charged carboxyl group of the sialyl-moiety is forced into close proximity to the hydrophobic side chain of Phel90. The presence of the bulky aromatic side chain of Phel90 imposes both steric and electrostatic hindrance for the binding of this distinct carboxyl group of the sialyl lactose acceptor. In Cst-II, the equivalent position to Phel90 is an alanine (Alal75) whose small methyl side chain would not impose the same steric and electrostatic burden on the binding of sialyl-lactose for the a-2,8 transfer. I l l Figure 4.6 Overlap of sialyl lactose from molecular modeling for Cst-II into the active site of Cst-I. The structure of Cst-I is shown in blue and the structure of Cst-II is shown in yellow. The donor analogue is shown in magenta and the sialyl lactose model is shown in green. The mesh represents the space filling diagram of the enclosed molecule. 112 4.4 DISCUSSION Three versions of Csts from Campylobacter jejuni have been identified thus far by activity screening of a plasmid library: two versions of Cst-II (monofunctional and bifunctional) and one version of Cst-I (monofunctional). The two Cst-II enzymes differ by only 8 amino acids, and represent a more attractive case for direct comparison between their structures. While work continues in order to solve the structure of monofunctional Cst-II (see Chapter 5), the Cst-I structure also provides significant insight for the substrate specificities, of these enzymes and further supports the kinetic mechanism proposed for GT-42 sialyltransferases. Our comparison of the apo and substrate-bound forms of Cst-I 1" 2 8 5, the first such analysis for a GT-42 sialyltransferase as the apo form of Cst-II was not amenable to crystallization, suggests that the donor substrate CMP-Neu5Ac binds initially, and upon binding, induces conformational changes that create the appropriate acceptor binding site. The disorder of the residues in the lid domain of the apo form effectively creates a wide open cavity which would serve to facilitate initial access of the C M P donor sugar to the active site. Upon binding the donor sugar, the flexible loop of the lid domain rigidifies through contacts with the carbohydrate, consequently creating the acceptor binding site and minimizing potential hydrolytic effects by effectively excluding bulk solvent from the reactive center in the enclosed form. Although this is the first such observation for the structurally distinct sialyltransferases, the equivalent ordering of active site residues has been observed in the structural analysis of other glycosyltransferases (e.g. SpsA (Charnock and Davies 1999), <x3GT (Boix, Swaminathan et al. 2001), and LgtC (Persson, L y et al. 2001) to name a few). 113 The conservation of the proposed catalytic residues (Argl44, Tyrl71, Tyrl77, and most importantly, His202) between the active sites of Cst-I and Cst-II further supports the previously proposed kinetic mechanism for GT-42 sialyltransferases. To recapitulate, we proposed that the arginine residue modulates the pKa value of the nearby histidine residue, which subsequently acts as the general base in the reaction to activate the incoming acceptor sugar by deprotonating the specific hydroxyl group. The oxygen of the acceptor can then nucleophilically attack the C2 anomeric center of the CMP-Neu5Ac donor sugar. The two tyrosine residues contribute to stabilization of the phosphate group of CMP. Recently, several studies on human sialyltransferases (ST3Gal I (Jeanneau, Chazalet et al. 2004), ST8Sia II and IV (Close, Mendiratta et al. 2003)) have also identified an invariant histidine residue that is critical for the catalytic activities of those enzymes. In addition, in the only other published sialyltransferase structure (from the divergent Pasteurella multocida enzyme in CAZy GT-80), a histidine residue (His311) is also found adjacent to CMP in the active site with a proximal arginine (Arg63) ~ 5 A • away from the imidazole side chain (Ni, Sun et al. 2006). It is unfortunate that extensive attempts to obtain ternary complex structures of either version of Cst were unsuccessful. Yet, molecular modeling can be performed with the structures of Cst-II and Cst-I in complex with inert donor sugar analogue CMP-3FNeu5Ac. Docking of their respective acceptor(s) to Cst-I and Cst-II active sites completes the picture for catalysis by these enzymes; the results show that the specific hydroxyl group on each acceptor (03 for galactose and 08 for sialyl-lactose) are favourably positioned in between the proposed general base (His 188 for Cst-II and His202 for Cst-I) and the anomeric center of the sialic acid. In addition, overlapping the 114 modeled sialyl lactose into the active site of Cst-I 1" 2 8 5 reveals a potential hindrance of binding this negatively charged sugar by a nearby aromatic residue Phel90. 1 T O C In conclusion, the structure of monofunctional Cst-I" from GT-42 provides further insights into the different substrate specificities of different isoforms of Campylobacter sialyltransferases within this family. Although no acceptor is observed in the active site of Cst-I 1" 2 8 5, its structure, together with the previously solved bifunctional Cst-II structure, enables us to compare and contrast the difference in the electrostatic environment of their acceptor binding sites. Molecular modeling of different acceptors to their respective active sites allows us to identify the key residues that are involved in defining acceptor substrate specificities. Although it is of utmost interest to characterize structurally and kinetically both single point and composite mutants of either Cst-I or Cst-II with switched substrate specificities, at this, moderate level of sequence identity (only -53% for the catalytic domains solved) this type of protein engineering is very difficult. Compensatory residues that surround these point differences in these active sites are not conserved. Thus, engineering of single point mutations that switch from mono- to bifunctional while maintaining kinetic competence activity is unlikely, as compensatory changes in the residues surrounding those point mutants will also be needed. 115 C H A P T E R 5 - CONCLUSIONS A N D F U T U R E DIRECTIONS 5.1 S U M M A R Y A N D S I G N I F I C A N C E OF R E S U L T S The main goal of this thesis was to investigate the structure and molecular mechanism of an important class of glycosyltransferases, the sialyltransferase class. To this end, the structural and biochemical properties of two Campylobacter jejuni sialyltransferase Csts have been examined. Collectively, this work represents the first detailed structural and kinetic analysis of a sialyltransferase from any source. The sialylated ganglioside mimics of C. jejuni play an important role in the virulence of this pathogen. This mimicry is believed to be the molecular basis for the development of autoimmune diseases such as Guillain Barre syndrome and Miller Fisher syndrome. Sialyltransferases from the pathogen are responsible for transferring a sialic acid moiety from the donor sugar CMP-Neu5Ac onto a terminal galactose residue on LOS. A bifunctional Cst, with both a-2,3- and a-2,8-sialyltransferase activity, was first characterized as part of this thesis work. The previous lack of structural information on sialyltransferases hindered the mechanistic understanding of this biologically important class of glycosyltransferases. With the structure of the bifunctional Cst-II solved in complex with either the end product CMP, the donor sugar analogue CMP-3FNeu5Ac, or the general inhibitor CDP, potential catalytic groups from both the enzyme and substrate could be analyzed to help elucidate the reaction mechanism. Together with kinetic studies of mutants of the active site residues, a catalytic mechanism for Cst-II is hypothesized. A conserved histidine is proposed to be the general base for this inverting glycosyltransferase, wherein it activates the acceptor sugar for the nucleophilic attack on the anomeric center of the donor. The observed conformation of the donor sugar, with 116 the carboxyl group in plane with the sugar ring, disfavors previous suggestions of a self-assisted intramolecular catalysis in which the carboxylate serves as the general base. Deriving from the first structural investigation of a sialyltransferase, this reaction mechanism serves as a model for other mammalian or bacterial sialyltransferases that use the same common sugar donor CMP-Neu5Ac, but also utilize diverse acceptor sugars. The sialylation mechanism of Cst-II could possibly be generalized to other important sialyltransferases with different physiological roles. A n example would be polysialyltransferases (GT-38) from pathogenic bacteria such as E. coli serogroup K l and Neisseria, which catalyze the formation of polysialic acid capsule on the surface (Roberts 1996; Whitfield and Roberts 1999). Since capsular polysaccharides containing polysialic acids are important virulence factors for many pathogenic bacteria, understanding the catalysis mechanism of these enzymes is valuable for to the development of antibacterial agents. Moreover, analysis of the interactions between Cst-II and both donor and acceptor analogues not only aids in understanding the binding motif for CMP-Neu5Ac (no previous structures in the protein database contain bound CMP-Neu5Ac), inhibitors with similar morphology to the sugars can also be designed. A CMP-binding motif (together with CDP) can also be generated from the Cst-II/CMP (and Cst-II/CDP) complex. Since a CMP moiety appears to be vital in terms of inhibitor design, other bacterial glycosyltransferases essential for virulence that contain this CMP-binding motif can also be potentially inhibited with the same group of inhibitors. Cst-II was used as the model system for the development of a high throughput screening methodology for directed evolution. Due to the nature of the glycosidic bond 117 formed by glycoyltransferases, detection of enzymatic activity in vivo is challenging and a HTS for high libraries has been lacking. The HTS methodology described in this study is not only sensitive and efficient in screening a large library (106 variants in less than two hours), but it also successfully identified a mutant, created by directed evolution using an error-prone D N A polymerase, with substantially higher catalytic proficiency toward the fluorescently labeled acceptor. This mutant was found to carry a single substitution, F91Y. Based on the structural characterization of this mutant and comparison to the wildtype, the increase in reaction rate can be explained. The replacement of the phenylalanine residue to tyrosine upsets the tight hydrophobic packing of phenylalanine in the core of the enzyme, results in the flipping of the tyrosine side chain to face the solvent, and the resultant exposure of a hydrophobic pocket. The geometry of this newly created pocket is complementary to the fused ring structure of the bodipy fluorescent dye. The potential for this HTS methodology is immense. It can be extended to screen glycosyltransferases from other families that utilize UDP-Gal or UDP-GalNAc as the donor sugar (both of which can act as the acceptor for Cst-II) by coupling the reaction with that of Cst-II. In addition to Cst-II, another member of GT-42, Cst-I was also studied. Cst-I is a monofunctional variant of the bifunctional Cst-II, which possesses solely a-2,3-transferase activity. Although the exact role of Cst-I in vivo is still unknown, its structure in both apo and in CMP-3FNeu5Ac complexed forms further supports the kinetic mechanism proposed for this family of sialyltransferases. Moreover, the CMP-3FNeu5Ac bound structure provides a framework for molecular modeling of different acceptor sugars into the active sites of both Cst-I and Cst-II. The comparison of this 118 modeling between the two enzymes gives insight into the different substrate specificities observed. 119 5.2 F U T U R E DIRECTIONS Attempts to capture both the donor and acceptor sugars in the active site of any of the Cst enzymes have been ongoing. Various donor and acceptor analogues have been used for co-crystallization and/or soaking experiments (Table 5.1). None of these attempts, however, yielded a ternary complex structure of the enzyme, a phenomenon that has plagued the analysis of other glycosyltransferase systems, presumably due to the relatively poor affinity of acceptor sugars. Table 5.1 Attempts for obtaining ternary complex structure for Csts Enzyme Donor/Donor analogue Acceptor/Acceptor analogue Cst-II CMP-Neu5Ac sialyl lactose CMP-3FNeu5Ac sialyl lactose 3'deoxylactose galactose DNP-lactose FCHASE- lactose C M P sialyl lactose galactose C D P sialyl lactose galactose DANA sialyl lactose C M P - 3 F e q N e u 5 A c * lactose Cst-II E235A CMP-3FNeu5Ac sialyl lactose lactose Cst-II F91Y CMP-3FNeu5Ac bodipy lactose bodipy galactose FCHASE- lactose C D P bodipy lactose 120 Enzyme Donor/Donor analogue Acceptor/Acceptor analogue sialyl lactose Cst-II H188N CMP-Neu5Ac lactose galactose sialyl lactose Cst-II H188Q CMP-Neu5Ac lactose galactose lactose Cst-I CMP-3FNeu5Ac galactose C M P - 3 F e q N e u 5 A c * lactose * C M P - 3 F e q N e u 5 A c is the inert donor analogue with the 3-Fluoro group in an equatorial position Observing the binding mode of acceptor sugar would significantly enhance our understanding for the kinetic action of sialyltransferases, and would serve as a starting point in the inhibitor designs. Hence, obtaining a ternary complex structure is of high priority. Soaking and co-crystallization experiments have been performed mostly with CMP-3FNeu5Ac with the 3'fluoro group in the axial position. However, during the modeling of bodipy-lactose into the active site of Cst-II F91Y, it appears that whilst the 3'fluoro substitution is effective in keeping the donor sugar intact, it also protrudes into the acceptor binding site and might impede with acceptor binding. Recently CMP-3F e qNeu5Ac with the 3'fluoro group in the equatorial position was synthesized. Preliminary trials with CMP-3F e qNeu5Ac have not yet been successful in improving acceptor sugar binding, although extensive screening with the variety of sugars available might prove otherwise. Of particular interest is the use of this CMP-3F e qNeu5Ac together with bodipy-lactose for Cst-II F91Y. Since this mutant was shown to bind the bodipy fluorescent label with much higher efficiency than the native sugar, this combination might have a higher chance of providing a ternary complex structure. 121 As structures of the enzyme in the presence of both donor and acceptor sugar substrates and/or substrate analogues are determined, this information can be used in the inhibitor design process. CMP appears to bind to the enzyme fairly tightly as the nucleotide is highly coordinated by the enzyme (Figure 2.5). Side chains of adjacent residues form extensive hydrogen bonds to the oxygen atoms in the phosphate. CDP has also been shown to act as an inhibitor for the enzyme (Schaub, Muller et al. 1998). These two molecules can serve as the starting points in the inhibitor design process. Other approaches to generate novel lead compounds for the inhibitor design process include automated computational docking methods such as AutoDock (Morris, Goodsell et al. 1998) and high-throughput screening of small molecule ligand libraries. Potential inhibitors can first be tested for altered phenotype or anti-bacterial activity in vitro. Since sialylated-LOS has been shown to associate with the serum resistance of C. jejuni (Guerry, Ewing et al. 2000), these compounds can ultimatley be tested in vivo to determine their anti-bacterial activity and the essential role of Cst in virulence. An N M R experiment to measure the pKa value of His 188 from wildtype Cst-II was proven to be unsuccessful. This is essentially due to the fact that Cst-II forms a tetramer of more than 120 kDa in solution. Currently, effort is being made to generate a Cst-II mutant that stays monomeric in solution by the introduction of charged residues into the hydrophobic interface where each Cst-II monomer interacts. Since the active site of each monomer is distant to the oligomerization interface, such a mutant is likely to retain catalytic activity. This mutant can then be subjected to further N M R analyses: to determine the pKa value of Hisl88 by titration, and to observe the dynamics in the loop of the lid domain upon substrate binding. 122 Although some enzymes involved in the different stages of LOS biosynthesis have been identified, the overall picture remains unclear. Since LOS biosynthesis is a complex process, questions arise about the interactions of these enzymes. One of the speculations is that since the in vivo concentration of CMP-Neu5Ac is low, there may be other proteins supplying the CMP-Neu5Ac to Cst through interacting with it. With the complete structural characterization of Cst, further study of protein-protein interactions can be performed. Immunoprecipitation studies with anti-Cst antibodies or affinity chromatography with immobilized CMP-Neu5Ac or Cst can be performed to extract other proteins that interact with Cst from a cell lysate. Since LOS synthesis is one of the most important pathogenesis mechanisms, identification of interactions between essential proteins involved is of considerable interest. In addition to GT-42, which is dominated by sialyltransferase Csts from Campylobacter jejuni, sialyltransferases from other sources were classified into four other GT families. These four families include GT-29 (all eukaryotic sialyltransferases), GT-38 (bacterial polysialyltransferase), GT-52 (sialyltransferases mostly from Neisseria and Salmonella), and GT-80 (sialyltransferases from Pasteurella multocida). Of these, only GT-42 and GT-80 have representative structures (Chiu, Watts et al. 2004; N i , Sun et al. 2006). Structures from these two families have been shown to be vastly different: GT-42 sialyltransferases belong to GT-A whereas the product of gene PM0188 from P. multocida belongs to GT-B (Figure 5.1). This difference in their folds might due to the multi-function nature of the P. multocida enzyme (a-2,3-sialyltransferase, a-2,6-sialyltransferase, a-2,3-sialidase and an a-2,3-trans-sialidase) (Ni, Sun et al. 2006). In order to understand the general sialylation mechanism across species, and to design 123 specific inhibitors that target only bacterial sialyltransferases, attempts have been made to characterize sialyltransferases from other G T families (summarized in Table 5.2). Figure 5.1 Structural compar ison of GT-42 and GT-80 (a) GT-A fold of Cst-II from GT-42 (PDB code 1ro7). (b) GT-B fold of Pasteurella multocida sialyltransferase (PDB code 2ex0) 5.2.1 Monofunctional Cst-II from C. jejuni Besides the bifunctional Cst-II and the monofunctional Cst-I, there is yet another variant of sialyltransferase from C. jejuni: a monofunctional Cst-II isolated from serotype 0:19 (Gilbert, Karwaski et al. 2002). This version of Cst shows only a-2,3-transferase activity (similar to Cst-I), however, its sequence is more related to the bifunctional Cst-II from serotype OH4384, which shows only eight amino acids difference. Solving the structure of Cst-IIo :i9 would be highly desirable as it represents a more direct comparison 124 between a mono- and bifunctional sialyltransferase. Consequently, a clearer picture for the difference in substrate specificities can be obtained. Work has been instigated for the characterization of this enzyme. It has been purified to homogeneity and crystallization trials have been set up. Micro-crystals were obtained and an X-ray diffraction dataset has been collected. However, the crystal is highly mosaic and data processing is not feasible. Nonetheless, further refinement of the crystallization condition to improve the crystals is possible. 5.2.2 Nst from N. meningitidis Another common mucosal pathogen, Neisseria meningitidis, also contains a sialyltransferase gene, Nst (GT-52), responsible for the biosynthesis of sialylated LOS on the surface of this bacterium (Rietschel, Brade et al. 1990; Gilbert, Watson et al. 1996). Similar to the ganglioside mimics in C. jejuni, these sialylated LOS of N. meningitidis are identical in structure to sialylated glycans found in mammalian glycolipids. This mimicry is also thought to play a role in the pathogenesis of this bacterium (Rest and Mandrell 1995; Smith, Parsons et al. 1995). Work has been instigated for the characterization of Nst. Various truncation constructs to remove the proposed N-terminal membrane association domain were made, and growth conditions for protein overexpression in E. coli were optimized. Detergents were found to be required for the solubility of the protein. A purification protocol was developed, and the enyzme was purified to homogeneity. Based on TLC analysis, purified Nst was shown to retain activity. From results of both static and dynamic light scattering, Nst was found to exist as a dimer in solution, in contrast to Csts. 125 Crystallization trials have been set up, and micro crystals observed. Optimization of crystallization conditions is currently underway. 5.2.3 ST3Gal I and ST3Gal III from Homo sapiens Mammalian sialyltransferases have long attracted much attention due to their enormously important role in biology. Up to 20 different human sialyltransferases have thus far been cloned and characterized, all of which have been placed in GT-29. Despite their critical role in the biosynthesis of sialylated glycoconjugates which in turn serve numerous roles in life, structural characterization of mammalian sialyltransferases is currently limited to identification of several sequence motifs. The lack of high resolution characterization of any mammalian sialyltransferases is possibly due to the difficulty involved in the obtaining a homogenous, soluble protein for crystallographic studies. Yet, with the availability of two bacterial sialyltransferases, structural characterization of mammalian sialyltransferases would be highly advantageous in understanding of sialyltransferase mechanisms, and for designing specific inhibitors of the bacterial sialyltransferases. Studies were initiated for two mammalian sialyltransferases, ST3Gal I and ST3Gal III, in addition to their bacterial counterparts. Genes for these two enzymes were obtained by RT-PCR of mRNA library from human intestinal cell line CACO-2 and cloned into a bacterial expression vector. Various E. coli strains were tested for the improvement of protein solubility. Both protein samples were overexpressed in E. coli, which resulted in inclusion bodies. Ultimately, various refolding methods of ST3Gal I from inclusion bodies were attempted. Trace amounts of properly folded protein were 126 obtained, their presence determined by slight activity observed on TLC plate methodology. Table 5.2 Summary for work in progress for characterization of various members of CAZy sialyltransferase families CAZy STase Source Specificity Constructs Work in progress 29 ST3Gal 1 H. sapiens <x2-3 to Ga ip i -3Ga lNAc ST3Gal Ic Expression Refolding purification ST3Gal III H. sapiens a2-3 to'GalB1-3/4GlcNAc ST3Gal lllb Expression a2-3 toGalp1-3GalNAc 42 Cst-I C. jejuni a2-3 to Gal Cst-95 structure solved CSt-l l 019 C. jejuni a2-3 to Gal Cst-109 Crystallization Cst-I loH4384 C. jejuni a2-3 to Gal Cst-83 structure solved a2-8 to Neu5Ac Lic3a H. influenzae 375 o2-3 to Gal Hst-08 -Lic3b H. influenzae 375 a2-3 to Gal . Hst-06 Trial purification limited proteolysis further truncation required a2-8 to Neu5Ac 38 P S T N. meningitidis 992B a2-8 to Neu5Ac Pst-07 Test expression P S T E. coli K1 a2-8 to Neu5Ac Pst-04 Test expression Pst-05 52 LSG-02 H. influenzae Eagan a2-3 toGalp1-3GlcNAc LSG-02E Test expression NST N. meningitidis a2-3 to Gal Nst-103 Purification, crystallization trial Nst-104 Purification, crystallization trial 127 R E F E R E N C E S Aharoni, A . , G. Amitai, et al. (2005). "High-throughput screening of enzyme libraries: thiolactonases evolved by fluorescence-activated sorting of single cells in emulsion compartments." Chem Biol 12(12): 1281-9. Aharoni, A. , A . D. Griffiths, et al. (2005). "High-throughput screens and selections of enzyme-encoding genes." Curr Opin Chem Biol 9(2): 210-6. Aharoni, A. , K . Thieme, et al. (2006). "High-throughput screening methodology for the directed evolution of glycosyltransferases." Nat Methods 3(8): 609-14. Alios, B. M . , F. T. Lippy, et al. (1998). "Campylobacter jejuni strains from patients with Guillain-Barre syndrome." Emerg Infect Pis 4(2): 263-8. Amaya, M . F., A . Buschiazzo, et al. (2003). "The high resolution structures of free and inhibitor-bound Trypanosoma rangeli sialidase and its comparison with T. cruzi trans-sialidase." Journal of Molecular Biology 325(4): 773-784. Antoine, T., A . Heyraud, et al. (2005). "Highly efficient biosynthesis of the oligosaccharide moiety of the GD3 ganglioside by using metabolically engineered Escherichia coli." Angew Chem Int Ed Engl 44(9): 1350-2. Arnold, F. H. , P. L. Wintrode, et al. (2001). "How enzymes adapt: lessons from directed evolution." Trends Biochem Sci 26(2): 100-6. Aspinall, G. O., S. Fujimoto, et al. (1994). "Lipopolysaccharides from Campylobacter jejuni associated with Guillain-Barre syndrome patients mimic human gangliosides in structure." Infect Immun 62(5): 2122-5. Aspinall, G. O., A . G. McDonald, et al. (1994). "Lipopolysaccharides of Campylobacter jejuni serotype 0:19: structures of core oligosaccharide regions from the serostrain 128 and two bacterial isolates from patients with the Guillain-Barre syndrome." Biochemistry 33(1): 241-9. Becker, S., H. U . Schmoldt, et al. (2004). "Ultra-high-throughput screening based on cell-surface display and fluorescence-activated cell sorting for the identification of novel biocatalysts." Curr Opin Biotechnol 15(4): 323-9. Blaser, M . J. (1997). "Epidemiologic and clinical features of Campylobacter jejuni' infections." J Infect Pis 176 Suppl 2: S103-5. Bochner, B. S., R. A . Alvarez, et al. (2005). "Glycan array screening reveals a candidate ligand for Siglec-8." J Biol Chem 280(6): 4307-12. Boix, E., G. J. Swaminathan, et al. (2001). "Structure of UDP complex of UDP-galactose:beta-galactoside-alpha -1,3-galactosyltransferase at 1.53-A resolution reveals a conformational change in the catalytically important C terminus." J Biol Chem 276(51): 48608-14. Breton, C , L . Snajdrova, et al. (2006). "Structures and mechanisms of glycosyltransferases." Glvcobiology 16(2): 29R-37R. Brunger, A . T., P. D. Adams, et al. (1998). "Crystallography & N M R system: A new software suite for macromolecular structure determination." Acta Crystallographica Section D 54(5): 905-921. Bruses, J. L. and U . Rutishauser (2001). "Roles, regulation, and mechanism of polysialic acid function during neural development." Biochimie 83(7): 635-43. Burkart, M . D., S. P. Vincent, et al. (2000). "Chemo-enzymatic synthesis of fluorinated sugar nucleotide: useful mechanistic probes for glycosyltransferases." Bioorganic & Medicinal Chemistry 8(8): 1937-1946. 129 Burmeister, W. P., R. W. Ruigrok, et al. (1992). "The 2.2 A resolution crystal structure of influenza B neuraminidase and its complex with sialic acid." Embo J 11(1): 49-56. Busch, C , F. Hofmann, et al. (1998). " A common motif of eukaryotic glycosyltransferases is essential for the enzyme activity of large clostridial cvtotoxins." J. Biol. Chem. 273(31): 19566-19572. Buschiazzo, A. , G. A . Tavares, et al. (2000). "Structural basis of sialyltransferase activity in trypanosomal sialidases." Embo J 19(1): 16-24. Charnock, S. J. and G. J. Davies (1999). "Structure of the nucleotide-diphospho-sugar transferase, SpsA from Bacillus subtilis, in native and nucleotide-complexed forms." Biochemistry 38(20): 6380-5. Chiu, C. P., A . G. Watts, et al. (2004). "Structural analysis of the sialyltransferase Cstll from Campylobacter jejuni in complex with a substrate analog." Nat Struct Mol Biol 11(2): 163-70. Close, B. E., S. S. Mendiratta, et al. (2003). "The minimal structural domains required for neural cell adhesion molecule polysialylation by PST/ST8Sia IV and STX/ST8Sia II." J Biol Chem 278(33): 30796-805. Cohen, G. E. (1997). "ALIGN: a program to superimpose protein coordinates, accounting for insertions and deletions." J Appl Crystallogr 30: 1160-1161. Coutinho, P. M . , E. Deleury, et al. (2003). "An evolving hierarchical family classification for glycosyltransferases." J Mol Biol 328(2): 307-17. Cowtan, K . (1994). " D M : An automated procedure for phase improvement by density modification." Joint CCP4 and E S F - E A C B M Newsletter on Protein Crystallography 31: 34-38. 130 Crennell, S., E. Garman, et al. (1994). "Crystal structure of Vibrio cholerae neuraminidase reveals dual lectin-like domains in addition to the catalytic domain." Structure 2(6): 535-544. Crennell, S., T. Takimoto, et al. (2000). "Crystal structure of the multifunctional paramyxovirus hemagglutinin-neuraminidase." Nature Structural Biology 7(11): 1068-1074. Crennell, S. J., E. F. Garman, et al. (1993). "Crystal structure of a bacterial sialidase (from Salmonella typhimurium LT2) shows the same fold as an influenza virus neuraminidase." Proc Natl Acad Sci U S A 90(21): 9852-6. Crennell, S. J., E. F. Garman, et al. (1996). "The structures of Salmonella typhimurium LT2 neuraminidase and its complexes with three inhibitors at high resolution." J of M o l Biol 259(2): 264-280. Crocker, P. R. and T. Feizi (1996). "Carbohydrate recognition systems: functional triads in cell-cell interactions." Curr Opin Struct Biol 6(5): 679-91. Dalby, P. A. (2003). "Optimising enzyme function by directed evolution." Curr Opin Struct Biol 13(4): 500-5. Datta, A . K . and J. C. Paulson (1995). "The sialyltransferase "sialylmotif' participates in binding the donor substrate CMP-NeuAc." J Biol Chem 270(4): 1497-500. Datta, A . K. , A . Sinha, et al. (1998). "Mutation of the sialyltransferase S-sialylmotif alters the kinetics of the donor and acceptor substrates." J Biol Chem 273(16): 9608-14. Delaglio, F., S. Grzesiek, et al. (1995). "NMRPipe: a multidimensional spectral processing system based on UNIX pipes." J Biomol N M R 6(3): 277-93. 131 Doublie, S. (1997). "Preparation of selenomethionyl proteins for phase determination." Methods Enzymol 276: 523-30. Elliott, S., J. Egrie, et al. (2004). "Control of rHuEPO biological activity: the role of carbohydrate." Exp Hematol 32(12): 1146-55. Endtz, H . P., C. W. Ang, et al. (2000). "Molecular characterization of Campylobacter jejuni from patients with Guillain-Barre and Miller-Fisher syndromes." J Clin Microbiol 38(6): 2297-301. Fernandez-Gacio, A. , M . Uguen, et al. (2003). "Phage display as a tool for the directed evolution of enzymes." Trends Biotechnol 21(9): 408-14. Fukuda, M . , N . Hiraoka, et al. (1999). "C-type lectins and sialyl Lewis X oligosaccharides. Versatile roles in cell-cell interaction." J Cell Biol 147(3): 467-70. Gaskell, A. , S. Crennell, et al. (1995). "The three domains of a bacterial sialidase: A beta-propeller, an immunoglobulin module and a galactose-binding jelly-roll." Structure 3(11): 1197-1205. Gebler, J. C , D. E. Trimbur, et al. (1995). "Substrate-induced inactivation of a crippled yff-glucosidase mutant: Identification of the lLabeled amino acid and mutagenic analysis of its role." Biochemistry 34: 14547-14553. Geremia, R. A. , A . Harduin-Lepers, et al. (1997). "Identification of two novel conserved amino acid residues in eukaryotic sialyltransferases: implications for their mechanism of action." Glvcobiology 7(2): v-vii. Gibbons, B. J., P. J. Roach, et al. (2002). "Crystal structure of the autocatalytic initiator of glycogen biosynthesis, glycogenin." J Mol Biol 319(2): 463-77. 132 Gibson, R. P., J. P. Turkenburg, et al. (2002). "Insights into trehalose synthesis provided by the structure of the retaining glucosyltransferase OtsA." Chem Biol 9(12): 1337-46. Gilbert, M . , J. R. Brisson, et al. (2000). "Biosynthesis of ganglioside mimics in Campylobacter jejuni OH4384. Identification of the glycosyltransferase genes, enzymatic synthesis of model compounds, and characterization of nanomole amounts by 600-mhz (1)H and (13)C N M R analysis." J Biol Chem 275(6): 3896-906. Gilbert, M . , M . F. Karwaski, et al. (2002). "The genetic bases for the variation in the lipo-oligosaccharide of the mucosal pathogen, Campylobacter jejuni. Biosynthesis of sialylated ganglioside mimics in the core oligosaccharide." J Biol Chem 277(1): 327-37. Gilbert, M . , D. C. Watson, et al. (1996). "Cloning of the lipooligosaccharide alpha-2,3-sialyltransferase from the bacterial pathogens Neisseria meningitidis and Neisseria gonorrhoeae." J Biol Chem 271(45): 28271-6. Goddard, J. P. and J. L. Reymond (2004). "Enzyme assays for high-throughput screening." Curr Opin Biotechnol 15(4): 314-22. Goddard, T. D. and D. G. Kneller " S P A R K Y 3." Gosselin, S., M . Alhussaini, et al. (1994). "A continuous spectrophotometric assay for glycosyltransferases." Anal Biochem 220(1): 92-7. Griffiths, A . D. and D. S. Tawfik (2003). "Directed evolution of an extremely fast phosphotriesterase by in vitro compartmentalization." Embo J 22(1): 24-35. Guerry, P., C. P. Ewing, et al (2000). "Sialylation of lipooligosaccharide cores affects irnmunogenicity and serum resistance of Campylobacter jejuni." Infect Immun 68(12): 6656-62. Guerry, P., C. M . Szymanski, et al. (2002). "Phase variation of Campylobacter jejuni 81-176 lipooligosaccharide affects ganglioside mimicry and invasiveness in vitro." Infect Immun 70(2): 787-93. Hagen, F. K. , B. Hazes, et al. (1999). "Structure-function analysis of the UDP-N-acetyl-D-galactosamine polypeptide A^-acetylgalactosaminyltransferase. Essential resides lie in a predicted active site cleft resembling a lactose repressor fold." J. Biol. Chem. 274(10): 6797-6803. Hao, Q., T. Saida, et al. (1998). "Antibodies to gangliosides and galactocerebroside in patients with Guillain-Barre syndrome with preceding Campylobacter jejuni and other identified infections." J Neuroimmunol 81(1-2): 116-26. Harduin-Lepers, A. , R. Mollicone, et al. (2005). "The animal sialyltransferases and sialyltransferase-related genes: a phylogenetic approach." Glvcobiology 15(8): 805-17. Harduin-Lepers, A. , V . Vallejo-Ruiz, et al. (2001). "The human sialyltransferase family." Biochimie 83(8): 727-37. Harvey, H . A. , N . Porat, et al. (2000). "Gonococcal lipooligosaccharide is a ligand for the asialoglycoprotein receptor on human sperm." M o l Microbiol 36(5): 1059-70. Horenstein, B. A. (1997). "Quantum mechanical analysis of an a-carboxylate-substituted oxocarbenium ion. Isotope effects for formation of the sialyl cation and the origin of an unusually large 1 4 C secondary isotope effect." J. Am. Chem. Soc. 119: 1101-1107. 134 Horenstein, B. A . and M . Bruner (1998). "The vV-acetyl neuraminyl oxocarbenium ion is an intermediate in the presence of anionic nucleophiles." J. Am. Chem. Soc. 120: 1357-1362. Horenstein, B. A . and M . Bruner (2002). "Characterization of alpha(2—>6)-sialyltransferase reaction intermediates: use of alternative substrates to unmask kinetic isotope effects." Methods Enzymol 354: 159-68. . Hye, J. B. and I. Nachamkin (2007). "Campylobacter jejuni cst-II polymorphisms and association with development of Guillain-Barre syndrome." Neurology 68(19): 1633-4. Jeanneau, C , V . Chazalet, et al. (2004). "Structure-function analysis of the human sialyltransferase ST3Gal I: role of 7V-glycosylation and a novel conserved sialylmotif." J Biol Chem 279(14): 13461-8. Kahler, C. M . and D. S. Stephens (1998). "Genetic basis for biosynthesis, structure, and function of meningococcal lipooligosaccharide (endotoxin)." Crit Rev Microbiol 24(4): 281-334. Karlyshev, A . V. , D. Linton, et al. (2000). "Genetic and biochemical evidence of a Campylobacter jejuni capsular polysaccharide that accounts for Penner serotype specificity." Mol Microbiol 35(3): 529-41. Karwaski, M . F., W. W. Wakarchuk, et al. (2002). "High-level expression of recombinant Neisseria CMP-sialic acid synthetase in Escherichia coli." Protein Expr Purif 25(2): 237-40. Kean, E. L. (1991). "Sialic acid activation." Glycobiology 1(5): 441-7. 135 Ketley, J. M . (1997). "Pathogenesis of enteric infection by Campylobacter." Microbiology 143 (Pt 1): 5-21. Kim, Y . W., S. S. Lee, et al. (2004). "Directed evolution of a glycosynthase from Agrobacterium sp. increases its catalytic activity dramatically and expands its substrate repertoire." J Biol Chem 279(41): 42787-93. Kiss, J. Z. and D. Muller (2001). "Contribution of the neural cell adhesion molecule to neuronal and synaptic plasticity." Rev Neurosci 12(4): 297-310. Kleywegt, G. J. and T. A . Jones (1997). "Model building and refinement practice." Methods Enzymol 277: 208-230. Ladisch, S., H . Becker, et al. (1992). "Immunosuppression by human gangliosides: I. Relationship of carbohydrate structure to the inhibition of T cell responses." Biochim Biophys Acta 1125(2): 180-8. Lariviere, L. , V . Gueguen-Chaignon, et al. (2003). "Crystal structures of the T4 phage beta-glucosyltransferase and the D100A mutant in complex with UDP-glucose: glucose binding and identification of the catalytic base for a direct displacement mechanism." J Mol Biol 330(5): 1077-86. Laskowski, R. A. , M . W. MacArthur, et al. (1993). "PROCHECK: a program to check the stereochemical quality of protein structures." Journal of Applied Crystallography 26(2): 283-291. Lin, H. , H . Tao, et al. (2004). "Directed evolution of a glycosynthase via chemical complementation." J A m Chem Soc 126(46): 15051-9. Lloyd, K . O. and K . Furukawa (1998). "Biosynthesis and functions of gangliosides: recent advances." Glycoconj J 15(7): 627-36. 136 Lowe, J. B. (2003). "Glycan-dependent leukocyte adhesion and recruitment in inflammation." Curr Opin Cell Biol 15(5): 531-8. Lowe, J. B. and J. D. Marth (2003). " A genetic approach to mammalian glycan function." Annu Rev Biochem 72: 643-91. Luo, Y . , S.-C. L i , et al. (1998). "The crystal structure of an intramolecular trans-sialidase with a NeuAc[alpha]2->3Gal specificity." Structure 6(4): 521-530. Luo, Y . , S.-C. L i , et al. (1999). "The 1.8 A Structures of leech intramolecular trans-sialidase complexes: evidence of its enzymatic mechanism." J Mol Biol 285(1): 323-332. Ly, H . D., B. Lougheed, et al. (2002). "Mechanistic studies of a retaining alpha-galactosyltransferase from Neisseria meningitidis." Biochemistry 41(16): 5075-85. Masserini, M . , E. Freire, et al. (1992). "Fuc-GMl ganglioside mimics the receptor function of GM1 for cholera toxin." Biochemistry 31(8): 2422-6. Mastrobattista, E., V . Taly, et al. (2005). "High-throughput screening of enzyme libraries: in vitro evolution of a beta-galactosidase by fluorescence-activated sorting of double emulsions." Chem Biol 12(12): 1291-300. Matyas, G. R. and D. J. Morre (1983). "Coupling of uridine-5'-diphosphate (UDP) formation and nicotinamide adenine dinucleotide (NAD+) reduction for cytochemical localization of glycosyltransferases." J Histochem Cytochem 31(10): 1175-82. McRee, D. E. (1999). "XtalView/Xfit-A Versatile Program for Manipulating Atomic Coordinates and Electron Density." Journal of Structural Biology 125(2-3): 156-165. Mishu, B. and M . J. Blaser (1993). "Role of infection due to Campylobacter jejuni in the initiation of Guillain-Barre syndrome." Clin Infect Pis 17(1): 104-8. Moran, A. P. and M . M . Prendergast (2001). "Molecular mimicry in Campylobacter jejuni and Helicobacter pylori lipopolysaccharides: contribution of gastrointestinal infections to autoimmunity." J Autoimmun 16(3): 241-56. Moran, A. P., M . M . Prendergast, et al. (1996). "Molecular mimicry of host structures by bacterial lipopolysaccharides and its contribution to disease." FEMS Immunol Med Microbiol 16(2): 105-15. Morris, G. M . , D. S. Goodsell, et al. (1998). "Automated Docking Using a Lamarckian Genetic Algorithm and and Empirical Binding Free Energy Function." J Computational Chemistry 19: 1639-1662. Murshudov, G. N., A. A. Vagin, et al. (1999). "Efficient anisotropic refinement of macromolecular structures using FFT." Acta Crystallogr D Biol Crystallogr 55(Pt 1): 247-55. Navaza, J. (1994). "AMoRe: an automated package for molecular replacement." Acta Crystallogr A 50: 157-163. Ni, L. , M . Sun, et al. (2006). "Cytidine 5'-monophosphate (CMP)-induced structural changes in a multifunctional sialyltransferase from Pasteurella multocida." Biochemistry 45(7): 2139-48. Oomes, P. G., B. C. Jacobs, et al. (1995). "Anti-GMl IgG antibodies and Campylobacter bacteria in Guillain-Barre syndrome: evidence of molecular mimicry." Ann Neurol 38(2): 170-5. Otwinowski, Z. and W. Minor (1997). "Processing of x-ray diffraction data collected in oscillation mode." Methods in Enzymology 276: 307-326. 138 Paulson, J. C. and K . J. Colley (1989). "Glycosyltransferases. Structure, localization, and control of cell type-specific glycosylation." J Biol Chem 264(30): 17615-8. Paulson, J. C , J. I. Rearick, et al. (1977). "Enzymatic properties of beta-D-galactoside alpha2 leads to 6 sialytransferase from bovine colostrum." J Biol Chem 252(7): 2363-71. Pedersen, L. C , K . Tsuchida, et al. (2000). "Heparan/chondroitin sulfate biosynthesis. Structure and mechanism of human glucuronyltransferase I." J Biol Chem 275(44): 34580-5. Penner, J. L. and G. O. Aspinall (1997). "Diversity of lipopolysaccharide structures in Campylobacter jejuni." J Infect Pis 176 Suppl 2: S135-8. Persson, K. , H . P . Ly, et al. (2001). "Crystal structure of the retaining galactosyltransferase LgtC from Neisseria meningitidis in complex with donor and acceptor sugar analogs." Nat Struct Biol 8(2): 166-75. Preston, A. , R. E. Mandrell, et al. (1996). "The lipooligosaccharides of pathogenic gram-negative bacteria." Crit Rev Microbiol 22(3): 139-80. Qasba, P. K. , B. Ramakrishnan, et al. (2005). "Substrate-induced conformational changes in glycosyltransferases." Trends Biochem Sci 30(1): 53-62. Raju, T. S., J. B . Briggs, et al. (2001). "Glycoengineering of therapeutic glycoproteins: in vitro galactosylation and sialylation of glycoproteins with terminal N-acetylglucosamine and galactose residues." Biochemistry 40(30): 8868-76. Raman, R., V . Sasisekharan, et al. (2005). "Structural insights into biological roles of protein-glycosaminoglycan interactions." Chem Biol 12(3): 267-77. Rearick, J. I., J. E. Sadler, et al. (1979). "Enzymatic characterization of beta D-galactoside alpha2 leads to 3 sialyltransferase from porcine submaxillary gland." J Biol Chem 254(11): 4444-51. Rest, R. F. and J. V . Frangipane (1992). "Growth of Neisseria gonorrhoeae in CMP-A 7-acetylneuraminic acid inhibits nonopsonic (opacity-associated outer membrane protein-mediated) interactions with human neutrophils." Infect Immun 60(3): 989-97. Rest, R. F. and R. E. Mandrell (1995). "Neisseria sialytransferases and their role in pathogenesis." Microb Pathog 19(6): 379-90. Rich, J. R., A . Szpacenko, et al. (2004). "Glycosyltransferase-catalyzed synthesis of thiooligosaccharides." Angew Chem Int Ed Engl 43(5): 613-5. Rietschel, E. T., L. Brade, et al. (1990). "Bacterial lipopolysaccharides: relationship of structure and conformation to endotoxic activity, serological specificity and biological function." Adv Exp Med Biol 256: 81-99. Roberts, I. S. (1996). "The biochemistry and genetics of capsular polysaccharide production in bacteria." Annu Rev Microbiol 50: 285-315. Rossmann, M . G. and P. Argos (1978). "The taxonomy of binding sites in proteins." Mol Cellular Biochemistry 21(3): 161-182. Rudd, P. M . , T. Elliott, et al. (2001). "Glycosylation and the immune system." Science 291(5512): 2370-6. Rutishauser, U . (1996). "Polysialic acid and the regulation of cell interactions." Curr Opin Cell Biol 8(5): 679-84. Sacks, D. and S. Kamhawi (2001). "Molecular aspects of parasite-vector and vector-host interactions in leishmaniasis." Annu Rev Microbiol 55: 453-83.. 140 Saida, T., S. Kuroki, et al. (1997). "Campylobacter jejuni isolates from Japanese patients with Guillain-Barre syndrome." J Infect Pis 176 Suppl 2: S129-34. Salloway, S., L. A . Mermel, et al. (1996). "Miller-Fisher syndrome associated with Campylobacter jejuni bearing lipopolysaccharide molecules that mimic human ganglioside GP3." Infect Immun 64(8): 2945-9. Schaub, C , B. Muller, et al. (1998). "New sialyltransferase inhibitors based on CMP-quinic acid: development of a new sialyltransferase assay." Glycoconj J 15(4): 345-54. Schuttelkopf, A . W. and P . M . F. van Aalten (2004). "PROPRG: a tool for high-throughput crystallography of protein-ligand complexes." Acta Crvstallographica Section P 60(8): 1355-1363. Sheikh, K. A. , T. W. Ho, et al. (1998). "Molecular mimicry in Guillain-Barre syndrome." Ann N Y Acad Sci 845: 307-21. Sheikh, K. A. , I. Nachamkin, et al. (1998). "Campylobacter jejuni lipopolysaccharides in Guillain-Barre syndrome: molecular mimicry and host susceptibility." Neurology 51(2): 371-8. Shibayama, K. , S. Ohsuka, et al. (1998). "Conserved structural regions involved in the catalytic mechanism of Escherichia coli K-12 WaaO (Rfal)." J. Bacteriol. 180(20): 5313-5318. Shin, J. E., S. Ackloo, et al. (1997). "Lipo-oligosaccharides of Campylobacter jejuni serotype O:10. Structures of core oligosaccharide regions from abacterial isolate from a patient with the Miller-Fisher syndrome and from the serotype reference strain." Carbohydr Res 305(2): 223-32. 141 Sinnott, M . L. and W. P. Jencks (1980). "Solvolysis of D-glucopyranosyl derivatives in mixtures of ethanol and 2,2,2-trifluoroethanol." J. Am. Chem. Soc. 102(6): 2026-2032. Smith, H. , N . J. Parsons, et al. (1995). "Sialylation of neisserial lipopolysaccharide: a major influence on pathogenicity." Microb PathoR 19(6): 365-77. Smith, P. L. , G. R. Bousfield, et al. (1993). "Equine lutropin and chorionic gonadotropin bear oligosaccharides terminating with S04-4-GalNAc and Sia alpha 2,3Gal, respectively." J Biol Chem 268(2): 795-802. Stockert, R. J. (1995). "The asialoglycoprotein receptor: relationships between structure, function, and expression." Physiol Rev 75(3): 591-609. Stura, E. A. , G. Zanotti, et al. (1983). "Comparison of A M P and N A D H binding to glycogen phosphorylase b." J Mol Biol 170(2): 529-65. Sujatha, M . S. and P. V . Balaji (2006). "Fold-recognition and comparative modeling of human alpha2,3-sialyltransferases reveal their sequence and structural similarities to Cstll from Campylobacter jejuni." B M C Struct Biol 6: 9. Svennerholm, L. (1994). "Designation and schematic structure of gangliosides and allied glycosphingolipids." Prog Brain Res 101: XI -XIY. Takashima, S., S. Tsuji, et al. (2002). "Characterization of the second type of human beta-galactoside alpha 2,6-sialyltransferase (ST6Gal II), which sialylates Galbeta l ,4GlcNAc structures on oligosaccharides preferentially. Genomic analysis of human sialyltransferase genes." J Biol Chem 277(48): 45719-28. Tao, H. and V. W. Cornish (2002). "Milestones in directed enzyme evolution." Curr Opin Chem Biol 6(6): 858-64. Terwilliger, T. C. and J. Berendzen (1999). "Automated M A D and MIR structure solution." Acta Crystallogr D Biol Crystallogr 55 (Pt 4): 849-61. Timm, D. E., J. Liu, et al. (2001). "Crystal structure of thiamin pyrophosphokinase." J Mol Biol 310(1): 195-204. Uitdehaag, J. C. M . , R. Mosi, Kalk, K . H . , et al. (1999). "X-Ray structures along the reaction pathway of cyclodextrin glycosyltransferase elucidate catalysis in the a-amylase family." Nat. Struct. Biol. 6(5): 432-436. Unligil, U . M . and J. M . Rini (2000). "Glycosyltransferase structure and mechanism." Curr Opin Struct Biol 10(5): 510-7. Unligil, U . M . , S. Zhou, et al. (2000). "X-ray crystal structure of rabbit N-acetylglucosaminyltransferase I: catalytic mechanism and a new protein superfamily." Embo J 19(20): 5269-80. Vagin, A . and A . Teplyakov (2000). "An approach to multi-copy search in molecular replacement." Acta Crystallogr D Biol Crystallogr 56(Pt 12): 1622-4. Varadarajan, N . , J. Gam, et al. (2005). "Engineering of protease variants exhibiting high catalytic activity and exquisite substrate selectivity." Proc Natl Acad Sci U S A 102(19): 6855-60. Vartanian, J. P., M . Henry, et al. (2001). "Simulating pseudogene evolution in vitro: determining the true number of mutations in a lineage." Proc Natl Acad Sci U S A 98(23): 13172-6. Vasella, A. , G. J. Davies, et al. (2002). "Glycosidase mechanisms." Curr Opin Chem Biol 6(5): 619-29. 143 Vogel, U . , H . Claus, et al. (1999). "Role of lipopolysaccharide sialylation in serum resistance of serogroup B and C meningococcal disease isolates." Infect Immun 67(2) 954-7. von Gunten, S. and H . U . Simon (2006). "Sialic acid binding immunoglobulin-like lectins may regulate innate immune responses by modulating the life span of granulocytes." FasebJ 20(6): 601-5. von Gunten, S., S. Yousefi, et al. (2005). "Siglec-9 transduces apoptotic and nonapoptotic death signals into neutrophils depending on the proinflammatory cytokine environment." Blood 106(4): 1423-31. Vrielink, A. , W. Ruger, et al. (1994). "Crystal structure of the D N A modifying enzyme beta-glucosyltransferase in the presence and absence of the substrate uridine diphosphoglucose." Embo J 13(15): 3413-22. Vriesendorp, F. J., B. Mishu, et al. (1993). "Serum antibodies to GM1, GDlb , peripheral nerve myelin, and Campylobacter jejuni in patients with Guillain-Barre syndrome and controls: correlation and prognosis." Ann Neurol 34(2): 130-5. Vyas, A . A . and R. L . Schnaar (2001). "Brain gangliosides: functional ligands for myelin stability and the control of nerve regeneration." Biochimie 83(7): 677-82. Watts, A . G., I. Damager, et al. (2003). "Trypanosoma cruzi trans-sialidase operates through a covalent sialyl-enzyme intermediate: tyrosine is the catalytic nucleophile." J A m Chem Soc 125(25): 7532-3. Weinhold, B. , R. Seidenfaden, et al. (2005). "Genetic ablation of polysialic acid causes severe neurodevelopmental defects rescued by deletion of the neural cell adhesion molecule." J Biol Chem 280(52): 42971-7. Wetzler, L. M . , K . Barry, et al. (1992). "Gonococcal lipooligosaccharide sialylation prevents complement-dependent killing by immune sera." Infect Immun 60(1): 39-43. Whitfield, C. and I. S. Roberts (1999). "Structure, assembly and regulation of expression of capsules in Escherichia coli." Mol Microbiol 31(5): 1307-19. Wiggins, C. A . R. and S. Munro (1998). "Activity of the yeast MNN1 alpha -1,3-mannosyltransferase requires a motif conserved in many other families of glycosyltransferases." PNAS 95(14): 7945-7950. Winn, M . D., M . N . Isupov, et al. (2001). "Use of TLS parameters to model anisotropic displacements in macromolecular refinement." Acta Crystallogr D Biol Crystallogr 57(Pt 1): 122-33. Witczak, Z. J. and J. M . Culhane (2005). "Thiosugars: new perspectives regarding availability and potential biochemical and medicinal applications." Appl Microbiol Biotechnol 69(3): 237-44. Yamamoto, F., H . Clausen, et al. (1990). "Molecular genetic basis of the histo-blood group A B O system." Nature 345(6272): 229-33. Yuki , N . , T. Taki, et al. (1994). 'Tenner's serotype 4 of Campylobacter jejuni has a lipopolysaccharide that bears a GM1 ganglioside epitope as well as one that bears a GDI a epitope." Infect Immun 62(5): 2101-3. Zhang, Y . , G. J. Swaminathan, et al. (2003). "Roles of individual enzyme-substrate interactions by alpha-1,3-galactosyltransferase in catalysis and specificity." Biochemistry 42(46): 13512-21. Appendix - Publications arising from graduate work First author publications arising from Ph.D. studies 2007 Chiu, C.P.C., Lairson, L .L . , Gilbert, M . , Wakarchuk, W.W., Withers, S.G., & Strynadka, N.C.J. "Structural analysis of the a-2,3-sialyltransferase Cst-I from Campylobacter jejuni in both apo and substrate-analog bound forms." Biochemistry. Accepted May 2007. 2004 Chiu, C.P.C., Watt, A. , Lairson, L .L . , Gilbert, M . , Lim, D., Wakarchuk, W.W., Withers, S.G., & Strynadka, N.C.J. "Structural analysis of the sialyltransferase Cst-II from Campylobacter jejuni in complex with a substrate analog." Nature Struct. Mol. Biol. 2004 Vol . 11, Issue 2, 163-170. Additional publications during Ph.D. studies 2006 Aharoni, A . , Thieme, K. , Chiu, C.P.C., Buchini, S., Lairson, L .L . , Chen, FL, Strynadka, N.C.J. , Wakarchuk, W.W., & Withers, S.G. "High throughput screening of glycosyltransferase libraries: Evolution of a sialyltransferase with a novel substrate specificity." Nature Methods 2006, 3, 609-614. 2004 Lairson, L .L . , Chiu, C.P.C., Ly, H.D., He,S., Wakarchuk, W W . , Strynadka, N.C.J. , & Withers, S.G. "Intermediate trapping on a mutant retaining alpha-galactosyltransferase identifies an unexpected aspartate residue." J. Biol. Chem., 2004, Vol . 279, Issue 27, 28339-28344. 


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items