Stephen Withers

Carbohydrates are abundant throughout nature, play key roles in diverse biological processes and form fundamental components of a wide range of functional materials. Their inherent complexity is a double edge sword in that it allows carbohydrates to adopt a diverse range of structures and functions, but in an industrial setting, that complexity can also make them especially difficult and costly to synthesize in a strictly defined manner. To help ensure uniform production, carbohydrate manufacturing processes are known to utilize Carbohydrate Active Enzymes (CAZymes), which have innate substrate specificities and conformational control. A class of CAZymes, known as glycoside phosphorylases (GPs) catalyze the reversible phosphorolysis of glycosidic bonds, releasing sugar 1-phosphates and have considerable potential as catalysts for the assembly of useful carbohydrates for products ranging from functional foods and prebiotics to novel materials. However, the substrate diversity of currently identified GPs is relatively small, limiting their practical applications. To address this limitation, I used a combined approach drawing upon rational, metagenomic and genomic exploration methods to discover new GPs. First, using a rational approach, I identified a new GP activity, the first reported with a β-retaining mechanism, while investigating a mechanistic oddity observed in the GH3 CAZy family. Next, I developed and deployed two metagenomic screening methodologies that targeted GP activity in the vast reservoir of uncultivated genetic diversity encoded in microbial communities inhabiting natural and engineered ecosystems. This approach yielded eight new GPs and established a screening paradigm that can be applied to an even greater range of GP activities. Lastly, a phylogenomically diverse, synthetic gene GP library was characterized, which led to the discovery of another previously undiscovered GP activity, a new biopolymer (that I named acholetin), and established a high throughput (HT) substrate specificity assay for GPs. In total, 11 new GPs were discovered, two of those representing previously unknown activities. The research presented in this dissertation will result in a better understanding of GP diversity, provide insights into how we can engineer GPs, establish a HT functional-characterization method, and provide a platform to explore genomic and metagenomic sequence space for more novel GP activities.

View record

Oxocarbenium ions play a pivotal role in the chemistry and biochemistry of glycosides. In thisthesis I prepared carbocyclic glycoside analogs that can generate allylic cations in the place ofoxocarbenium ions. In these unsaturated cyclitols the C5-O5 bond of a pyranoside is replaced bya C-C double bond. Phenolic leaving groups were introduced at the pseudo anomeric carbon toform allyl ether linkages analogous to glycosidic linkages. Spontaneous heterolysis of these arylcarbasugars was shown, by measurement of secondary kinetic isotope effects, to proceed via anallylic cation. Not only was this mechanism of cleavage similar to that seen for glycosides but therate constants for solvolysis of aryl glycosides and unsaturated aryl carbasugars were found to benearly identical. To explore if these close similarities in structure and inherent reactivities wouldallow glycoside hydrolases to act on non-glycosidic linkages I screened a library of over 150glycoside hydrolases with a fluorogenic allyl-carbasugar substrate. This screen identified severalbeta-retaining glycosidases capable of unsaturated cyclitol ether hydrolysis. Detailed analysis of amodel enzyme demonstrated that the same mechanism was operative for cleavage of unsaturatedcyclitol ethers as glycosidic linkages. Once again, kinetic isotope effects implicated an allyliccation at the rate-limiting transition state for the enzymatic reaction. The observed kineticparameters for aryl carbasugar cleavage by glycosidases served as a reminder that these enzymes have evolved to act upon glycosides, and not carbocyclic imitators. Turnover constants were 100 – 1000-fold lower for aryl carbasugar hydrolysis compared to hydrolysis of aryl glycosides. Directed evolution has begun to bridge this gap in activity and yield glycosidases better capable of cleaving aryl unsaturated cyclitol linkages. The mechanistic insights gained by these studies have proven useful in preparing novel glycosidase inactivators. Specifically, unsaturatedpseudosugars, with the vinyl hydrogen replaced by electron withdrawing halogens, react with retaining glycosidases to accumulate the covalent species and accomplish enzyme inactivation.Taken together these studies highlight the parallels between allylic cations and oxocarbenium ions.Not only do allylic carbasugars have similar inherent reactivities to glycosides but glycosidasesare capable of allylic cation stabilization at their transition states.

View record

The parasite Trypanosoma cruzi displays a trans-sialidase (TcTS) on its surface that is hypothesized to be a therapeutic target for Chagas disease. TcTS transfers sialic acid from the cells of infected hosts to the surface of the pathogenic parasite, masking it from immune recognition and enhancing cellular invasion. The design of TcTS inhibitors has been largely unsuccessful to date. Accordingly, this work aims to identify new TcTS inhibitors, both by modifying existing inhibitors and by screening natural product and peptide libraries to discover new chemical scaffolds for this purpose. First, analogues of the established mechanism-based inhibitor, difluorosialic acid (DFSA), were investigated in search of increased potency and selectivity for TcTS. The synthesis and kinetic analysis of nine C9 amide-linked DFSAs and seven N-acyl modified DFSAs was explored to this end. One candidate was identified that inhibited TcTS 10-fold better than the unmodified precursor. Further, TcTS showed a tolerance for the C5 functionalized inhibitors, a fact that can be leveraged – together with C9 substitution – for specificity versus human neuraminidases. Next, a library of ~1000 marine organism extracts was screened for TcTS inhibition, from which five hits were selected. Bioassay-guided isolation and structural determination of the active chemicals afforded two new natural product inhibitors of TcTS with IC50 values
View record

Human pancreatic α-amylase (HPA) catalyzes a key step in the degradation of ingested starch. Accordingly HPA activity has been positively correlated to post-prandial blood glucose levels and has been identified as a viable target for inhibition and the development of therapeutics towards the treatment of diabetes and obesity. This work directs the hunt away from traditional saccharide-based inhibitors, which represent all carbohydrate metabolism therapeutics currently in use, to novel inhibitors with improved selectivity and potency. In Chapter 2, the synthesis of flavonol-based HPA inhibitors based upon the structure of Montbretin A, a complex flavonol glycoside, is explored. Through the synthesis of a library of Montbretin A analogues we were able to identify an inhibitor of HPA with a KI of 44 nM that formed new interactions within the amylase active site. Chapter 3 details work on the peptide-based HPA inhibitor helianthamide previously isolated from the Caribbean Sea anemone Stichodactyla helianthus. Recombinant expression of helianthamide as a fusion peptide was achieved in Escherichia coli and Pichia pastoris. Kinetic analysis indicated that recombinant helianthamide is one of the most potent HPA inhibitors known to date, with a KI of 0.01 nM. Structural analysis of the recombinant material indicated that it contained three disulfide bonds in a 1-5, 2-4, 3-6 pattern. Site-directed mutagenesis of helianthamide indicated that disruption of disulfide bonds led to a large decrease in potency, while alanine variants of residues forming polar contacts with HPA’s active site residues led to smaller decreases in potency, indicating that the intact tertiary structure of helianthamide is necessary for blockage of the amylase active site. Small peptides were synthesized based on the sequence of helianthamide, but most showed modest or no inhibition of HPA.

View record

Plant biomass offers a sustainable source for energy and materials and an alternative to fossil fuels. However, the industrial scale production or biorefining of fermentable sugars from plant biomass is currently limited by the lack of cost effective and efficient biocatalysts. Microbes, the earth's master chemists - employing biocatalytic solutions to harvest energy, and transform this energy into useful molecules - offer a potential solution to this problem. However, a majority of microbes remain uncultured, limiting our access to the genetic potential encoded within their genomes. This has spurred the development of culture independent methods, termed metagenomics. In this thesis I harnessed high-throughput functional metagenomic screening to discover biomass deconstructing biocatalysts from uncultured microbial communities. Towards this goal, twenty-two clone libraries containing DNA sourced from diverse microbial communities inhabiting terrestrial and aquatic ecosystems were screened with 4-methylumbelliferyl cellobioside to detect glycoside hydrolase activity. This revealed 178 active clones containing glycoside hydrolases, often in gene clusters. This set of active clones was consolidated and further characterized through sequencing and rapid, plate-based, biochemical assays. Additionally, libraries sourced from beaver fecal and gut microbiomes were screened with four fluorogenic probes (6-chloro-4-methylumbelliferyl derivatives of cellobiose, xylobiose, xylose and mannose) for glycoside hydrolase activity. This revealed a total of 247 active fosmid-harbouring clones, that encoded many polysaccharide-degrading genes and gene cassettes. Specific candidate genes from the fecal library were sub-cloned, and the resulting purified enzymes were shown to be involved in synergistic degradation of arabinoxylan oligomers. The clone libraries that were generated through functional metagenomic screening were then employed to reveal the promiscuity of glycoside hydrolases towards unnatural azido- and aminoglycosides. Promiscuous enzymes identified from metagenomic and synthetic clone libraries were then used as a starting point for the generation of new glycosynthases capable of incorporating modified glucosides and galactosides. The resulting set of eight new glycosynthases are capable of synthesizing di- and trisaccharides, glycolipids and inhibitors such as 2,4-dinitrophenyl 4'-amino-2,4'-dideoxy-2-fluoro-cellobioside. Taken together this work has exploited the power of functional metagenomics to reveal new modes of biocatalysis and develop new synthetic tools.

View record

Carbohydrate-modifying enzymes can be used for the enzymatic synthesis or cleavage of complex glycans. In this study, various high-throughput assays were assessed as methods for the discovery of glycoside hydrolases (GHs) and glycosyltransferases (GTs) within environmental samples. Fluorescence-activated cell sorting (FACS) was evaluated as an approach for the functional enrichment of sialyltransferase (ST) genes in environmental samples. A model ST from Campylobacter jejuni, CstI, was successfully enriched from a mixture of genomic DNA (proof-of-principle), but active STs could not be isolated from a set of metagenomic libraries. The same FACS screen was applied to the directed evolution of multifunctional sialyltransferase from the Pasteurella multocida, PmST1, in order to isolate mutants with reduced sialidase activity and improved synthetic efficiency over the wild-type enzyme. Sialidase activity (kcat/KM) of the best mutants was reduced approximately 2-fold, with an approximately 2.5-fold increase in the sialyltransferase activity (kcat/KM). Despite these improvements, the maximum product yield of the mutants did not increase appreciably. While engineering PmST1, a study of the sialidase and trans-sialidase mechanisms of PmST1 and other STs from the glycosyltransferase family 80 was also undertaken. The mechanisms of both these activities were found to follow a reversible sialylation path, varying from that previously proposed in literature. A high-throughput plate-based assay was also evaluated as a functional screen for the identification of blood antigen-cleaving enzymes within the human gut microbiome. One GH enzyme from Bacteroides vulgatus (BvGH109) was found to be capable of converting the blood type A antigen into blood type O, offering a new enzyme for the engineering of universal donor blood. Two enzymes with α-N-acetylgalactosaminidase activity were also isolated and determined to represent a new sub-family of the GH family 31.

View record

Human pancreatic alpha-amylase (HPA) is the enzyme responsible for hydrolyzing starch within the gut into shorter oligosaccharides. Selective inhibition targeted at only HPA could be used to modulate blood glucose levels for the treatment of diabetes and obesity.Montbretin A (MbA) is a potent (Ki = 8.1 nM) and specific inhibitor of HPA. Controlled degradation studies on MbA, coupled with inhibition analysis, identified an essential high-affinity core structure comprising the myricetin and caffeic acid moieties linked via a disaccharide, mini-MbA. X-ray structural analyses of the complex of MbA-HPA confirmed the importance of this core structure and revealed a novel mode of glycosidase inhibition wherein internal pi-stacking interactions between the myricetin and caffeic acid organize their ring hydroxyls for optimal hydrogen bonding to the catalytic residues of HPA. The simplified analogue mini-MbA therefore offers potential for new strategies for glycosidase inhibition and therapeutic development.As part of a search for selective, mechanism-based covalent inhibitors of HPA, chemo-enzymatic syntheses of oligoglycosyl epi-cyclophellitols are described. Alpha-1,4 glucosyl epi-cyclophellitol, synthesized from epi-cyclophellitol by coupling of a glucosyl moiety using maltose phosphorylase, inactivated HPA stoichiometrically. X-ray crystallographic analysis of the covalent derivative so formed confirmed its reaction at the active site with the catalytic nucleophile Asp197. Another trisaccharide analogue 4’-O-methyl-alpha-maltosyl epi-cyclophellitol was synthesized enzymatically or by in situ elongation by HPA. Both of the inhibitors showed time-dependent inactivation of HPA, with the trisaccharide version being a better inactivator. This new class of mechanism-based inhibitors will be useful as activity-based probes for amylases. Several potential starch binding sites have been identified on the surface of HPA by crystallography, but their role, if any, in starch degradation is unknown. Through analysis of the binding of HPA mutants, modified individually at each site, to soluble and granule starch, two of these surface binding sites (SBSs) were shown to play a role in starch granule binding. A quite separate site was shown to be important for binding to, and cleavage of, soluble starch. Binding at SBSs was distinguished from binding to the active site by blocking the active site with the glycosyl epi-cyclophellitol mechanism-based inactivators.

View record

Glycosaminoglycans are the main structural polysaccharides of vertebrates, and represent a major barrier to the spread of both bacterial infection and tumours. The enzymes by which mammals and pathogens degrade these polysaccharides use very different mechanisms, and may represent suitable therapeutic targets. In this thesis, work is presented towards an understanding of the mechanisms of Clostridium perfringens unsaturated glucuronyl hydrolase (UGL), the second enzyme in the bacterial pathway for degradation of glycosaminoglycans, and human heparanase, the enzyme by which the abundant glycosaminoglycan heparan sulfate is remodelled.For UGL, evidence was presented for a hydration reaction scheme that had previously been proposed on the basis of crystallographic evidence. This was shown by characterisation of products formed by reaction in D₂O and 10 % methanol, and by demonstrating hydrolysis of three compounds that are only expected to be turned over by the enzyme if this reaction is correct. Investigation of the effects of substituents on the transition state stability, by measurement of a linear free-energy relationship for a series of aryl glycosides, kinetic isotope effects, and rate determination for heteroatom-substituted substrates, led to the proposal of alternate mechanisms. Attempts to verify these mechanisms were made by testing of potential inhibitors, rescue of a catalytic-residue mutant, trapping of a covalent glycosyl-enzyme intermediate, or synthesis of a potential intermediate, but without success. The mechanism that appears most likely proceeds through protonation of the substrate C4-C5 double bond, with the resulting C5 positive charge being quenched by opening of the pyranose ring to give a C5 ketone and a C1-C2 epoxide. Subsequent hydration of the ketone and opening of this epoxide reforms the pyranose ring and gives the same product as direct hydration, but through a lower energy path.For mammalian heparanase, several new potential substrates and a potential inactivator were synthesized and tested. While this work was largely unsuccessful, it indicated that optimisation of sulfation patterns without modification of the aglycone is likely a futile strategy. A redesigned aglycone was proposed, representing a new path towards the goal of studying this enzyme for its eventual use as a therapeutic target in cancer therapy.

View record

The overall goal of this thesis was to investigate the structures and enzymatic mechanisms of glycosyltransferases using NMR spectroscopy and enzyme kinetic measurements. The bifunctional sialyltransferase CstII from Campylobacter jejuni and the α-1,4-galactosyltransferase LgtC from Neisseria meningitidis were chosen to be the model inverting and retaining enzymes, respectively. By systematically introducing point mutations at the subunit interfaces of CstII, two active monomeric variants were obtained and characterized. In contrast to the wild-type tetramer, the monomeric CstII variants yielded good quality amide ¹H/¹⁵N-HSQC and methyl-TROSY NMR spectra. However, the absence of signals from approximately one half of the amides in the ¹H/¹⁵N-HSQC spectra of both monomeric forms suggests that the enzyme undergoes substantial conformational exchange on a msec-µsec time-scale. The histidine pKa values of CstII-F121D in its apo form were measured by monitoring the pH-dependent chemical shifts of biosynthetically incorporated [¹³Cε¹]-histidine. Consistent with its proposed catalytic role, the site-specific pKa value ~ 6.6 for His188 matches the apparent pKa value ~ 6.5 governing the pH-dependence of kcat/Km for CstII towards CMP-Neu5Ac. The enzymatic mechanism of the retaining glycosyltransferase LgtC appears to involve a “front-side attack” SNi or SNi-like mechanism with a short-lived oxocarbenium-phosphate ion pair intermediate. Furthermore, based upon X-ray crystallographic studies, two flexible loops were proposed to become ordered over the active site of LgtC upon sugar donor binding. Accordingly, NMR spectroscopy was used to investigate the dynamic properties of the enzyme with an emphasis on delineating the possible roles of these motions. The amide ¹H/¹⁵N-TROSY-HSQC and methyl-TROSY spectra of LgtC were partially assigned using a variety of NMR spectroscopic approaches, combined with mutagenesis of all the isoleucine residues. More than the expected number of methyl signals was observed, indicating that LgtC adopts multiple conformational states in equilibrium on a seconds time-scale, and that their relative populations change upon mutation and substrate binding. Furthermore, relaxation dispersion studies indicated substantial msec-µsec time-scale motions of methyl groups both within and distal to the active site in apo and substrate-bound forms of LgtC. Thus LgtC exhibits a range of dynamic behaviours potentially linked to its catalytic function. They were studies in this thesis using NMR spectroscopy and kinetic studies.

View record

The ABO blood groups are vitally important in blood transfusion and organ transplantation. Transfusion with an incorrect blood type results in destruction of the incompatible blood cells, which can result in death. In my thesis, the catalytic mechanisms of three enzymes, two of which can directly be used on red blood cells (RBCs), were investigated in detail as follows.YesZ, a family GH 42 β-galactosidase (retaining), was used as a model system for the identification of catalytic residues. The mechanism-based inhibitor, 2,4-dinitrophenyl 2-deoxy-2-fluoro-β-D-galactopyranoside was synthesized and used to inactivate YesZ via trapping of a reaction intermediate. Subsequent proteolytic digestion and comparative MS analysis identified the labeled peptide which, combined with, sequence alignments identified the catalytic nucleophile, a glutamate in the sequence ETSPSYAASL. Use of the acid/base mutant for trapping experiments provided support for its role thereby providing experimental verification of the identities of the catalytic residues in Family GH42.EABase, a family GH98 endo-β-galactosidase, cleaves blood group A and B trisaccharides from glycoconjugates and RBCs. The mechanism of Family 98 glycosidases was unknown but inferred to be retaining. The DNP-β-A-trisaccharide substrate was synthesized by in vivo enzymatic and subsequent chemical methods and direct 1H NMR analysis of its hydrolysis by EABase revealed that EABase is an inverting glycosidase. Both activated and nonactivated substrates were used to kinetically characterize EABase and its mutants, revealing that D453 and/or E506 act as the base catalyst and that E354 is the acid catalyst. EABase was used, in collaboration with Dr. Kizhakkedathu’s lab, to generate “universal blood cells” from type-B blood. Several α-L-fucosidases from family GH29 (retaining), which cleave α(1,2)-fucose from glycoconjugates were kinetically characterized in the hope of identifying the acid/base residue which is not conserved by sequence. A combination of modeling, sequence comparisons and phylogenetic tree analysis was used to identify candidate acid/base residues and further subgroup GH29 fucosidases based on these comparisons. The identity of the acid/base residue in four fucosidases is supported by kinetic characterization of a series of mutants of candidate residues and can now be predicted for all Family GH29 fucosidases.

View record

The majority of retaining α-glycosidases are believed to adopt the classical double displacement mechanism to catalyze their reactions, which features a catalytic nucleophilic residue, a general acid/base residue, two oxocarbenium-ion like transition states and one covalent glycosyl-enzyme intermediate. In my thesis, the catalytic mechanisms of three retaining alpha-glycosidases were investigated in detail as follows. HPA is an enzyme which is responsible for hydrolyzing starch into shorter oligosaccharides. Several 2-deoxy-2,2-dihalo maltosyl chlorides were synthesized and tested as potential mechanism-based inhibitors of HPA, in the hope of trapping its covalent glycosyl-enzyme intermediate for crystallographic studies. Unfortunately, none of newly-synthesized compounds could cause time-dependent inactivation of HPA. By employing our newly developed in situ elongation strategy, 5-fluoro-α-D-glucopyranosyl fluoride and 5-fluoro-β-L-idopyranosyl fluoride showed kinetic behavior consistent with the proposed in situ elongation-inactivation process, allowing the trapping and further kinetic and structural analysis of the covalent intermediate of HPA. These structures provide interesting mechanistic insights into the catalytic mechanism of HPA. TreS is an enzyme which catalyzes the reversible interconversion of maltose and trehalose. 5-Fluoro glycosyl fluorides were shown to be mechanism-based inhibitors of this enzyme by accumulating the covalent glycosyl-enzyme intermediate. The trapped intermediate was subjected to protease digestion followed by MS analysis of the resultant peptides to identify the catalytic nucleophile residue as D230. The inability of TreS to carry out transglycosylation reactions onto exogenously added acceptors establishes the intramolecular nature of the rearrangement reaction, consistent with previous studies on other TreS enzymes. All studies support a double displacement mechanism involving an intramolecular “glucose flipping” step as the catalytic mechanism of this enzyme. SpGH101 is an enzyme which specifically removes an O-linked disaccharide Gal-β-1,3-GalNAc-α from glycoproteins. Using the recently solved 3-dimensional structure of this protein as a guide, we carried out a detailed mechanistic investigation of this retaining α-glycosidase using a combination of synthetic and natural substrates. Based on a model of the substrate complex of SpGH101, we proposed D764 and E796 as the nucleophile and general acid/base residues, respectively. These roles were confirmed by kinetic and mechanistic analysis of mutants at those positions using synthetic substrates and anion rescue experiments.

View record

Activated fluorosugars are covalent inactivators for a number of glycosidases, functioning through accumulation of stable glycosyl-enzyme intermediates. Two approaches were taken in designing new inactivators: more highly fluorinated sugars that could form more stable intermediates, and fluorosugars bearing novel aglycones that could be manipulated to improve selectivity and efficiency. Six novel difluorosugar fluorides were synthesized and evaluated as covalent inactivators. Four of the compounds were tested as time-dependent inactivators of the β-glucosidase from Agrobacterium sp. (Abg) and, while they were shown to behave as reversible competitive inhibitors, the only time-dependent inactivation was traced to the presence of an extremely small amount (
View record